Download stimulant treatment history predicts frontal-striatal

 1 The research in this thesis was performed at the University Medical Center Groningen (UMCG), Department of Psychiatry, The Netherlands, and at the University of California, San Diego (UCSD), Department of Psychiatry, United States of America. The PhD program was embedded in the Research School of Behavioral and Cognitive Neurosciences (BCN), a multi-­‐disciplinary research school within the University of Groningen. The research was supported by the National Institutes of Health Grant R01MH62873 awarded to prof. S. Faraone, the Netherlands Organization for Scientific Research (NWO) Large Investment Grant 1750102007010 awarded to prof. J. Buitelaar, and the ZonMW Priority Medicines for Children Grant 113202005 awarded to prof. P. Hoekstra. The research performed at the UCSD was further supported by the National Institute on Alcohol Abuse and Alcoholism of the National Institutes of Health R01 AA13419, U01 AA021692, and U01 DA041089 awarded to prof. S. Tapert, K12 DA031794 awarded to L. Squeglia, PhD, and the National Institute on Drug Abuse F32 DA032188 awarded to J. Jacobus. ISBN 978-­‐90-­‐367-­‐9055-­‐0 (printed version) ISBN 978-­‐90-­‐367-­‐9054-­‐3 (electronic version) Printed by Ipskamp Printing, Enschede, The Netherlands Copyright © 2016 by Lizanne JS Schweren No part of this thesis may be reproduced, distributed, or transmitted in any form without prior written permission from the author. 2 STIMULANTS AND THE DEVELOPING BRAIN
PHD THESIS
to obtain the degree of PhD at the University of Groningen on the authority of the Rector Magnificus Prof. E. Sterken and in accordance with the decision by the College of Deans. This thesis will be defended in public on 14 December 2016 at 11 hours by Lizanne Johanna Stephanie Schweren born on 28 December 1986 in Geldrop, The Netherlands 3 SUPERVISORS
Prof. Pieter Hoekstra Prof. Jan Buitelaar CO-SUPERVISOR
Catharina Hartman, PhD ASSESSMENT COMMITTEE
Prof. André Aleman Prof. Katya Rubia Prof. Roshan Cools 4 CONTENTS
Chapter 1 Chapter 2 GENERAL INTRODUCTION BRAIN STRUCTURE AND FUNCTION IN ADHD Chapter 3 47 67 85 AGE AND DRD4 GENOTYPE MODERATE ASSOCIATIONS BETWEEN STIMULANT TREATMENT HISTORY AND CORTEX STRUCTURE IN ADHD Chapter 7 21 STIMULANT TREATMENT TRAJECTORIES ARE ASSOCIATED WITH NEURAL
REWARD PROCESSING IN ADHD Chapter 6 7 STIMULANT TREATMENT HISTORY PREDICTS FRONTAL-STRIATAL
STRUCTURAL CONNECTIVITY IN ADOLESCENTS WITH ADHD Chapter 5 THINNER MEDIAL TEMPORAL CORTEX IN ADOLESCENTS WITH ADHD AND
THE EFFECTS OF STIMULANTS Chapter 4 MR IMAGING OF THE EFFECTS OF THE EFFECTS OF METHYLPHENIDATE ON
105 COMBINED STIMULANT AND ANTIPSYCHOTIC TREATMENT IN
ADOLESCENTS WITH ADHD: A CROSS-SECTIONAL OBSERVATIONAL
STRUCTURAL MRI STUDY Chapter 8 127 145 SYMPTOMS, SOCIAL-EMOTIONAL FUNCTIONING, OR COGNITION MEMORY PERFORMANCE AND HIPPOCAMPUS STRUCTURE AFTER
INFREQUENT RECREATIONAL STIMULANT USE IN YOUTH Chapter 9 NO LONG-TERM EFFECTS OF STIMULANT TREATMENT ON ADHD
163 177 NEDERLANDSE SAMENVATTING 195 221 235 239 Chapter 10 GENERAL DISCUSSION ACKNOWLEDGEMENTS / DANKWOORD REFERENCES CURRICULUM VITAE 5 6 Chapter 1 GENERAL INTRODUCTION
7 8 As I am writing this chapter, the Dutch public broadcasting association KRO-­‐
NCRV is airing its second episode in a special series about ‘the ADHD-­‐epidemic’. Within one week of announcing the making of this series earlier this year, KRO-­‐
NCRV’s editorial office had received over two-­‐hundred emails from parents, teachers, and psychiatrists. One week later, members of Parliament were publicly expressing their concerns about (over-­‐) diagnosis and, perhaps even more so, (over-­‐)medicating of youth with ADHD, causing intense debate. Distress about the potential neurotoxic effect of ADHD-­‐medications is bolstered by popular reports comparing them with drugs of abuse such as cocaine and ecstasy. Understandably, people are wary: do we really want to expose our children to ADHD-­‐medications, not knowing whether and how they may affect brain development in the long term? In response to parliamentary questions, the Secretary of the Ministry of Health, Welfare and Sport assured members of parliament that an investigation into the long-­‐term effects of ADHD-­‐medications on the developing brain was underway. In this thesis, I present the first results of that investigation. In this introductory chapter, I first provide an overview of the scientific state-­‐
of-­‐affairs regarding ADHD, stimulant treatment, and the developing brain. Next, I identify questions and challenges that have not satisfactorily been addressed in existing studies. At the end of this first chapter, remaining questions are translated into specific research aims and objectives. ATTENTION-DEFICIT/HYPERACTIVITY DISORDER
Attention-­‐deficit/hyperactivity disorder (ADHD) is the most common developmental disorder in childhood, estimated to affect 3-­‐7% of school-­‐age children and 2.5% of adults globally (Polanczyk et al., 2007; Simon et al., 2009). ADHD is characterized by age-­‐inappropriate attention problems and/or impulsivity and hyperactivity (American Psychiatric Association, 2000 and 2013). The clinical presentation of ADHD is heterogeneous and comorbidity with externalizing disorders such as oppositional defiant disorder, internalizing disorders such as mood and anxiety disorders, and other neurodevelopmental disorders such as autism spectrum disorder is frequent (Antshel et al., 2013; Connor et al., 2010; Meinzer et al., 2014). ADHD is regarded as an etiologically multifactorial disorder, with both genetic and environmental factors influencing the clinical presentation and course (Thapar & Cooper, 2016). Neuropsychological deficits are frequent in ADHD as well. Perhaps the most thoroughly investigated domain is that of executive functioning, a broad concept referring to higher-­‐order processes required for goal-­‐directed behavior and self-­‐
regulation. Individuals with ADHD often present with prominent impairment in 9 executive functioning (Coghill et al., 2014a), which may be reflected in impaired performance on a spectrum of cognitive tasks. Especially poor performance on cognitive control tasks, measuring the ability to inhibit inappropriate responses in favor of more appropriate ones, and on working memory tasks, measuring the capacity to temporarily remember, process, and manipulate information, has consistently been implicated in ADHD (Doyle, 2006; Kasper et al., 2012). Apart from executive functioning impairments, poor performance on tasks measuring sustained attention and vigilance (Huang-­‐Pollock et al., 2012), motivation and reward sensitivity (Luman et al., 2005; Volkow et al., 2011), and timing (Noreika et al., 2013) has also frequently been reported, as well as on other domains. Although individuals with ADHD as a group show deficits in each of these domains, individual patients are more likely to present with problems in some but not all of these domains (van Hulst et al., 2015). Approximately 25-­‐30% of children with ADHD appear not to have neuropsychological impairments (Coghill et al., 2014a; Sonuga-­‐Barke et al., 2010). NEUROBIOLOGY OF ADHD
The dopamine and norepinephrine systems have been implicated in ADHD (Del Campo et al., 2011; Swanson et al., 2007b). Dopamine especially has extensively been studied in this regard. In the human brain, dopamine is abundant in the frontal-­‐
striatal pathways, connecting the striatum (caudate nucleus, putamen, and nucleus accumbens) to the orbitofrontal, dorsolateral, and ventrolateral prefrontal cortices, and the pre-­‐ and supplementary motor areas (Leh et al., 2007). Neuroimaging studies consistently provide evidence of changes in the frontal-­‐striatal pathways of children and adults with ADHD as compared to their typically developing peers. Such changes entail structural differences, such as local volume reduction in the caudate and putamen (Frodl & Skokauskas, 2012; Nakao et al., 2011), reduced grey matter volume and cortical thickness in prefrontal areas (Depue et al., 2010; Greven et al., 2015; Shaw et al., 2006), and compromised structural integrity of white matter fibers connecting the striatal and frontal regions (Ashtari et al., 2005; Casey et al., 2007; Pavaluri et al. 2009; van Ewijk et al., 2014b). Furthermore, ADHD has been associated with altered activation patterns in frontal-­‐striatal circuits during various cognitive tasks (meta-­‐analysis by Dickstein et al., 2006; review by Cubillo et al., 2012a). Radiotracer techniques such as positron emission tomography (PET) and single photon emission computed tomography (SPECT) have provided more direct evidence of altered dopamine binding patterns in the striatum of patients with ADHD (Zimmer, 2009). Changes in the frontal-­‐striatal system have been linked to severity of ADHD symptoms (Cubillo et al., 2011; Shaw et al., 2011; Yang et al., 2016) and cognitive impairment (Casey et al., 2007; Depue et al., 2010; Valera et al., 2010). Of note, ADHD 10 has also been associated with more global brain changes (i.e., decrease in total brain volume), as well as with localized brain changes in areas outside the frontal-­‐striatal circuits such as the parietal cortices, thalamus, amygdala, and cerebellum, and altered activation patterns within other networks such as the default-­‐mode network. Such global changes might be associated with altered levels or metabolism of norepinephrine, a neurotransmitter that is widely distributed across the brain, but technological drawbacks have hampered radiotracer studies of norepinephrine metabolism in humans (Zimmer, 2009). For an accessible review on structural and functional brain changes associated with ADHD, see (Cortese, 2012). STIMULANT TREATMENT
ADHD treatment typically consists of behavioral interventions (e.g., cognitive behavioral therapy, parent training) and/or pharmacological interventions. Treatment with stimulants such as methylphenidate and d-­‐amphetamine is the pharmacological intervention of first choice in ADHD (NHS NICE guideline, 2008). Robust acute treatment effects, including reduction of hyperactivity symptoms and attention problems, occur in the majority of patients (Chan et al., 2016; Faraone & Buitelaar, 2010; MTA Cooperative Group, 1999). Moreover, stimulant treatment is generally well tolerated and adverse events are rare (e.g. Barbaresi et al., 2006; Storebo et al., 2016; Wilens et al., 2006). Perhaps not surprisingly, treatment with stimulants has become increasingly customary in recent years, resulting in a large and growing number of children, adolescents and adults worldwide being exposed (Dalsgaard et al., 2013; Trip et al., 2009). Stimulant treatment is often initiated shortly after diagnosis, typically between the ages of 6 and 11 (van den Ban et al., 2015). Oftentimes, short-­‐acting, immediate-­‐release formulations are prescribed at first (e.g., generic methylphenidate, Ritalin®, Dexedrine®, or Adderall®), and longer-­‐acting or extended-­‐release formulations are prescribed only at second instance (e.g., Concerta®, Equasym®, or Adderall-­‐XR®). When patients respond well to treatment, stimulant use is often continued throughout childhood and (early) adolescence (van den Ban et al., 2010; Wong et al., 2009). The public and scientific debate about long-­‐term outcomes of stimulant treatment has been intense, especially after the large-­‐scale MTA consortium reported that, in the long term, stimulant treatment was not superior compared to behavioral treatment or even to community care (Jensen et al., 2007; Molina et al., 2009). Individual treatment characteristics such as age of treatment onset, treatment duration, and dose, vary substantially from one patient to the other. There have been reports suggesting that such individual differences with regard to treatment patterns may predict long-­‐term outcomes (Dalsgaard et al., 2014; Mannuzza et al., 2008). 11 ACUTE STIMULANT EFFECTS IN THE BRAIN
The immediate effects of stimulants on the brain are relatively well known. Within one hour after a single clinical dose, stimulants block dopamine and norepinephrine reuptake in the presynaptic terminal. As a result, dopamine and norepinephrine levels in the synaptic cleft increase, which in turn enhances stimulation of postsynaptic monoamine receptors (e.g. Arnsten, 2006; Kuczenski & Segal, 1997; Volkow et al., 2001). Magnetic resonance imaging (MRI) techniques offer non-­‐invasive, accessible, and versatile tools to study stimulant effects in the developing human brain. Especially functional MRI studies investigating acute stimulant effects on neuronal activation patterns, measurable as the blood oxygenation level dependent (BOLD-­‐) response that indicates cerebral blood flow, are abundant. In such studies, participants with ADHD typically perform a cognitive task while in the scanner (generally a task of which performance is known to improve upon stimulant intake) on two occasions: once shortly after intake of a single clinical dose of methylphenidate, and once while unmedicated or on placebo. A fairly consistent overall picture emerges across tasks. During tasks tapping cognitive control, methylphenidate appears to normalize brain activation patterns in children with ADHD by enhancing activation in frontostriatal circuits (Epstein et al., 2007; Lee et al., 2010; Rubia et al., 2011a; Vaidya et al., 1998) and parietal brain regions (Rubia et al., 2011b). Methylphenidate has been found to restore default-­‐mode network suppression towards normative levels during cognitive control as well (Liddle et al., 2011; Peterson et al., 2009). Normalization of prefrontal cortex activation has also been observed during emotional processing (Posner et al., 2011). During tasks measuring different aspects of attention, methylphenidate appears to, at least partially, normalize brain activation and functional connectivity patterns (Rubia et al., 2009; Shafritz et al., 2004). Less consistent are the findings of acute methylphenidate effects during working memory tasks. For example, prefrontal cortex activation was found to increase, decrease, or remain unchanged after methylphenidate administration (Kobel et al., 2009; Prehn-­‐Kristensen et al., 2011; Sheridan et al., 2010; Wong & Stevens, 2012). Finally, normalizing effects of a single dose of methylphenidate have also been reported when no tasks were performed, i.e., during so-­‐called resting-­‐state scans (An et al., 2013; Anderson et al., 2002; Teicher et al., 2000). For a review about stimulant-­‐induced changes in brain activation patterns in patients with ADHD, see Rubia et al., 2014. 12 LONG-TERM TREATMENT EFFECTS IN THE BRAIN
Considering the substantial acute effects of a single dose of methylphenidate in the brain, it may be expected that repeated exposure to stimulants could cause lasting brain changes as well. Different mechanisms may underlie such lasting changes, two of which I wish to highlight. On a positive account, repeated manifestation of typical, age-­‐appropriate behaviors (i.e., paying attention in class, making non-­‐impulsive decisions) could increase the strength of neural networks underlying such behaviors through a process called activity-­‐induced neuronal plasticity. Similar to how, for instance, juggling practice is known to result in increased grey matter volume in the motor cortex, repeated ‘practice’ of non-­‐ADHD behavior may lead to attenuation of brain changes associated with untreated ADHD (Kasparek et al., 2015). Repeated positive reinforcement and appraisal of non-­‐ADHD behavior may strengthen such effects. Alternatively, on a negative account, lasting brain changes may represent neurochemical damage from repeated exposure to dopaminergic agents, similar to a scar. Dopaminergic nerve terminals may suffer detrimental effects, generally referred to as neurotoxicity, resulting from the accumulation of reactive oxygen species (i.e., oxidative stress) occurring when levels of extracellular dopamine are high and vesicular dopamine storage is disrupted (Berman et al., 2009). Evidence for neurotoxic stimulant effects derives almost exclusively from animal studies in which very high ‘binge’ (pharmacological) doses of stimulants are administered. Translated to humans, such binge patterns may be more compatible with stimulant abuse (e.g., methamphetamine addiction) or recreational stimulant use (e.g., MDMA/ecstasy) rather than stimulant treatment for ADHD as offered in clinical practice (Lynch et al., 2010). In patients with ADHD, two radiotracer studies have investigated long-­‐term consequences of stimulant treatment for dopamine metabolism in the brain. The results of both studies suggested long-­‐term effects. In the first, patients with ADHD, and especially those with a history of childhood stimulant treatment, presented with down-­‐regulated dopamine metabolism in the striatum, amygdala and midbrain compared to controls (Ludolph et al., 2008). In the second study, individuals with ADHD presented with normative levels of striatal dopamine transporter availability prior to treatment, but showed higher levels after one year of methylphenidate treatment (Wang et al., 2013). Few MRI studies have evaluated long-­‐term consequences of stimulant treatment. Lasting brain changes after stimulant treatment can be studied using structural MRI, functional MRI, or diffusion MRI, but the latter two techniques have very rarely been employed. Structural MRI studies reporting on grey matter changes have yielded mixed results. Two meta-­‐analyses report indirect evidence that 13 stimulant treatment is associated with more normative basal ganglia volumes: across different studies, findings of caudate nucleus and putamen volume reduction were more pronounced when predominantly medication-­‐naïve individuals were included, as compared to when more previously stimulant-­‐treated individuals were included (Frodl & Skokauskas, 2012; Nakao et al., 2011). However, large-­‐scale original studies have failed to detect volume differences in the striatum between treated and untreated patients, or associations between treatment duration and striatal volumes (Greven et al., 2015; Onnink et al., 2014; Semrud-­‐Clikeman et al., 2014; Shaw et al., 2014). In other brain regions, including the lateral prefrontal cortex, anterior cingulate cortex, parieto-­‐occipital cortices, thalamus, and cerebellum, treatment effects suggestive of normalization have been reported (Bledsoe et al., 2009; Ivanov et al., 2010; Semrud-­‐Clikeman et al., 2006 and 2012; Shaw et al., 2009). Here, treatment-­‐
naive children with ADHD, but not those with a history of stimulant treatment, showed significant volume reductions compared to their typically developing peers. Note that findings of long-­‐term structural brain changes after stimulant treatment are, with one exception (Shaw et al., 2009) based on observational and cross-­‐sectional data, are confined to specific and small portions of the cortex while leaving other abnormalities associated with ADHD unchanged, and have typically not been replicated. Stimulant treatment may also, in the long-­‐term, affect structural integrity of white matter pathways in the brain, which can be assessed using diffusion MRI. Only two studies, each with few participants, have investigated associations between treatment and white matter integrity. The first compared integrity of the frontal-­‐
striatal tracts between children with a relatively short versus a relatively long history of stimulant treatment, and found no differences (De Zeeuw et al., 2012). The second performed a whole-­‐brain hypothesis-­‐free comparison of treatment-­‐naïve children with ADHD, stimulant-­‐treated children with ADHD, and healthy control children (De Luis-­‐García et al., 2015). In this study, stimulant treatment was found to be associated with enhanced structural connectivity in several major white matter bundles, including those of orbitofrontal-­‐striatal pathways. Functional MRI studies into lasting stimulant treatment effects have been equally rare. Overall, after medication wash-­‐out, a history of stimulant treatment appears not to translate into lasting alterations of brain activation patterns during cognitive control, although some effects restricted to highly specific task conditions have been observed (Konrad et al., 2007; Pliszka et al., 2006). In contrast, during emotion and reward processing, stimulant-­‐naïve adults with ADHD have shown functional brain changes compared to controls, that were not seen in adult patients with a history of childhood stimulant treatment, which may suggest lasting normalization of activation patterns (Schlochtermeier et al., 2011; Stoy et al., 2011). 14 Thus, in short, there is preliminary evidence for stimulant treatment-­‐induced normalization of specific structural brain changes associated with ADHD, including grey matter volume reductions in the frontal and anterior cingulate cortex. Findings in the basal ganglia have been inconclusive. Studies that evaluate lasting treatment effects on white matter structural connectivity and brain activation patterns are markedly underrepresented and hence do not allow for drawing firm conclusions. CHALLENGES AHEAD
The available literature provides important directions for the work in thesis. Here, I discuss several questions and issues that remain unanswered in the existing literature, which have guided our research aims and objectives described in the next section. First, more research attention is warranted regarding long-­‐term effects of stimulant treatment on non-­‐volumetric brain outcomes, such as activation patterns and structural and functional connectivity, but also advanced cortical surface measures such as cortical thickness and shape, deserve more research attention. Especially functional neuroimaging can be informative, as it has the potential to bridge the gap between brain structure on the one hand, and clinical outcomes of ADHD and treatment on the other. After all, intuitively, the suggested normalization of structural brain changes after long-­‐term stimulant treatment (Frodl & Skokauskas, 2012; Nakao et al., 2011) appears at odds with the rapid return of ADHD symptoms when stimulant treatment is temporarily ceased. Second, a developmental perspective will be critical. The brain undergoes rapid developmental changes throughout childhood, adolescence, and young adulthood. Moreover, different brain regions and neural networks show distinct developmental trajectories (Raznahan et al., 2011; Wierenga et al., 2014). As a result, stimulant treatment effects are likely to be both developmentally sensitive (i.e., dependent on age of treatment, as has previously been shown in animal studies [van der Marel et al., 2014 and 2015]) and regionally specific. Moreover, treatment effects may become apparent only after adolescence or young adulthood, when neurodevelopment enters a more stable phase, which emphasizes the importance of long-­‐term follow-­‐up and developmentally matched reference groups, including typically developing adolescents. Third, differences between patients that may predispose lasting treatment effects, such as genetic variation, remain largely unexplored. Similar to clinical or behavioral response to treatment, the brain’s response to stimulants is likely to differ between individuals. As one example directly linked to the biological substrates of ADHD, an acute dose of methylphenidate was found to attenuate neural response to reward in adults with ADHD carrying the 9R risk allele of the dopamine transporter 15 gene, but not in those without the risk allele (Aarts et al., 2015). Lasting brain changes may be predisposed by genetic makeup in a similar fashion. Investigating these and other sources of interindividual differences (such as age, gender, and symptom severity) will not only advance our understanding of the neurobiological substrates of stimulant treatment, but may also potentially be of prognostic value to patients in the future. Finally, fourth, the distinction between previously-­‐treated and stimulant-­‐
naive patients with ADHD is a practical and often used yet overly simplistic representation of complex stimulant treatment trajectories. In practice, patients with ADHD initiate and discontinue treatment at different ages, require different dosages, do or do not receive other psychoactive medications, etc. More specifically, animal studies have shown that lasting treatment effects may be more pronounced when treatment occurred at younger age (van der Marel et al., 2014). Moreover, acute beneficial effects of stimulants were found to be reversed by concurrent antipsychotic exposure in rats (Cheng & Li, 2013). Studying long-­‐term treatment effects thus requires consideration of individual differences in treatment trajectories of stimulant and non-­‐stimulant medications. Taken one step further, recreational stimulant use patterns may be informative as well, as these are typically characterized by infrequent, irregular, high-­‐dose exposure, which is very different from the everyday low-­‐dose stimulant exposure in ADHD treatment. Here, too, different trajectories of stimulant exposure and their associated neural and behavioral outcomes are likely to provide insight into the working mechanisms of stimulant treatment. Furthermore, such information could in the future contribute to optimized clinical decision-­‐making. AIMS, OBJECTIVES, AND CHAPTER OUTLINE
The first and foremost overall aim of this thesis is to describe long-­‐term effects, or the absence thereof, of stimulant treatment on the developing brain in children, adolescents and young adults with ADHD. The second aim of this thesis is to advance our knowledge about mechanisms underlying stimulant effects in the developing brain of young people affected with ADHD. To these aims, we formulated the following objectives: 1) to investigate structural brain correlates of long-­‐term stimulant treatment, with a special focus on grey and white matter aspects of the frontal-­‐striatal circuits; 2) to explore functional brain correlates of long-­‐term stimulant treatment; 3) to investigate how individual differences in relevant patient characteristics (e.g., gender, age, genetic makeup) may predict long-­‐term treatment outcomes; 4) to explore which treatment characteristics (e.g., dose, age of treatment onset, concurrent non-­‐stimulant treatment) are predictive of long-­‐term treatment 16 outcomes; and 5) to evaluate long-­‐term clinical and cognitive effects of stimulant treatment beyond the level of acute changes that are well-­‐known. These objectives are addressed in eight chapters. CHAPTER 2 provides an extensive overview of the available literature regarding acute and long-­‐term stimulant treatment effects on the brain in children and adolescents with ADHD. Chapter two is followed by seven empirical chapters, all except one (chapter eight) based on data from the NeuroIMAGE study (BOX 1). First, we investigate associations between long-­‐term stimulant treatment and brain structure and function, while taking into account individual differences regarding patient characteristics and treatment patterns. In CHAPTER 3, we compare grey matter cortical thickness across the cortical mantle between individuals with and without ADHD, followed by a detailed evaluation of the effects of cumulative stimulant dose, age of stimulant treatment onset, and other treatment parameters. In CHAPTER 4 we describe changes in structural connectivity of the frontal-­‐striatal white matter pathways in patients, and how these pathways may be affected by long-­‐term stimulant treatment. Next, in CHAPTER 5, we investigate whether reward processes in the brain are influenced by long-­‐term stimulant treatment. To this end, we evaluate whether brain activation patterns during anticipation and/or reception of a monetary reward are associated with history of stimulant treatment even while off-­‐medication. In chapters six through eight, more focus is placed on patient-­‐ and treatment-­‐
related features that may mediate or moderate treatment effects in the brain. In CHAPTER 6, we evaluate how patient characteristics, i.e., age and genetic variability in two dopaminergic genes, may interact to affect outcome of long-­‐term stimulant treatment at the level of brain structure. In CHAPTER 7, we investigate whether stimulant treatment effects in the basal ganglia may be altered by augmentation with low-­‐dose atypical antipsychotics such as risperidone, a dopaminergic antagonist with neurochemical properties opposing those of stimulants. In CHAPTER 8 we take a sidestep, and look at the effects of non-­‐medical, recreational use of stimulants, which may include prescription medications but mainly entailed cocaine and ecstasy (3,4-­‐
methylenedioxymethamphetamine, MDMA). Recreational stimulant exposure patterns are very different from typical ADHD treatment patterns. In the Youth At Risk (YAR) sample, based in the University of California, San Diego (BOX 2), we evaluate the effect of incidental, high-­‐dose stimulant exposure on the development of hippocampus structure and memory performance. Finally, after having explored the neural correlates of stimulant exposure in multiple ways, we turn to the analysis of clinical and cognitive outcomes of stimulant treatment in CHAPTER 9. Here, we make use of the valuable longitudinal aspect of the 17 NeuroIMAGE sample, and relate changes in clinical and cognitive outcomes over six years to stimulant intake in that exact same timeframe. BOX 1. The NeuroIMAGE study The NeuroIMAGE study is a follow-­‐up phase (2009-­‐2012) of the Dutch part of the International Multicenter ADHD Genetics project (IMAGE; 2003-­‐2006). It is a multi-­‐site family-­‐based cohort study designed to investigate the course of ADHD and its cognitive and neurobiological underpinnings. More than 1,000 children, adolescents, and young adults from almost 500 families participated in NeuroIMAGE (70.3% ADHD, 56% male, average age = 17.0 years old). Data collection included diagnostic interviews, behavioral questionnaires, cognitive assessment, structural and functional neuroimaging, collection of lifetime pharmacy transcripts, and genotyping. More information about the NeuroIMAGE study can be found online (www.neuroimage.nl) and in von Rhein et al., 2015a. Chapters three through seven and chapter nine are based in the NeuroIMAGE cohort. BOX 2. The YAR study The Youth At Risk (YAR) study is a prospective longitudinal neuroimaging study of adolescents at elevated risk for substance use problems. The study started in 2002 and is based in San Diego, California. At baseline, a total of 295 healthy boys and girls between twelve and fourteen years of age, with no or minimal exposure to alcohol or substances, were recruited. The sample was enriched with children who had a family history of substance use disorders, which put them at high risk for developing substance use problems themselves. Initial assessment included substance use interviews, behavioral questionnaires, cognitive testing, and structural and functional neuroimaging. After enrollment, participants were administered substance use interviews every six months, and were invited for complete assessment including an MRI scan every year. The YAR-­‐
study is now in its 12th year of funding, and participants are currently in their mid-­‐
twenties. Chapter eight is based in the YAR cohort. 18 19 20 Chapter 2 MR IMAGING OF THE EFFECTS OF
METHYLPHENIDATE ON BRAIN STRUCTURE AND
FUNCTION IN ADHD
Published as: Schweren LJS, de Zeeuw P, Durston S. MR imaging of the effects of methylphenidate on brain structure and function in attention-­‐deficit/hyperactivity disorder. Eur Neuropsychopharmacol. 2013;23(10):1151-­‐1164. 21 ABSTRACT
Methylphenidate is the first-­‐choice pharmacological intervention for the treatment of attention-­‐deficit/hyperactivity disorder (ADHD). The pharmacological and behavioral effects of methylphenidate are well described, but less is known about neurochemical brain changes induced by methylphenidate. This level of analysis may be informative on how the behavioral effects of methylphenidate are established. This paper reviews structural and functional MRI studies that have investigated effects of methylphenidate in children with ADHD. Structural MRI studies provide evidence that long-­‐term stimulant treatment may normalize structural brain changes found in the white matter, the anterior cingulate cortex, the thalamus, and the cerebellum in ADHD. Moreover, preliminary evidence suggests that methylphenidate treatment may normalize the trajectory of cortical development in ADHD. Functional MRI has provided evidence that methylphenidate administration has acute effects on brain functioning, and even suggests that methylphenidate may normalize brain activation patterns as well as functional connectivity in children with ADHD during cognitive control, attention, and during rest. The effects of methylphenidate on the developing brain appear highly specific and dependent on numerous factors, including biological factors such as genetic predispositions, subject-­‐related factors such as age and symptom severity, and task-­‐related factors such as task difficulty. Future studies on structural and functional brain changes in ADHD may benefit from inclusion strategies guided by current medication status and medication history. Further studies on the effects of methylphenidate treatment on structural and functional MRI parameters are needed to address unresolved issues of the long-­‐term effects of treatment, as well as the mechanism through which medication-­‐induced brain changes bring about clinical improvement. 22 INTRODUCTION
Attention-­‐deficit/hyperactivity disorder (ADHD) is the most common developmental disorder in childhood, estimated to affect three to seven percent of school-­‐age children and 2.5 percent of adults globally (Polanczyk et al., 2007; Simon et al., 2009). ADHD is characterized by age-­‐inappropriate attention problems, and/or impulsivity and hyperactivity (American Psychiatric Association, 2000). The clinical presentation of ADHD is heterogeneous and comorbidity with oppositional defiant disorder (ODD), conduct disorder (CD), other neurodevelopmental disorders, mood and anxiety disorders is frequent (Pliszka, 2000). ADHD is regarded as an etiologically multifactorial disorder, with most likely both genetic and environmental factors influencing the clinical presentation (APA, 2000; Plomp et al., 2009). Stimulant treatment remains the most common pharmacological intervention in ADHD and is effective for the majority of affected individuals (e.g., Barbaresi et al., 2006). Methylphenidate, available in various forms of delivery, is commonly regarded as the first-­‐choice stimulant intervention (Meijer et al., 2009). Treatment is generally well tolerated and side-­‐effects are typically mild (Greenhill et al., 2002; Wilens et al., 2006). Robust behavioral effects of stimulants have been reported, and include reduced symptoms of hyperactivity and inattention (Conners, 2002; MTA Cooperative Group, 1999; Spencer et al., 1996; Swanson et al., 1993). The pharmacological properties of methylphenidate have been well investigated (Heal et al., 2009). Changes in the levels of dopamine and norepinephrine in the brain have frequently been associated with ADHD (Del Campo et al., 2011). A growing body of evidence suggests that methylphenidate increases the levels of dopamine and norepinephrine in the synaptic cleft, thereby stimulating their receptors (e.g., Arnsten, 2006; Kuczenski et al., 1997; Volkow et al., 2001). However, the question remains how these neurochemical methylphenidate-­‐induced changes in the synapse result in the clinical and behavioral effects of methylphenidate treatment (Winstanley et al., 2006). Moreover, much remains unknown about the manifestation of such neurochemical effects in structural brain development, and its behavioral correlates. One approach to addressing these questions is to make use of modern neuroimaging techniques such as magnetic resonance imaging (MRI). Neuroimaging has the potential to provide researchers with relevant biomarkers, or clinically relevant characteristics that may be objectively measured as an indicator of pharmacological responses to therapeutic intervention (Minzenberg, 2012). In recent years, structural and functional neuroimaging has been used to investigate the effects of medication on macroscopic brain measures in a variety of psychiatric disorders, including depression and schizophrenia (Honey & Bullmore, 2004). Not only have these studies contributed to our understanding of the brain 23 mechanisms which may drive the clinical effect of pharmacological treatments for these psychiatric disorders, but they have also advanced our understanding of the mechanisms underlying these disorders themselves (Doyle et al., 2005). Whereas direct pharmacological effects of methylphenidate have been investigated using the more invasive methods of positron emission tomography (PET) and SPECT, magnetic resonance imaging (MRI) is a tool more apt to studying macroscopic brain effects of methylphenidate. MRI has undergone rapid technological progression in recent years, allowing for the investigation of both structural and functional correlates of pharmacological treatment. In addition, MRI is a noninvasive tool, permitting its use in children, and is available in many research institutions (Honey & Bullmore, 2004). Since the advent of MRI, several reports have appeared on the effects of methylphenidate on brain measures derived from MRI. Recent extensive and high-­‐
quality studies of methylphenidate effects (e.g., Rubia et al., 2009, 2011a and 2011b) have shown significant effects of methylphenidate. Owing to the high degree of specialization of such investigations, an increasingly detailed and complex picture emerges of the effects of methylphenidate on the developing brain. In an attempt to integrate such detailed descriptions of methylphenidate effects, this paper reviews the effects of childhood methylphenidate treatment on structural and functional MRI measures. Both acute and long-­‐term effects of psychostimulant treatment will be addressed. METHODS This paper provides a systematic narrative review of the currently available MRI studies addressing the macroscopic effects of methylphenidate on the developing brain. The online databases of the US National Library of Medicine and the National Institutes of Health (PubMed), of the American Psychological Association (PsychInfo), and Web of Science were searched for relevant articles. In addition, reference lists of relevant articles were searched. Keywords for the search included ADHD, methylphenidate, MRI, fMRI, and neuroimaging. All abstracts that met these search criteria were read, and we selected those articles that (1) were written in English, (2) addressed the effects of methylphenidate administration during childhood (as opposed to drug administration during adulthood) in individuals with ADHD (as opposed to in other diagnostic groups), and (3) used magnetic resonance imaging (MRI) as a neuroimaging tool. When eligibility for inclusion in the review was doubted, the authors discussed and included the paper only upon consensus. Our search resulted in a total of 27 articles published between November 1998 and March 2012, including 26 original research articles and one meta-­‐analysis/review article. 24 RESULTS
The effects of methylphenidate on brain structure in children with ADHD A large and growing number of structural MRI studies document neuroanatomical changes between children with ADHD and typically developing controls (Durston et al., 2009; Seidman et al., 2005, Valera et al., 2007). A decrease in global brain volume has been reported in children and adolescents with ADHD compared to typically developing children, as well as local volume reductions in the frontal cortex, caudate nucleus, the cerebellum, and the corpus callosum, among other brain regions (Seidman et al., 2005; Valera et al., 2007). However, the long-­‐term effects of stimulant treatment on these structural changes have received little attention. Studies that have specifically targeted such analyses are summarized in Table 1. Importantly, all studies investigating the effects of stimulant treatment on structural MRI measures are naturalistic in nature, and all except one (Shaw et al., 2009) employ a cross-­‐sectional design (Table 1). Although this approach may be the most feasible approach to studying long-­‐term treatment effects in children, the limitations of naturalistic, cross-­‐sectional study designs need to be kept in mind when interpreting the results. A small number of volumetric studies have compared volumes of specific brain regions (usually gray matter) between children with ADHD who were medication-­‐naïve, children with ADHD who had been using medication, and typically developing children. An early study investigated total cerebral and cerebellar gray matter volume, as well as gray matter volume in the four major lobes, in these three groups. Compared to typically developing children, both medicated and medication-­‐
naïve subjects with ADHD showed reduced gray matter volumes, suggesting that gray matter volumes may not be affected by methylphenidate treatment (Castellanos et al., 2002). More recently, the majority of studies have turned to a hypothesis-­‐driven region of interest (ROI) approach, where the volume of an a priori defined brain region was compared between groups. With this approach, medication effects have been found in the anterior cingulate cortex (ACC; Semrud-­‐Clikeman et al., 2006), the pulvinar nucleus of the thalamus (Ivanov et al., 2010), and in the posterior inferior cerebellum (Bledsoe et al., 2009). Compared to typically developing children, children with ADHD who had not been treated with stimulants had significant volume reductions in these regions. These ADHD-­‐related volume reductions did not occur, or were attenuated, in children who had been medicated with psychostimulants. No such effect of stimulant treatment was found in the caudate nucleus (Castellanos et al., 2002), although one study investigating a small sample of children with ADHD, the majority of them diagnosed with comorbid CD, suggested smaller caudate nucleus 25 volumes in previously medicated children with ADHD compared to medication-­‐naïve children with ADHD (Bussing et al., 2002). TABLE 1. Structural magnetic resonance imaging studies investigating the effects of methylphenidate treatment in children with ADHD. Study Main findings Castellanos et al. (2002) Design: TDC vs. ADHDon vs. ADHDoff. Volumetric study of the gray and white matter in frontal, temporal, parietal and occipital lobes, basal ganglia and cerebellum Subjects: NTDC=139, NADHD-­‐off=103, NADHD-­‐on=49; age range 5–19 Findings: Children with ADHD had smaller cerebral volumes, smaller cerebellar volumes, and smaller temporal gray matter volumes than TDC. MPH affected WM but not GM volumes. WM volumes in non-­‐medicated children with ADHD were smaller than in TDC, while medicated children with ADHD did not show WM volume reductions Limitations: 4, 5 Bussing et al. (2002) Design: ADHDoff vs. ADHDon Subjects: NTDC=19, NADHD-­‐off=7, NADHD-­‐on=5; age range 8–12 Findings: Caudate nucleus volumes were smaller in medicated children with ADHD when compared to medication-­‐naïve children with ADHD Limitations: 1, 2, 4, 5, 6 Semrud-­‐Clikeman et al. Design: TDC vs. ADHDoff vs. ADHDon (2006) Subjects: NTDC=21, NADHD-­‐off=14, NADHD-­‐on=16; age range 9–15 Findings: Caudate nucleus and ACC volume were investigated. MPH did not affect caudate nucleus volume. Compared to in TDC, right ACC volume was decreased in medication-­‐naïve children with ADHD, but not in medicated children with ADHD. A similar trend was found in the left ACC Limitations: 4, 5, 6 Bledsoe et al. (2009) Design: TDC vs. ADHDoff vs. ADHDon Subjects: NTDC=15, NADHD-­‐off=14, NADHD-­‐on=18; mean age=11.5(2.5) Findings: Overall cerebellar volume was the same in all three groups. There were no volumetric group differences in the anterior or posterior superior vermis of the cerebellum. The inferior vermis of the cerebellum was smaller in non-­‐medicated children with ADHD compared to both medicated children with ADHD and TDC Limitations: 4, 5, 6 Shaw et al. (2009) Design: ADHDoff vs. ADHDon vs. template of cortical development based on TDC participants Subjects: NTDC=294, NADHD-­‐off=19, NADHD-­‐on=24, age range 9–20 Findings: At baseline, no group differences in cortical thickness were 26 found between the three groups, but both ADHD groups contained children with a history of stimulant medication. At follow-­‐up, cortical thickness did not differ between controls and medicated children with ADHD, while non-­‐medicated children with ADHD showed a higher rate of cortical thinning in the IFG, the right precentral gyrus, and right parieto-­‐occipital gyrus Limitations: 3, 5, 7 Ivanov et al. (2010) Design: TDC vs. ADHDoff vs. ADHDon Subjects: NTDC=59, NADHD-­‐off=15, NADHD-­‐on=31, age range 8–18 Findings: Children with ADHD had reduced regional volumes of thalamic surfaces, including the pulvinar nucleus. Medicated children with ADHD had larger pulvinar volumes than their non-­‐
medicated peers Limitations: 4, 5, 6, 7 De Zeeuw et al. (2012) Design: TDC vs. ADHD, and ADHDlong-­‐medication vs. ADHDshort-­‐medication Subjects: NTDC=34, NADHD=30, NADHD-­‐long-­‐medication=13, NADHD-­‐short-­‐medication=13, age range 6–16 Findings: Microstructural abnormalities were observed in frontostriatal WM in children with ADHD. Medication duration (long vs. short) was calculated relative to the children's age. Exploratory analyses showed no effect of treatment duration on these abnormalities Limitations: 4, 5, 6 Nakao et al. (2011) Design: TDC vs. ADHD, meta-­‐regression of 9 pediatric VBM datasets Subjects: NTDC=198, NADHD=202 Findings: In a VBM meta-­‐analysis of regional GM volumes, children with ADHD showed GM volume decrease in the right caudate nucleus, and GM volume increase in the left precuneus cortex. In studies including a larger percentage of medicated children with ADHD, right caudate nucleus volume was larger, thus more normal, compared to in studies including a smaller percentage of medicated children with ADHD. No medication effect was found in the precuneus cortex Limitations: 4, 5, 7 ADHDon, children with ADHD receiving stimulant treatment; ADHDoff, children with ADHD not receiving stimulant treatment; TDC, typically developing controls; MPH, methylphenidate; CD, conduct disorder; GM, gray matter; WM, white matter; ACC, anterior cingulate cortex; IFG, inferior frontal gyrus; VBM, voxel-­‐based morphometry. Study limitations are coded as follows: 1=small sample size (N<45 or n per group/condition <15); 2=no direct between-­‐group comparison between TDC, ADHDmedicated and ADHDnon-­‐medicated is reported; 3=medication history prior to study participation is unknown/not reported; 4=cross-­‐sectional study design: no within-­‐subject analysis of medication effects; 5=naturalistic study design: no randomized clinical trial; 6=selective analyses of brain regions; 7=medication group/condition includes pharmacological treatment other than MPH 27 Another approach to investigating structural changes in the brain involves data-­‐driven methods, where large brain areas are investigated without an a-­‐priori hypothesis, such as with voxel-­‐based morphometry (VBM) or cortical thickness measurements. In VBM, individual scans are normalized to a template brain, effectively comparing gray and white matter volume between groups in each voxel of the brain, rather than in specific anatomical brain regions of interest (Ashburner & Friston, 2000). In a meta-­‐analysis of VBM studies in ADHD, a medication effect was found in the right basal ganglia (Nakao et al., 2011). Across individuals with ADHD, gray matter volume in the right caudate nucleus was decreased. However, the percentage of medicated subjects in the studies included in this meta-­‐analysis correlated positively with gray matter volume in the right caudate nucleus. In studies with a larger percentage of medicated children, right caudate nucleus volume was larger, thus more normal, compared to in studies including a smaller percentage of medicated children. No such medication effect was found in the left precuneus that showed increased volume in children with ADHD compared to controls (Nakao et al., 2011). Cortical thickness analyses provide a more direct measure of the thickness of the outer gray matter layer of the cerebrum than volumetric analyses can provide. Shaw et al. (2009) applied such analyses in a unique longitudinal study, and found that unmedicated children with ADHD showed excessive cortical thinning compared to typically developing children during adolescence, whereas children with ADHD receiving stimulant medication during the period of study did not. This study suggests that stimulant treatment may reduce the rate of cortical thinning in frontal and parieto-­‐occipital regions during adolescence. Methylphenidate treatment may also affect measures of white matter structure (Table 1). For example, cerebral white matter volume was found to be reduced in medication-­‐naïve children with ADHD compared to typically developing children, but also compared to previously medicated children with ADHD. In fact, children with ADHD with a history of psychostimulant use showed no changes in total white matter volume, or in white matter volume in any of the four major lobes, suggesting that methylphenidate treatment may normalize white matter volume reductions in children with ADHD (Castellanos et al., 2002). In line with this, two studies investigating white matter volume that included previously medicated subjects with ADHD only, reported no evidence for white matter volume reduction (Batty et al., 2010; Carmona et al., 2005). Unfortunately, in a meta-­‐analysis of volumetric studies, Valera et al. (2007) were unable to assess possible effects of stimulant medication on white matter volume, due to a lack of information regarding medication status and history. A recent study that investigated the microstructure of frontal-­‐striatal white matter tracts in children with ADHD using diffusion tensor imaging (DTI) found decreased structural connectivity within these tracts (De Zeeuw 28 et al., 2012). An exploratory analysis of the effects of stimulant medication compared subjects that had been medicated for a relatively long period with subjects that had been medicated for a shorter period. No significant effect of treatment duration on microstructural abnormalities was found (De Zeeuw et al., 2012). Thus, white matter volume may be affected by the use of stimulant medication, but the exact nature of these changes requires further investigation. In conclusion, structural MRI studies have provided evidence that stimulant treatment may normalize specific, but not all, structural brain abnormalities found in children with ADHD, in both gray and white matter. The effects of methylphenidate on patterns of brain activation in children with ADHD In contrast to the few structural MRI studies investigating the effects of methylphenidate, there have been numerous functional MRI (fMRI) studies on the acute effects of methylphenidate (Table 2). Such studies can be divided into those investigating spontaneous brain activation while at rest (resting-­‐state fMRI), and those investigating task-­‐related brain activation. Two resting-­‐state fMRI studies, employing T2 relaxometry as an indirect measure of cerebral blood volume, suggested that methylphenidate may have an acute normalizing effect on striatal (Teicher et al., 2000) and cerebellar (Anderson et al., 2002) activation in boys with ADHD during rest. Task-­‐related fMRI studies on the effects of methylphenidate have been more numerous and have often focused on cognitive domains implicated in ADHD, and where performance is known to improve with methylphenidate administration. A fairly consistent overall picture of the acute effects of methylphenidate administration emerges across various experimental designs. During tasks tapping cognitive control, the ability to inhibit inappropriate responses in favor of more appropriate ones, methylphenidate appeared to normalize brain activation patterns in children with ADHD by enhancing activation in frontostriatal circuits (Epstein et al., 2007; Lee et al., 2010; Rubia et al., 2011b; Vaidya et al., 1998). A similar upregulation of activation towards activation levels of typically developing children was found in parietal brain regions (Rubia et al., 2011a). Furthermore, methylphenidate administration was found to overcome deficits in the suppression of default mode network (DMN) activity, or activity within a network of brain regions typically deactivated during goal-­‐directed behavior: during cognitive control tasks, children with ADHD were shown to fail to suppress activation within this network, and methylphenidate was found to restore DMN suppression towards the levels of typically developing children (Peterson et al., 2009). The effects of methylphenidate on behavioral measures of attention problems are well established (e.g., MTA Cooperative Group, 1999). Cognitive tasks measuring 29 aspects of attention, such as the ability to sustain attention over time, or to divide attentional resources amongst subtasks, are frequently used in fMRI studies of ADHD. Methylphenidate appeared to at least partially normalize brain activation patterns in children with ADHD during divided attention (Rubia et al., 2009; Shafritz et al., 2004). In addition, methylphenidate has been shown to restore functional connectivity, or synchronized activity, in disparate but functionally related brain regions, in networks involved in attention in children with ADHD (Rubia et al., 2009). Interestingly, methylphenidate administration may also elicit non-­‐normalizing, additional activity in the brain. During vigilance, a basic form of sustained attention, methylphenidate administration was found to elicit compensatory activation in the prefrontal cortex of children with ADHD that was not evident in typically developing children (Rubia et al., 2009). By contrast, studies investigating the effects of methylphenidate on activation patterns during working memory tasks have yielded inconsistent results. During working memory, stimulant administration elicited increases of activation in the frontal cortex (Prehn-­‐Kristensen et al., 2011) and in frontal-­‐parietal networks (Wong & Stevens, 2012), but not in the striatum (Prehn-­‐Kristensen et al., 2011). However, opposing findings of reduction of prefrontal cortex activity by methylphenidate administration (Sheridan et al., 2010) or no effect of medication on frontal-­‐striatal activation (Kobel et al., 2009) have also been reported. The lack of a control group in some of these studies (Sheridan et al., 2010; Wong & Stevens, 2012) and the absence of a placebo condition in others (Kobel et al., 2009; Prehn-­‐Kristensen et al., 2011), may contribute to the conflicting findings, and limits the interpretation of the results. Studies investigating stimulant effects on regional functional connectivity during working memory tasks have similarly yielded inconsistent results. In a pilot study in adolescent girls with ADHD, stimulant medication decreased functional connectivity during a working memory task (Sheridan et al., 2010). By contrast, Wong and Stevens (2012) found that stimulant medication increased functional connectivity between frontostriatal brain structures and other brain regions in children with ADHD during a working memory task. However, since both studies did not include a control group, it is unclear how these methylphenidate-­‐induced changes compare to functional brain connectivity during working memory in typically developing children. In contrast to the numerous studies investigating the acute effects of methylphenidate administration, only two studies have investigated the effects of long-­‐term methylphenidate treatment on brain activation patterns in children with ADHD. Overall, after wash-­‐out, methylphenidate treatment appeared not to translate into long-­‐lasting alterations of brain activation patterns, nor into normalization of functional brain development. During tasks tapping aspects of cognitive control (Konrad et al., 2007; Pliszka et al., 2006), no changes were found in overall brain 30 31 Study Teicher et al., 2000 Anderson et al., 2002 Vaidya et al., 1998 Zang et al., 2005 Domain Resting state Response inhibition MPH tended to restore the neuronal Stroop effect in children with ADHD towards the pattern seen in typically developing children Typically developing children showed a ‘neuronal Stroop effect’ in brain activation patterns, in which brain activation during non-­‐congruent Stroop-­‐trials was larger than during congruent Stroop-­‐trials. Children with ADHD showed the opposite pattern, congruent trials eliciting more brain activation than non-­‐congruent trials ADHD vs. TDC: MPH effects: NTDC=9; NADHD=9; Age range 9–15 1,4,7 Subjects: Limitations: MPH increased striatal activation in children with ADHD, but decreased striatal activation in typically developing subjects. MPH increased frontal activation in both children with ADHD and typically developing children Children with ADHD showed decreased striatal activation, and increased frontal activation MPH effects: NTDC=6; NADHD=10; Age range 8–13 ADHD vs. TDC: 1.3.4.5.7 Subjects: Limitations: MPH decreased activity of the cerebellar vermis, but not the cerebellar hemispheres, in hyperactive children with ADHD. MPH increased activity of the cerebellar vermis in non-­‐
hyperactive children with ADHD No analyses ADHD vs. TDC: MPH effects: NTDC=6; NADHD=10; Mean age 9.6(1.6) Subjects: Limitations: 1,4,5,7 Children with ADHD showed lower bilateral putamen activation, but normal thalamic and caudate activation MPH increased putamen activation in hyperactive children with ADHD, but decreased putamen activation in non-­‐hyperactive children with ADHD. MPH did not affect thalamic or caudate activation ADHD vs. TDC: MPH effects: NTDC=6; NADHD=11; Age range 6–12 Subjects: Main findings TABLE 2 Functional magnetic resonance imaging studies investigating the effects of methylphenidate administration in childhood ADHD
32 Study Pliszka et al., 2006 Epstein et al., 2007 Peterson et al., 2009 Lee et al., 2010 Domain TABLE 2 Continued
Limitations: MPH effects: Subjects: Limitations: During interference suppression, MPH increased activation in the right PFC. There were no effects of MPH during response inhibition 1,2,3,4,9 NADHD=8; Age range 9–11 MPH restored task-­‐related deactivations in the DMN, and increased functional connectivity between the LPFC and the ACC 4 During a Stroop task, children with ADHD showed less prominent task-­‐related deactivation of the DMN than typically developing subjects. Functional connectivity between the Lateral PFC and the ACC was reduced in children with ADHD ADHD vs. TDC: MPH effects: NTDC=20; NADHD=16; Age range 7–18 Subjects: Limitations: MPH increased activation in the frontal lobe, the ACC, the inferior parietal cortex, the caudate nucleus and the cerebellum 1,3,7 Children with ADHD showed decreased activation in the frontal lobe, the ACC, the inferior parietal cortex, and caudate nucleus ADHD vs. TDC: MPH effects: NTDC=9; NADHD=20; Age range 7–9 1,5,6,7 Subjects: Limitations: One year MPH treatment increased ACC activation during incorrect responses, but not to TDC levels Overall, comparable activation was found in the ACC, dorsolateral PFC, and ventrolateral PFC, in typically developing children, medication-­‐naïve children with ADHD, and previously medicated children with ADHD. During incorrect responses, typically developing children showed increased ACC activation compared to during correct responses, while medication-­‐naïve children with ADHD showed the opposite pattern ADHD vs. TDC: MPH effects: NTDC=15; NADHD=17; Age range 9–15 1,3,7,9 Subjects: Limitations: Main findings 33 Study Rubia et al., 2011a Kobel et al., 2009 Sheridan et al., 2010 Prehn-­‐Kristensen et al., 2011 Wong & Stevens, 2012 Domain Working Memory TABLE 2 Continued MPH effects: Subjects: Limitations: Stimulant medication increased the magnitude of frontoparietal network activation during a working memory task. Moreover, stimulant medication increased regional functional connectivity between frontoparietal network structures and other brain regions NADHD=18; Age range 11–17 1,4,10 MPH restored frontal activation during the non-­‐distracted task to normal levels. MPH did not restore caudate nucleus activation during the distracted task In the non-­‐distracted task, children with ADHD showed reduced frontal activation compared to typically developing children. In the distracted task, children with ADHD showed a lack of caudate nucleus activation ADHD vs. TDC: MPH effects: NTDC=12; NADHD=12; Age range 10–17 1,2,3,4,7,8,10 MPH decreased PFC activation and functional connectivity between frontal and subcortical regions. MPH increased connectivity between the MFG and the cerebellar vermis NADHD=5; Age range 11–17 Subjects: Limitations: MPH effects: Subjects: Limitations: 1,4,10 Children with ADHD showed less activation in the frontal and parietal regions, and failed to recruit the cerebellum. Left sided frontal and parietal activation was more reduced than right sided activation No effects of MPH were found ADHD vs. TDC: MPH effects: NTDC=12; NADHD=14; Age range 9–13 Subjects: Limitations: MPH administration resulted in an upregulation of activity in the right inferior prefrontal cortex and in premotor areas in children with ADHD. While on MPH, children with ADHD no longer showed activity reduction in the right PFC or striato-­‐thalamic areas 1 During interference inhibition, boys with ADHD showed reduced activity in the right inferior PFC, left striatum and thalamus, mid-­‐cingulate/SMA region, and left superior temporal lobe MPH effects: NTDC=13; NADHD=13; Age range 10–16 ADHD vs. TDC Subjects: Main findings 34 Study Shafritz et al., 2004 Konrad et al., 2007 Rubia et al., 2009 Rubia et al., 2009 Domain Attention Reward processing TABLE 2 Continued Limitations: 1 MPH decreased orbito-­‐frontal activation During reward processing, children with ADHD showed increased orbito-­‐frontal and superior temporal activation MPH effects: NTDC=13; NADHD=13; Age range 10–16 ADHD vs. TDC: 1 Subjects: Limitations: MPH increased fronto-­‐striato-­‐cerebellar and parieto-­‐temporal activation, as well as fronto-­‐striatal and fronto-­‐cerebellar connectivity During selective attention, children with ADHD showed reduced activation and functional connectivity in fronto-­‐striato-­‐parieto-­‐cerebellar networks ADHD vs. TDC: MPH effects: NTDC=13; NADHD=13; Age range 10–16 1,3,6 Subjects: Limitations: MPH did not elicit developmentally appropriate increases in ACC and TPJ activation. However, whereas non-­‐medicated children showed developmentally inappropriate activation of the insula and striatum during reorienting of attention, this possibly compensatory activation was not present in children that received one year of stimulant treatment Whereas typically developing children showed increase in ACC and TPJ activation during attention processes over the course of one year, children with ADHD (regardless of medication status) did not MPH effects: NTDC=14; NADHD=16; Mean age 11.3(1.3) ADHD vs. TDC: -­‐ Subjects: Limitations: MPH increased activation in the left basal ganglia, but not in the middle temporal gyrus During tasks of selective and divided attention, children with ADHD showed less activation in the left basal ganglia and the middle temporal gyrus ADHD vs. TDC: MPH effects: NTDC=14; NADHD=27; Age range 12–20 2,3,4 Subjects: Limitations: Main findings Stoy et al., 2011 Rubia et al., 2011b Schlochtermeier et al., 2011 Error processing Emotion processing Limitations: MPH-­‐naïve adults with childhood ADHD, but not adults with ADHD who had been treated with MPH during childhood, showed decreased activation in the subgenual cingulate and the ventral striatum compared to TDC 1,5,6,7 No analyses ADHD vs. TDC: MPH effects: NTDC=10; NADHD-­‐untreated=10; NADHD-­‐treated=10; Mean age 28.4(5.2) Subjects: Limitations: During both failed and successful inhibition, MPH restored activation patterns to the levels of typically developing children. No differences between brain activation patterns of children with ADHD on MPH and typically developing children were observed 1 During failed response inhibition, or error monitoring, boys with ADHD showed reduced brain activation in the dorsomedial and left ventrolateral PFC, the thalamus, cingulate, and parietal areas. During successful response inhibition, boys with ADHD showed underactivation in the right medial temporal and inferior parietal lobe, the precuneus and the cerebellum MPH effects: NTDC=13; NADHD=12; Age range 10–16 ADHD vs. TDC: 1,5,6,7 Subjects: Limitations: During loss avoidance, insula activation was lower in drug-­‐naïve subjects in comparison to both TDC and previously treated groups Activation in the ventral striatum and OFC were equal in TDC adults, adults with ADHD who had been treated during childhood, and adults with ADHD who had not MPH effects: NTDC=12; NADHD-­‐untreated=12; NADHD-­‐treated=11; Mean age 27.5(4.6) ADHD vs. TDC: Subjects: Main findings ADHDon, subjects with ADHD receiving stimulant treatment; ADHDoff, subjects with ADHD not receiving stimulant treatment; TDC, typically developing control subjects; MPH, methylphenidate; ACC, anterior cingulate cortex; VBM, voxel-­‐based morphometry; TPJ, temporo-­‐parietal junction; PFC, prefrontal cortex; DMN, default mode network; RI, response inhibition; MFG, medial frontal gyrus. Study limitations are coded as follows: 1=small sample size (n per group/condition<15); 2=absence of control group/condition; 3=no direct between-­‐group comparison between TDC, ADHDon-­‐medication and ADHDoff-­‐medication is reported; 4=medication history prior to study participation is unknown/not reported; 5=cross-­‐sectional study design: no within-­‐subject analysis of medication effects; 6=naturalistic study design: no randomized clinical trial; 7=selective analyses of brain regions; 8=medication group/condition includes pharmacological treatment other than MPH; 9=statistical testing is not reported for all contrasts of interest; 10=absence of placebo group/ condition Study Domain TABLE 2 Continued 35 activation patterns between children who had previously received long-­‐term methylphenidate treatment, and children who had not. Observed long-­‐term methylphenidate treatment effects were restricted to highly specific tasks and brain regions. By contrast, two recent fMRI studies in currently non-­‐medicated adults with ADHD that specifically investigated the effect of stimulant treatment history during childhood did find evidence of lasting medication effects on brain activation patterns. Methylphenidate-­‐naïve adult subjects with ADHD showed changes in brain activation during emotional processing (Schlochtermeier et al., 2011) and during reward processing (Stoy et al., 2011) compared to typical control subjects, whereas adult subjects with ADHD who had received methylphenidate treatment during childhood did not. This suggests that childhood methylphenidate treatment may result in normalization of brain activation patterns in adulthood, similar to the acute functional normalization after methylphenidate administration in children with ADHD. However, opposing findings of larger functional brain abnormalities in adult ADHD subjects with a history of stimulant treatment, compared to subjects without a history of stimulant treatment, have also been reported (Ludolph et al., 2008). Importantly, these studies have limitations (Table 2). First, for many cognitive domains, the effects of methylphenidate have only been investigated once or twice. As such, replication is clearly needed. Second, sample sizes are typically small, with the majority of studies investigating fewer than fifteen subjects per group or condition. Third, only a small minority of these studies are randomized, double-­‐blind, placebo-­‐controlled trials (Rubia et al., 2011a and 2011b; Wong & Stevens 2012), limiting the inferences that can be drawn from the literature currently available. Fourth, studies tend to range widely in terms of inclusion criteria and the selection of cognitive tasks, further complicating direct comparison of their results and this may well contribute to the discrepant outcomes between studies. For instance, a subject's medication history (medication-­‐naïve vs. previously medicated, as well as medication duration and dosage in the past) may influence the effects of acute methylphenidate administration on the brain, but remains undisclosed in several of the fMRI studies (e.g., Peterson et al., 2009; Teicher et al., 2000; Vaidya et al., 1998). Finally, although all studies investigating acute effects of methylphenidate administration ensured a wash-­‐out period prior to study participation, acute rebound effects after sudden cessation of medication cannot be controlled for. 36 DISCUSSION
Methylphenidate effects are likely to be highly specific and rate-­‐dependent Based on the literature to date, methylphenidate administration appears to elicit functional and structural changes throughout the brain. The results of the studies reported in Table 1 and Table 2 suggest that the direction of methylphenidate effects may be related to the specific neural changes in ADHD. Different cognitive paradigms elicit different brain activation patterns, and therefore are associated with differing changes in brain activity with methylphenidate administration. Functional neuroimaging studies have suggested that methylphenidate induces acute normalization of activation patterns in widespread regions of the brain, including brain regions that have been associated with structural and functional changes in ADHD. However, the direction of these methylphenidate-­‐induced functional changes varied. For example, reduced ACC activity was found in children with ADHD during a cognitive control task, which increased with methylphenidate administration (Epstein et al., 2007), while another study found that methylphenidate enhanced task-­‐related deactivation in the ACC in children with ADHD (Peterson et al., 2009). A third study, by contrast, has found that typically developing children, but not children with ADHD, showed increased ACC activity during incongruent Stroop-­‐trials compared to during neutral trials, and that this lack of increase in ACC activity in ADHD was restored by methylphenidate administration (Zang et al., 2005). Thus, whereas all three studies report acute normalization of ACC activation after methylphenidate administration, the effects of methylphenidate on the ACC differed between studies. The consistency of the normalization of brain activity patterns by methylphenidate suggests that its seemingly discrepant acute effects on brain activation patterns across studies do not merely represent conflicting findings. Rather, the effects of methylphenidate may be highly specific, depending on the presence or absence of ADHD-­‐related functional changes in specific brain regions during the performance of specific cognitive tasks. Since unmedicated children with ADHD show regionally specific and task-­‐dependent deficits in brain activation, the modulating effects of methylphenidate treatment on these functional changes may also be specific. In sum, the studies to date suggest that methylphenidate modulates activity levels towards the levels of typically developing children, regardless of whether the ADHD-­‐related changes (i.e., without medication) constituted decreased or increased activity compared to typically developing controls. The rate-­‐dependency hypothesis suggests that the effects of methylphenidate treatment on brain activation may be dependent on the amount of ‘baseline’ activation (or activation while not on stimulant medication) in a particular brain region (Andersen, 2005). Thereby, methylphenidate may have a different effect on the 37 brain of subjects with ADHD than in typical controls. This issue is subject to intense debate, and is hard to address as ethical constraints disallow the administration of drugs to typically developing children. In the only study to address methylphenidate effects in both children with ADHD and typically developing children, Vaidya et al. (1998) used a cognitive control paradigm. Children with ADHD showed decreased striatal activation compared to typically developing children without methylphenidate, but showed an increase in striatal activation after methylphenidate administration. By contrast, typically developing children showed decreased activation of the striatum after methylphenidate administration (Vaidya et al., 1998). In typical adults, methylphenidate administration has been found to change brain activation patterns during a variety of cognitive functions, including working memory, attention and reversal learning (e.g., Dodds et al., 2008; Tomasi et al., 2011), but such changes have not been compared directly to methylphenidate-­‐induced brain changes in adult subjects with ADHD. Methylphenidate effects may be influenced by various biological, subject-­‐related and task-­‐related factors Adding to the complexity of the rate-­‐dependency hypothesis are numerous other factors, including biological, subject-­‐related demographic, and task-­‐related ones. All may affect baseline brain activation. By affecting brain activation patterns, these factors may also influence the effect of methylphenidate administration on these patterns. First, several biological factors may be at play. For example, neural activity levels while not on medication may be modulated by the density of neuronal dopamine receptors, which, in turn, seems to be genetically determined, at least partially (e.g., Bédard et al., 2010; Durston et al., 2008; Heinz et al., 2000). In typical adults, regional D2 receptor availability was found to be correlated to metabolic changes in the cerebellum, frontal and temporal regions induced by methylphenidate administration (Volkow et al., 1997). Thus, biological factors including genetic predispositions may determine, at least to some extent, how the brain responds to methylphenidate administration, via modulation of dopamine neurotransmission characteristics in an unmedicated state. In addition to biological factors, subject-­‐related demographic factors such as age and gender may contribute to regional brain activation patterns. Such factors could in turn influence the neural effects of methylphenidate treatment in a highly specific manner. Animal studies suggest that age at initiation of stimulant treatment is of particular importance (e.g., Canese et al., 2009). The brain undergoes drastic changes in structural and functional organization during development. Although the mechanism of stimulant action is likely to be similar in the immature and mature 38 brain, the substrate that stimulants act on, i.e., the catecholaminergic system, develops with age (Andersen, 2005). This may result in different patterns of brain response to methylphenidate administration at different ages. For example, methylphenidate was found to increase frontostriatal and cerebellar activation in children with ADHD, whereas it increased striatal activation but decreased prefrontal and right parietal activation in adults with ADHD (Epstein et al., 2007). The effects of methylphenidate administration may differ not only between children and adults, but also over childhood. Moreover, each brain region has a unique developmental trajectory, which leads to a complex picture of brain regions each being more or less sensitive to stimulant treatment at different time points in development (Rapoport & Gogtay, 2008). Finally, task-­‐related factors may influence the effect of methylphenidate on the brain through the brain activation patterns they evoke. For example, a working memory task elicits different patterns of brain activation than a cognitive control task, and will therefore show functional changes in ADHD that are specific to that cognitive task or domain. As a result, possible methylphenidate-­‐induced functional changes during a working memory task will differ from methylphenidate-­‐induced changes during a cognitive control task. The same principle may apply to task difficulty. It has been suggested that functional brain changes in children with ADHD are more pronounced during cognitively demanding tasks than during less demanding tasks (e.g., Kobel et al., 2009; Vaidya et al., 1998; Zang et al., 2005). Consequently, the rate-­‐
dependency hypothesis implies larger methylphenidate effects during more difficult cognitive challenges. Indeed, normalization of brain activation patterns with methylphenidate appears more pronounced in cognitively more demanding tasks than in less demanding tasks (Shafritz et al., 2004; Vaidya et al., 1998). Interestingly, a recent study suggests that the effects of methylphenidate across cognitive tasks of increasing difficulty are finite (Prehn-­‐Kristensen et al., 2011). In a relatively non-­‐
demanding working memory task, children with ADHD showed a lack of predominantly frontal activation compared to typically developing children, which was resolved by the administration of methylphenidate. When task difficulty was increased by adding distractors, typically developing children, but not children with ADHD, recruited the caudate nucleus. Whereas increasing task difficulty resulted in the additional recruitment of the caudate nucleus in typically developing children, this strategy was not brought about by methylphenidate administration in children with ADHD (Prehn-­‐Kristensen et al., 2011). Thus, task characteristics, including task difficulty, may affect the detection of methylphenidate effects on the brain. In sum, methylphenidate-­‐induced changes in brain activation patterns may be dependent on the level of ‘baseline’ activation in the brain, which in turn is determined by a multitude of factors. As such, biological factors, subject-­‐related 39 factors and task-­‐related factors, amongst many, may influence methylphenidate-­‐
induced changes in the brain of children with ADHD. Thus, the available evidence suggests that methylphenidate treatment does not necessarily affect brain activity in the same way for all children with ADHD, similar to the way methylphenidate treatment also affects behavior differentially between children with ADHD (e.g., MTA Cooperative Group, 1999). Behavioral correlates of methylphenidate-­‐induced neural changes This paper aimed to decrease the gap between the well-­‐described physiological effects of methylphenidate in the synapse, and the beneficial effects of methylphenidate on ADHD symptoms. A growing body of work shows that methylphenidate affects MRI-­‐based brain measures in children with ADHD, by normalizing both structural and functional changes observed in unmedicated subjects with ADHD. However, to investigate the mechanisms by which methylphenidate changes behavior it is important to relate these brain changes to behavioral outcome measures, such as ADHD symptom severity or task performance during functional MRI experiments. The important issue of whether methylphenidate-­‐induced brain changes in children with ADHD are accompanied by behavioral changes has rarely been addressed directly, and results have been inconsistent. Ivanov et al. (2010) found an association between behavioral ratings of hyperactivity and thalamus volume reduction in ADHD. Both ADHD symptoms and volume reductions were less severe in children who had been treated with psychostimulants. By contrast, Shaw et al. (2009) found that the excessive cortical thinning that occurred in children with ADHD that stopped taking psychostimulant medication, but not in children with ADHD that continued taking psychostimulant medication, was not associated with differences in clinical outcome between these two groups. Several methodological differences between studies, such as age of the studied sample, make these results hard to interpret. In their meta-­‐analysis, Nakao et al. (2011) were unable to assess whether methylphenidate-­‐induced structural changes were related to behavioral changes in children with ADHD, due to methodological discrepancies between studies. Diverse results also emerge from functional MRI studies. Whereas in some studies functional normalization was accompanied by symptom reduction or improved task performance (e.g., Anderson et al., 2002; Lee et al., 2010; Peterson et al., 2009; Prehn-­‐
Kristensen et al., 2011; Rubia et al., 2009; Teicher et al., 2000; Vaidya et al., 1998), others did not find such a relation (e.g., Epstein et al., 2007; Kobel et al., 2009; Rubia et al., 2011a; Shafritz et al., 2004). 40 Alternatively, one could regard behavioral measures while not on medication, such as symptom severity, as a subject-­‐related factor associated with ‘baseline’ brain activity. Subjects with ADHD who are more severely behaviorally affected could present with more functional changes in the brain (Depue et al., 2010), which could hypothetically result in more pronounced changes in brain activation patterns after methylphenidate administration. As such, behavioral or clinical measures may be a possible predictor of methylphenidate-­‐induced changes in brain activation patterns. Indeed, differential effects of methylphenidate administration on resting-­‐state brain activity have been found for behaviorally hyperactive vs. non-­‐hyperactive children with ADHD. More hyperactive children with ADHD showed larger effects of methylphenidate administration on resting-­‐state activity in the striatum and cerebellum than less hyperactive children with ADHD (Anderson et al., 2002; Teicher et al., 2000). Clearly, more evidence is needed in order to understand the association between the structural and functional brain changes after methylphenidate administration on the one hand, and the clinical effects of psychostimulant treatment on the other. Implications for neurobiological research into ADHD A large proportion of children with ADHD receive stimulant treatment at some point in their life. As a result, researchers investigating the neural correlates of ADHD face a major methodological challenge: whereas including children with a history of stimulant treatment introduces a potential biasing effect of methylphenidate treatment into the study sample, excluding children with a history of stimulant treatment would greatly reduce the number of potential study participants, and would reduce generalizability of study results to the population of all individuals affected with ADHD. In practice, neuroimaging studies include subjects with or without a history of stimulant treatment, or a mixture of these groups. The effects of methylphenidate on structural and functional MRI highlight the importance of monitoring medication status and medication history of study participants. Including medicated subjects with ADHD in structural MRI studies may result in reduced power to find structural brain changes that are relevant to the phenotype, given that treatment with methylphenidate appears to normalize brain development to some extent. Therefore, studies investigating neuroanatomical changes between subjects with ADHD and typically developing subjects may benefit from the inclusion of medication-­‐naïve children with ADHD only. Alternatively, addressing the possible confounding effects of treatment history in a sample including participants with variable medication histories (e.g., De Zeeuw et al., 2012), may further our understanding of long-­‐term methylphenidate effects. The acute 41 normalizing effects of methylphenidate on fMRI measures imply that functional deficits underlying ADHD symptoms may be less pronounced in an ADHD sample including medicated individuals, emphasizing the importance of medication wash-­‐out prior to functional MRI scanning. Until today, no conclusive evidence of lasting functional alterations after methylphenidate treatment has been reported in children with ADHD. The long-­‐term effects of stimulant treatment on functional MRI measures thus need further investigation. Several hypotheses have been proposed on mechanisms underlying the brain effects of long-­‐term methylphenidate treatment. It has been suggested that the effects of methylphenidate treatment on structural MRI measures could be an example of activity-­‐induced neuronal plasticity (Ivanov et al., 2010; Semrud-­‐Clikeman et al., 2006; Shaw et al., 2009). Stimulants are thought to increase catecholaminergic neurotransmission, which may influence cellular morphology (Robinson & Kolb, 2004). In addition, the increased catecholaminergic neurotransmission may lower the threshold for learning, or plasticity, allowing the brain to form new connections during cognitive tasks, resulting in volumetric changes. However, such hypotheses regarding possible biological mechanisms for long-­‐term structural changes remain speculative. More direct measures of catecholaminergic neurotransmission, such as SPECT and PET, may be more informative than MRI in this regard (e.g., Fusar-­‐Poli et al., 2012). Limitations and future perspectives This paper reviews the currently available studies on the effects of stimulant treatment on brain structure and functioning as assessed using MRI-­‐based techniques, in children with ADHD. Although a meta-­‐analytic review has benefits over a narrative review such as this one, the methods applied in the different studies vary strongly at this time. As such, the descriptive nature of this review reflects the current state of the literature. Any conclusions regarding the specificity of the effects of stimulant treatment on the developing brain therefore remain speculative, and are limited by the caveats inherent to the original research articles presented in Table 1 and Table 2. Each of the reported studies has their own methodological drawbacks, but some of these are common across the literature. We mention five main issues. First, the majority of structural MRI studies have (of necessity) adopted a naturalistic, cross-­‐sectional design. This does not allow for any conclusions on cause and effect or a more informative within-­‐subject evaluation of stimulant treatment effects. Second, a major concern that applies to all but two functional MRI studies is small sample size, with sometimes less than fifteen subjects per group or condition. Third, several functional MRI studies have not provided information on the type or duration of 42 stimulant treatment. Fourth, a less common limitation, which is more common to structural MRI studies, has been the inclusion of children (albeit a small percentage of participants) who were being treated with non-­‐stimulant medication, such as atomoxetine (e.g., Ivanov et al., 2010; Shaw et al., 2009). Fifth, both structural and, to a lesser extent, functional MRI studies have often investigated specific brain regions of interest, and as such could have failed to detect effects of stimulant treatment on other regions. Finally, signs of considerable publication bias have been reported in clinical treatment literature, where null-­‐findings or conflicting findings are less likely to be published (Dwan et al., 2008; McGauran et al., 2010). We cannot exclude the possibility that a similar issue may be at play in the literature on brain changes induced by methylphenidate and as such may limit our understanding of its effects. A number of issues are of great importance for future work. First, an unresolved issue is the link between structural and functional brain changes induced by methylphenidate treatment on the one hand, and clinical response to methylphenidate treatment on the other. Not all children with ADHD respond to stimulant treatment in the same way, both at the neural and behavioral level. One promising approach would be to compare responders to non-­‐responders. This could provide researchers with valuable clues as to which regions or processes in the brain are related to the beneficial effects to stimulant treatment. Secondly, the long-­‐term effects of stimulant treatment on the brain functioning are largely unknown. The long-­‐term normalizing effects of stimulant treatment on brain structure in children with ADHD, appear paradoxical to the rapid return of ADHD symptoms after stimulant treatment is ceased, and do not correspond with the putative evidence of the absence of long-­‐term functional brain changes with stimulant treatment. In future studies, a developmental perspective will be critical, since brain development may well add to the existing variability in methylphenidate-­‐
induced brain changes in children with ADHD. Moreover, long-­‐term methylphenidate effects may become apparent only after adolescence or young adulthood, when neurodevelopment enters a more stable phase (Andersen, 2005). Long-­‐term effects of methylphenidate treatment on the developing brain should ideally be investigated in a large prospective cohort of children with ADHD as well as typically developing control subjects, with both structural and functional neuroimaging sessions before stimulant treatment is initiated, and follow-­‐up continuing into adulthood. MR imaging of methylphenidate effects could benefit from adding genetic markers to relate to individual differences in treatment response. Neuroimaging of the effects of methylphenidate could serve as an endophenotype between genetic predisposition and clinical amelioration with methylphenidate treatment, by comparing methylphenidate-­‐induced alterations in brain functioning and behavior between groups that differed in genotype for genetic polymorphisms relevant to 43 ADHD. Such an approach may relate the molecular basis of the clinical effects of the treatment, to neurobiological substrate of ADHD. In addition, such studies may provide geneticists with valuable biological markers in pursuit of identifying genes involved in brain structure and function in ADHD (Wood & Neale, 2010). Several promising papers have appeared combining pharmacogenetics with neuroimaging methods other than MRI, such as SPECT and EEG (e.g., Loo et al., 2003; Rohde et al., 2003). In sum, a modest literature suggests that the effects of methylphenidate treatment on brain development in ADHD are highly specific and dependent on numerous factors, including biological factors such as genetic predispositions, subject-­‐related factors such as age and symptom severity, and task-­‐related factors such as task difficulty. Future studies on structural and functional brain changes in ADHD would benefit from inclusion strategies guided by current medication status and medication history and will need to address unresolved issues including the effects of the long-­‐term treatment, and the mechanism underlying its therapeutic effects. 44 45 46 Chapter 3 THINNER MEDIAL TEMPORAL CORTEX IN
CHILDREN WITH ADHD AND THE EFFECTS OF
STIMULANTS
Published as: Schweren LJS, Hartman CA, Heslenfeld DJ, van der Meer D, Franke B, Oosterlaan J, Buitelaar JK, Faraone SV, Hoekstra PJ. Thinner medial temporal cortex in adolescents with attention-­‐deficit/hyperactivity disorder and the effects of stimulants. J Am Acad Child Adolesc Psychiatry. 2015;54(8):660-­‐667. 47 ABSTRACT
Objective: Attention-­‐deficit/hyperactivity disorder (ADHD) has been associated with widespread changes in cortical thickness (CT). Findings have been inconsistent, however, possibly due to age differences between samples. Cortical changes have also been suggested to be reduced or disappear with stimulant treatment. We investigated differences in CT between adolescents/young adults with and without ADHD, in the largest ADHD sample to date, the NeuroIMAGE sample. Second, we investigated how such differences were related to age and stimulant treatment. Methods: Participants (ADHD=306; healthy controls=184, 61% male, 8-­‐28 years old, mean age=17) underwent structural magnetic resonance imaging. Participants and pharmacies provided detailed information regarding lifetime stimulant treatment, including cumulative intake and age of treatment initiation and cessation. Vertex-­‐wise statistics were performed in Freesurfer, modeling the main effect of diagnosis on CT and its interaction with age. Effects of stimulant treatment parameters on CT were modeled within the ADHD sample. Results: After correction for multiple comparisons, participants with ADHD showed decreased medial temporal CT in both left (pCLUSTER=0.008) and right (pCLUSTER=0.038) hemispheres. These differences were present across different ages, and were associated with symptoms of hyperactivity and prosocial behavior. There were no age-­‐by-­‐diagnosis interaction effects. None of the treatment parameters predicted CT within ADHD. Conclusions: Individuals with ADHD showed thinner bilateral medial temporal cortex throughout adolescence and young adulthood compared to healthy controls. We found no association between CT and stimulant treatment. The cross-­‐sectional design of the current study warrants cautious interpretation of the findings. 48 INTRODUCTION Magnetic resonance imaging (MRI) has revealed structural and functional brain changes associated with attention-­‐deficit/hyperactivity disorder (ADHD) (Bush et al., 2005; Frodl & Skokauskas, 2012; Nakao et al., 2011). Surface-­‐based reconstruction of the cortical sheet allows quantification of different features of cortical structure, including volume, thickness, surface area, and curvature. Such features may represent distinct developmental processes having separate developmental trajectories (Raznahan et al., 2011). Changes in different features may be associated with distinct forms of psychopathology (Shaw et al., 2012). Volumetric studies have consistently reported global cortical volume reduction in individuals with ADHD (Greven et al., 2015; Nakao et al., 2011). Widespread reductions of cortical thickness (CT) have also been implicated in ADHD. Children and adults with ADHD have shown decreased CT in frontal cortex (Almeida et al., 2010; Hoekzema et al., 2012; Proal et al., 2011; Shaw et al., 2013; Yang et al., 2015), inferior and superior parietal cortex (Hoekzema et al., 2012; Narr et al., 2009; Proal et al., 2011), temporal pole and medial temporal cortex (Langevin et al., 2014; Proal et al., 2011). However, patterns of ADHD-­‐related cortical changes differ widely across studies. There have been multiple reports of increased rather than decreased CT in individuals with ADHD (Almeida Montes et al., 2012; Duerden et al., 2012), and other studies have found no association between CT and clinical features of ADHD (Narr et al., 2009; Yang et al., 2015). Discrepant patterns of CT changes in ADHD between studies may result from age differences in groups under study. ADHD often persists into adulthood (Copeland et al., 2013), typically showing reduced hyperactivity but persistent inattention throughout adolescence. In typical development, CT increases during childhood to reach its peak in early adolescence, after which it decreases again. The “maturational delay” hypothesis of ADHD proposes that CT changes observed in children with ADHD reflect the ADHD group lagging behind the typically developing group and reaching peak CT at a later age (Shaw et al., 2007a). As they grow older, adolescents with ADHD are proposed to “catch up” with their unaffected peers, resulting in fewer or no cortical changes along with a decline in clinical symptoms at later age (remission). The hypothesis is supported by an impressive longitudinal sample of children and adolescents, with an average age of twelve (Shaw et al., 2007a). A substantial proportion of children with ADHD, however, continues to have symptoms in late adolescence and adulthood (Faraone et al., 2006). Differences in CT in adults with ADHD have also been reported (Almeida Montes et al., 2012; Duerden et al., 2012), suggesting that cases of persistent ADHD do not show cortical normalization during late adolescence. Unfortunately, the majority of studies focused on either children or 49 adults, and the development of CT in (late) adolescent ADHD has not extensively been documented. One cross-­‐sectional study found both increases and decreases in CT in older adolescents/young adults with ADHD (Almeida Montes et al., 2012). Zooming in on the late adolescent phase could aid in further elaboration of cortical development in ADHD. A substantial proportion of individuals with ADHD are prescribed stimulants. MRI studies investigating the effect of methylphenidate treatment on brain volume and function in children with ADHD have suggested at least partially normalizing effects (Frodl & Skokauskas ,2012; Nakao et al., 2011; Rubia et al., 2014; Spencer et al., 2013). Very few have studied the effect of stimulants on CT. In a longitudinal study, Shaw et al. (2009) showed normalized developmental trajectories of CT in stimulant-­‐
treated, but not in non-­‐treated children with ADHD. Treatment effects were local rather than global, affecting CT in the left dorsolateral prefrontal cortex, and right motor and posterior parietal cortex. By contrast, other studies have reported greater CT abnormalities in previously medicated patients (Narr et al., 2009), or observed no differences between stimulant-­‐naïve and stimulant-­‐treated patients (Hoekzema et al., 2012). The investigation of long-­‐term treatment effects in pediatric groups is complex. Long-­‐term effects (spanning multiple years) may only be assessed in observational studies, in which ADHD cases have not been randomized over stimulant and non-­‐stimulant treatment. This creates the possibility of confound by indication, i.e. non-­‐stimulant treated cases may be less severe, or may differ from stimulant-­‐
treated cases in other ways. An advantage of observational studies, however, is that study samples are typically representative of the study population. To investigate stimulant treatment effects on brain structure, “treated” and “untreated” individuals with ADHD are typically compared. However, this distinction is rather crude, and neglects between-­‐subject variation in treatment history. Whereas some classify past users as “treated” (e.g., Shaw et al., 2009), others may classify them as “untreated” (e.g., Amico et al., 2011) or exclude such participants (e.g., Onnink et al., 2014). Investigating treatment heterogeneity in more detail may reveal mechanisms by which stimulant treatment may affect brain structure. In the current study we compared CT in a large sample of adolescents/young adults with ADHD (n=306) to that of a healthy control sample (n=184). Further, the linear and nonlinear effects of age on changes in CT associated with ADHD (if any) were investigated. Last, we tested the effect of multiple well-­‐defined stimulant treatment parameters. The current study adds to the previous volumetric findings of our group of ADHD being associated with global rather than local volume reductions (Greven et al., 2015). Other neuroimaging studies based on the same sample investigated volumetric features (O’Dwyer et al., 2014; Schweren et al., 2015b, van 50 der Meer et al., 2015), structural connectivity (Francx et al., 2015; van Ewijk et al., 2014b and 2015), or functional MRI (Pruim et al., 2015a and 2015b; van Rooij et al., 2015a and 2015b). To the best of our knowledge CT has not previously been studied in an ADHD sample of this size. METHODS Participants Participants were selected from the Dutch follow-­‐up phase of the International Multicenter ADHD Genetics study (IMAGE; Mueller et al., 2011; Rommelse et al., 2008). ADHD diagnosis, ADHD severity, and presence of comorbid disorders were established using an algorithm based on both the Schedule for Affective Disorders and Schizophrenia for School-­‐Age Children (K-­‐SADS; Kaufman et al., 1997) and Conners’ ADHD questionnaires for parents, teachers, and adult participants (Conners et al., 1998a, 1998b, and 1999). See Von Rhein et al. (2015) and Supplement S1, for more details and relevant publications regarding the sample and diagnostic algorithm. IQ was estimated from the subtests “vocabulary” and “block design” of the Wechsler Intelligence Scale for Children – Version III (participants ≤ 16 years old) or the Wechsler Adult Intelligence Scale – Version III (participants > 16 years old; Wechsler, 2000 and 2002). The subtest “digit span” was administered as an indication of working memory capacity. In addition, the strengths and difficulties questionnaire for children was administered (SDQ; Van Widenfelt et al., 2003). Socio-­‐
economic status (SES) was calculated as the average (of both parents) number of years of education. Participants withheld use of psychoactive drugs for 48 hours prior to their visit. Informed consent was signed by all participants and parents (parents signed informed consent for participants < 12 years old). Testing took place at the University Medical Center of either Amsterdam or Nijmegen. The study was approved by the local ethical committee. The final sample consisted of 306 participants with ADHD and 184 healthy control subjects, between the ages of 8.3 and 27.8 years old (M=17.05, SD=3.33). Assessment of medication history Lifetime medication transcripts from pharmacies were available for 74%, and covered lifespan for 25% of participants with ADHD. In addition, a questionnaire was administered to all participants and parents assessing lifetime history of psychoactive medication. When pharmacy transcripts did not fully cover the self-­‐reported treatment period, medication parameters of the missing period(s) were calculated 51 from the questionnaire data, and were added to the measures derived from the pharmacy. Retrospective assessment of ADHD medication has shown good to excellent concordance between parent-­‐ and physician-­‐report even after multiple years (Kuriyan et al., 2014). The following indices of stimulant treatment (methylphenidate immediate/extended release, and d-­‐amphetamine preparations) were calculated: history of treatment (stimulant-­‐exposed vs. stimulant-­‐naïve); start age; stop age; median age of exposure (age in years at the median of all exposed days); treatment duration corrected for age (treatment duration divided by [age minus the minimum start-­‐age within the sample, i.e. age 2.3]); mean daily dose (average dose in mg for all exposed days; d-­‐amphetamine dose was multiplied by two); cumulative intake corrected for age (corrected treatment duration multiplied by mean daily dose); and time since last treatment (age minus stop age). For stimulant-­‐
naïve patients, mean daily dose, treatment duration and cumulative intake were zero; start age was imputed as the participant’s age at scan (mimicking late initiation), and stop age was imputed as age 2.3 (mimicking early cessation). MRI acquisition and analysis MRI data was acquired at 1.5T on a Siemens Sonata scanner at the University Medical Center in Amsterdam, and on a Siemens Avanto scanner in Nijmegen, with an identical 8-­‐channel phased array coil and identical acquisition parameters. There were no major hardware upgrades on either of the scanners during the study. Comparability of MRI data from the two sites has extensively been described elsewhere (Von Rhein et al., 2015). Scanning parameters and quality assurance procedures are described in Supplement S1. Cortical reconstruction was performed with Freesurfer (Dale et al., 1999; Fischl et al., 1999 and 2004). Freesurfer is an automated technique to create a 3D reconstruction of the cortical sheet that uses both intensity and continuity information, with good test-­‐retest reliability across scanner stations (Han et al., 2006). CT was calculated for each vertex on the reconstructed cortical sheet, and defined as the closest distance from the gray/white boundary to the gray/CSF boundary (Fischl & Dale, 2000). Cortical surface area, used in post-­‐hoc analyses, was measured at the geometric middle between the inner and outer cortical surfaces. A 10mm FWHM surface-­‐based smoothing kernel was applied. Average CT per subject was calculated across all vertices. Total brain volume was calculated as the sum of Freesurfer estimated total grey and white matter volume. Data analysis 52 Statistical modeling was performed with the glmfit-­‐tool embedded in Freesurfer, and in second instance in SPSS version 20.0.0.0 (IBM, 2013). The effects of diagnostic group (healthy controls vs. participants with ADHD) and stimulant exposure (stimulant-­‐exposed vs. stimulant-­‐naïve) on CT were analyzed in a linear main effects model including gender, scanner location, and SES as covariates, and age and age2 as optional per-­‐vertex covariates. Optimal modeling of age as a covariate across the cortex was obtained in a two-­‐step approach: First, between-­‐group differences were evaluated with both age and age2 in the model in all vertices where age2 significantly contributed to the prediction of CT. Second, in all other vertices, age was kept in the model only where it significantly contributed to the prediction of CT. As a result, each vertex contained either a quadratic, a linear, or no effect of age (Figure S2). IQ was not added as a covariate in the primary analyses, since we consider lower IQ to be part of the ADHD phenotype (Dennis et al., 2009). In two additional vertex-­‐wise models, we tested age-­‐by-­‐diagnosis and age2-­‐by-­‐diagnosis interactions. Comparing stimulant-­‐exposed to stimulant-­‐naïve participants allowed detection of between-­‐group differences of medium effect size (NEXPOSED=270, NNAÏVE=36; two-­‐tailed alpha=0.05, power=0.80, smallest detectable Cohen’s d effect size=0.50). We further investigated treatment effects by vertex-­‐wise linear modeling of continuous treatment variables within the ADHD sample, i.e. treatment duration corrected for age, mean daily dose, cumulative intake corrected for age, start age, stop age, median age of exposure, and time since last treatment. These parameters were initially tested in seven separate models, predicting CT with gender, scanner location, SES, age and age2 as covariates (Bonferroni correction, cluster-­‐wise alpha/7), and then simultaneously for those treatment parameters significantly predicting CT. Unlike in the case-­‐control analyses, linear and quadratic age-­‐terms were included for each vertex, as they were expected to be correlated with the predictor variables. With this approach regression coefficients of small to medium effect size could be detected (NTOTAL=290; per vertex: two-­‐tailed alpha=0.007, power=0.80, smallest detectable Cohen’s ƒ2 effect size =0.067). We applied Monte Carlo simulation testing (10.000 iterations, vertex-­‐wise threshold p<0.01, cluster-­‐wise threshold p<0.05) to correct for multiple comparisons. Within each significant cluster, mean CT and surface area were extracted for each participant in standard space, to perform post-­‐hoc and sensitivity analyses in SPSS. We reported cluster size and p-­‐value from the Monte Carlo simulation testing in Freesurfer, and estimated marginal mean CT per group and Cohen’s d effect size from the SPSS analyses. Exploratory post-­‐hoc analyses were performed to investigate clinical correlates of case-­‐control differences or treatment effects within participants with 53 ADHD (n=306). In separate linear mixed effects models, mean CT within each cluster was predicted by number of hyperactivity symptoms, number of inattention symptoms (both derived from the K-­‐SADS interview and Conner’s questionnaires), four subscales of the SDQ (conduct problems, emotional problems, peer problems, and prosocial behavior), working memory capacity (maximum digit span backwards), and IQ. Gender, scanner site, SES, and if appropriate age and age2, were used as covariates. Second, we tested whether cortical surface area was affected in clusters of significant between-­‐group or treatment effects. Last, for each significant cluster we tested age-­‐by-­‐diagnosis, age2-­‐by-­‐diagnosis, and age-­‐quintiles-­‐by-­‐diagnosis interactions effects. Sensitivity analyses were performed to investigate the robustness of our findings. First, a random intercept per family was added to the model to account for dependencies among participants from the same family. Further sensitivity analyses entailed repeating each analysis with IQ, average CT, and total brain volume as additional covariates, respectively, and repeating each analysis within subgroups, i.e. within each of the two scanning sites, within boys and girls, within five age quintiles (age<14.05; age=14.06-­‐16.21; age=16.22-­‐18.01; age=18.02-­‐20.04; and age>20.04), within participants who had never received psychoactive treatment other than stimulants, and within participants without any comorbid diagnoses (Table S3). Furthermore, vertex-­‐wise analyses in Freesurfer were repeated with IQ, average CT, and total brain volume as covariates to allow detection of additional clusters (Table S4 and Figure S4). TABLE 1. Demographic and clinical information of partici-­‐
pants with and without ADHD. HC ADHD n % n % Participants 184 37.6 306 62.4 Male 92 50.0 209 68.3 0.001 Amsterdam 116 63.0 135 44.1 0.001 M SD M SD p Age 16.77 3.15 17.23 3.43 0.138 IQ 106.16 13.75 97.05 15.24 0.001 SES 13.33 2.50 11.61 2.23 0.001 HC, healthy controls; SES, socio-­‐economic status. 54 p RESULTS Demographic and clinical information Compared to healthy controls, participants with ADHD were more likely to be male, to have participated in Nijmegen, and had lower SES and IQ (Table 1). Forty-­‐
four percent of participants with ADHD were of combined type (n=134). Thirty-­‐three percent of participants with ADHD had a comorbid disorder (n=100), mostly oppositional defiant disorder and/or conduct disorder (n=91, 29.7%) but also tic disorders (n=3, 1.0%) and anxiety/depression (n=11, 3.6%). Eighty-­‐eight percent (n=254) of participants with ADHD had received stimulant treatment at some point in their lives, including immediate-­‐release (n=245, 84.5%) and/or extended-­‐release (n=201, 69.3%) methylphenidate preparations and/or dexamphetamine (n=25, 8.6%). Compared to stimulant-­‐naïve participants, stimulant-­‐exposed participants were more likely to be male, to have participated in Nijmegen, were younger, and had lower IQ and more hyperactivity-­‐impulsivity symptoms (Table 2). Medication parameters could be calculated for the majority of participants with ADHD (N=290, 94.5%; including 254 stimulant-­‐exposed participants, 87.6%). On average stimulant-­‐exposed participants had received 4.9 years of stimulant treatment (SD=3.19; range 0.05-­‐14.17) corresponding to 33% of their lives. They started stimulant treatment, on average, at age 8.5 (SD=2.75; range 2.30-­‐20.61), and received a mean dose of 34 mg per day (SD=12.47; range 10.00-­‐78.52). Forty-­‐nine percent (n=125) of stimulant-­‐exposed participants had ceased treatment at least three months and on average 1.6 years prior to study participation, with an average stop age of 15.5 years (SD=3.27; range 4.86-­‐23.38). Twenty-­‐eight percent (n=81) of all participants had received psychoactive medication other than stimulants, including atomoxetine (n=39, 13.5%), clonidine (n=18, 6.2%), antidepressants (n=16, 5.5%), atypical antipsychotics (n=48, 16.6%), and benzodiazepines/anxiolytics (n=15, 5.2%). CT in participants with ADHD vs healthy controls Participants with ADHD showed decreased CT in the medial temporal cortex in both left (cluster size=468mm2; pCLUSTER=0.008; Cohen’s d effect size=0.443; CTHC=3.323mm; CTADHD=3.182 mm) and right hemisphere (cluster size=368mm2; pCLUSTER=0.038; Cohen’s d effect size=0.445; CTHC=3.224mm; CTADHD=3.113mm; Figure 1). These case-­‐control differences were significant after accounting for dependencies among participants from the same family, were present in both testing 55 TABLE 2. Characteristics of the exposed and unexposed participants with ADHD. Exposed Unexposed n % n % Participants p 270 88.2 36 11.8 Male 192 71.1 17 47.2 0.004 Amsterdam 104 38.5 31 86.1 0.001 Combined type 122 45.2 12 33.3 0.178 Comorbid disorder 89 33.0 11 30.6 0.772 ODD/CD 82 30.4 9 25.0 0.508 Tic disorder 3 1.1 0 0.0 0.525 Anxiety / Depression 9 3.3 2 5.6 0.501 M SD M SD p Age 17.04 3.23 18.61 4.45 0.048 IQ 96.42 14.84 101.75 17.43 0.049 Number of symptoms 13.36 2.93 11.55 3.13 0.001 Inattentive 7.34 1.71 6.75 1.59 0.053 Hyperactive-­‐impulsive 6.03 2.30 4.89 2.80 0.024 SES 11.60 2.25 11.69 2.12 0.813 ODD/CD, oppositional defiant disorder/conduct disorder; SES, socio-­‐
economic status. sites and both genders, remained significant when participants with comorbid diagnoses or psychoactive medication other than stimulants were excluded, and when IQ, total brain volume, and average CT (respectively) were added to the model as additional covariates (Table S3). In vertex-­‐wise analyses with IQ and average CT as an additional covariate, a left superior parietal cluster of increased CT in participants with ADHD reached significance as well (Table S4 and Figure S4). In the primary analyses, the same pattern was observed but failed to reach significance after correction for multiple testing (data not shown). 56 Age2 did not contribute to the prediction of CT in either of the medial temporal clusters, and the linear age term contributed in the right but not the left hemisphere cluster (Figure S2). CT of the ADHD and healthy control groups within each cluster was plotted in five age quintiles (Figure 2). The direction of effect remained unchanged in all age groups, and there were no age-­‐by-­‐diagnosis (pLEFT=0.137, pRIGHT=0.328) or age-­‐quintile-­‐by-­‐diagnosis (pLEFT=0.085, pRIGHT=0.135) interaction effects. In accordance, we found no age/age2-­‐by-­‐diagnosis interaction effects in vertex-­‐wise analyses. There was no between-­‐group difference in cortical surface area within the left (p=0.241) or right (p=0.166) cluster. Main effects of gender, site, and SES are in Table S2. FIGURE 1. Regions of significant decrease in cortical thickness (cluster-­‐
wise p-­‐value < 0.05, corrected for multiple comparisons using Monte Carlo simulation testing), in participants with ADHD compared to healthy controls, indicated in red and projected on the pial surface of a standard brain template (fsaverage). There were no regions of increased cortical thickness in participants with ADHD. Stimulant exposure There were no differences in CT between stimulant-­‐treated and stimulant-­‐
naïve participants with ADHD. Treatment duration corrected for age, mean daily dose, cumulative intake corrected for age, start age, stop age, median age of exposure, and time since last treatment did not predict CT within the ADHD sample. 57 FIGURE 2. Cortical thickness (CT) in participants with attention-­‐deficit/hyperactivity disorder (ADHD) and healthy controls (HC), stratified by age quintiles (Q1<14.05y; Q2=14.06-­‐16.21y; Q3=16.22-­‐18.01y; Q4=18.02-­‐20.04y; Q5>20.04y) within the medial temporal clusters of case-­‐control difference. Age-­‐quintile-­‐
by-­‐diagnosis interaction effects are not significant. Error bars represent standard deviations. Post-­‐hoc analyses: clinical correlates Exploratory post-­‐hoc analyses indicated that in participants with ADHD, CT within the left medial temporal cluster was related to number of hyperactivity symptoms (β=-­‐0.039; p=0.020), but not to number of inattention symptoms (p=0.571), conduct problems (p=0.183), emotional problems (p=0.200), peer problems (p=0.562), prosocial behavior (p=0.647), working memory capacity (p=0.651), or IQ (p=0.730). Within the right medial temporal cluster, CT was related to prosocial behavior (β=0.031; p=0.034) but not to symptoms of inattention (p=0.985), hyperactivity (p=0.246), conduct problems (p=0.979), emotional problems (p=0.971), peer problems (p=0.768), working memory capacity (p=0.789) or IQ (p=0.817). DISCUSSION The current study investigated CT among adolescents and young adults with ADHD, and its associations with age and stimulant treatment. We found bilateral decreased medial temporal CT in participants with ADHD compared to healthy control participants. These differences were present across different ages, were not accompanied by changes in cortical surface area, were not driven by global brain changes, and were associated with symptoms of hyperactivity and prosocial behavior. Despite having the largest ADHD sample to date with substantial within-­‐subject treatment variability, we found no association between CT and stimulant treatment history. Reduced CT in medial temporal regions, including the hippocampus, amygdala, and parahippocampal cortex has previously been reported in pediatric 58 (Fernández-­‐Jaén et al., 2014; Narr et al., 2009; Shaw et al., 2006) and adult (Proal et al., 2011) ADHD groups. Smaller medial temporal volumes have been associated with impaired response inhibition in individuals with ADHD (McAlonan et al., 2009), and structural changes of the hippocampus and amygdala have been associated with emotional dysregulation (Frodl et al., 2010; Posner et al., 2014). In a volumetric study of the current sample, a decrease in overall grey matter volume but no changes in hippocampal or amygdalar volumes were detected in participants with ADHD (Greven et al., 2015). Discrepant findings may be expected, however, since cortical volume is determined by cortical thickness as well as other parameters (i.e., surface area and gyrification). In addition, analyses of regional cortical volumes (including the hippocampus) in the volumetric study were corrected for global brain changes. Smaller hippocampal volumes may have been masked by the reduction in total brain volume in participants with ADHD (Greven et al., 2015). In the current study, adding global brain measures did not change our findings, suggesting that decreased medial temporal CT may not be related to global changes. Our findings add to the growing body of evidence suggesting that regions outside the frontal-­‐striatal circuits may be important in the pathophysiology of ADHD (Kobel et al., 2010). Our exploratory and preliminary post-­‐hoc analyses suggest a link between left medial temporal CT and hyperactivity symptoms, a core feature of ADHD. The clinical relevance of decreased medial temporal CT is to be further elaborated in future studies, in which hyperactivity and prosocial behavior but also typical medial temporal functions such as memory should be addressed. The current study being cross-­‐sectional, any findings regarding developmental changes or age effects should be interpreted with caution. Nevertheless, our study provides several interesting findings regarding age and ADHD. First, both clusters of case-­‐control difference occurred (at least partially) in regions where CT was not related to age (Figure 2). In most other vertices, CT decreases with increasing age (Figure S2). Case-­‐control differences thus occur in the absence of developmental changes in CT. Second, we found no age-­‐by-­‐diagnosis or age2-­‐by-­‐diagnosis interaction effects. Thus, the medial temporal case-­‐control differences are equally driven by younger and older participants. The developmental delay hypothesis proposes that, later in development, some children with ADHD “catch up” with their typically developing peers (Shaw et al., 2007a), resulting in smaller cortical abnormalities accompanied by (at least partial) clinical remission. This hypothesis could not be tested in the current study, since no cases of remittent ADHD were included. We emphasize again the cross-­‐sectional nature of the current study. As a group, the older participants with ADHD may differ from the younger ones. A sizeable portion of participants within the younger ADHD groups may remit during adolescence, whereas this has not occurred in the older ADHD groups. This more 59 heterogeneous composition of the younger age groups may have masked any age-­‐by-­‐
diagnosis interaction effects. There is a clear need for long-­‐term longitudinal studies to characterize cortical development associated with persistence and remission of ADHD during late adolescence/young adulthood. Despite having sufficient power to detect even small effects, we found no associations between stimulant treatment and CT. Any treatment parameter, regardless of its correlations with the other parameters, would have shown its individual effect (if any) in our initial approach of modeling each parameter separately. The absence of stimulant treatment effects has two implications for our findings. First, it aids the interpretation of the case-­‐control differences. As the ADHD sample consisted largely of stimulant-­‐exposed participants with an average treatment duration of almost five years, any case-­‐control differences we observed may have been the result of stimulant treatment rather than associated with the ADHD phenotype. Two recent studies both reported hippocampal volume reduction in adults with ADHD who had during childhood been treated with stimulants, but not in stimulant-­‐naïve adults with ADHD (Frodl et al., 2010; Onnink et al., 2014). The lack of association between stimulant treatment and CT within our ADHD group, however, renders this explanation less plausible. Second, our findings do not support with the hypothesis of CT normalization with stimulant treatment. Most previous studies suggesting structural normalization with stimulant treatment reported cortical volume rather than thickness, of which two recent studies found evidence in meta-­‐regression analyses (Frodl & Skokauskas, 2012; Nakao et al., 2011). In one study, development of CT over time was found to be normalized in participants with ADHD who received stimulant-­‐treatment (n=24) compared to those who did not (n=19). These effects were confined to specific brain regions, including the left dorsolateral prefrontal cortex (Shaw et al., 2009). In a larger study of the same group, however, no stimulant treatment effects were found (Shaw et al., 2013). We found no evidence of stimulant treatment being associated with CT. Possibly, long-­‐term stimulant treatment affects cortical volume but not thickness. Long-­‐term treatment effects across different cortical features are an interesting opportunity for future studies. The current study had several strengths. First, our sample comprised older adolescents and young adults, an age group that has received very little attention in previous studies. Second, as pediatric long-­‐term treatment effects cannot be studied in randomized clinical trials, the current study took advantage of its observational nature. This resulted in a large and representative sample, allowing a detailed investigation of between-­‐subject variation in treatment history. Third, access to pharmacy records allowed exact quantification of lifetime stimulant exposure. This extent of detail has rarely been accomplished in previous studies. Our study had 60 limitations too. The study was cross-­‐sectional. An optimal design to investigate long-­‐
term outcomes would be longitudinal and include a pre-­‐treatment measurement. In accordance, an optimal study design would include individuals with remitted ADHD as well. Second, few participants with ADHD were naïve to stimulants, and the average treatment duration of the ADHD sample was relatively long. Future studies of treatment effects would benefit from targeted inclusion of additional stimulant-­‐naïve individuals. Third, the large sample size did not allow manual editing of the Freesurfer segmentations, which may have affected reconstruction of the cortical surface especially in the anterior temporal lobes. However, we expect such distortions, if any, to be small and randomly distributed across the participant groups. In conclusion, we found reduced CT in bilateral medial temporal cortex in youths with ADHD compared to healthy controls. There were no age-­‐by-­‐diagnosis interaction effects. These findings suggest ADHD-­‐related changes in CT existing throughout adolescence and young adulthood, and add to our prior report of overall grey matter volume reduction. In the largest ADHD sample to date, we found no evidence that CT was affected by stimulant treatment. Our cross-­‐sectional findings suggest the importance of medial temporal regions in adolescent ADHD, and highlight the need for longitudinal studies of ADHD extending into late adolescence and young adulthood. 61 SUPPLEMENTAL INFORMATION
S1 – The IMAGE-­‐NeuroIMAGE sample and MR parameters Three-­‐hundred-­‐thirty-­‐one ADHD families and 153 control families participated in a diagnostic interview, questionnaires, and extensive MRI scanning. The following inclusion criteria applied for all participants in the current study: participants had to be (1) between 8-­‐30 years old at follow-­‐up, (2) of European Caucasian descent, (3) have an IQ ≥ 70, (4) have no diagnosis of epilepsy, general learning difficulties, brain disorders and known genetic disorders (such as Down syndrome), (5) have no contraindication to MR scanning, and (6) show no incidental findings on the MRI scan. Healthy control participants had to fulfill the following additional criteria: no current or past mental health care utilization, no sibling(s) with any past or current psychiatric diagnosis, and no current or past psychoactive medication use. As recruitment was family-­‐based, multiple members of one family could be included in the same diagnostic group. Unaffected siblings of participants with ADHD were excluded. Previous relevant publications from our group regarding the same sample that are not in the reference list included a study focusing on working memory (van Ewijk et al., 2014a) and another on the risk of developing substance use disorder in relation to stimulant treatment (Groenman et al., 2013). Structural MRI acquisition consisted of two T1-­‐weighted 3D MP-­‐RAGE scans (TI = 1000 ms, TR = 2730 ms, TE = 2.95 ms, FA = 7°; Parallel imaging by generalized autocalibrating partially parallel acquisition (GRAPPA); 176 sagittal slices, voxel size 1 x 1 x 1 mm, FOV = 256 x 256 x 176 mm). For each participant, the structural acquisition of highest quality was selected by visual inspection (Blumenthal et al., 2002), accepting only scans with no/mild distortions. To assure Freesurfer reconstruction quality, the following reconstructions were subjected to visual inspection to detect regions of “flattened” or “spiky” surface and surface-­‐holes: (1) twenty percent (randomly selected) of the sample; (2) all reconstructions based on a structural scan with mild distortions. Reconstructions that did not meet quality criteria were excluded from all analyses; no manual edits were made. S2 -­‐ Covariates FIGURE S2.
Clusters of significant main effects of the linear and quadratic age terms (light-­‐ and dark-­‐blue, respectively; corrected for multiple comparisons using Monte Carlo simulation testing). Increasing age was associated with decreasing cortical thickness. There were no regions of increasing cortical thickness with increasing age. The two medial temporal clusters of case-­‐control difference are delineated in red. 62 TABLE S2. Gender, scanner and socio-­‐economic status. Clusters of significant main effects of covariates gender, scanner site and socio-­‐economic status (pvertex=0.01, pcluster=0.05, corrected for multiple testing), tested in the full model (cortical thickness is predicted by diagnostic status, scanner site, gender, socio-­‐economic status, age and age2). Covariate Direction Hemi Gender Boys > Girls R Size T M A X P C L U S T E R Lingual cortex 522.67 -­‐4.578 0.00480 L Precentral cortex 374.89 -­‐6.175 0.03100 R Insula 439.79 -­‐5.401 0.01300 R Superior temporal cortex 475.71 -­‐4.982 0.00820 L Middle frontal cortex 353.99 -­‐4.629 0.04050 Girls > Boys R Posterior cingulate cortex 499.60 6.114 0.00610 L Postcentral cortex 577.36 4.980 0.00230 R Precentral cortex 360.60 3.398 0.04190 R Inferior parietal cortex 410.96 5.105 0.01930 R Medial orbitofrontal 458.62 6.134 0.00960 Neg R Lateral occipital cortex 434.31 -­‐3.013 0.01340 Scanner site AMS < NIJM R Middle temporal cortex 1196.71 -­‐15.618 0.00010 L Middle temporal cortex 3396.30 -­‐17.608 0.00010 R Middle frontal cortex 7402.56 -­‐12.716 0.00010 L Middle frontal cortex 9300.90 -­‐14.960 0.00010 NIJM < AMS R Inferior parietal cortex 529.24 3.065 0.00400 L Superior parietal cortex 8971.47 7.844 0.00010 R Middle frontal cortex 896.99 4.627 0.00010 L Middle frontal cortex 2012.35 6.515 0.00010 R Precuneus cortex 3257.91 8.543 0.00010 R Supramarginal cortex 894.47 4.618 0.00010 SES Region R = right, L = left, Size = cluster size in mm , PCLUSTER = cluster-­‐wise p-­‐value after correction for multiple comparisons, AMS = scanner in Amsterdam, NIJM = scanner in Nijmegen, Pos = positive correlation, Neg = negative correlation, SES = socio-­‐economic status. 2
63 S3 – Sensitivity analyses TABLE S3. Sensitivity analyses. Estimated marginal mean cortical thickness in subsamples of healthy control participants and participants with ADHD, and associated p-­‐values, within the left and right medial temporal cluster of significant case-­‐control difference. LH RH n EMM H C EMM A D H D p EMM H C EMM A D H D p Original analyses / all subjects 490 3.323 3.182 0.001 3.224 3.113 0.001 Within Amsterdam 251 3.347 3.221 0.003 3.207 3.136 0.028 Within Nijmegen 239 3.306 3.140 0.001 3.256 3.092 0.001 Within boys 301 3.342 3.191 0.001 3.222 3.110 0.001 Within girls 189 3.304 3.178 0.014 3.238 3.113 0.002 Within age < 14.05 99 3.252 3.201 0.526 3.198 3.092 0.120 Within age 14.05-­‐16.21 98 3.347 3.242 0.060 3.278 3.144 0.007 Within age 16.21-­‐18.01 98 3.335 3.197 0.037 3.211 3.142 0.162 Within age 18.01-­‐20.04 97 3.416 3.119 0.001 3.304 3.094 0.001 Within age > 20.04 98 3.336 3.173 0.031 3.156 3.100 0.334 Excluding co-­‐medication 401 3.326 3.184 0.001 3.221 3.117 0.001 Excluding co-­‐morbidity 389 3.326 3.193 0.001 3.224 3.115 0.001 Additional covariate: IQ 490 3.321 3.184 0.001 3.227 3.112 0.001 Additional covariate: TBV 490 3.323 3.183 0.001 3.224 3.113 0.001 Additional covariate: average CT 490 3.318 3.185 0.001 3.223 3.116 0.001 RH = right hemisphere, LH = left hemisphere, EMM = estimated marginal mean cortical thickness in mm, HC = healthy control participants, ADHD = participants with attention-­‐deficit/hyperactivity disorder, p = cluster-­‐wise p-­‐value after correction for multiple comparisons, TBV = total brain volume, CT = cortical thickness 64 S4 – Vertex-­‐wise analyses with additional covariates. TABLE S4. Participants with ADHD vs. healthy control participants. Regions of significant increased and decreased cortical thickness (cluster-­‐wise p-­‐value < 0.05, corrected for multiple comparisons using Monte Carlo simulation testing), in participants with ADHD compared to healthy control participants, in a statistical model including estimated IQ, total brain volume or average cortical thickness as an additional covariate. Additional covariate Region Cluster size PCLUSTER Cohen’s d EMMHC EMMADHD IQ L medial temporal 435.07 0.012 0.415 3.355 3.214 R medial temporal 357.37 0.043 0.436 3.235 3.117 L superior parietal 359.33 0.037 -­‐0.434 2.091 2.204 Average CT L medial temporal 440.36 0.006 0.417 3.334 3.199 R medial temporal 340.71 0.032 0.449 3.229 3.117 L superior parietal 385.43 0.014 -­‐0.475 2.094 2.200 TBV L medial temporal 475.00 0.009 0.425 3.349 3.207 R medial temporal 390.75 0.028 0.419 3.184 3.073 CT = cortical thickness, TBV = total brain volume, R = right, L = left, PCLUSTER = cluster-­‐wise p-­‐value after correction for multiple comparisons, EMM = estimated marginal mean cortical thickness in mm, HC = healthy control participants, ADHD = participants with attention-­‐deficit/hyperactivity disorder. FIGURE S4. Regions of significant decreased cortical thickness in red and increased cortical thickness in blue (cluster-­‐wise p-­‐value < 0.05, corrected for multiple comparisons using Monte Carlo simulation testing), in participants with ADHD compared to healthy control participants, in a statistical model including estimated IQ as an additional covariate. 65 66 Chapter 4 STIMULANT TREATMENT HISTORY PREDICTS
FRONTAL-STRIATAL STRUCTURAL CONNECTIVITY
IN ADOLESCENTS WITH ADHD
Published as: Schweren LJS, Hartman CA, Zwiers MP, Heslenfeld DJ, Franke B, Oosterlaan J, Buitelaar JK, Hoekstra PJ. Stimulant treatment history predicts frontal-­‐striatal structural connectivity in adolescents with attention-­‐deficit/hyperactivity disorder. Eur Neuropsychopharmacol. 2016;26(4):674-­‐683. 67 ABSTRACT
Objective: Diffusion tensor imaging (DTI) has revealed white matter abnormalities in individuals with attention-­‐deficit/hyperactivity disorder (ADHD). Stimulant treatment may affect such abnormalities. The current study investigated associations between long-­‐term stimulant treatment and white matter integrity within the frontal-­‐
striatal and mesolimbic pathways, in a large sample of children, adolescents and young adults with ADHD. Methods: Participants with ADHD (N=172; mean age 17, range 9-­‐26) underwent diffusion-­‐weighted MRI scanning, along with an age-­‐ and gender-­‐matched group of 96 control participants. Five study-­‐specific white matter tract masks (orbitofrontal-­‐
striatal, orbitofrontal-­‐amygdalar, amygdalar-­‐striatal, dorsolateral-­‐prefrontal-­‐striatal and medial-­‐prefrontal-­‐striatal) were created. First we analyzed case-­‐control differences in fractional anisotropy (FA) and mean diffusivity (MD) within each tract. Second, FA and MD in each tract was predicted from cumulative stimulant intake within the ADHD group. Results: After correction for multiple testing, participants with ADHD showed reduced FA in the orbitofrontal-­‐striatal pathway (p=0.010, effect size=0.269). Within the ADHD group, higher cumulative stimulant intake was associated with lower MD in the same pathway (p=0.011, effect size=-­‐0.164), but not with FA. The association between stimulant treatment and orbitofrontal-­‐striatal MD was of modest effect size. It fell short of significance after adding ADHD severity or ADHD type to the model (p=0.036 and p=0.094, respectively), while the effect size changed little. Conclusions: Our findings are compatible with stimulant treatment enhancing orbitofrontal-­‐striatal white matter connectivity, and emphasize the importance of the orbitofrontal cortex and its connections in ADHD. Longitudinal studies including a drug-­‐naïve baseline assessment are needed to distinguish between-­‐subject variability in ADHD severity from treatment effects. 68 INTRODUCTION
Diffusion tensor imaging (DTI) has revealed abnormalities in white matter integrity, or structural connectivity, in individuals with attention-­‐deficit/hyperactivity disorder (ADHD) (for extensive reviews, see van Ewijk et al. 2012; Konrad & Eickhoff, 2010). Multiple parameters of white matter integrity can be derived from DTI, including fractional anisotropy (FA) and mean diffusivity (MD). FA measures directionality of water diffusion and is typically high in organized structures such as densely packed white matter bundles, as water is more likely to diffuse along the axons rather than perpendicular to the axons. MD measures the amount of water diffusion in any direction and is high in areas with few natural barriers to water diffusion, such as the ventricles. Less commonly reported are axial diffusivity (AD; measuring water diffusion along the main diffusion direction) and radial diffusivity (RD; measuring water diffusion perpendicular to the main diffusion direction). It is important to note that the interpretation of altered diffusion parameters is complex, especially in psychiatric disorders where changes are mostly subtle. Increased MD and decreased FA are often regarded as indications of impaired or decreased structural connectivity (Thomason & Thompson, 2011), but the neuropathological processes underlying such changes are largely unknown (Jones et al., 2013). To date, findings on structural connectivity in individuals with ADHD compared to healthy controls have been mixed. Whereas some studies reported decreased FA and/or increased MD in individuals with ADHD compared to controls (Ashtari et al., 2005; Cao et al., 2010; Hamilton et al., 2008; Pavuluri et al., 2009), suggesting decreased structural connectivity in ADHD, others found increased FA and/or decreased MD (Li et al., 2010; Peterson et al., 2011; Silk et al., 2009a). Null findings of no changes in structural connectivity have also been reported (Silk et al., 2009a; Silk et al., 2009b). In recent work from our group, Van Ewijk et al. found widespread FA reduction in both participants with ADHD and their unaffected siblings, compared to healthy control participants, suggesting that reduced FA may represent a genetic vulnerability to ADHD. In addition, higher FA and lower MD were observed in more severely affected compared to less severely affected individuals with ADHD, which may reflect a second, distinct, mechanism associated with ADHD symptom severity (Van Ewijk et al., 2014b). Inconsistent findings in previous studies may partially be explained by these two seemingly opposing mechanisms being at play. Inconsistent findings may also reflect differences between the ADHD samples with regard to stimulant treatment history. Individuals with ADHD often take stimulants for prolonged periods of time. Studies investigating long-­‐term stimulant treatment effects on brain structure have almost exclusively focused on grey matter 69 and/or subcortical structures. Several such studies (but not all) have suggested structural normalization with long-­‐term stimulant treatment (Nakao et al., 2011; Shaw et al., 2009; Sobel et al., 2010), i.e. abnormalities typically associated with ADHD were smaller or absent in individuals with ADHD who had been treated with stimulants. Stimulant-­‐induced changes in grey matter might be accompanied by changes in white matter. Only few studies have explored long-­‐term stimulant effects on white matter integrity quantified by DTI in individuals with ADHD. One study applied deterministic tractography to delineate the frontal-­‐striatal tracts, and compared average FA within these tracts between children with a relatively short versus a relatively long history of stimulant treatment (De Zeeuw et al., 2012). No differences between the two groups were detected. Small sample size (n=13 per group) and using average FA across all frontal-­‐striatal tracts as the primary outcome measure limits the interpretation of this negative finding. A second study used both tract-­‐based spatial statistics (TBSS) and whole-­‐brain deterministic tractography, to perform a hypothesis-­‐free search for differences in FA or MD between young treatment-­‐naïve children with ADHD, children with ADHD who had been treated with stimulants, and healthy control children (n=16, n=24, and n=26, respectively) (De Luis-­‐García et al., 2015). Stimulant treatment was associated with decreased MD in several major white matter bundles, including the uncinate fasciculus connecting the medial temporal limbic structures to the orbitofrontal cortex. Importantly, differences in pre-­‐treatment ADHD severity between children with and without stimulant treatment were not assessed, and may have confounded results. In a prior study of our own group on the association between structural connectivity and symptom severity, results did not change when history of stimulant treatment (treated/untreated) was taken into account (Van Ewijk et al., 2014b). In the current report, we investigated the association between stimulant treatment history and structural connectivity in a large sample of children, adolescents and young adults with ADHD. This investigation adds to the previous study from our group in two ways. First, in the current study we assessed stimulant treatment history to detail, and performed dimensional analyses of lifetime cumulative stimulant dose. Second, we applied a sensitive hypothesis-­‐driven region-­‐
of-­‐interest (ROI) approach based on the dopaminergic working mechanism of stimulants. Stimulants generate their clinical effects, at least partially, by enhancing dopaminergic neurotransmission in the striatum (Volkow et al., 2012). Two major dopaminergic pathways connect the striatum to other brain regions: the frontal-­‐
striatal pathway, connecting the striatum to the medial and dorsolateral prefrontal cortex; and the mesolimbic pathway, connecting the striatum to the limbic system including the amygdala and orbitofrontal cortex. We used probabilistic tractography 70 to quantify white matter microstructure within these pathways. A healthy control group was included for reference. We hypothesized that participants with ADHD would present with reduced FA and/or increased MD in frontal-­‐striatal pathways, indicative of lower structural connectivity. Second, in line with observations of grey matter structural normalization with stimulant treatment, we hypothesized that FA would be higher and MD would be lower (both indicative of enhanced structural connectivity) in participants with a history of high cumulative stimulant intake, compared to those with a history of no or less stimulant intake. METHODS Participants An ADHD and control sample were selected from the NeuroIMAGE cohort, a family-­‐based cohort that includes 415 families with one or more probands with ADHD, as well as 141 healthy control families (Von Rhein et al., 2015a). For inclusion, participants had to meet the following criteria: (1) age between 8 and 30 years old, (2) IQ > 70, (3) no diagnosis of epilepsy, general learning difficulties, or known genetic disorders, and (4) availability of a good quality diffusion scan and T1 structural scan. Participants with ADHD had to meet diagnostic criteria for ADHD (see below). The following additional inclusion criteria applied to healthy control participants: (1) no past or present mental health care utilization, (2) no past or present psychiatric disorders (ADHD or otherwise) in first-­‐degree relatives, and (3) no past or present psychoactive medication use reported by either the pharmacy or the participant/parents (incidental use was allowed). All subjects who did not fulfill criteria for either the ADHD or the control group were excluded, thus excluding unaffected siblings of individuals with ADHD. A group-­‐matched (on age, gender, and scanner location) healthy control sample was drawn. The final sample consisted of 172 participants with ADHD and 96 healthy control participants (for reference), between the ages of nine and twenty-­‐six years (M=17.2, SD=3.1). Informed consent was signed by all participants (parents signed informed consent for participants under 12 years of age) and their parents (for participants under 18 years of age), and the study was approved by the ethical committees of participating institutions. Diagnostic Assessment All participants were assessed using a combination of a semi-­‐structured diagnostic interview and Conners' ADHD questionnaires. For participants using medication, ratings of participants functioning were done off medication. All 71 participants were administered the Dutch translation of the Schedule for Affective Disorders and Schizophrenia for School-­‐Age Children -­‐ Present and Lifetime Version (Kaufman et al., 1997), carried out by trained professionals. Both the parents and the participant, if ≥ 12 years old, were interviewed separately and were initially administered the ADHD screening interview. Participants with elevated scores on any of the screening items were administered the full ADHD section. In addition, each participant was assessed with a parent-­‐rated questionnaire (Conners' Parent Rating Scale -­‐ Revised: Long version, Conners et al., 1998b) combined with either a teacher-­‐
rated questionnaire for children < 18 years (Conners' Teacher Rating Scale -­‐ Revised: Long version; Conners et al., 1998a) or a self-­‐report questionnaire for participants ≥ 18 years (Conners' Adult ADHD Rating Scales -­‐ Self-­‐Report: Long version; Conners et al., 1999). Participants with ≥ six symptoms of hyperactive/impulsive behavior and/or inattentive behavior were diagnosed with ADHD, provided they: a) met the DSM-­‐IV criteria for pervasiveness and impact of the disorder; b) had an onset-­‐age before 12, and c) scored T ≥ 63 on at least one of the ADHD scales on either one of the Conners' ADHD questionnaires. Healthy control participants were required to score T < 63 on each of the ADHD scales of each of the Conners’ questionnaires, and have ≤ 3 combined symptoms. Criteria were adapted for participants of 18 years or older, such that five symptoms were sufficient for a diagnosis and ≤ two symptoms were allowed for healthy control participants. Additional assessments included the Children’s Social Behavior Questionnaire (Hartman et al., 2006) to measure symptoms of autism spectrum disorders, the block-­‐design and vocabulary subtests of the Wechsler Intelligence Scale for Children (Wechsler, 2002) for participants <18, or the Wechsler Adult intelligence Scale (Wechsler, 2000) for participants ≥18 to estimate IQ, and the Children’s Global Assessment Scale (Shaffer et al., 2014) to assess functional impairment in daily life. Assessment of medication history All participants provided written consent to obtain lifetime medication transcripts from the pharmacy. For participants under the age of twelve, permission was obtained from one or both of the parents. In addition, an extensive questionnaire was used to assess lifetime history of psychoactive medication for all participants. The questionnaire was administered during the testing day, either by the parents or by the participant (> age 18). For healthy control participants, the pharmacy transcripts and questionnaires were used to ascertain a negative history of any type of psychoactive medication. For participants with ADHD, cumulative stimulant intake was calculated from the pharmacy transcripts as the lifetime total stimulant dose in mg. D-­‐
amphetamine dose was multiplied by two in this calculation. If pharmacy transcripts 72 did not cover the medicated period according to the medication questionnaire (n=48, 29%), stimulant intake during the missing period(s) were calculated from the questionnaire data, and were added to the cumulative intake derived from the pharmacy data. Cumulative stimulant intake was divided by participant’s age minus 2.3 (the minimum stimulant start age within our sample), to obtain a measure of cumulative intake independent of age (CSI), which was used in all analyses. MRI acquisition Participants were asked to withhold use of psychoactive drugs for 48 hours before scanning. MRI data was acquired at 1.5T on a Siemens Sonata scanner at the VU Medical Center (Amsterdam, the Netherlands) and on a Siemens Avanto scanner at the Donders Center for Cognitive Neuroimaging, (Nijmegen, the Netherlands). An identical 8-­‐channel phased array coil was used at both sites and all scan parameters were matched as closely as possible. The scan protocol included one eddy-­‐current compensated diffusion-­‐weighted SE-­‐EPI sequence (5 volumes without directional weighting, followed by 60 diffusion-­‐weighted volumes of 60 interleaved transverse slices each, slice thickness 2.2 mm, FOV = 256 mm, TR = 8500 ms, TE = 97 ms; b-­‐
value = 1000 s/mm2, GRAPPA-­‐acceleration 2), and two T1-­‐weighted MP-­‐RAGE scans (TI = 1000 ms, TR = 2730 ms, TE = 2.95 ms, FA = 7°; 176 sagittal slices, voxel size 1 x 1 x 1 mm, FOV = 256 x 256 x 176 mm; GRAPPA acceleration 2). MRI analyses For each subject, residual eddy current correction and realignment of all denoised diffusion-­‐weighted images per subject were performed in SPM8 (Wellcome Trust Centre for Neuroimaging). Denoising was performed using a local PCA procedure and a dedicated robust estimation algorithm (PATCH; Zwiers, 2010) was used to correct for motion-­‐induced artifacts and tensor estimation. Realignment parameters were comparable for participants with and without ADHD, indicating similar levels of motion-­‐distortion in the two groups. B0 diffusion images of each participant were registered to their own T1 structural scan, which in turn was registered to the a 2mm template (specific for NeuroIMAGE sample) using FNIRT in FSL, creating warp fields to transform images from subject space to standard space and vice versa. First, the following masks were created from the AAL atlas in standard space: left and right dorsolateral prefrontal cortex (dlPFC; AAL 7+11+13 and 8+12+14), left and right medial prefrontal cortex (mPFC; AAL 23 and 24), left and right orbitofrontal cortex (OFC; AAL 5+9+15+25+27 and 6+10+16+26+28), left and 73 A B C FIGURE 1. Study-­‐specific white matter tractography masks, projected on the study-­‐
specific standard brain template. Coordinates are in MNI-­‐space. A. orbitofrontal-­‐
striatal tract (red), orbitofrontal-­‐amygdalar tract (pink) and amygdalar-­‐striatal tract (green; X=-­‐26, Y=0, Z=14) B. dorsolateral-­‐prefrontal-­‐striatal tract (light blue; X=32, Y=14, Z=16) C. medial-­‐prefrontal-­‐striatal tract (dark blue; X=-­‐16, Y=36, Z=2). right striatum (AAL 71+73 and 72+74), and left and right amygdala (AAL 41 and 42). Three additional exclusion masks were registered: X<-­‐1 (excluding the left hemisphere), X>1 (excluding the right hemisphere) and Y<-­‐30 (excluding all voxels posterior to the tail of the caudate nucleus). Next, all masks were inversely warped to subject space and used to specify in-­‐ and exclusion criteria for reconstructed white matter tracts. The probtrackx2 tool in FSL performed probabilistic tractography propagating 5000 streamlines per voxel within the seed mask, including all streams that reach the waypoint-­‐mask, and excluding all streams that ran through an exclusion-­‐mask. The following white matter tracts were reconstructed for each hemisphere: dorsolateral-­‐prefrontal-­‐striatal (seed=dlPFC; waypoint=striatum), medial-­‐prefrontal-­‐striatal, (seed=mPFC; waypoint=striatum), orbitofrontal-­‐striatal, (seed=OFC; waypoint=striatum), orbitofrontal-­‐amygdalar (seed=amygdala; waypoint=OFC), and amygdalar-­‐striatal (seed=amygdala; waypoint=striatum). All ROI-­‐masks other than the seed-­‐ and waypoint-­‐mask were specified as exclusion 74 masks. Resulting distribution images were thresholded to include only voxels that were hit by at least 1% of all streamlines generated by probtrackx, and binarized to create a subject-­‐specific mask of each white matter tract. These maps were then warped to standard space, summed up across all subjects, thresholded such that voxels were included if at least 75% of all subjects had a white matter tract in that voxel, and binarized to create a study-­‐specific mask of each white matter tract in standard space (Figure 1). Last, FA and MD images of each participant were warped to standard space, and FA and MD within each tract-­‐mask were extracted. For each tract we calculated left, right, and average FA and MD. For follow-­‐up analyses we also warped axial diffusivity (AD) and radial diffusivity (RD) maps to standard space and extracted average AD and RD for each tract and participant. Statistical analysis Statistical analyses were performed in SPSS version 22.0.0.0 (IBM, 2013). First, we compared average FA and MD within each white matter tract between participants with ADHD and healthy control participants using General Linear Mixed (GLM) modeling, taking into account gender, scanner site, age, and age-­‐squared and including a random intercept to account for the clustered family data. We also examined possible interaction effects between diagnosis and age, age-­‐squared, and gender. Second, the same statistical model was applied to investigate the effect of CSI on FA and MD (only within participants with ADHD), as well as possible interaction effects between CSI and age, age-­‐squared, and gender. For the within-­‐ADHD analyses, participants with missing stimulant treatment data (n=8) were excluded, as well as one participant who was an outlier on cumulative stimulant intake (z=5.6). We applied a two-­‐step method to correct for multiple comparisons. Given that we had five positively correlated tracts per DTI-­‐measure, we first calculated the effective number of tests (Meff; Moskvina & Schmidt, 2008) based on bivariate correlations between the five outcome measures (ranging from r=0.185 to r=0.887). Next, Meff was applied to the Dunn-­‐Sidak familywise error rate correction (Šidàk, 1967). For the analyses of FA and MD, Meff was 3.81 and 3.40, and the adjusted alpha-­‐
levels derived from Dunn-­‐Sidak-­‐correction were 0.013 and 0.015, respectively. Significant effects of treatment or diagnosis were ensued by follow-­‐up analyses: (1) analyzing the left and right hemisphere separately, (2) analyzing AD and RD separately, and (3) adding to the model any clinical or demographic variable that was associated with CSI, as these may have been confounders (for treatment effects). The same adjusted alpha-­‐levels were used in the follow-­‐up analyses. Last, we tested the robustness of our findings by repeating our analyses within various subsamples, namely boys (n=172) and girls (n=96), participants who had been tested in 75 Nijmegen (n=143) and in Amsterdam (n=125), participants with no history of dexamphetamine treatment (n=249), and participants stratified into age quartiles (n<14.96=67, n14.96-­‐17.50=67, n17.50-­‐19.57=68, n>19.57=66; Supplement S1). RESULTS
The ADHD and control samples did not differ in age (HC: M=16.96, SD=3.26; ADHD: M=17.39, SD=3.05; t=-­‐1.096, p=0.274), gender (HC: NMALE=56, 58%; ADHD: NMALE=116, 67%; Chi2=2.223, p=0.136), or scanner location (HC: NNIJMEGEN=44, 46%; ADHD: NNIJMEGEN=99, 58%; Chi2=3.403, p=0.065). Participants with ADHD had lower IQ compared to control participants (HC: M=106.47, SD=14.09; ADHD: M=96.62, SD=13.67; t=5.582; p=0.001). The difference in IQ was considered part of the ADHD phenotype (Dennis et al., 2009), thus IQ was not added as a covariate. Cumulative stimulant intake was available for 163 participants with ADHD. Participants had inattentive type (n=82, 50%), combined type (n=66, 41%), or hyperactive type ADHD (n=15, 9%). Thirty-­‐two percent (n=52) had a comorbid diagnosis, mostly oppositional defiant disorder or conduct disorder (n=47, 29%), but also tic disorders (n=3, 2%) and anxiety/depression (n=4, 3%). Treatment characteristics of the ADHD group are summarized in Table 1. The vast majority had at some point in their lives been treated with stimulants, including immediate release methylphenidate (n=141, 87%), extended-­‐release methylphenidate (n=112, 69%), and dexamphetamine (n=18, 11%). Average cumulative stimulant intake was 62396 mg (equal to 5.7 years of 30 mg per day), ranging from zero mg (stimulant-­‐naïve) to 289000 mg (equal to 13.2 years of 60 mg per day). Forty percent (n=65) of participants with ADHD had received stimulant treatment within three months prior to scanning; the other participants had ceased treatment prior to study participation. Psychoactive medication other than stimulants was frequent, and included atomoxetine (n=20, 12%), atypical antipsychotics (n=22, 14%), benzodiazepines (n=10, 6%), and antidepressants (n=7, 4%). Individual differences in clinical characteristics or stimulant treatment history (other than CSI) were analyzed as potential confounders (see below). Participants with ADHD vs. healthy control participants Compared to healthy control participants, participants with ADHD had lower FA in white matter tracts connecting the striatum and the orbitofrontal cortex (Cohen’s d=0.269, p=0.010). The direction of effect remained unchanged when participants were stratified by gender, scanner site, or age quartiles, and when participants treated with d-­‐amphetamine preparations were excluded (Supplement 76 S1). Follow-­‐up analyses showed that lower OFC-­‐striatal FA was present in both hemispheres, and significant after correction for multiple testing in the left hemisphere. There were no case-­‐control differences in axial (p=0.132) or radial diffusivity (p=0.218) in this pathway. Lower FA in the dorsolateral-­‐prefrontal-­‐striatal pathway was nominally significant, but failed to reach the alpha-­‐level adjusted for multiple testing (Cohen’s d=0.289, p=0.018). In the mPFC-­‐striatal tract, amygdalar-­‐
striatal tract, and the OFC-­‐amygdalar tract we found no FA differences between participants with ADHD and healthy controls (Table 2). Further, we found no diagnosis-­‐by-­‐gender (p>0.432), diagnosis-­‐by-­‐age (p>0.128), or diagnosis-­‐by-­‐age-­‐
squared (p>0.289) interaction effects on FA in any tract. TABLE 1. Treatment characteristics of the ADHD group (n=163). n % Stimulant-­‐naïve 18 11 Current users (<3 months prior to scan) 65 40 Other psychoactive medication (any) 40 25 M SD Stimulant treatment duration (years) 4.31 3.39 Stimulant age of initiation (years) * 8.66 2.63 Stimulant age of cessation (years) † 12.79 5.78 62395.95 55929.65 31.31 16.06 Stimulant cumulative stimulant intake (mg) Stimulant mean daily dose (mg) * within participants who had received stimulant treatment (n=145). † within participants who had ceased treatment > 3 months prior to scanning (n=98). There were no between-­‐group differences in MD in any of the tracts (p>0.503, Table 2). Moreover, we found no diagnosis-­‐by-­‐gender (p>0.142), diagnosis-­‐by-­‐age (p>0.490), or diagnosis-­‐by-­‐age-­‐squared (p>0.186) interaction effects on MD. Finally, across all participants, FA and MD were negatively correlated in the mPFC-­‐striatal tract (Pearson’s r=-­‐0.212, p<0.001), but not in the other tracts. 77 TABLE 2. Main effects of diagnosis (ADHD or healthy control) on FA and MD for each tract FA b p OFC – Striatum 0.008 Right Left MD Sign b p Sign 0.010 ** -­‐0.001 0.876 0.008 0.032 * 0.008 0.009 ** dlPFC – Striatum 0.008 0.018 * < -­‐0.001 0.951 mPFC – Striatum < -­‐0.001 0.912 < -­‐0.001 0.910 Amygdala – Striatum 0.004 0.253 0.001 0.880 OFC – Amygdala 0.001 0.617 0.003 0.503 * p<0.05; ** significant after correction for multiple testing, αFA=0.013, αMD=0.015; positive b-­‐values indicate reduced FA/MD in participants with ADHD compared to healthy controls. FA, fractional anisotropy; MD, mean diffusivity; OFC, orbitofrontal cortex; dlPFC, dorsolateral prefrontal cortex; mPFC, medial prefrontal cortex. Cumulative stimulant treatment Within the ADHD group, CSI did not predict FA values in any of the white matter tracts (p>0.355, Table 3). Furthermore, there were no CSI-­‐by-­‐gender (p>0.103), CSI-­‐by-­‐age (p>0.060), or CSI-­‐by-­‐age-­‐squared (p>0.293) interaction effects on FA. Cumulative stimulant intake was negatively associated with MD in orbitofrontal-­‐striatal pathway (Cohen’s d=-­‐0.164, p=0.011). The direction of effect remained unchanged when participants were stratified by gender, scanner site, or age quartiles, and when participants treated with d-­‐amphetamine preparations were excluded (Supplement S1). Follow-­‐up analyses revealed that higher CSI was associated with lower MD in both hemispheres, and that higher CSI was associated with lower AD (Cohen’s d=-­‐0.107, p=0.046) and lower RD (Cohen’s d=-­‐0.165, p=0.029). None of the follow-­‐up analyses reached the alpha-­‐level adjusted for multiple testing (α=0.015). In addition, we found nominally significant lower MD in the orbitofrontal-­‐amygdalar (Cohen’s d=-­‐0.120, p=0.016) and medial-­‐prefrontal-­‐
striatal pathways (Cohen’s d=-­‐0.188, p=0.021), but neither met the adjusted alpha-­‐
level. CSI was not associated with MD in the dlPFC-­‐striatal or amygdalar-­‐striatal tracts (Table 3). There were no CSI-­‐by-­‐gender interaction effects (p>0.115), CSI-­‐by-­‐age (p>0.368) or CSI-­‐by-­‐age-­‐squared (p>0.350) interaction effects on MD. Within participants with ADHD, there was a negative correlation between FA and MD in the mPFC-­‐striatal tract (Pearson’s r=-­‐0.211, p=0.007), but not in the other tracts.
78 Cumulative stimulant intake was positively correlated with the number of ADHD symptoms (r=0.271, p=0.001), and was different for the three ADHD types (F=8.380, p=0.001). CSI was higher in the combined type group compared to the hyperactive group (MCOMBINED=14.6, MHYPERACTIVE=5.9, p=0.007) and to the inattentive group (MCOMBINED=14.6, MINATTENTIVE=8.8, p=0.002). CSI was not related to functioning in daily life (p=0.325), presence of comorbid disorders (0.291), treatment with non-­‐
stimulant ADHD medication (p=0.730), treatment with medication other than for ADHD (p=0.206), or symptoms of autism spectrum disorder (p=0.258). The analyses of orbitofrontal-­‐striatal MD were repeated including the number of ADHD symptoms as a covariate. Number of symptoms did not predict orbitofrontal-­‐striatal MD (p=0.360). Adding the number of symptoms to the model caused the effect of CSI to fall short of significance, although the size of the effect changed little (Cohen’s d=-­‐
0.144, p=0.036). Similarly, ADHD type, added to the model as two dummy variables, did not predict OFC-­‐striatal MD (pCOMBINEDvs.HYPERACTIVE=0.090; pCOMBINEDvs.INATTENTIVE= 0.109), but adding this covariate caused the effect of CSI to fall short of significance (Cohen’s d=-­‐0.114, p=0.094). These analyses indicate that although ADHD severity, ADHD type, and stimulant exposure are overlapping (resulting in larger standard errors of the estimated regression coefficients when modeled simultaneously, reducing statistical significance), this overlap has little impact on the effect size of the association between orbitofrontal-­‐striatal MD and CSI. TABLE 3. Main effects of age-­‐independent cumulative stimulant intake on FA and MD for each tract. FA OFC – Striatum MD b p Sign b p Sign 0.001 0.501 -­‐0.005 0.011 ** Right -­‐0.006 0.016 * Left -­‐0.004 0.027 * mPFC – Striatum -­‐0.001 0.713 -­‐0.005 0.021 * OFC – Amygdala -­‐0.001 0.389 -­‐0.005 0.016 * dlPFC– Striatum 0.001 0.713 -­‐0.004 0.055 Amygdala – Striatum -­‐0.002 0.355 -­‐0.003 0.119 * p<0.05; ** significant after correction for multiple testing, αFA=0.013, αMD=0.015; positive b-­‐
values indicate reduced FA/MD in participants with ADHD compared to healthy controls. FA, fractional anisotropy; MD, mean diffusivity; OFC, orbitofrontal cortex; dlPFC, dorsolateral prefrontal cortex; mPFC, medial prefrontal cortex. 79 DISCUSSION The current study investigated associations between stimulant treatment history and white matter microstructural abnormalities in frontal-­‐striatal and mesolimbic pathways, in a large sample of children, adolescents, and young adults with ADHD. We had hypothesized that FA would be lower and MD would be higher, both indicative of impaired structural connectivity, in participants with ADHD compared to healthy control participants. Indeed we found lower FA in the orbitofrontal-­‐striatal tracts in participants with ADHD. There were no case-­‐control differences in MD. Second, we had hypothesized that white matter microstructural abnormalities would be more prominent in participants with ADHD who had received little or no stimulant treatment. We found that cumulative stimulant intake was negatively correlated with orbitofrontal-­‐striatal MD, suggesting higher structural connectivity with more and/or longer stimulant treatment. We found no correlations between stimulant treatment and frontal-­‐striatal or mesolimbic FA. Our findings are in line with those of De Luis-­‐Garcia et al. (2015), who found lower MD in children who had been treated with stimulants compared to those who had not. In their sample, lower MD in stimulant-­‐treated children was present in white matter tracts connecting the orbitofrontal cortex, and this was not accompanied by higher FA. The interpretation of subtle FA and MD changes is equivocal. Increased MD and decreased FA may be regarded as indications of impaired structural connectivity, and often, but not always, co-­‐occur (Thomason & Thompson, 2011). It remains speculative, however, which neurodevelopmental or neuropathological processes underlie such alterations in FA and MD (Jones et al., 2013). Changes in FA and MD in ADHD may represent two distinct mechanisms, i.e. reduced FA may represent a familiar (e.g. genetic) vulnerability to ADHD, whereas MD may be a clinical state marker (Van Ewijk et al., 2014b). The current associations between stimulant treatment and orbitofrontal-­‐striatal MD but not FA are in line with this hypothesis. Alternatively, the discrepancy between FA and MD findings could suggest that stimulant treatment may interact with a non-­‐dysfunctional feature of white matter connectivity. We wish to additionally emphasize that confirmation in an independent sample is needed to exclude the possibility of a Type 1 error thereby avoiding over-­‐
interpretation, especially given the modest effect size of the finding in MD. Stimulant intake correlated with white matter abnormalities in pathways connecting the orbitofrontal cortex to the striatum, and, albeit only nominally significant, to the amygdala. Altered structural connectivity within orbitofrontal-­‐
striatal pathways in individuals with ADHD has been related to impulsivity (Konrad & Eickhoff, 2010), impaired school functioning (Wu et al., 2014) and neuropsychological deficits (Shang et al., 2013). Very little is known about the long-­‐term effects (multiple 80 years) of stimulant treatment on either mesolimbic structures or on behaviors mediated by these structures. In previous studies of the current sample, we found no associations between stimulant treatment and striatal or amygdalar volumes (Greven et al., 2015) or orbitofrontal cortical thickness (Schweren et al., 2015a). Other structural and functional neuroimaging studies, including a positron-­‐emission tomography (PET) study, have indicated long-­‐term stimulant treatment effects on the striatum and amygdala (Ludolph et al., 2008; Onnink et al., 2014), but there have also been null findings (Schlochtermeier et al., 2011; Stoy et al., 2011). Like the current study, each of these studies in adults has been observational and lacked pretreatment measurement. Nevertheless, previous studies in conjunction with the current study support the importance of the orbitofrontal cortex and its striatal and limbic connections in ADHD, and suggest that structural connectivity of the orbitofrontal cortex may be affected by stimulant treatment. As we expected, stimulant treatment was positively associated with ADHD severity: participants with a history of higher cumulative stimulant intake presented with more ADHD symptoms compared to patients with little or no stimulant treatment history. Similarly, patients with different ADHD types presented with different treatment histories (i.e. higher stimulant intake in the combined group compared to the hyperactive and inattentive groups). Disentangling the individual contributions of stimulant treatment and ADHD severity or type is a challenge. Confounding by indication, where more severely affected individuals are likely to receive more treatment, is inevitable when studying long-­‐term treatment effects in children. We attempted to address the confounding effect of clinical differences by entering ADHD severity and ADHD type as covariates in the model predicting MD from cumulative stimulant intake. These clinical variables were measured post-­‐
treatment, thus at best representing a proxy of pretreatment clinical differences. The treatment effect on orbitofrontal-­‐striatal MD changed little when differences in ADHD severity and/or type were accounted for, i.e., the effect was small with and without these additional covariates. However, the effect was no longer significant, which indicates an increased standard error around the estimated effect. When simultaneously analyzed, the effect of treatment was more significant than the effect of either of the clinical variables (p=0.036 versus p=0.360 for number of symptoms, and p=0.094 versus p=0.134 for ADHD type), which indicates that stimulant treatment history is an important predictor and more predictive than ADHD severity or type. In addition, when ADHD severity and type were each separately analyzed without cumulative stimulant intake, neither significantly predicted orbitofrontal-­‐
striatal MD (data not shown). We conclude that although confounding by indication cannot be excluded, our findings support the importance of stimulant treatment history for orbitofrontal-­‐striatal MD. 81 The current study had the following limitations. Studies investigating long-­‐
term treatment effects in children, including the current study, are inevitably naturalistic by design. In addition our study lacked pre-­‐treatment assessment. As a result, we cannot exclude the possibility that white matter differences may have led to stimulant prescription (possibly through more severe ADHD behavior) instead of vice versa. Second, some regions of our interest, including the orbitofrontal cortex, are relatively susceptible to DTI scanning artifacts (e.g. image distortion). Consequently, we used group templates of white matter tracts, as opposed to using each participant’s individual tracts, which resulted in relatively small study-­‐specific regions of interest per white matter tract. ROI selection of frontal-­‐striatal pathways also precluded finding potential changes in other white matter pathways affected by non-­‐
dopaminergic stimulant effects. For example, noradrenergic stimulant effects are known to occur both within and beyond the PFC, and may result in white matter changes in the posterior lobes and cerebellum that were not studied here. Finally, performing tractography using masks of functionally defined striatal subregions (e.g., Di Martino et al., 2008) may in future studies enhance anatomical specificity of white matter tracts, and aid in understanding behavioral correlates of neural changes. There are several strengths to our study as well. This is the first study to date investigating long-­‐term stimulant effects on white matter integrity in a large and representative ADHD sample. In addition, our sample with its wide age-­‐range allowed the investigation of long-­‐term treatment effects spanning multiple years. Moreover, using lifetime pharmacy transcripts stimulant treatment history was assessed to a level of detail that has not previously been achieved. In conclusion, participants with ADHD showed white matter microstructural abnormalities in orbitofrontal-­‐striatal pathways, and stimulant treatment was associated with white matter microstructure in this same pathway. Whereas stimulant treatment was related to MD, case-­‐control differences were found in FA but not MD. These findings could be interpreted to suggest that differences in FA and MD may represent two distinct pathophysiological processes in ADHD, and/or that stimulant treatment may act through enhancing a non-­‐dysfunctional feature of white matter connectivity. Both hypotheses need further investigation in independent samples. Our findings support the importance of the orbitofrontal cortex and its connections in the pathophysiology of ADHD. 82 SUPPLEMENTAL INFORMATION
S1 -­‐ Robustness analyses TABLE S1. Analyses resulting in significant findings were repeated in various sub-­‐
samples of the original sample. HC vs. ADHD OFC-­‐Striatum (FA) CSI OFC-­‐Striatum (MD) n b p n b p Original sample 268 0.008 0.010 163 -­‐0.005 0.015 Male 172 0.008 0.016 112 -­‐0.005 0.028 Female 96 0.013 0.022 51 -­‐0.003 0.344 Amsterdam 125 0.008 0.075 70 -­‐0.005 0.119 Nijmegen 143 0.009 0.068 93 -­‐0.004 0.105 Age quartile 1 (<14.96 years) 67 0.012 0.053 37 -­‐0.009 0.055 Age quartile 2 (14.96-­‐17.50 years) 67 0.011 0.091 40 -­‐0.003 0.515 Age quartile 3 (17.50-­‐19.57 years) 68 0.012 0.045 44 -­‐0.003 0.559 Age quartile 4 (>19.57 years) 66 0.004 0.562 42 -­‐0.003 0.489 Excl. d-­‐amphetamine 249 0.008 0.007 154 -­‐0.006 0.010 Positive b-­‐values in column three indicate reduced FA in participants with ADHD compared to healthy controls. Negative b-­‐values in column six indicate a negative correlation between MD and CSI. ADHD, attention-­‐deficit/hyperactivity disorder; HC, healthy controls; CSI, age-­‐independent cumulative stimulant intake; FA, fractional anisotropy; MD, mean diffusivity; OFC, orbitofrontal cortex. 83 84 Chapter 5 STIMULANT TREATMENT TRAJECTORIES ARE
ASSOCIATED WITH NEURAL REWARD
PROCESSING IN ADHD
In press as: Schweren LJS, Groenman A, von Rhein D, Weeda W, Faraone SV, Luman M, van Ewijk H, Heslenfeld DJ, Franke B, Buitelaar JK, Oosterlaan J, Hoekstra PJ, Hartman CA. Stimulant treatment trajectories are associated with neural reward processing in attention-­‐deficit/hyperactivity disorder. J Clin Psychiatry, 2016. 85 ABSTRACT
Objective: The past decades have seen a surge in stimulant prescriptions for the treatment of attention-­‐deficit/hyperactivity disorder (ADHD). Stimulants acutely alleviate symptoms and cognitive deficits associated with ADHD by modulating striatal dopamine neurotransmission, and induce therapeutic changes in brain activation patterns. Long-­‐term functional changes after treatment are unknown, as long-­‐term studies are scarce and have focused on brain structure. In this observational study (2009-­‐2012), we investigated associations between lifetime stimulant treatment history and neural activity during reward processing. Methods: Participants fulfilling DSM-­‐5 criteria for ADHD (n=269) were classified according to stimulant treatment trajectory. Of those, 124 performed a monetary incentive delay task during magnetic resonance imaging, all in their non-­‐medicated state (NEARLY&INTENSE=51; NLATE&MODERATE=49; NEARLY&MODERATE=9; NNAIVE=15; mean age=17.4 years, range 10-­‐26 years). Whole-­‐brain analyses were performed with additional focus on the striatum, concentrating on the two largest treatment groups. Results: Compared to the ‘late-­‐and-­‐moderate’ treatment group, the ‘early-­‐and-­‐intense’ treatment group showed more activation in the supplementary motor area and dorsal anterior cingulate cortex (SMA/dACC) during reward outcome (cluster size=8696 mm3; pCLUSTER<0.001). SMA/dACC activation of the control group fell in between the two treatment groups. Treatment history was not associated with striatal activation during reward processing. Conclusions: Our findings are compatible with previous reports of acute increases of SMA/dACC activity in individuals with ADHD after stimulant administration. Higher SMA/dACC activity may indicate that patients with a history of intensive stimulant treatment, but currently off-­‐medication, recruit brain regions for cognitive control and/or decision-­‐making upon being rewarded. No striatal or structural changes were found. CLINICAL POINTS
•
•
Stimulant treatment is regarded a safe and effective treatment for ADHD symptoms, yet their long-­‐term effects on brain activation patterns in children and adolescents are largely unknown. Early and intense stimulant treatment may result in increased activation of cognitive control areas during rewarding situations, even when patients are non-­‐
medicated at that time. 86 INTRODUCTION Stimulant treatment is the medical intervention of first choice for children and adolescents with attention-­‐deficit/hyperactivity disorder (ADHD). The past decades have seen a surge in stimulant prescription rates (Trip et al., 2009). Alleviation of symptoms and cognitive deficits associated with ADHD appears – in general – not to last after medication is discontinued, and there is little evidence of long-­‐term improved functioning (Jensen et al., 2007; Molina et al., 2009; Van De Loo-­‐
Neus et al., 2011). The absence of conclusive evidence regarding potential long-­‐term effects of stimulant treatment, either positive or negative, has unsettled parents, patients, and society at large. Studies of long-­‐term stimulant treatment effects on brain structure have yielded mixed results. Two meta-­‐analyses found that striatal volume was more reduced in patients compared to controls when the ADHD sample included more treatment-­‐naive patients (Frodl & Skokauskas, 2012; Nakao et al., 2011), suggesting that striatal volume reduction observed in ADHD is driven by untreated rather than stimulant-­‐treated patients. However, a large-­‐scale longitudinal study, which employed the optimal design for the study of long-­‐term treatment effects, did not find such treatment effects (Shaw et al., 2014), nor did previous analyses in our own sample (Greven et al., 2015; Schweren et al., 2015a). The literature on long-­‐term treatment effects in the human brain has, with few exceptions, focused on brain structure, while studies of acute stimulant effects focused on brain activation patterns. A single dose of methylphenidate has repeatedly been found to alter brain activation patterns in ADHD patients; case-­‐control differences in blood-­‐oxygen level dependent (BOLD)-­‐response to cognitive/motivational tasks became smaller or disappeared when patients were on stimulant medication (Cubillo et al., 2012a). Little is known about whether acute functional changes translate into long-­‐term functional changes as well. Adults with a history of untreated childhood ADHD showed blunted ventral-­‐striatal activation compared to controls when exposed to emotional pictures, whereas adults with a history of ADHD who had received stimulant treatment during childhood did not (Schlochtermeier et al., 2011). During reward processing, the same group of treatment-­‐naive adults showed lower insula activation compared to controls and childhood stimulant-­‐treated adults (Stoy et al., 2011). These findings may suggest enduring functional therapeutic changes. In a meta-­‐analysis of attention tasks, striatal activity was particularly reduced in studies including mostly stimulant-­‐naive patients (Hart et al., 2013). Radioligand studies, however, have reported exacerbated rather than attenuated deficits in striatal dopamine neurotransmission after long-­‐term stimulant treatment in adults with ADHD (Fusar-­‐Poli et al., 2012; Ludolph et al., 87 2008). Summarizing, stimulant treatment may be associated with persistent changes in brain activation patterns and/or dopamine metabolism, but the evidence is very limited and it remains unclear to what extent such changes may be therapeutic or disadvantageous. The striatum is of particular interest when studying stimulant treatment effects in ADHD. Reduced striatal volumes (Frodl & Skokauskas, 2012; Nakao et al., 2011), lower striatal activity during reward anticipation and higher striatal activity during outcome of reward (Aarts et al., 2015; Paloyelis et al., 2012; Von Rhein et al., 2015b) have repeatedly been found in ADHD. Moreover, the striatum is rich in dopamine transporters, an important molecular target of stimulant treatment. Hence, long-­‐term stimulant treatment effects may be expected to occur in the striatum. However, acute stimulant-­‐induced changes in activation patterns have also been reported in supplementary motor areas (SMA), frontal cortex, anterior and posterior cingulate cortex, and precuneus cortex (e.g., Peterson et al., 2009, Prehn-­‐Kristensen et al., 2011, Rubia et al., 2009). We investigated associations between lifetime stimulant treatment history and neural activity during reward processing, using magnetic resonance (MRI) data from a large observational study. An innovative data-­‐driven classification method was used to identify patient subgroups with distinct treatment trajectories. In our cohort, Groenman et al., (2015) found these trajectories to be clinically relevant for the development of substance use disorder (unpublished data). Moreover, treatment timing and dose have been found to moderate long-­‐term stimulant treatment effects in the rat brain (e.g., van der Marel et al., 2014). In prior work, our group showed higher striatal BOLD-­‐response to reward outcome in ADHD patients compared to controls (Von Rhein et al., 2015b). In the current study, we hypothesized that patients who had received more intense treatment would show reduced striatal BOLD-­‐
response (i.e., more similar to controls) to reward outcome compared to those who had received less intense treatment. Second, we hypothesized that between-­‐group differences in other brain regions, if any, would show a similar pattern. METHODS Participants Participants with ADHD were selected from the family-­‐based IMAGE-­‐
NeuroIMAGE cohort (2009-­‐2012) (von Rhein et al., 2015a). Children, adolescents, and young adults participated in diagnostic interviews, questionnaires, DNA collection, and an MRI session, taking place at two sites. Informed consent was signed by all participants ≥12 years old and all parents of participants <18 years old. The study 88 was approved by the local ethical committees of each participating site. Inclusion criteria were: IQ≥70, age 8-­‐30 years, no diagnosis of classical autism, learning difficulties, brain disorders, or genetic disorders, and no contra-­‐indication for MRI scanning. ADHD diagnosis (any type) was confirmed in accordance with the Diagnostic and Statistical Manual of Mental Disorders (DSM-­‐IV; ADHD, 2013), operationalized as six or more symptoms on the Schedule for Affective Disorders and Schizophrenia for School-­‐Age Children (K-­‐SADS; Kaufman et al., 1997) and t>63 on the Conners parent-­‐, teacher-­‐, and/or self-­‐rated ADHD scales (Conners et al., 1998a, 1998b and 1999), rated while participants were off-­‐medication. Five K-­‐SADS symptoms were sufficient for diagnosis in participants age 16 or older, in line with DSM-­‐5 revised criteria. The initial ADHD sample consisted of 269 participants. Functional MRI data were available for 124 patients (mean age=17.4 years, range 10-­‐
26 years). Control participants were required to have no scores in the (sub)clinical range on any of the ADHD rating scales or interviews, no current or past psychiatric diagnosis or treatment, and no first-­‐degree relative with ADHD. The initial control sample consisted of 187 participants. Functional MRI data was available for 97 controls (mean age=17.0 years, range 10-­‐23 years). Stimulant treatment History of psychoactive treatment was assessed using pharmacy prescription records containing delivery date, substance name, dose, quantity, and frequency of use for each delivery between date-­‐of-­‐birth and date-­‐of-­‐scan. In addition, patients and parents participated in face-­‐to-­‐face semi-­‐structured interviews to reconstruct lifetime treatment history. Self-­‐report data was highly compatible with data derived from pharmacies (data not shown), with reliability estimates similar as those reported by Kuriyan et al. (2014) Self-­‐report data was used only when pharmacy data was incomplete. Stimulant intake in mg (immediate-­‐ and extended-­‐release methylphenidate preparations, and dexamphetamine preparations) was reconstructed for each day between date-­‐of-­‐birth and date-­‐of-­‐scan. Daily intake in mg was averaged for every month of the participant’s life. Stimulant start age, stop age, and lifetime cumulative stimulant dose were calculated from this reconstruction. A smooth generalized additive model curve was fitted to each participant’s reconstruction, allowing estimation of three additional treatment parameters that were more sensitive to noise, i.e., treatment duration (estimated stop age minus estimated start age), treatment variability (standard deviation of the fitted curve), and the lifetime maximum dose. Treatment duration and cumulative stimulant dose were adjusted for current age. 89 Community detection algorithm The six stimulant treatment parameters (start age, stop age, total dose, estimated duration, estimated maximum daily dose, and estimated variability) were entered in an automated, optimization-­‐based, weight-­‐conserving community detection algorithm (Rubinov & Sporns, 2011). The algorithm categorizes participants into mutually exclusive communities (groups), segregating groups such that within-­‐
group positive/negative correlations are maximal while between-­‐group correlations are minimal. The modularity statistic Q (range 0-­‐1) quantifies the degree to which participants may be subdivided into clearly delineated groups. The algorithm terminates when Q no longer increases from one iteration to the next. Robustness of the optimal community structure was confirmed using non-­‐parametric bootstrap procedures (Supplement S1). TABLE 1. Stimulant treatment characteristics [mean(SD)] for participants in each treatment group. ‘early-­‐and-­‐intense’ ‘late-­‐and-­‐moderate’ ‘early-­‐and-­‐moderate’ (41.3%) (35.7%) (7.4%) Start age 6.89(1.29) 11.19(2.62) 8.25(1.48) Stop age 15.65(2.97) 16.37(3.12) 11.37(2.01) Treatment duration a 9.10(2.63) 3.99(2.47) 2.40(1.47) Variability 320.76(459.27) 110.42(154.94) 66.43(88.48) Cumulative dose in mg a 67480(55751) 38413(29986) 18437(17474) 47.37(23.36) 24.45(16.60) 17.53(12.61) Maximum daily dose in mg Classification was based on an age-­‐adjusted measure; the value reported here is calculated based on the non-­‐smoothed trajectories. a
The data-­‐driven classification method produces more reliable results in larger samples, hence all participants with ADHD were included in this step (n=269). Stimulant-­‐naive participants were a priori defined as a separate category (n=42, 15.1%). For stimulant-­‐treated participants, the optimal solution yielded three treatment groups (Q=0.580; Table 1). The first group (n=111, 41.3%, ‘early-­‐and-­‐
intense’) was characterized by early treatment onset, long duration, and a high maximum and total dose. The second group (n=96, 35.7%; ‘late-­‐and-­‐moderate’) was characterized by older age at treatment onset, shorter duration, and lower maximum and total dose. The third group (n=20, 7.4%; ‘early-­‐and-­‐moderate’) was characterized by early treatment onset, medium duration, and low maximum and total dose. As few 90 participants were classified to the ‘early-­‐and-­‐moderate’ group or were stimulant-­‐
naive, ‘early-­‐and-­‐intense’-­‐vs.-­‐‘late-­‐and-­‐moderate’ was our primary contrast of interest. As shown in Table 1, the ‘early-­‐and-­‐intense’ and ‘late-­‐and-­‐moderate’ groups differed in stimulant start age, treatment duration, variability, maximum dose, and total dose, but not in stop age. Reward task A modified version of the monetary incentive delayed task was performed in the scanner (Von Rhein et al., 2015b). Participants were asked to respond as quickly as possible to a target by pressing a button. Before this target, a cue indicated the possibility of gaining a reward after a button press within a given time-­‐window. Every trial ended with a feedback screen informing about the outcome of the current trial. Depending on the participant’s performance, the response-­‐window for a correct response was adapted in the next trial, resulting in an expected hit-­‐rate of 33%. The experiment lasted 12 minutes, and a total of €5 could be gained. At the end of the experiment, the awarded money was paid to the participant. Compared with the original task, our version differed on two main aspects: hit-­‐rate (33% versus 66%) and reward magnitude (€0.20 versus $5). The rationale behind these adaptations was firstly to increase the demands of the task with stronger task engagement as a result. Secondly, our adaptations aimed at meeting the practical constraints of our study. Considering that we limited ourselves to rewarded and neutral conditions, rewarding participants according to the original task parameters would have led to disproportionate monetary rewards (approximately €80), which was a concern for us and our ethical review board. Reaction time reward sensitivity was calculated as the mean reaction time across non-­‐rewarded trials minus the mean reaction time across rewarded trials, with higher values indicating higher sensitivity to reward. Functional MRI processing and analyses Acquisition parameters, preprocessing steps, and first-­‐level analyses were identical to those in our previous publication (Von Rhein et al., 2015b; Supplement S2). Second-­‐level analyses for each task condition (reward anticipation and outcome) comprised both region of interest (ROI) and whole-­‐brain analyses in FMRIB Software Library (FSL; Smith et al., 2004). First, main task effects were identified in a one-­‐
sample t-­‐test, with scanner, age, gender, and three motion parameters as regressors of no interest. For the ROI analyses, average parameter estimate was extracted for each participant from the (warped) task-­‐activated voxels within a binary mask of the striatum (caudate, putamen, and accumbens). In a linear mixed effect regression 91 model in SPSS (IBM, 2013), striatal activation was predicted from treatment group (primary contrast: ‘early-­‐and-­‐intense’-­‐vs.-­‐‘late-­‐and-­‐moderate’; secondary contrasts: ‘stimulant-­‐naive’-­‐vs.-­‐‘early-­‐and-­‐intense’, ‘stimulant-­‐naive’-­‐vs.-­‐‘late-­‐and-­‐moderate’, ‘stimulant-­‐naive’-­‐vs.-­‐‘early-­‐and-­‐moderate’, ‘early-­‐and-­‐moderate’-­‐vs.-­‐‘early-­‐and-­‐
intense’, ‘early-­‐and-­‐moderate’-­‐vs.-­‐‘late-­‐and-­‐moderate’). Gender, scanner, age, and age2 (to account for non-­‐linear developmental trajectories of reward-­‐related striatal activation) were added as covariates, along with a random intercept per family to account for relatedness within the sample. Alpha was adjusted for analyzing one primary and five secondary contrasts and two task conditions (α=0.05/6/2=0.004). Normalized first-­‐level b-­‐maps were also entered into whole-­‐brain second-­‐level mixed effect analyses. Treatment group was entered as a predictor along with scanner, gender, age, and three movement parameters (ZVOXEL>2.3; αCLUSTER=0.004). Structural MR images were also acquired, to assess structural correlates of long-­‐term functional changes, if any (Supplement S3). Follow-­‐up and sensitivity analyses For each whole-­‐brain significant cluster, average parameter estimate was extracted per participant for follow-­‐up analyses in SPSS. Treatment groups were data-­‐
driven, hence not matched with regard to clinical and demographic variables. Potential confounders other than age and gender (i.e., IQ, SES, ADHD symptoms, ADHD-­‐type, comorbidity, and history of non-­‐stimulant psychoactive medication) were added to the model. Moreover, analyses were repeated within one-­‐to-­‐one age-­‐, gender-­‐, and ADHD symptom count-­‐matched subsamples (n=25 per group). To exclude acute withdrawal/rebound effects, each significant effect was re-­‐
estimated separately for participants who were on active stimulant treatment within two weeks prior to scanning and those who had ceased treatment more than two weeks prior to scanning. Main reward task effects and case-­‐control differences in the current cohort have previously been reported (Von Rhein et al., 2015b), hence are not addressed here. For reference only, the control sample mean for each outcome measure was estimated in a covariate-­‐only model. RESULTS Sample characteristics The ADHD sample consisted of 83 males (66.9%) and 41 females (33.1%), with an average age of 17.4 years (SD=3.0, range 10-­‐26 years; Table 2). Of those, 51 92 TABLE 2. Characteristics of the ‘early-­‐and-­‐intense’ and ‘late-­‐and-­‐moderate’ treatment groups [N(%), unless otherwise specified]. ADHD E&I L&M N=124 N=51 N=49 Male 83(66.9) 46(90.2) 26(53.1) Chi2=17.1* Site Nijmegen 79(63.7) 32(62.7) 36(73.5) Chi2=1.3 Age M(SD) 17.4(3.0) 17.1(2.4) 18.1(3.0) F=3.2 IQ M(SD)a 99.2(15.0) 98.6(14.4) 100.2(14.6) F=0.3 Current stimulant users 46(42.2) 30(58.8) 13(26.5) Chi2=10.6* Symptoms of inattention M(SD) 7.2(1.8) 7.8(1.3) 6.6(2.0) F=11.5* Symptoms of hyperactivity/impulsivity M(SD) 6.0(2.4) 6.7(2.3) 5.7(2.2) F=5.4 ADHD-­‐type Inattentive 56(45.2) 18(35.3) 24(49.0) Chi2=1.9 Hyperactive/impulsive 17(13.7) 6(11.8) 9(18.4) Chi2=0.9 Combined 51(41.1) 27(52.9) 16(32.7) Chi2=4.2 Comorbidity ODD-­‐CD 30(24.2) 17(33.3) 8(16.3) Chi2=3.9 Tic disorder 1(0.8) 1(2.0) 0(0.0) Chi2=1.0 Anxiety/depression 3(2.4) 1(2.0) 1(2.0) Chi2<0.1 24(19.4) 8(15.7) 14(28.6) Chi2=2.4 Substance use disorderb Non-­‐stimulant medication Atomoxetine 19(15.3) 10(19.6) 8(16.3) Chi2=0.2 Antipsychotics 23(18.5) 16(31.4) 5(10.2) Chi2=6.8 Anxiolytics 8(6.5) 3(6.1) 3(6.1) Chi2<0.1 Antidepressants 8(6.5) 4(7.8) 2(4.1) Chi2=0.6 E&I, early-­‐and-­‐intense; L&M, late-­‐and-­‐moderate; ODD-­‐CD, oppositional defiant disorder-­‐conduct disorder; a estimated based on the ‘vocabulary’ and ‘block design’ subtests of the Wechsler intelligence scales for children/adults. b assessed approximately two years prior to participation in the current study. *p<0.004. 93 participants were assigned to the ‘early-­‐and-­‐intense’ treatment group (46.8%), and 49 to the ‘late-­‐and-­‐moderate’ group (45.0%). Compared to the ‘late-­‐and-­‐moderate’ treatment group, the ‘early-­‐and-­‐intense’ group contained more males and more participants on active stimulant treatment, and had more inattention problems. The two groups did not differ with regard to age, socio-­‐economic status, IQ, ADHD-­‐type, hyperactivity/impulsivity symptoms, comorbidity, or history of non-­‐stimulant medication. The control sample (n=97; age M=17.0 years, SD=2.9, range 10-­‐23 years) contained fewer males compared to the ADHD sample (44.3% vs. 66.9%; p=0.001). For the stimulant-­‐naive (n=15) and ‘early-­‐and-­‐moderate’ (n=9) groups, see Supplement S4. FIGURE 1. Striatal activation during reward anticipation in yellow-­‐red, and during reward outcome in blue-­‐light blue, across all participants. Reward processing The striatum was activated by both task conditions (Figure 1). There were no differences in striatal BOLD-­‐response between the ‘early-­‐and-­‐intense’ and ‘late-­‐and-­‐
moderate’ treatment groups during reward anticipation (MEARLY&INTENSE=360.7, MLATE&MODERATE=394.8, MCONTROL=299.7, p=0.784), nor during reward outcome (MEARLY&INTENSE=362.1, MLATE&MODERATE=677.5, MCONTROL=414.9; p=0.180). Whole-­‐brain functional MRI analyses did not yield any clusters of significant difference between the ‘early-­‐and-­‐intense’ and the ‘late-­‐and-­‐moderate’ groups during reward anticipation. In the reward outcome condition, the ‘late-­‐and-­‐moderate’ group showed lower activity compared to the ‘early-­‐and-­‐intense’ group in a cluster located in the SMA, extending into the dorsal anterior cingulate cortex (dACC) and paracingulate gyrus (Figure 2; MEARLY&INTENSE=635.1, MLATE&MODERATE=-­‐813.9, MCONTROL=35.5, cluster size=8696 mm3, B=-­‐1449.0, pCLUSTER<0.001). Gender (B=964.6, p=0.014), scanner (B=179.0, p=0.604), age (B=-­‐285.8, p=0.087), and age2 (B=153.8, p=0.087) were not associated with activation in this cluster, nor were any of the additional covariates when added to the model while the effect of treatment history remained unchanged. Moreover, the pattern was consistently observed in past 94 users (MEARLY&INTENSE=374.9, MLATE&MODERATE=-­‐687.3) and current users (MEARLY&INTENSE=785.0, MLATE&MODERATE=-­‐1323.6), and within the age-­‐, gender-­‐, and symptom-­‐matched subsamples (MEARLY&INTENSE=721.5, MLATE&MODERATE=-­‐395.5). There was no behavioral (i.e., reaction time) difference in reward sensitivity between the ‘early-­‐and-­‐intense’ and ‘late-­‐and-­‐moderate’ groups (MEARLY&INTENSE= 35.0ms, MLATE&MODERATE=29.4ms, MCONTROL=25.7ms, p=0.559). Moreover, reaction time reward sensitivity was not associated with striatal activity during reward anticipation (Pearson r=0.173, p=0.055) or reward outcome (Pearson r=0.014, p=0.879), nor with activity within the SMA/dACC cluster (Pearson r=0.177, p=0.050). There were no structural brain differences between the two groups. For findings involving the ‘early-­‐and-­‐moderate’ and stimulant-­‐naive groups, see supplement S4. FIGURE 2. Higher activation in the ‘early-­‐and-­‐intense’ group compared to the ‘late-­‐and-­‐moderate’ group during reward outcome. DISCUSSION In a large sample of children, adolescents and young adults with ADHD, we investigated whether characteristics of stimulant treatment history were associated with brain activation patterns during reward processing while off medication. Stimulant treatment history was not associated with BOLD-­‐response to reward anticipation or outcome in the striatum. In the SMA/dACC, individuals with a history of moderate treatment showed lower activity during reward outcome compared to those with a history of intense treatment. While activity in the moderately treated group was reduced compared to controls, activity in the intensely treated group was higher compared to controls. Our findings thus suggest compensatory SMA/dACC recruitment in individuals with a history of intense stimulant treatment. The effect is likely driven by treatment duration and dose rather than recency of treatment discontinuation, since stop age did not differ between the two groups. Higher striatal BOLD-­‐response to reward outcome has consistently been reported in ADHD (e.g., Paloyelis et al., 2012; Von Rhein et al., 2015b). As such changes have been shown to disappear after stimulant administration (Aarts et al., 95 2015; Knutson et al., 2004), we had hypothesized that participants with a history of intense treatment would show lower striatal BOLD-­‐response to reward outcome compared to those with a history of less intense treatment. We found no evidence for such an effect. Moreover, there was no association between treatment history and striatal activity during reward anticipation. Our findings may indicate that the acute changes in striatal activity in response to stimulants do not translate into lasting functional changes in this region during reward processing. This finding is consistent with Stoy et al. (2011) who, in a small adult sample, also reported no changes in striatal activation during reward outcome after childhood stimulant treatment. We found a large cluster of lower activity during reward outcome in the moderately treated subgroup compared to the intensely treated subgroup, located in the bilateral SMA and dACC, extending into the precuneus and posterior cingulate cortex. Dorsal and mid-­‐cingulate regions project to the ventral striatum, and are important for monitoring incentive-­‐based behavioral responses (Haber & Knutson, 2010; Shima et al., 1998). Hypo-­‐activation has previously been reported in medication-­‐naive ADHD patients during reward outcome (Carmona et al., 2012). Acute stimulant effects in the SMA/dACC during reward processing have been reported as well (Aarts et al., 2015), although most fMRI studies of reward reported no acute stimulant effects in this region (e.g., Knutson et al., 2004; Rubia et al., 2009). Lower activity in the SMA/dACC in ADHD patients has also been associated with cognitive processes other than reward processing. Higher SMA/dACC activation may represent recruitment of a cognitive process enhancing feedback-­‐based decision-­‐
making, even when a motor response is not required (Ullsperger & Von Cramon, 2003; Vassena et al., 2015), as was the case in the reward outcome phase of our task. ADHD patients have shown lower SMA activity when selection of a non-­‐habitual response was required (Hart et al., 2013; Suskauer et al., 2008). Higher SMA/dACC, PCC, and precuneus activity has been reported after a single dose of stimulants during tasks requiring feedback-­‐based modulation of motor responses (Bush et al., 2008; Cubillo et al., 2012b; Pliszka et al., 2007), but acute effects in the opposite direction have also been reported (Ivanov et al., 2014; Rubia et al., 2014). Enhanced cognitive decision-­‐making upon reward in intensely treated individuals is consistent with the lower rate of SUD in this group (Groenman, unpublished data, 2015), although the difference in SUD rate in the current (smaller) fMRI sample was not significant. Summarizing, higher SMA/dACC activity may indicate enhanced cognitive decision-­‐
making following reward after early and high-­‐dose stimulant treatment. Note that this proposition is not supported in behavioral data, as our paradigm required no response following reward outcome. Alternatively, higher SMA/dACC activity may represent increased salience network activity, enhancing attention allocation to emotional, rewarding, or 96 surprising events (Menon, 2015). Stimulant-­‐induced improvement in cognitive performance has been shown to be mediated by enhanced salience (Jolles et al., 2011; Ter Huurne et al., 2015). Stimulant treatment history may be associated with greater focus on the task. Yet, increased task focus may be expected to occur throughout the task as opposed to during the outcome phase only, and may result in improved task performance which we did not observe. Finally, higher SMA/dACC activity may entail enhanced ‘readiness to act’ upon reward outcome, as the SMA is embedded in the task-­‐positive motor network (Volkow et al., 2004). However, we found no association between SMA/dACC activation and reaction times. The current study has several strengths. First, only a handful of prior studies investigated functional rather than anatomical long-­‐term neural changes in relation to stimulant treatment in ADHD. Of those, the current sample is by far the largest. Second, the data-­‐driven classification of participants with ADHD based on multiple treatment characteristics is novel and clinically relevant. The current study has limitations as well. Long-­‐term treatment effects can only be studied observationally. Although findings have been statistically adjusted for group differences, confounding by indication could not be excluded. Moreover, few participants were stimulant-­‐naïve (in accordance with high prescription rates), and data-­‐driven classification of stimulant-­‐treated participants yielded unbalanced groups. This allowed powerful analysis of participants in the two largest groups, but restricted analyses of stimulant-­‐
naive participants and those with early-­‐and-­‐moderate treatment. Finally, no data was collected regarding behavioral treatment, which, according to guidelines, should be offered in conjunction with pharmacological treatment; hence, pharmacological and behavioral treatment effects cannot be distinguished in our study. The recruitment of compensatory cognitive control areas may reflect the application of cognitive strategies learned during behavioral treatment. We conclude that ADHD patients with a history of early-­‐onset high-­‐dose stimulant treatment showed more SMA/dACC activation during reward outcome, compared to those with a history of late-­‐onset moderate-­‐dose stimulant treatment. Higher SMA/dACC activity may represent a compensatory mechanism of enhanced higher-­‐level processing of reward information in the intensely treated group. Stimulant treatment history was not associated with striatal BOLD-­‐response to reward processing. Understanding long-­‐term risk and benefits of stimulant treatment could be further enhanced by evaluating functional rather than neuroanatomical brain changes in future studies. 97 SUPPLEMENTAL INFORMATION
S1. Robustness of the community detection algorithm The community detection algorithm was rerun one thousand times, while randomly selecting 227 times one participant with replacement for each run (bootstrapping). Participants could be duplicated within a bootstrapped sample. The algorithm produced a three-­‐class solution in 793 runs (79.3%), with an average modularity Q of 0.58 (SD=0.02; confidence interval=0.579-­‐0.582). When a three-­‐class solution was found, the largest class contained on average 46.0% of participants (SD=3.5%; confidence interval=45.7% -­‐ 46.3%), while the second largest class contained on average 40.9% of participants (SD=3.0%; confidence interval=40.7% -­‐ 41.1%) and the smallest class contained on average 13.2% of participants (SD=5.4%; confidence interval=12.8% -­‐ 13.5%). Thus, the average distribution of classes across 1000 runs strongly resembled the solution reported in the paper. Moreover, the 95% confidence intervals are narrow and standard deviations are low, indicating that the percentage of participants per class tend to be stable across runs. As expected, the percentage of participants in the smallest class is most susceptible to random variations (SD=5.4%). The algorithm produced a four-­‐class solution in 133 runs (13.1%). The four classes contained on average 43.0% (SD=3.4), 38.7% (SD=3.3), 11.6% (SD=3.7) and 6.6% (SD=3.2) of participants, respectively. In 76 runs, the algorithm produced a two-­‐class solution (7.6%), with classes containing on average 54.1% (SD=2.9) and 45.9% (SD=2.9) of participants, respectively. Thus, the three-­‐class solution reported in the paper resembles the distribution of participants across the three largest classes of the four-­‐
class solution, as well as the distribution of participants in the two-­‐class solution. This underlines the stability of the three-­‐class model. S2. Functional MRI -­‐ acquisition and preprocessing MRI data was acquired on two Siemens 1.5 Tesla scanners (Erlangen, Germany) with matched head coils and acquisition parameters. Participants were randomly assigned a combination of three or four functional acquisitions (a diffusion weighted scan, resting-­‐state scan, reward task, working memory task, and/or response inhibition task). Reward task functional MRI data was thus available for a random subset of participants (nADHD=124, nHC=97). Whole brain functional imaging was performed using a gradient-­‐echo echo-­‐planar scanning (EPI) sequence (37 axial slices, TR=2340ms, TE=40ms, voxel size=3.5x3.5x3.0mm, inter-­‐slice gap=0.5mm, FOV=224mm, FA=90°). Participants with more than three head movements of ≥4mm during the task were excluded. Functional MRI preprocessing steps included spatial realigning, nuisance regression, spatial smoothing at FWHM=6mm. First-­‐level statistical parametric maps (b-­‐maps) were estimated for each participant, including 6 regressors of interest (onset times of non-­‐rewarded and rewarded cues, hits, and misses), and 6 regressors of no interest (onset times of non-­‐rewarded and rewarded targets, onset times of targets, cues, and outcomes followed by incorrect responses, and a motion regressor identifying and excluding events affected by excessive movement). Participants with less than five occurrences of one or more event types (rewarded hits, rewarded misses, non-­‐rewarded hits, or non-­‐rewarded misses) were excluded. All regressors and their temporal derivatives were convolved with a canonical hemodynamic response function. For reward anticipation, response maps for rewarded cues were contrasted with response maps for non-­‐rewarded cues. Activation during reward outcome was assessed by the interaction of accuracy (hits versus misses) and reward (rewarded versus non-­‐rewarded trials). First-­‐level b-­‐maps were registered using non-­‐linear transforms to a study-­‐specific template. S3. Structural MRI – acquisition, processing, and analyses 98 The MRI session included at least one T1-­‐weighted structural acquisition (3D MP-­‐RAGE; 176 sagittal slices, TR=2730ms, TE=2.95ms, voxel size=1x1x1mm, FOV=256mm, FA=7°, parallel imaging by generalized auto-­‐calibrating partially parallel acquisition [GRAPPA]). For each participant, the structural acquisition of highest quality was selected by visual inspection, accepting only scans with no or minimal distortions. Structural MRI analyses were performed in the initial samples (nADHD=269; nHC=187), to increase power. Total striatal volume (sum of left and right putamen, caudate, and accumbens) was calculated using FSL FIRST with default settings (Patenaude et al., 2011). ROI analyses were performed in SPSS, predicting striatal volume from treatment group, with covariates gender, scanner, age, age2, total brain volume (TBV, calculated with the VBM8 toolbox in SPM; Ashburner & Friston, 2005), and a random intercept per family. Whole-­‐brain structural analyses included volumetric analyses of additional subcortical structures (globus pallidus, amygdala, hippocampus, and thalamus; sum of left and right hemisphere; α=0.008/4=0.002) and vertex-­‐wise analysis of cortical thickness and surface area. Freesurfer with default settings was used to reconstruct the cortical surface of each participant (Dale et al., 1999; Fischl et al., 1999; Fischl & Dale, 2000). Statistical maps were computed for the ‘early-­‐and-­‐intense’ vs. ‘late-­‐and-­‐moderate’ contrast and the five secondary contrasts, with age, gender, and scanner as covariates, and age2 as an additional per-­‐
vertex-­‐regressor. Normalized statistical maps were thresholded per vertex (Z > 2.3). Cluster-­‐level thresholding was based on Monte Carlo simulation testing, with αCLUSTER adjusted for testing six contrasts and two hemispheres (α=0.004). For each significant cluster, a random intercept per family was added to the model in SPSS. S4 -­‐ Secondary contrasts FIGURE S2.1. Whole-­‐brain functional and structural MRI results for secondary contrasts. Left panel: significant between-­‐group differences in BOLD-­‐response during reward anticipation. I: ‘early and moderate’ < stimulant naïve in red, ‘early and moderate’ < ‘late and moderate’ in yellow, ‘early and moderate’ < ‘early and intense’ in green. II: ‘early and moderate’ < stimulant naïve in purple, ‘late and moderate’ < stimulant naïve in blue. Right panel (III): cluster of decreased cortical surface area in the ‘early and moderate’ compared to the ‘early and intense’ treatment group. Follow-­‐up analyses showed that surface area in this cluster was also associated with gender (data not shown). 99 TABLE S4.1. Sample characteristics and significant confounders in the five secondary contrasts. L&M N=49 E&I N=51 E&M N=9 NAIVE N=15 Male N(%) I, II 26 (53.1) 46 (90.2) 4 (44.4) 7 (46.7) Nijmegen N(%) III 36 (73.5) 32 (62.7) 6 (66.7) 5 (33.3) Age in years M(SD) IV 18.1 (3.0) 17.1 (2.4) 14.9 (2.5) 17.5 (4.0) Estimated IQ M(SD) a 100.2 (14.6) 98.6 (14.4) 101.1 (15.8) 97.3 (18.8) SES M(SD) 11.4 (2.2) 11.6 (2.1) 10.6 (1.7) 12.1 (2.5) Current stimulant users N(%) 13 (26.5) 30 (58.8) 3 (33.3) 0 (0.0) Inattention sympt. M(SD) 6.6 (2.0) 7.8 (1.3) 7.8 (0.8) 6.7 (2.1) Hyperactive/impulsive sympt. M(SD) II 5.7 (2.2) 6.7 (2.3) 5.4 (2.0) 4.5 (3.0) Inattentive N(%) 24 (49.0) 18 (35.3) 6 (66.7) 8 (53.3) Hyperactive/impulsive N(%) 9 (18.4) 6 (11.8) 0 (0.0) 2 (13.3) Combined N(%) 16 (32.7) 27 (52.9) 3 (33.3) 5 (33.3) ODD-­‐CD N(%) 8 (16.3) 17 (33.3) 2 (22.2) 3 (20.0) Tic disorder N(%) 0 (0.0) 1 (2.0) 0 (0.0) 0 (0.0) Anxiety/depression N(%) 1 (2.0) 1 (2.0) 1 (11.1) 0 (0.0) 14 (28.6) 8 (15.7) 1 (11.1) 1 (6.7) Atomoxetine N(%) 8 (16.3) 10 (19.6) 1 (11.1) 0 (0.0) Atypical antipsychotics N(%) 5 (10.2) 16 (31.4) 2 (22.2) 0 (0.0) Anxiolytics N(%) 3 (6.1) 3 (6.1) 1 (11.1) 1 (8.3) Antidepressants N(%) 2 (4.1) 4 (7.8) 0 (0.0) 1 (8.3) ADHD type Comorbidity Substance use disorder N(%) b Non-­‐stimulant treatment L&M, late & moderate; E&I, early & intense; E&M, early and moderate; ODD-­‐CD, oppositional defiant disorder-­‐conduct disorder; SES, socio-­‐economic status; a estimated based on the ‘vocabulary’ and ‘block design’ subtests of the Wechsler intelligence scales for children/adults. b assessed approximately two years prior to participation in the current study. Significant differences between groups (α<0.004): I early & intense vs. early & moderate; II naïve vs. early & intense; III naïve vs. late & moderate; IV early & moderate vs. late & moderate. 100 TABLE S4.2. Mean reaction time reward sensitivity and striatal region-­‐of-­‐interest results for each secondary contrast. RT-­‐RS in ms. BOLDANTICIPATION BOLDOUTCOME Volume (in mL) M M M M NAÏVE 35.5 540.7 671.8 19.8 L&M 29.4 394.8 362.1 20.1 E&M 45.1 180.2 1035.7 20.1 E&I 35.0 360.7 677.5 20.3 B p B p B p B p NAIVE vs. L&M 6.1 0.647 145.8 0.403 309.7 0.345 -­‐0.3 0.253 NAIVE vs. E&M -­‐9.6 0.606 360.4 0.140 -­‐363.8 0.427 -­‐0.3 0.511 NAIVE vs. E&I 0.5 0.970 180.0 0.332 -­‐5.7 0.987 -­‐0.5 0.098 E&M vs. L&M 15.6 0.331 -­‐214.6 0.310 673.6 0.091 -­‐0.1 0.893 E&M vs. E&I 10.1 0.535 -­‐180.4 0.398 358.1 0.372 -­‐0.2 0.567 RT-­‐RS, reaction time reward sensitivity; L&M, late & moderate; E&M, early & moderate; E&I, early & intense; M, estimated marginal means in a model with covariates gender, scanner, age, and age2 (and total brain volume for volumetric analyses). 101 102 L ACC a ACC, OFC a Precuneus Sup. parietal None None Inf. temporal Hippocampus Amygdala Thalamus Gl. Pallidus BOLD OUTCOME Cortical thichness Surface area Volume 3.7 16.7 2.7 7.9 1300.9 -­‐ -­‐ 54.2 255.6 322.7 123.6 172.3 MNAIVE 3.7 16.7 2.6 7.8 1385.8 -­‐ -­‐ -­‐1094.1 -­‐366.1 -­‐158.2 -­‐244.0 65.3 ML&M 3.8 16.5 2.7 7.6 1171.0 -­‐ -­‐ -­‐1061.3 -­‐1294.2 -­‐1196.9 -­‐1452.7 -­‐1510.0 ME&M 3.8 17.0 2.7 7.9 1395.4 -­‐ -­‐ -­‐539.8 -­‐102.0 -­‐224.7 -­‐101.1 -­‐214.6 ME&I -­‐ -­‐ -­‐ -­‐ None None None -­‐224.4 -­‐ -­‐ -­‐1148.3 -­‐1549.8 -­‐1519.6 -­‐1351.6 -­‐1575.3 B None E&M < E&I None None L&M < NAÏVE E&M < NAÏVE E&M < NAIVE E&M < E&I E&M < L&M Sign. contrast -­‐ -­‐ -­‐ -­‐ 995 mm2 -­‐ -­‐ 7128 mm3 16400 mm3 9032 mm3 7696 mm3 6408 mm3 Cluster size -­‐ -­‐ -­‐ -­‐ 0.00340 -­‐ -­‐ 0.00007 0.00001 0.00074 0.00157 0.00399 pCLUSTER L&M, late & moderate; E&M, early & moderate; E&I, early & intense; OFC, orbitofrontal cortex; L, left hemisphere; R, right hemisphere; B, bilateral; M, estimated marginal means in a model with covariates gender, scanner, age, age2 (and total brain volume for volumetric analyses). a clusters partially overlap B B B B L B B B B ACC a BOLD ANTICIPATION Brain region Condition/measure TABLE S4.3. Whole-­‐brain functional and structural MRI results for each secondary contrast 103 104 Chapter 6 AGE AND DRD4 GENOTYPE MODERATE
ASSOCIATIONS BETWEEN STIMULANT TREATMENT
HISTORY AND CORTEX STRUCTURE IN ADHD
Published as: Schweren LJS, Hartman CA, Heslenfeld DJ, Groenman A, Franke B, Oosterlaan J, Buitelaar JK, Hoekstra PJ. Age and DRD4 genotype moderate associations between stimulant treatment history and cortex structure in ADHD. J Am Acad Child Adolesc Psychiatry. 2016;55(10):577-­‐585 105 ABSTRACT
Objective: Attention-­‐deficit/hyperactivity disorder (ADHD) has been associated with dopaminergic imbalance and subtle volume reductions in the brain. Stimulants acutely enhance dopaminergic neurotransmission. Long-­‐term effects of chronic manipulation of the dopaminergic system on brain structure remain poorly understood; they could be beneficial or unfavorable, and may be moderated by common genetic variants and/or age. Methods: In a large observational cohort study (n =316), we evaluated the effects of cumulative stimulant treatment, genotype (for DAT1 haplotype and DRD4 variants), and treatment-­‐by-­‐genotype interactions on striatal, frontal, and hippocampal volumes, as well as their interactions with age. Results: We found no main effects of treatment. Associations between treatment and bilateral frontal and left hippocampal volume depended on DRD4 genotype and age. At younger age and lower treatment-­‐levels, but not at younger age and higher treatment levels, carriers of the DRD4 7R-­‐allele showed decreased frontal cortex volumes. At older age, both carriers and non-­‐carriers showed lower frontal volumes irrespective of treatment history. Left hippocampal volume was similar to controls at average treatment levels, and increased with treatment only in carriers of the DRD4 risk allele and at younger age. No interaction effects were found in the striatum. Conclusions: Carriers of the DRD4 risk allele may at younger age be sensitive to cortical remodeling after stimulant treatment. The cross-­‐sectional nature of our study warrants cautious interpretation of age effects. Our findings, although of small effect size, may ultimately contribute to optimal care for individuals with ADHD. ADHD
106 INTRODUCTION Attention-­‐deficit/hyperactivity disorder (ADHD) has been associated with widespread subtle changes in brain structure. Total gray matter volume is reduced by 2-­‐3% in children with ADHD compared to typically developing children, with more pronounced reduction and atypical age-­‐related changes in the frontal-­‐striatal system (Frodl & Skokauskas, 2012; Greven et al., 2015; Nakao et al., 2011; Semrud-­‐Clikeman et al., 2006). The striatum and its frontal connections are rich in dopaminergic neurons, and ADHD symptoms are thought to, at least partially, stem from dopaminergic and noradrenergic imbalances (Swanson et al., 2007b). Stimulants such as methylphenidate enhance dopaminergic and noradrenergic neurotransmission by binding to the dopamine transporter, thereby inhibiting presynaptic dopamine reuptake, and increasing extracellular dopamine availability (Volkow et al., 2001). Long-­‐term effects of stimulant treatment on the developing brain remain poorly understood. Although several studies have suggested fewer structural abnormalities in individuals with ADHD after long-­‐term stimulant treatment (e.g. Semrud-­‐Clikeman et al., 2006; Shaw et al., 2009), findings are equivocal. Meta-­‐
analyses found larger (thus more normative) striatal volumes in studies including a higher percentage of stimulant-­‐treated patients compared to studies with lower percentages (Frodl & Skokauskas, 2012; Nakao et al., 2011). However, recent large-­‐
scale original studies did not find evidence of structural normalization in the striatum (Greven et al., 2015; Shaw et al., 2014). In the frontal cortex, disproportionate cortical thinning has been found in non-­‐treated children but not in stimulant-­‐treated children with ADHD (Shaw et al., 2009), while others have found reduced middle frontal cortex volumes in stimulant-­‐treated compared to stimulant-­‐naive patients (Villemonteix et al., 2015). Prior analyses of the current sample found no treatment effects on frontal cortical thickness (Schweren et al., 2015a). Thus, conclusive evidence of long-­‐term treatment effects on frontal-­‐striatal brain structures is missing. Genetic make-­‐up may predispose potential brain changes after stimulant treatment. The 3'-­‐untranslated region (3’UTR) of the gene encoding the dopamine transporter (SLC6A3/DAT1) contains a variable number of tandem repeat (VNTR) polymorphism influencing presynaptic dopamine transporter density, especially in the striatum where gene expression is high. The 9-­‐ and 10-­‐repeat alleles are most frequently encountered in the population. In children and adolescents, the 10-­‐repeat allele has been associated with increased risk of ADHD (Faraone et al., 2014), smaller striatal volumes (Shook et al., 2011), and distinct striatal activity patterns (Bédard et al., 2010; Durston et al., 2008), but not with clinical treatment response (Brookes et al., 2006; Contini et al., 2013). Recent studies have performed association analyses based on a haplotype of the 3’UTR VNTR and a second VNTR of the DAT1 gene located 107 in intron 8 (Brookes et al., 2006). The 10-­‐6 haplotype (10-­‐repeat allele in the 3’UTR VNTR, 6-­‐repeat allele in the intron 8 VNTR) has been identified as the risk haplotype for ADHD in children and adolescents (Bellgrove et al., 2009), whereas the 9-­‐6 haplotype has been associated with adult ADHD (Franke et al., 2010). Associations between the DAT1 haplotype, stimulant treatment, and brain structure have not yet been investigated. A second dopaminergic gene, the dopamine receptor D4 (DRD4) gene, encodes the postsynaptic dopamine D4 receptor, and is highly expressed in the frontal cortex and hippocampus (Defagot et al., 1997, Hawrylycz et al., 2012). The 7-­‐repeat allele of a VNTR in exon 3 (DRD4 7R) has been identified as the risk allele for ADHD, but has also been associated with better clinical outcome in late adolescence (Shaw et al., 2007b). In ADHD-­‐enriched samples, carriers of the 7-­‐repeat allele have shown reduced frontal cortex volume and thickness (Monuteaux et al., 2008; Shaw et al., 2007b). Moreover, DRD4 genotype modulated prefrontal cortex activation during various tasks (Gillsbach et al., 2012; Mulligan et al., 2014). In the current sample, DRD4 genotype and social environment together, but not DRD4 genotype alone, influenced prefrontal cortex activation during response inhibition (Richards et al., 2016). It has been suggested that the DRD4 polymorphism may be linked to attention problems (Kebir & Joober, 2011). Most treatment studies failed to predict clinical treatment response from DRD4 7R-­‐carriership (e.g., Contini et al., 2012; Kooij et al., 2008), although modest genotype-­‐by-­‐dose interaction effects have been reported (e.g., Froehlich et al., 2011; McGough et al., 2009). With DAT1 and DRD4 genes affecting presynaptic dopamine transporters in the striatum and postsynaptic dopamine receptors in the frontal cortex and hippocampus, respectively, inter-­‐individual genetic differences may predispose treatment effects in these brain regions. Pharmacological neuroimaging studies have shown different acute striatal responses to methylphenidate in individuals with different DAT1 3’UTR genotypes (Aarts et al., 2015; Kasparbauer et al., 2015). Furthermore, the dopaminergic system undergoes changes during development, hence long-­‐term treatment and genetic effects on brain structure may be different at different ages. Rapidly developing brain regions are particularly sensitive to external influences such as stimulant exposure (Andersen & Navalta, 2011). Support for age-­‐
dependent long-­‐term stimulant treatment effects comes from animal studies showing striatal volume reduction and hippocampal shape deformations after chronic juvenile exposure but not following treatment in adulthood (Martins et al., 2006; Van Der Marel et al., 2015). Long-­‐term structural changes after stimulant treatment may reflect dopamine-­‐dependent long-­‐term plasticity, a process of structural remodeling to which the hippocampus is known to be particularly sensitive (Jay, 2003). Thus, 108 differential neural susceptibility to acute methylphenidate effects may translate into differential sensitivity to long-­‐term stimulant treatment effects on brain structure. Identifying sources of neural sensitivity to long-­‐term treatment effects is important, and may ultimately influence therapeutic decisions. Here, we investigated associations between stimulant treatment, genetic predispositions, age, and brain structure. We hypothesized that stimulant treatment would be associated with larger (more normative) striatal volume, and that this association would be more pronounced in DAT1 10-­‐6 risk haplotype carriers and at younger age. Second, we hypothesized that stimulant treatment would be associated with larger frontal and hippocampal volume, especially in DRD4 7R-­‐allele carriers and at younger age. We investigated these hypotheses in a large cross-­‐sectional sample of children, adolescents and young adults with ADHD. The current paper adds to prior studies of our group that focused on case-­‐control differences in brain structure (Greven et al., 2015; Schweren et al., 2015a). Individual differences in neural sensitivity to long-­‐term treatment effects, due to age and/or genetic makeup, have not been addressed in prior studies in our sample. METHODS Participants Participants with ADHD (n=316, mean age=17.2 years, 69.3% male) and control participants (n=187, mean age=16.5 years, 52.4% male) were selected from the Dutch family-­‐based follow-­‐up phase (NeuroIMAGE) of the International Multisite ADHD Genetics (IMAGE) study. The protocol included diagnostic interviews, questionnaires for participants, parents, and teachers, DNA collection, and a magnetic resonance imaging (MRI) session, taking place at two testing sites in The Netherlands (Amsterdam and Nijmegen). Informed consent was signed by participants ≥ 12 years and parents of participants < 18 years. The study was approved by the ethical committee of each site. ADHD diagnosis and type (inattentive, hyperactive/impulsive, or combined type), ADHD severity, and axis-­‐I comorbidity were obtained from a diagnostic interview (Kaufman et al., 1997) and Conners ADHD questionnaires (Conners et al., 1998a, 1998b, and 1999), rated while participants were off-­‐
medication. Controls were required to have no first-­‐degree relative with psychiatric problems, i.e., unaffected siblings of participants with ADHD were excluded. All participants were of European Caucasian descent. For a detailed description of inclusion and diagnostic criteria, see von Rhein et al. (2015a). 109 Treatment history Pharmacy transcripts and self/parent-­‐report questionnaires were combined to assess treatment history. For each stimulant-­‐treated participant (immediate and/or extended release methylphenidate preparations and/or d-­‐amphetamine preparations), a dose-­‐by-­‐age trajectory from age=0 to age at scan was reconstructed (Figure 1). The area under the curve equals cumulative stimulant intake. Cumulative intake was divided by participant’s age minus 2.3 (minimum stimulant start age within the cohort) to obtain an age-­‐adjusted treatment variable (CSI ) in mg/year which was subsequently standardized into a z-­‐score. One extreme outlier (Z >4) was excluded (for details, see Supplement S1). Since CSI is a composite parameter capturing dose, start age, treatment duration, and time since last treatment, these alternative parameters were evaluated post-­‐hoc. Genotyping DNA was extracted from blood or saliva samples (for details, see Thissen et al., 2015). No deviations from Hardy-­‐Weinberg Equilibrium were found (p =0.15, p
=0.78, p
8=0.55). DAT1 haplotypes were calculated using the HaPloStats package (R version 2.12.0; Schaid et al., 2002). Participants with zero, one, or two copies of the DAT1 10-­‐6 risk haplotype were distinguished. We performed pairwise testing to avoid imposing a linear model (0vs.1 copy, 0vs.2 copies, and 1vs.2 copies). For the DRD4 7R-­‐allele, we differentiated between non-­‐carriers (0 copies) and carriers (1 or 2 copies), in line with the literature. Allele frequencies are in Table 1. Magnetic Resonance Imaging MRI data was acquired on two 1.5T Siemens scanners (Siemens, Germany), equipped with product 8-­‐channel phased-­‐array head coils using equivalent acquisition parameters. The session consisted of multiple acquisitions, including two T1-­‐weighted 3D-­‐MPRAGE scans (TI=1000 ms, TR=2730 ms, TE=2.95 ms, FA=7°; 176 sagittal slices, 1x1x1 mm voxels). For each participant, the structural acquisition of highest quality was selected by visual inspection, accepting only scans with no/mild distortions (Blumenthal et al., 2002). For data quality and compatibility between sites, see Von Rhein et al. (2015a). Striatal and hippocampal volumes were obtained with FMRIB’s Integrated Registration and Segmentation Tool (FSL FIRST; Patenaude et al., 2011). As our hypothesis regarding treatment effects was the same across striatal structures, striatal volume was calculated as the sum of caudate nucleus, putamen, and nucleus ADJ
CSI-­‐ADJ
ADJ
DRD4
DAT1-­‐3’UTR
110 DAT1-­‐INTRON
FIGURE 1.
Participants’ stimulant treatment trajectories. Daily stimulant dose (y-­‐
axis) is plotted as a function of age in years (x-­‐axis), between age=0 and age at study participation. Participants are stratified in age quartiles; bold markers represent the average daily dose across participants within each quartile. Age at study participation was significantly associated with treatment duration (r=0.202), start age (r=0.298), and time since last treatment (r=0.377), and participants on active treatment are younger (M=15.9) compared to participants who had discontinued treatment (M=18.4). accumbens volume; individual structures were evaluated only post-­‐hoc. Frontal cortex volume was derived by multiplying cortical thickness and surface area of the medial and lateral orbitofrontal, inferior frontal, caudal and rostral medial frontal, superior frontal, and frontal pole cortex, reconstructed using the automated Freesurfer pipeline (Dale et al., 1999; Desikan et al., 2006). Total brain volume (TBV) was acquired using SPM (VBM8.1 toolbox, http://dbm.neuro.uni-­‐jena.de/vbm/) as the sum of gray and white matter tissue probability maps. Analyses All analyses were performed separately for three regions of interest (ROI; striatum, frontal cortex, hippocampus) in two hemispheres (left, right). ROI volumes 111 were predicted from CSI , genotype (DAT1 haplotype for striatum, DRD4 genotype for frontal cortex and hippocampus), and CSI -­‐by-­‐genotype interaction, in linear mixed effects models with covariates gender, site, age, age , TBV, and a random intercept per family to correct for relatedness within the sample (‘initial models’). ADJ
ADJ
2
volume ~ α + β*covariates + β*CSI + β*genotype + β*CSI *genotype ADJ
ADJ
Age and CSI were standardized, such that main effects of the predictors of interest are conditional to average age and treatment; non-­‐genotype effects are also conditional to a reference category, i.e., DAT1 10-­‐6 homozygotes and DRD4 7R-­‐
carriers. Next, the initial models were extended to allow interactions with age (‘age-­‐
interaction models’. ADJ
volume ~ α + β*covariates + β*CSI + β*genotype + β*CSI *genotype + β*age*CSI + β*age*genotype + β*age*CSI *genotype + β*age *CSI + β*age *genotype + β*age *CSI *genotype) ADJ
ADJ
ADJ
2
ADJ
ADJ
2
2
ADJ
Non-­‐significant interaction terms were dropped from the model one-­‐by-­‐one, each time eliminating the highest-­‐level and least-­‐predictive interaction term and re-­‐
estimating the regression coefficients. Predictors and covariates from the initial model and lower-­‐level interaction terms conditional to significant higher-­‐level interaction terms were never removed. Alpha was divided by six (0.05/3/2=0.008). Clinical variables associated with stimulant treatment (e.g., severity) may drive spurious associations between brain volume and CSI . Therefore, each variable associated with CSI was post-­‐hoc evaluated as a potential confounder. First, the confounder (and its interactions with age and genotype) was tested in a model identical to the significant CSI model. If significant, CSI and the confounder (and their age-­‐ and genotype-­‐interactions) were modeled competitively. The same procedure was adopted to disentangle treatment parameters contributing to CSI , i.e., non-­‐adjusted cumulative intake, active versus past treatment, start age, treatment duration, and time since last treatment. Case-­‐control comparisons of subcortical volumes (Greven et al., 2015) and frontal cortex structure (Schweren et al., 2015a) have previously been reported. In the current study, controls served only to estimate reference volumes. All other analyses are based on participants with ADHD only. Since the full control sample differed from the ADHD sample in terms of age (M =16.5, M =17.2), gender (Male =52.4%, Male =69.3%), and site (Nijmegen =38.5%, Nijmegen =56.3%), reference volumes were also estimated for a group-­‐matched control sample (n=151). ADJ
ADJ
ADJ
ADJ
ADJ
HC
HC
112 ADHD
HC
ADHD
ADHD
RESULTS The majority of participants with ADHD were male (n=219, 69%), and most had inattentive (n=141, 45%) or combined type ADHD (n=138, 44%). Age ranged from 8 to 28 years (M[SD]=17.2[3.4] years). Comorbidities included oppositional defiant disorder or conduct disorder (n=97, 31%), anxiety/depression (n=11, 4%), and tic disorders (n=3, 1%). Age-­‐adjusted cumulative stimulant intake ranged from 0 mg/year (treatment-­‐naïve, n=38, 12%) to 15766 mg/year. Treatment start age ranged from 2.3 to 20.6 years, and 146 participants (46.2%) were on active stimulant treatment within three months prior to study participation. Figure 1 shows treatment trajectories over time per age quartile. At older age, more participants had ceased treatment. Eighty-­‐one participants (25.6%) had been treated with non-­‐stimulant psychoactive medication. CSI was higher in participants with combined type compared to inattentive or hyperactive/impulsive type ADHD (p=0.001). Furthermore, CSI was marginally associated with DAT1 10-­‐6 haplotype (M >M
>M
, p=0.029). CSI was not associated with parent-­‐rated inattention or hyperactivity/impulsivity symptoms, IQ, SES, DRD4 genotype, comorbidity, or non-­‐stimulant medication. As expected, CSI correlated positively with treatment duration, and negatively with start age and time since last treatment, and was higher in participants on active treatment compared to those who had discontinued (for details, see Supplement S1). Striatal volume and DAT1 Left striatal volume was reduced in participants with ADHD carrying one 10-­‐
6 risk allele compared to 10-­‐6 homozygotes (M
=10.30 mL, M =9.89 mL, M
=10.14 mL; p vs. =0.005, p vs. =0.018, p vs. =0.352; effect size β[95% confidence interval] vs =0.243[0.075-­‐0.410]), with a similar trend on the right (p vs. =0.030, p vs. =0.145, p vs. =0.719). This effect was most pronounced in the left putamen (p=0.009) and accumbens (p=0.013), and less prominent in the caudate (p=0.090). Covariates site, gender, and age or age were not associated with striatal volume, but, as expected, total brain volume was (Table S2). Participants with and without ADHD did not differ with regard to striatal volume (p =0.531; p =0.531). Treatment (CSI ) was not associated with left or right striatal volume as a main effect, nor in interaction with DAT1 haplotype, age, or age (Table 2). ADJ
ADJ
1COPY
0COPIES
2COPIES
ADJ
ADJ
0COPIES
2COPIES
1
2
0
1
0
1COPY
2
1 .2
0
1
1
0
2
2
2
LEFT
RIGHT
ADJ
2
113 TABLE 1. Demographic and clinical characteristics of the ADHD sample, and their associations with stimulant treatment. n % Gender=male 219 69.3 Site=Nijmegen 178 56.3 Age (M,SD) 17.2 3.4 CPRS inattention (M,SD) 64.4 14.3 CPRS hyperactivity/impulsivity (M,SD) 68.2 17.4 DAT1 10-­‐6 risk allele 20 6.4 1 copy (heterozygote) 134 42.7 2 copies (homozygote) 160 51.0 208 65.8 1 or 2 copies (carrier) 108 34.2 141 44.6 Hyperactive/impulsive 37 11.7 Combined 138 43.7 278 88.0 Treatment duration in years (M,SD) 4.1 3.3 Start age (M,SD) 8.5 2.8 Years since last treatment (M,SD) 1.5 2.3 Currently on active treatment 148 46.8 65 20.6 History of non-­‐stimulant medication (y/n) F 29.8 7.0 * 3.6 ADJa
r -­‐0.01 0.05 0.05 1.2 10.9 Inattentive History of stimulant treatment (y/n) 0 copies (non-­‐carrier) ADHD-­‐type 0 copies (non-­‐carrier) DRD4 7R risk allele Association with CSI
* N/A 0.713 * -­‐0.496 * -­‐0.417 * b
b
59.8 b
2.2 * * p<0.008; age-­‐adjusted cumulative stimulant intake; within stimulant-­‐treated (i.e., non-­‐naïve) participants. a
114 b
Frontal cortex volume and DRD4 In the initial models, neither CSI , nor DRD4 genotype or their interaction was associated with frontal cortex volume (Table 2, Table S2). When age and age were allowed to interact with stimulant treatment and genotype in the age-­‐
interaction models, however, significant age -­‐by-­‐CSI -­‐by-­‐DRD4 interaction effects were found in both hemispheres (p =0.003, β[CI]=0.187[0.062-­‐0.311]; p <0.001, β[CI]=0.220[0.093-­‐0.346]; Figure 2). At younger age (plotted at 1SD below the mean, 13.9 years), frontal cortex volume increased with increasing CSI in carriers of the 7R-­‐allele. No such association was found in carriers of the 7R-­‐allele at older age (plotted at 1SD above the mean, 20.5 years), nor in non-­‐carriers at older or younger age. As a consequence of the three-­‐way interactions, the lower-­‐level age -­‐by-­‐CSI interaction effect reached significance as well, as did the age -­‐by-­‐DRD4 interaction effect on right frontal cortex volume (Table 2). CSI was higher in combined type ADHD compared to inattentive or hyperactive/impulsive type. Models were re-­‐estimated replacing CSI with ADHD-­‐
type. Age -­‐by-­‐ADHD-­‐type-­‐by-­‐DRD4 reached nominal significance in both hemispheres for inattentive versus combined type ADHD (p =0.030, p =0.045); when CSI and ADHD-­‐type (and their interactions with age and genotype) were modeled competitively, the age -­‐by-­‐CSI -­‐by-­‐DRD4 interaction term remained significant (p =0.002, p =<0.001), while the age -­‐by-­‐ADHD-­‐type-­‐by-­‐DRD4 interaction term was marginally significant (inattentive versus combined type, p =0.011, p =0.009; hyperactive/impulsive versus combined type, p =0.062, p =0.044). Finally, frontal cortex volume was reduced in participants with ADHD compared to control participants at trend level (right: M =81.53 mL, M
=82.78 mL; p=0.010; left: M =82.24 mL, M
=83.11 mL; p=0.073). Volume reduction in the right frontal cortex was significant when compared to the matched control group (Table S3). Hippocampus volume and DRD4 Neither CSI nor DRD4 genotype was associated with hippocampus volume in the initial models, and there were no CSI -­‐by-­‐DRD4 interaction effects (Table 2). When treatment and genotype were allowed to interact with age in the age-­‐
interaction models, however, a significant age-­‐by-­‐CSI -­‐by-­‐DRD4 interaction effect was found in the left hippocampus (p=0.008; β[CI]=0.323[0.086-­‐0.561]; Figure 2). Irrespective of genotype, there was little association between CSI and left hippocampal volume at older age, whereas at younger age a negative association was ADJ
2
2
ADJ
LEFT
RIGHT
ADJ
2
ADJ
2
ADJ
ADJ
2
LEFT
2
LEFT
RIGHT
ADJ
ADJ
2
RIGHT
LEFT
LEFT
RIGHT
ADHD
ADHD
RIGHT
CONTROL
CONTROL
ADJ
ADJ
ADJ
ADJ
115 found in 7R-­‐non-­‐carriers and a positive association was found in 7R-­‐carriers. The association was strongest in the 7R-­‐carriers and at younger age. A similar but non-­‐
significant trend was found in the right hippocampus (p
DRD4=0.066). Age-­‐by-­‐ADHD-­‐type-­‐by-­‐DRD4 (i.e., replacing CSI by ADHD-­‐type) was not associated with left hippocampal volume. Finally, participants with and without ADHD did not differ in hippocampal volume (p =0.838; p =0.277). AGE-­‐BY-­‐CSI-­‐ADJ-­‐BY-­‐
ADJ
LEFT
RIGHT
TABLE 2. Regression weights of treatment, genotype, and treatment-­‐by-­‐genotype, and their interactions with age and age . Parameters of covariates and lower-­‐level terms are available online. 2
Striatum L Initial model CSI
ADJa
DAT1
0-­‐COPIES
DAT1
1-­‐COPY
DRD4
NON-­‐CARRIER
CSI x DAT1
0-­‐COPIES
CSI x DAT1
1-­‐COPY
ADJ
ADJ
CSI x DRD4
ADJ
Age-­‐interaction model Age x CSI x DAT1
0-­‐COPIES
Age x CSI x DAT1
1-­‐COPY
ADJ
ADJ
Age x CSI x DRD4
ADJ
0-­‐COPIES
Age x CSI x DAT1
1-­‐COPY
ADJ
Age x CSI x DRD4
2
a
-­‐0.252* -­‐0.186 ADJ
-­‐0.081 ns ns ns ns ns ns 116 0.067 0.002 0.236 -­‐0.001 0.023 -­‐0.240 -­‐0.085 -­‐0.008 0.756 ns 0.031 R Age-­‐adjusted cumulative stimulant intake. 0.040 L -­‐0.458 ns -­‐0.111 NON-­‐CARRIER
-­‐0.086 Hippocampus R 0.007 -­‐0.170 NON-­‐CARRIER
ADJ
2
0.060 Age x CSI x DAT1
2
0.162 L 0.051 NON-­‐CARRIER
R 0.073 Frontal cortex 0.458 0.158* ns -­‐1.717* -­‐2.030* ns ns FIGURE 2. Three-­‐way interaction effects between CSI , DRD4 genotype, and age or age , on volumes 2
ADJ
of the left and right frontal cortex and left hippocampus. For display, volumes based on the regression function are estimated at younger age (age=1SD below the mean, age =-­‐1SD*-­‐1SD) and at older age (age=1SD above the mean, age =+1SD*+1SD), and separately for carriers and non-­‐
carriers of the DRD4 7R-­‐risk allele. In the absence of stimulant treatment in control participants, estimated average volume for controls is presented as a dashed horizontal line (+/-­‐ 1SD shaded). CSI =age-­‐adjusted cumulative stimulant intake; mg/y = milligrams per year. 2
2
ADJ
117 Alternative treatment parameters In an attempt to disentangle treatment parameters contributing to CSI , significant models were re-­‐estimated replacing CSI with non-­‐adjusted cumulative dose, treatment duration, start age, current treatment (y/n), and time since last treatment. Age -­‐by-­‐treatment-­‐by-­‐DRD4 interaction effects on the frontal cortex, significant when the treatment parameter was CSI , were not significant when the treatment parameter was start age (pLEFT=0.676, pRIGHT=0.924), time since last treatment (pLEFT=0.157, pRIGHT=0.064), or current treatment (y/n) (pLEFT=0.659, pRIGHT=0.259). By contrast, when CSI was replaced by non-­‐adjusted cumulative intake or treatment duration, the effect changed very little (non-­‐adjusted CSI: p =0.003, p =0.001; duration: p =0.009, p =0.002). When CSI and treatment duration were modeled competitively, neither of the interaction terms reached significance, suggesting that the effects of CSI and treatment duration at least partially overlap. In the left hippocampus, replacing CSI by non-­‐adjusted cumulative dose, treatment duration, or start age yielded age-­‐by-­‐treatment-­‐by-­‐DRD4 interaction effects similar to those of CSI , but none reached significance according to the multiple comparisons threshold (p=0.013, p=0.014, and p=0.024, respectively). Current use and time since last treatment did not show such effects. DISCUSSION We investigated associations between stimulant treatment and striatal, frontal, and hippocampal volumes, and potential moderating effects of genotype and age, in children, adolescents, and young adults with ADHD. There were three main findings. First, stimulant treatment was not associated with striatal volume. Second, associations between stimulant treatment and bilateral frontal and left hippocampal volume depended on DRD4 genotype and age. Associations were particularly pronounced in carriers of the DRD4 7R-­‐allele at younger age, and were not accounted for by clinical and/or demographic confounders. Finally, irrespective of stimulant treatment, left striatal volume was reduced in carriers of one DAT1 10-­‐6 risk allele compared to 10-­‐6 homozygotes. We had hypothesized that stimulant treatment would be associated with more normative regional brain volumes, especially at younger age and in carriers of DAT1 10-­‐6 and/or DRD4 7R-­‐risk alleles. Striatal volumes were not altered in participants with ADHD compared to controls, which, as previously reported in the our sample (Greven et al., 2015), may be due to volume reduction becoming less apparent at late-­‐adolescent/early-­‐adult age. Furthermore, we found no indication of ADJ
ADJ
2
ADJ
ADJ
LEFT
RIGHT
LEFT
RIGHT
ADJ
ADJ
ADJ
118 ADJ stimulant treatment positively affecting striatal volume, as had been suggested by two meta-­‐analyses both including only slightly more participants compared to the current sample (Frodl & Skokauskas, 2012; Nakao et al., 2011). In contrast, our findings in the frontal cortex are consistent with age-­‐ and genotype-­‐specific volumetric changes toward more normative levels after stimulant treatment. Frontal cortex volume was reduced in participants with ADHD compared to controls. In carriers of the DRD4 7R-­‐
allele, more intense stimulant treatment (either higher dose or longer duration) was associated with increased frontal cortex volumes at younger age. Such associations were not observed at older age, or in individuals not carrying the 7R-­‐allele. Although our observational study design is inconclusive as to whether associations between stimulant treatment and frontal cortex volume constitute treatment effects, it is worthwhile to explore possible underlying mechanisms. The frontal cortex of DRD4 7R-­‐carriers may at younger age, e.g., in late childhood when D4 receptor density in the frontal cortex peaks (Tseng & O’Donnell, 2007), exhibit postsynaptic characteristics allowing for long-­‐term neural plasticity in the event of exposure to stimulants. Long-­‐term plasticity occurs only when tonic dopamine levels, maintained by continuous background firing of dopaminergic neurons, are within an optimal range, i.e., neither too high nor too low (Goto et al., 2010). Fine-­‐tuning of tonic dopamine levels is managed through inhibitory feedback mechanisms involving D4 receptors (Padmanabhan & Luna, 2013) and has been associated with DRD4 7R-­‐
carriership (Asghari et al., 1995). Thus, early age, presence of the 7R-­‐allele, and stimulant treatment may together adjust tonic dopamine levels to enable structural remodeling in the frontal cortex. Alternatively, the older participants may represent a specific patient population (i.e., persistent ADHD) that could be less likely to show genotype-­‐by-­‐treatment interaction effects, compared to the potentially more mixed younger group (i.e., this group likely includes participants who will remit during adolescence). This interpretation is plausible given the cross-­‐sectional nature of our study. As another possibility, brain changes in 7R-­‐carriers may result from an enhanced acute frontal cortex response to stimulant treatment at younger age, increasing the likelihood of long-­‐term changes in this group. Finally, the lack of association between treatment and brain changes at later age may result from treatment discontinuation, i.e., lasting treatment effects may require ongoing treatment. Post-­‐hoc analysis did not indicate significant contributions of current treatment (y/n) or time since last treatment, but individual and combined effects of various treatment parameters can only be disentangled in rigorously designed intervention studies. Note that speculations about potential micro-­‐level mechanisms can only be tested using alternative approaches (e.g., animal or radioligand studies). Moreover, our findings await replication in an independent sample. 119 Similar to the frontal cortex findings, the association between stimulant treatment and left hippocampal volume was strongest in DRD4 7R-­‐carriers and at younger age. Notably, there were no case-­‐control differences in hippocampal volume. In 7R-­‐carriers, especially at younger age, hippocampal volume appeared to deviate from the controls with more intense stimulant treatment. Similar but non-­‐significant effects were found for treatment duration, non-­‐adjusted cumulative dose, and start age, suggesting that these parameters each contribute to the effect of CSI . Treatment-­‐related hippocampal volume reduction has previously been reported in individuals with ADHD (Frodl et al., 2010; Onnink et al., 2014). Further investigation of long-­‐term stimulant treatment effects on hippocampal development is warranted. A final noteworthy finding was the larger striatal volume in DAT1 10-­‐6 risk allele homozygotes compared to carriers of only one risk allele. At thend level, the non-­‐carriers also differed from the heterozygotes, but not from the homozygotes., which could indicate a non-­‐linear association (0copies > 1 copy < 2 copies). Other studies have reported smaller striatal volumes in 3’UTR VNTR 10-­‐repeat homozygotes compared to heterozygotes and/or non-­‐carriers (Durston et al., 2005; Shook et al., 2011). This seems at variance with the current findings, but note that these studies classified participants based on the 3’-­‐UTR VNTR alone rather than on DAT1 haplotype. Moreover, non-­‐carriers were not examined in these previous studies. In follow-­‐up analyses we found no association between striatal volume and the 3’-­‐UTR 10/10 polymorphism alone (data not shown). Participants’ age may also contribute to divergent findings; comparing striatal volumes of children/adolescents (including the current sample) and adults with different DAT1 haplotypes, our group found that the 9-­‐6 variant was associated with larger striatal volume in adults but not adolescents with ADHD (Onnink et al., 2016). It is noteworthy that in the current study we did not find this gene-­‐by-­‐age interaction effects on striatal volume, nor was the 9-­‐6 variant associated with striatal volume (data not shown). Our findin of a putative non-­‐linear pattern should not be over-­‐interpreted; it did not reach our adjusted significance level, was not hypothesized a priori, and could not be related to the existing literature. Replication of the DAT1 haplotype findings in an independent sample is needed. Essential features of the current study could not have been achieved in a randomized controlled design. Our sample covered a wide treatment-­‐ and age-­‐range, allowing for the study of within-­‐group heterogeneity. The focus on late-­‐adolescence further allowed the investigation of potential treatment effects occurring only after multiple years of treatment, and/or occurring long after treatment had been discontinued. Notwithstanding, the current observational design brings caveats regarding causal inference. Unmeasured pre-­‐existing factors associated with treatment (e.g., pre-­‐treatment symptom severity) or simultaneously occurring events ADJ
120 (e.g., concurrent behavioral treatment) may have contributed to the observed associations. Second, age effects should be interpreted with appropriate caution given our cross-­‐sectional design. Participants included at older age may represent a different population (e.g., persistent ADHD) compared to those included at younger age. A longitudinal study design is required to allow conclusions about developmental effects. Finally, we wish to emphasize that clinical application of our findings of small effect size is still several steps away, e.g., behavioral correlates of subtle brain changes require further investigation. In sum, we investigated associations between stimulant treatment and regional brain volumes, and potential moderating effects of age and genotype, in a cross-­‐sectional ADHD cohort. We found that frontal cortex volume was associated with stimulant treatment in carriers of the DRD4 7R-­‐allele at younger age, possibly suggesting normalizing effects in these participants. Striatal volume was associated with DAT1 haplotype, but not with treatment. We propose that neural sensitivity to long-­‐term treatment effects may arise from genotype-­‐ and age-­‐specific characteristics of postsynaptic dopamine receptors, allowing for long-­‐term plasticity when exposed to stimulant treatment. The clinical relevance of subtle brain changes of small effect size is expected to be modest and requires further investigation; nevertheless, our findings may ultimately contribute to optimal care for individuals with ADHD. 121 SUPPLEMENTAL INFORMATION
S1 -­‐ Supplementary Methods Pharmacy prescription transcripts covering lifetime (>90%) could be obtained for 118 participants with ADHD (37.3%). Parents and participants self-­‐reported lifetime history of psychoactive medication to complement partially available (53.5%) or missing (9.2%) pharmacy transcripts. Retrospective self-­‐assessment of ADHD medication has shown to be reliable even after multiple years (Kuriyan et al., 2014). For the majority of stimulant treated participants (n=189, 68.0%), pharmacy transcripts covered the self/parent-­‐reported treatment period, hence CSIADJ could be calculated based entirely on these transcripts. For 31 participants (11.2%), CSIADJ was calculated based entirely on self/parent-­‐report, and for the remaining 58 participants (20.9%) CSIADJ was calculated based on a combination of pharmacy transcripts and self-­‐report. For each stimulant-­‐treated participant (immediate and/or extended release methylphenidate preparations and/or d-­‐amphetamine preparations), a dose-­‐by-­‐
age trajectory from age=0 to age at scan was reconstructed. Missing dose information was imputed to the group mean per stimulant type, and d-­‐amphetamine dose was multiplied by two in line with clinical guidelines (n=27 d-­‐amphetamine users; Russell et al., 1998). The area under the curve equals cumulative stimulant intake. Cumulative intake was divided by participant’s age minus 2.3 (minimum stimulant start age within the cohort) to obtain a continuous age-­‐adjusted treatment variable (CSIADJ) in mg/year, and standardized (z-­‐scores). One extreme outlier (ZCSI-­‐ADJ>4) was excluded. For the distribution of CSIADJ, see Figure S1. As expected, CSIADJ correlated positively with treatment duration (r=0.713, p<0.001), and negatively with start age (r=-­‐0.496, p<0.001) and time since stop (r=-­‐0.417, p<0.001). Moreover, CSIADJ was higher in participants on active treatment compared to those who had discontinued (MACTIVE-­‐
TREATMENT=5996 mg/year, MPAST-­‐TREATMENT=3039 mg/year; p<0.001) S2 -­‐ Supplementary tables and figures FIGURE S1. Distribution of CSIADJ within stimulant-­‐treated participants. CSIADJ = age-­‐adjusted cumulative stimulant intake in mg/year. 122 TABLE S2. Regression weights of all covariates and parameters of interest in the initial and developmental models, for each regional brain volume. Effects are conditional to the genotype reference category (DAT1HOMOZYGOTE or DRD4 7R-­‐carrier). Striatum L Initial model parameters R Frontal cortex L R Hippocampus L R Site 0.134 0.210 3.675* 3.721* -­‐0.006 0.017 Gender 0.013 0.075 0.513 0.366 0.020 0.004 Age 0.026 -­‐0.004 -­‐3.799* -­‐3.794* 0.042 0.041 Age2 0.036 0.029 0.473 0.612* -­‐0.030 -­‐0.016 TBV 0.686* 0.714* 7.080* 7.095* 0.233* 0.209* CSIADJ a 0.073 0.051 0.040 -­‐0.031 0.067 0.002 DAT10 COPIES 0.162 0.060 DAT11 COPY -­‐0.252* -­‐0.186 0.007 0.236 -­‐0.001 0.023 CSIADJ x DAT10 COPIES -­‐0.169 -­‐0.086 CSIADJ x DAT11 COPY -­‐0.111 -­‐0.081 -­‐0.458 -­‐0.240 -­‐0.085 -­‐0.008 Site 0.134 0.210 3.636* 3.711* -­‐0.015 0.017 Gender 0.013 0.075 -­‐0.066 0.233 0.008 0.004 Age 0.026 -­‐0.004 -­‐4.191* -­‐4.171* 0.022 0.041 Age2 0.036 0.029 1.460* 1.760* -­‐0.033 -­‐0.016 TBV 0.686* 0.714* 7.352* 7.362* 0.241* 0.209* CSIADJ 0.073 0.051 -­‐1.035 -­‐1.026 0.046 0.002 DAT10 COPIES 0.162 0.060 DAT11 COPY -­‐0.252* -­‐0.186 0.630 1.083 <-­‐0.001 0.023 DRD4NON-­‐CARRIER CSIADJ x DRD4NON-­‐CARRIER Age-­‐interaction model parameters DRD4NON-­‐CARRIER 123 Striatum L R CSIADJ x DAT10 COPIES -­‐0.169 -­‐0.086 CSIADJ x DAT11 COPY -­‐0.111 -­‐0.081 Age x CSI ns ns Age x DAT10 COPIES ns ns Age x DAT11 COPY ns ns CSIADJ x DRD4NON-­‐CARRIER Age x DRD4NON-­‐CARRIER Age x CSIADJ x DAT10 COPIES ns ns Age x CSIADJ x DAT11 COPY ns ns Age x CSIADJ x DRD4NON-­‐CARRIER Age2 x CSI ns ns Age2 x DAT10 COPIES ns ns Age2 x DAT11 COPY ns ns Age2 x CSIADJ x DAT10 COPIES ns ns Age2 x CSIADJ x DAT11 COPY ns ns Age2 x DRD4NON-­‐CARRIER Age2 x CSIADJ x DRD4NON-­‐CARRIER a
Frontal cortex L R L R 0.667 0.996 -­‐0.065 -­‐0.008 -­‐0.889 -­‐0.571 -­‐0.111 ns 0.613 0.697 0.029 ns 0.756 0.458 0.158* ns 1.641* 1.754* ns ns -­‐1.192 -­‐1.533* ns ns ns ns -­‐1.716* -­‐2.030* age-­‐adjusted cumulative stimulant intake; * p<0.008; ns=not significant. 124 Hippocampus TABLE S3. Normative volumes in mL based on a group-­‐level age-­‐, gender-­‐, and site matched subsample of control participants (n=151). For regions of significant age/age2-­‐by-­‐
genotype-­‐by-­‐treatment effects within the ADHD sample, normative volumes are also provided stratified by genotype and estimated at younger and older age. Left Right EMM SD EMM SD Striatum 10.08 0.86 10.12 0.86 Frontal cortex 83.58 5.41 83.21 5.28 DRD4NON-­‐CARRIER , younger 88.80 6.19 88.44 6.19 DRD4NON-­‐CARRIER , older 81.82 6.19 81.81 6.19 DRD4CARRIER , younger 88.78 6.21 88.52 6.29 DRD4CARRIER , older 83.75 6.21 82.48 6.29 3.88 0.49 4.05 0.49 DRD4NON-­‐CARRIER , younger 3.89 0.51 -­‐ -­‐ DRD4NON-­‐CARRIER , older 3.98 0.51 -­‐ -­‐ DRD4CARRIER , younger 3.75 0.50 -­‐ -­‐ DRD4CARRIER , older 3.89 0.50 -­‐ -­‐ Hippocampus EMM, estimated marginal means 125 126 Chapter 7 COMBINED STIMULANT AND ANTIPSYCHOTIC
TREATMENT IN ADOLESCENTS WITH ADHD:
A CROSS-SECTIONAL OBSERVATIONAL
STRUCTURAL MRI STUDY
Published as: Schweren LJS, Hartman CA, Zwiers MP, Heslenfeld DJ, van der Meer D, Franke B, Oosterlaan J, Buitelaar JK, Hoekstra PJ. Combined stimulant and antipsychotic treatment in adolescents with attention-­‐deficit/hyperactivity disorder: a cross-­‐
sectional observational structural MRI study. Eur Child Adolesc Psychiatry. 2015;24(8):959-­‐968. 127 ABSTRACT
Objective: Meta-­‐analyses suggest normalizing effects of methylphenidate on structural fronto-­‐striatal abnormalities in patients with attention-­‐deficit/hyperactivity disorder (ADHD). A subgroup of patients receives atypical antipsychotics concurrent with methylphenidate. Long-­‐term safety and efficacy of combined treatment are unknown. The current study provides an initial investigation of structural brain correlates of combined methylphenidate and antipsychotic treatment in patients with ADHD. Methods: Structural magnetic resonance imaging was obtained in 31 patients who had received combined methylphenidate and antipsychotic treatment, 31 matched patients who had received methylphenidate but not antipsychotics, and 31 healthy controls (M age 16.7 years). We analyzed between-­‐group effects in total cortical and subcortical volume, and in seven frontal cortical and eight subcortical-­‐limbic volumes of interest, each involved in dopaminergic neurotransmission. Results: Patients in the combined treatment group, but not those in the methylphenidate only group, showed a reduction in total cortical volume compared to healthy controls (Cohen’s d=0.69, p<0.004), which was apparent in most frontal volumes of interest. Further, the combined treatment group, but not the methylphenidate group, showed volume reduction in bilateral ventral diencephalon (Left: Cohen’s d=0.48, p<0.04; Right: Cohen’s d=0.46, p<0.05) and the left thalamus (Cohen’s d=0.47, p <0.04). Conclusions: These findings may indicate antipsychotic treatment counteracting the normalizing effects of methylphenidate on brain structure. However, it cannot be ruled out that pre-­‐existing clinical differences between both patient groups may have resulted in anatomical differences at the time of scanning. The absence of an untreated ADHD group hinder unequivocal interpretation and implications of our findings.
128 INTRODUCTION
Attention-­‐deficit/hyperactivity disorder (ADHD) is characterized by age-­‐
inappropriate hyperactivity, impulsivity, and/or inattention. Subtle though widespread differences in brain morphology have been found in patients with ADHD, the most replicated being reduced volumes of the basal ganglia including the caudate, putamen, and globus pallidus, and of frontal regions (Frodl & Skokauskas, 2012; Nakao et al., 2011; Valera et al., 2007). Treatment with methylphenidate is the medical intervention of first choice (NHS NICE, 2008). Neuroimaging studies investigating the effect of methylphenidate on brain structure and function in children with ADHD consistently suggest normalizing effect. At least partial normalization has been reported for volumes of the anterior cingulate cortex, thalamus, inferior frontal gyrus, right precentral gyrus, right parieto-­‐occipital gyrus, and the cerebellar vermis (Schweren et al., 2013). In a subgroup of patients with ADHD, stimulant treatment is combined with atypical antipsychotics such as risperidone or pipamperone. Antipsychotics have been recommended by an expert group for the treatment of comorbid disruptive behavior and severe aggression in ADHD (Kutcher et al., 2004). Besides disruptive behavior, comorbid pervasive developmental disorder (PDD) has been found to be predictive of the prescription of antipsychotics (Faber et al., 2010). In the Netherlands, atypical antipsychotics are prescribed to 8% of stimulant-­‐treated children with ADHD (Faber et al., 2010). Increasing prescription rates of atypical antipsychotics to children and adolescents (Bachmann et al., 2014) have raised concerns (Harrison et al., 2012). Abnormalities in dopaminergic neurotransmission have been reported in patients with ADHD, and include increased levels of striatal dopamine auto-­‐receptors and reduced dopamine metabolism in the frontal cortex (Del Campo et al., 2011). Both methylphenidate and atypical antipsychotics exert their effects by interacting with dopaminergic neurotransmission, albeit with opposite modes of action. Methylphenidate blocks dopamine reuptake and stimulates dopamine release from the presynaptic cell, resulting in increased synaptic levels of dopamine (Volkow et al., 2001). Most atypical antipsychotics, by contrast, are dopamine antagonists blocking the effects of dopamine in the synapse by occupying the postsynaptic dopamine D2 receptors. It has been suggested that combined treatment with methylphenidate and antipsychotics may compromise the effects of each of the individual agents (Yanofski, 2010). Large-­‐scale studies on long-­‐term safety and efficacy of combined treatment have not yet been performed (Linton et al., 2013). Little is known about the possible effects of atypical antipsychotic treatment on structural brain development in children. The few studies investigating the neural effects of antipsychotics in pediatric populations have been limited to childhood-­‐onset 129 schizophrenia and pediatric bipolar disorder. Frazier et al. (1996) reported a trend of normalizing subcortical volumes with clozapine treatment in the children with childhood-­‐onset schizophrenia, but others found no such effect (Mattai et al., 2010). Despite increasing prescription rates, no studies have yet investigated the effects of atypical antipsychotic treatment on brain development in children with ADHD. The current cross-­‐sectional, observational MRI study investigated brain correlates of combined methylphenidate and atypical antipsychotics treatment in patients with ADHD, in comparison to patients who had received methylphenidate only, and to healthy control subjects. In the absence of a medication-­‐naïve ADHD group and of pretreatment measurements, we were unable to directly investigate whether concurrent antipsychotic treatment would counteract any normalizing effects of methylphenidate treatment on brain structure. However, based on the opposing synaptic effects of the two substances, we expected to find volume reductions in frontal-­‐striatal regions in patients who had received combined treatment compared to healthy control participants, and that such reductions would be smaller or absent in patients who had received methylphenidate treatment only. METHODS
Participants This study was part of NeuroIMAGE (von Rhein et al., 2015a), the follow-­‐up of the Dutch part of the International Multicenter ADHD Genetics (IMAGE) study (Nijmeijer et al., 2009). The NeuroIMAGE sample consists of 1045 children from 330 ADHD and 154 control families, who met the following inclusion criteria: age between 5-­‐30 years, of European Caucasian descent, an IQ ≥ 70, and no diagnosis of autistic disorder, general learning difficulties, brain disorders, or known genetic disorders. All subjects who successfully underwent diagnostic assessment and structural MRI scanning were considered for inclusion in the current study. First, all participants with ADHD who received combined treatment with (1) any methylphenidate preparation and (2) atypical antipsychotics, either in the past or currently and for a minimum duration of thirty days, were included in the study sample (MPH+AAP group). Next, two one-­‐to-­‐one age-­‐ and gender-­‐matched control samples were drawn: a methylphenidate group (MPH group) consisting of participants with ADHD with current or past methylphenidate treatment with a minimum duration of 30 days and no treatment with antipsychotics and a healthy control group (HC group) consisting of participants with no psychiatric diagnosis and no current or past treatment with psychoactive medication of any type. Informed consent was signed by all participants (and parents) and the study had been approved by the local ethics committees. 130 Detailed demographic, clinical and treatment characteristics of the three participant groups are presented in the results section and in Table 1. Diagnostic assessment The Dutch translation of the Schedule for Affective Disorders and Schizophrenia for School-­‐Age Children -­‐ Present and Lifetime Version (K-­‐SADS; Kaufman et al., 1997) was administered. In addition, all participants were administered the Conners Parent Rating Scale -­‐ Revised (Long version, CPRS-­‐R:L; Conners et al., 1998a) combined with either the Conners Teacher Rating Scale -­‐ Revised (Long version, CTRS-­‐R:L; Conners et al., 1998b) for participants < 18 years old, or the Conners Adult ADHD Rating Scales -­‐ Self-­‐Report (Long Version, CAARS-­‐S:L; Conners et al., 1999) for participants ≥ 18 years old. For participants using medication, ratings reflected their functioning while they were off medication. Scores on the K-­‐SADS interview and Conners questionnaires were restructured to match the DSM-­‐5 criteria for ADHD. Participants with ADHD had to fulfill the following criteria: (1) six or more symptoms of hyperactivity/impulsivity and/or inattentiveness (five for participants ≥ 18 old), (2) meet DSM 5 criteria for pervasiveness of symptoms and impact on daily functioning, (3) symptom onset before the age of 12 years, and (4) T ≥ 63 on at least one of the ADHD scales on either one of the Conners questionnaires. Two participants who fulfilled all criteria but one for a full ADHD diagnosis were classified as mild ADHD cases. ADHD type (predominantly hyperactive/impulsive, predominantly inattentive, or combined type), impairment in daily functioning (Children’s Global Assessment Scale, CGAS; Shaffer et al., 2014) and comorbidity were assessed using the K-­‐SADS interview. Mild PDD symptoms were assessed with the Children’s Social Behavior Questionnaire (CSBQ; Hartman et al., 2006). Healthy control participants were required to have less than three ADHD symptoms (two for participants ≥ 18 years old) and T < 63 on each of the scales of all Conners’ questionnaires. From pharmacy transcripts, the following parameters were obtained for each type of psychoactive medication used: treatment duration, age of treatment initiation and cessation, mean daily dose, and current vs. past user. If pharmacy transcripts were incomplete (n=19, 31%), information from parent-­‐report questionnaires was used. 131 TABLE 1. Clinical and phenotypical characteristics of both ADHD groups. Groups, including the HC group, are matched one-­‐to-­‐one on gender and age (M age = 16.7 years; 84 % male; no differences between groups). ADHD M P H (N = 31) ADHD M P H + A A P (N = 31) Mean SD Mean SD F p* Total number of ADHD Symptoms 13.03 3.13 14.74 2.91 4.971 0.030* IQ 99.71 14.65 92.52 15.28 3.580 0.063* CGAS 63.23 7.91 59.68 9.48 2.560 0.041* MPH treatment duration (years) 5.11 3.12 5.71 3.47 0.505 0.480* MPH age of initiation (years) 9.35 3.33 8.35 3.39 1.340 0.252* MPH age of cessation (years) 15.91 1.80 14.58 1.66 0.298 0.591* MPH dose 32.60 2.44 30.89 2.40 0.250 0.619* CSBQ score 28.38 17.86 32.77 16.09 1.005 0.320* n % n % Chi 2 p* 2.761 0.430 Combined type 13 41.9 19 61.3 Hyperactive type 3 9.7 3 9.7 Inattentive type 14 45.2 8 25.8 Subthreshold/mild ADHD 1 3.2 1 3.2 Current users (<3 months prior to scan) 20 64.5 18 58.1 0.272 0.602 Co-­‐morbid diagnosis (any) 12 38.7 15 48.4 0.590 0.304 Co-­‐morbid diagnosis (ODD/CD) 10 32.3 14 45.2 1.088 0.297 Medication other than MPH or AAP 17 54.8 24 77.4 3.528 0.060 ADHD type *= p<0.05; MPH: methylphenidate; CGAS: Children’s Global Assessment Scale; CSBQ: Children’s Social Behavior Questionnaire; ODD: oppositional defiant disorder; CD: conduct disorder. MRI acquisition and analysis MRI data was acquired at 1.5 Tesla on a Siemens Sonata scanner at the VU Medical Centre in Amsterdam and on a Siemens Avanto scanner at the Donders Centre for Cognitive Neuroimaging in Nijmegen (Siemens, Germany). A standard identical 8-­‐
132 channel phased array coil model was used at both sites and all scan parameters were matched as closely as possible. A T1-­‐weighted 3D MP-­‐RAGE scan was acquired with parallel imaging by generalized auto-­‐calibrating partially parallel acquisition (GRAPPA; 176 sagittal slices, voxel size 1 x 1 x 1 mm, FOV = 256 x 256 x 176 mm). Cortical reconstruction and volumetric segmentation was performed with FreeSurfer software version 5.3 with default settings. FreeSurfer is an image processing pipeline including a volume-­‐based route to subcortical segmentation (Fischl et al., 2004) and a surface-­‐based route to create a 3D reconstruction and parcellation of the cortical sheet (Dale et al., 1999). From FreeSurfer parcellations (Desikan et al., 2006) and segmentations, we calculated total cortical and subcortical volume, and selected eight bilateral subcortical and limbic volumes of interest (VOIs; bilateral ventral diencephalon, putamen, caudate nucleus, globus pallidus, nucleus accumbens area, hippocampus, amygdala, and thalamus) and seven bilateral frontal cortical VOIs (inferior frontal gyrus [IFG; sum of pars orbitalis, pars triangularis, and pars opercularis], orbitofrontal gyrus [OFG; sum of medial orbitofrontal gyrus and lateral orbitofrontal gyrus], middle frontal gyrus [MFG; sum of caudal middle frontal gyrus and rostral middle frontal gyrus], superior frontal gyrus [SFG], anterior cingulate gyrus [ACC; sum of caudal anterior cingulate cortex and rostral anterior cingulate cortex], precentral gyrus, and the frontal pole), all implicated in dopaminergic pathways. Statistical analyses Treatment group (HC, MPH, or MPH+AAP) was entered in two univariate linear mixed regression models predicting standardized total cortical and subcortical volume. Dummy variables modeled between-­‐group contrasts (MPH+AAP vs. HC, MPH vs. HC, and MPH+AAP vs. MPH; the third contrast was tested in a second run of the model). Age, gender, and scanner location were entered as fixed covariates. To correct for family relatedness within the sample, a random family intercept was modeled. Restricted Maximum Likelihood (REML) was applied for model estimation. We analyzed between-­‐group effects in (1) the subcortical-­‐limbic VOIs, (2) the right, and (3) left frontal cortical VOIs in three multivariate linear mixed regression models, containing the same covariates and dummy variables for between-­‐group contrasts. The VOI analyses were initially ran without total cortical or subcortical volume as a covariate. If there was a significant (α<0.05) effect of treatment group on total cortical or subcortical volume, totals were added to the model to investigate the local effects that could not be accounted for by global effects. For each significant between-­‐
group effect, effect size (Cohen’s d; Cohen, 1992) was calculated as the difference between the estimated marginal means divided by their pooled standard deviations. 133 False discovery rate (FDR) procedures (maximum acceptable FDR of 5%) accounted for multiple hypothesis testing (Benjamini & Hochberg, 1995). Structural brain differences between groups may be mediated by pre-­‐existing clinical differences between groups (e.g. in symptom severity), but adding such measures to the model as covariates eliminates variance of interest (Miller & Chapman, 2001). Continuous variables of significant difference between both treatment groups were thus entered as covariates only in a secondary step to provide an exploratory analysis of possible confounders. The contribution of the covariates to between-­‐group differences was assessed for each contrast by calculating the range of changes in effect size (Cohen’s d
–Cohen’s d
) and p-­‐values, and the average of absolute changes in effect size and p-­‐value, within brain volumes affected by treatment group. For categorical factors that may have confounded the results, sensitivity analyses were performed by repeating the analyses in each subgroup (e.g., patients with and without a history of atomoxetine treatment). RESULTS model with covariate
initial model
Clinical sample characteristics The sample consisted of 93 participants from 87 families, between the ages of 10 and 24 years with no age differences between the three participant groups (HC: M=16.7, SD=3.3, range=10.6-­‐24.8; ADHD : M=16.6, SD=3.0, range=10.6-­‐22.3; ADHD
: M=16.7, SD=3.3, range=10.2-­‐24.2). Eighty-­‐four percent of participants were male and 50% participated in Amsterdam, both variables being equally distributed over the three groups (Chi =0.000, p=1.00; Chi
=2.409, p<0.300). IQ in the ADHD sample was lower than in the healthy control sample (M =103.26, M =96.11, t=2.235, p=0.028), as was CGAS-­‐score of daily functioning (M =89.67, M =61.45, t=17.254 p=0.001). Patients in the MPH+AAP group had more ADHD symptoms than the MPH group. We found no other clinical differences (i.e., IQ, ADHD type, presence of comorbid diagnoses, scores on an autism spectrum questionnaire) between both ADHD groups (Table 1). Comorbid diagnoses included ODD/CD (n =10; n
=14; Chi =1.088, p=0.297), anxiety disorders (n =2; n
=0; Chi =2.067, p=0.151), and tic disorders (n =1; n
=1; Chi =0.000, p=1.000), which were equally distributed across the two groups. Within all patients who had been medicated, 81% had been prescribed immediate-­‐release methylphenidate preparations (n=50), 90% extended-­‐release methylphenidate preparations (n=56), and 15% d-­‐amphetamine preparations (n=9). The majority of patients had a treatment history of more than one stimulant type (n=46, 74%). Sixty-­‐one percent (n=38) of patients received stimulant treatment MPH
MPH+AAP
2gender
2location
HC
ADHD
HC
ADHD
MPH
MPH+AAP
2
MPH
MPH
134 2
MPH+AAP
2
MPH+AAP
within three months prior to scan (current users). Stimulant treatment duration ranged from 0.1 to 12.1 years with a mean(SD) of 5.4(3.3) years. We found no between-­‐group differences regarding stimulant treatment duration, age of stimulant treatment onset, age of treatment cessation, mean daily dose, and the proportion of current users (Table 1). Although not significant, more patients in the MPH+AAP group had a history of psychotropic medication treatment other than MPH+AAP. Medication other than MPH or AAP included clonidine (n =1; n
=5), atomoxetine (n =3; n
=15), melatonin (n =13; n
=19), antidepressants (n =1; n
=5), and anxiolytics (n =1; n
=1). A history of atomoxetine use was significantly more prevalent in the combined treatment group compared to in the MPH only group (Chi =11.3, p=0.001). In the MPH+AAP group, most patients had been prescribed risperidone (n=24), with a mean daily dose of 1.2 mg; other antipsychotics were pipamperone (n=8), quetiapine (n=1), olanzapine (n=1), and aripiprazole (n=1). Four participants had a history of two antipsychotic agents. Antipsychotic treatment duration ranged from 0.2 to 10.9 years (mean=3.7, SD=2.9), and age of antipsychotic treatment initiation ranged from 2.6 to 17.2 years (mean=10.7, SD=3.7). Fifty-­‐eight percent of patients (n=18) had ceased AAP treatment at least three months, and on average 3.8 years (SD=2.7), prior to scan. Subcortical-­‐limbic volumes There were no differences in total subcortical volume between the three groups. Local volume reductions approaching moderate effect sizes were found in the MPH+AAP group compared to the HC group (Figure 1), in the left (β=-­‐0.5204, p
<0.04, Cohen’s d=0.48) and right (β=-­‐0.4919, p
=0.05, Cohen’s d=0.46) ventral diencephalon and the left thalamus (β=-­‐0.5118, p
=0.04, Cohen’s d=0.47). A similar effect size was found in the right thalamus, but this effect only approached significance (β=-­‐0.471, p
=0.058, Cohen’s d=0.44). No between-­‐
group effects survived FDR-­‐correction for multiple testing. There were no local volume differences between the MPH and HC group, or between the MPH and MPH+AAP group. Cortical volumes Total cortical volume was reduced in the MPH+AAP group compared to the HC group with moderate effect size (p
=0.004; Cohen’s d=0.69; p
<0.075), but not between the MPH and HC group or between the MPH and MPH+AAP group. Results from the cortical VOI analyses are summarized in Table 2 and Figure 2A. The MPH
MPH
MPH+AAP
MPH
MPH+AAP
MPH
MPH+AAP
MPH+AAP
MPH
MPH+AAP
2
uncorrected
uncorrected
uncorrected
uncorrected
uncorrected
FDR-­‐corrected
135 MPH+AAP group showed significant volume reductions compared to the HC group in bilateral precentral gyrus, IFG, OFG, SFG, MFG, and left ACC, with effect sizes ranging from small (Cohen’s d=0.47 in left MFG) to large (Cohen’s d=0.80 in left ACC and right precentral gyrus). Effects in the right precentral gyrus (p
=0.028) and left ACC (p
=0.028) survived correction for multiple testing. The MPH group showed volume reductions compared to HC in the right SFG, the right OFG, and the left ACC, with moderate effect sizes ranging from 0.47 to 0.77, of which left ACC volume reduction survived correction for multiple comparisons (p
=0.028). Comparing MPH to MPH+AAP yielded a significant difference of moderate effect size in the right precentral gyrus (Cohen’s d=0.58), which did not survive FDR-­‐correction. Total cortical volume was added to the model to assess local rather than global effects. After adding total cortical volume, left ACC volume was reduced in both the MPH+AAP group (p=0.001; Cohen’s d=0.72) and the MPH group (p=0.03; Cohen’s d=0.49) compared to the HC group. Volume reduction in the MPH+AAP group survived FDR-­‐
correction (p
=0.042). FDR-­‐corrected
FDR-­‐corrected
FDR-­‐corrected
FDR-­‐corrected
FIGURE 1. Mean left and right ventral diencephalon volume (left graph) and left and right thalamus volume (right graph) of the three groups, with their standard errors. Uncorrected volumes are displayed in mm . The MPH+AAP group (light grey), but not the MPH group (dark grey), showed significant volume decrease compared to the HC group (black). 3
Exploratory analyses of clinical confounders Since the MPH+AAP group displayed more ADHD symptoms, greater functional impairment, and more patients with a history of atomoxetine treatment than the MPH group, we explored the contribution of these factors to structural brain differences. Adding total symptom count to the model affected the effect sizes and p-­‐
values in each contrast and in each brain region (Figure 2B, Table 3). Average absolute change in effect sizes (Cohen’s d
–Cohen’s d
) across brain regions affected by treatment group was 0.23 for the MPH+AAP vs. HC contrast (range: -­‐0.66 to 0.21), 0.18 for the MPH vs. HC contrast (range: -­‐0.61 to 0.29) and 0.05 (range: -­‐0.02 to 0.10) for the MPH vs. MPH+AAP contrast. The absolute average model with symptom count
136 initial model
change in p-­‐values was 0.25 and 0.30 for MPH vs. MPH+AAP and MPH vs. HC, respectively, and 0.06 for the MPH vs. MPH+AAP contrast. Adding CGAS scores for daily functioning to the model had a very similar effect: average absolute change in effect size was 0.20 for MPH+AAP vs. HC, 0.14 for MPH vs. HC and 0.05 for MPH vs. MPH+AAP. Average absolute change in p-­‐values was 0.21 for MPH+AAP vs. HC, 0.24 for MPH vs. HC and 0.03 for MPH vs. MPH+AAP. Thus, adding symptom count or functional impairment scores to the model influenced effect sizes and p-­‐values in contrasts involving HC subjects, but had a minimal impact on the contrast between the two treatment groups (Table 3). Last, the combined treatment group contained more patients with a history of atomoxetine treatment. To evaluate the possible confounding effect of atomoxetine treatment history, all tests with significant results were repeated in atomoxetine-­‐
naïve patients only. Our findings remained unchanged: excluding atomoxetine users did not change the direction of effect in any VOI, and all but three (left thalamus in the MPH+AAP vs. HC contrast, right superior frontal gyrus in the MPH vs. HC contrast, and right precentral gyrus in the MPH vs. MPH+AAP contrast) p-­‐values remained significant. No other clinical differences between the two treatment groups were found. 137 TABLE 2. Between-­‐group contrasts of total cortical volume and cortical volumes of interest. β-­‐estimates and corresponding p-­‐values for each between-­‐group contrast for total cortical volume and cortical VOIs. Negative β-­‐estimates indicate smaller volumes in the group first mentioned. Orbitofrontal gyrus Inferior frontal gyrus Precentral gyrus Superior frontal gyrus Anterior cingulate gyrus Frontal pole * = p
FDR-­‐uncorrected
< 0.05; † = p
MPH+AAP vs. HC MPH+AAP vs. MPH β p*† β p*† β p* -­‐0.32 0.10*† -­‐0.60 0.01*† 0.28 0.16* R -­‐0.32 0.15*† -­‐0.55 0.02*† 0.23 0.31* L -­‐0.38 0.09*† -­‐0.46 0.04*† 0.08 0.71* R -­‐0.55 0.01*† -­‐0.54 0.02*† -­‐0.01 0.95* L -­‐0.22 0.32*† -­‐0.54 0.02*† 0.32 0.16* R -­‐0.17 0.45*† -­‐0.58 0.01*† 0.42 0.06* L -­‐0.18 0.41*† -­‐0.59 0.01*† 0.40 0.07* R -­‐0.22 0.31*† -­‐0.79 0.01*† 0.57 0.01* L -­‐0.38 0.09*† -­‐0.64 0.01*† 0.27 0.23* R -­‐0.45 0.04*† -­‐0.53 0.02*† 0.08 0.73* L -­‐0.16 0.47*† -­‐0.55 0.02*† 0.39 0.08* R 0.04 0.85*† -­‐0.30 0.18*† 0.35 0.12* L -­‐0.75 0.01*† -­‐0.79 0.01*† 0.04 0.85* R -­‐0.29 0.19*† -­‐0.37 0.10*† 0.08 0.71* L -­‐0.12 0.59*† -­‐0.21 0.36*† 0.09 0.69* Total cortical volume Middle frontal gyrus MPH vs. HC FDR-­‐corrected
< 0.05 R: right; L: left; VOI: Volume of interest. DISCUSSION This study intended to provide an initial investigation of long-­‐term structural brain correlates of combined methylphenidate and atypical antipsychotics treatment in adolescent patients with ADHD. Compared to unaffected peers, patients who had received combined treatment showed reduced total cortical volume, which was reflected in volume reductions across the frontal cortex. In addition, these patients showed reduced local volumes of the bilateral ventral diencephalon and the left 138 TABLE 3. Exploratory analyses of clinical between-­‐group differences. Range of change (∆ -­‐ ∆ ) and min
max
average absolute change (µμ|∆|) in Cohen’s d effect sizes and p-­‐values across all fifteen volumes of interest* where a significant (FDR-­‐uncorrected) difference was found between any of the three groups, upon adding clinical, possibly confounding, factors to the regression model (model with clinical variable -­‐ initial model). ADHDMPH+AAP vs. HC ADHDMPH vs. HC ADHDMPH+AAP vs. ADHDMPH + total symptom count + functional impairment Range Abs. average Range Abs. average ∆ β -­‐0.66 – 0.21 0.23 -­‐0.56 – 0.09 0.20 ∆ p -­‐0.03 – 0.77 0.25 0.03 – 0.60 0.21 ∆ β -­‐0.61 – 0.29 0.18 -­‐0.50 – 0.15 0.14 ∆ p -­‐0.32 – 0.74 0.30 -­‐0.15 – 0.44 0.24 ∆ β -­‐0.02 – 0.10 0.05 -­‐0.14 – 0.04 0.05 ∆ p -­‐0.20 – 0.12 0.06 -­‐0.09 – 0.05 0.03 * total cortical volume, left thalamic volume, bilateral diencephalic volume, bilateral middle frontal, inferior frontal, superior frontal, orbitofrontal, and precentral volume, and left anterior cingulate cortex volume. ADHD, attention-­‐deficit/hyperactivity disorder; MPH, methylphenidate; AAP, atypical antipsychotics; HC, healthy control thalamus. Patients treated with methylphenidate solely, by contrast, showed no reduction in total cortical or subcortical volume compared to unaffected peers, nor in any of the subcortical-­‐limbic volumes of interest. Patients in the MPH+AAP group displayed more ADHD symptoms, functional impairment, and comprised a higher incidence of atomoxetine users compared to patients in the MPH only group. Adding these covariates to the model had minimal impact on differences in brain structure between the two ADHD groups. Reduced total cortical volume and frontal cortical volumes are among the most replicated findings in ADHD and have repeatedly been shown to be normalized in patients using psychostimulants (Castellanos et al., 2002; Frodl & Skokauskas, 2012; Shaw et al., 2009). Within the frontal cortex, we found little evidence of spatial specificity. The reduction of total cortical volume in the MPH+AAP group is reflected in all frontal regions of interest except the frontal poles. All frontal volume reductions in the MPH+AAP group were driven by total cortical volume reduction, with the exception of left ACC volume reduction. However, left ACC volume reduction was also found in the MPH group. Cortical volume reduction associated with combined methylphenidate and antipsychotic treatment thus appears to be global rather than local. This is in line with previous studies suggesting that ADHD itself may be associated with global rather than local cortical changes (Nakao et al., 2011). It is less 139 clear whether stimulants have local or global effects on brain structure, since the majority of previous studies adopted a regions of interest approach. Future investigations of the effects of stimulants, antipsychotics, and combined treatment may benefit from a whole-­‐brain approach. In the subcortical-­‐limbic regions, we found volume reduction in the bilateral ventral diencephalon and left thalamus in the combined treatment group, but not in the methylphenidate only group. The ventral diencephalon includes the subthalamic nuclei and substantia nigra. The subthalamic nuclei are strongly connected within the basal ganglia, and have been attributed an important role in response inhibition (Ray et al., 2012). The substantia nigra is the largest dopaminergic nucleus in the human brain, strongly connected to the striatum, and is thought to play an important role in reward (Krebs et al., 2011) and movement (Obeso et al., 2008). The ventrolateral portion of the thalamus relays and modulates neurotransmission in the frontostriatal circuits, connecting the cerebellum and basal ganglia to the cortical motor areas. Although these thalamic and subthalamic functions are highly relevant to the clinical presentation of ADHD, the role of thalamic and subthalamic nuclei in ADHD pathophysiology remains poorly understood. Thalamic volume reduction (Ivanov et al., 2010), altered left thalamic morphology and structural connectivity (Xia et al., 2012), and normalized thalamic morphology with methylphenidate treatment (Ivanov et al., 2010) have previously been reported in ADHD. We found no previous reports of thalamic or subthalamic changes with childhood antipsychotic treatment. Further investigation of thalamic and subthalamic structures in ADHD and their susceptibility to treatment, is needed to interpret the thalamic and subthalamic changes we observed in the combined treatment group. The current study was explorative and we warrant cautious interpretation of our findings. Participants in the MPH+AAP group showed more pronounced structural brain changes relative to healthy controls than stimulant-­‐treated patients with ADHD, thereby resembling the non-­‐medicated patient groups in previous studies (Spencer et al., 2013). Volume reductions in the combined treatment group but not in the methylphenidate only group could be indicative of counteractive effects of methylphenidate and atypical antipsychotics. That is, the opposing mechanism of action may compromise the synaptic effects of both individual agents, resulting in attenuation of structural normalization typically occurring with methylphenidate treatment. However, the current study did not allow direct investigation of this hypothesis and multiple interpretations are possible. Two shortcomings in study design prevent us from rejecting two plausible alternative interpretations. First, our study was observational and cross-­‐sectional. Our study is in many ways different from randomized controlled trials (RCTs), which are considered the gold standard when investigating treatment effects. First, in the current study no pre-­‐
140 treatment measurement was performed. The structural changes in the combined treatment group may be the cause rather than the result of medication intake (i.e., antipsychotic treatment may be assigned to patients at higher risk for cortical volume reduction). Pre-­‐treatment measurements are needed to exclude this possibility. As a substitute to pre-­‐treatment assessment, we explored the contribution of several clinical measures at time of scan. Our findings suggested that adding such variables to the model only minimally affected differences between the combined treatment group and the methylphenidate only group. However, the possibility of pre-­‐treatment differences confounding treatment effects cannot be excluded, and the exploratory analyses we applied are not conclusive in this respect. Second, due to the absence of a matched stimulant-­‐naïve patient sample, we were unable replicate previous findings of structural normalization with methylphenidate treatment in the current sample. Future studies would benefit from the inclusion of a stimulant naïve patient group, to enable the interpretation of the individual contributions of both stimulant and antipsychotic treatment on brain structure. Besides these two essential caveats for correct interpretation of our findings, several other limitations should be kept in mind. The naturalistic study design resulted in a heterogeneous patient sample, and the age-­‐range of participants was very wide. The combined treatment group included patients who were receiving treatment within a week prior to scanning as well as patients who had ceased treatment years before study participation, and patients who initiated treatment before the age of four as well as patients who initiated treatment after the age of sixteen. Due to limited power, we were unable to assess the effects of factors such as treatment duration, timing of treatment, co-­‐medication and daily dose within the combined treatment group. Increased variance within our sample may have limited our ability to detect clinically meaningful effects. Apart from being heterogeneous, our sample was also relatively small, and subtle brain changes of small effect size may therefore have gone undetected. In line with this, most of our findings did not survive FDR correction for multiple testing. Last, as we were unable to obtain complete medication transcripts from all pharmacies, our data may have been subject to recall bias. Although a recent report suggested that recall bias may be limited in ADHD (Kuriyan et al., 2014), it cannot be fully excluded, given our long-­‐term retrospective study design. At the same time, there are several advantages to our study design. Large-­‐
scale observational cohort studies, such as the current study, investigate patients that are representative of the heterogeneous clinical population, which enhances generalizability. Moreover, observational studies allow the investigation of complex treatment patterns in vulnerable patient groups, such as children with ADHD (Silverman, 2009). Results from observational studies are generally more consistent 141 than results of RCTs and susceptibility bias does not typically result in an overestimation of treatment effects (Concato et al., 2000; Concato, 2013). Moreover, neuroimaging is a valid tool to investigate treatment safety and efficacy in observational study designs (Buitelaar & Coghill, 2013). Therefore, while awaiting replication, the current study may have implications. Our findings, although preliminary, raise concern and stress the need for clinical guidelines for the prescription of these agents (Panagiotopoulos et al., 2010). Furthermore, they emphasize that future studies investigating the effects of methylphenidate on the developing brain should carefully document co-­‐medication with dopaminergic agents, including atypical antipsychotics. In conclusion, we found reduced total cortical volume, that was reflected in local frontal volume reductions, as well as volume reductions in the bilateral ventral diencephalon and left thalamus, in patients with ADHD who had received combined methylphenidate and atypical antipsychotic treatment. These structural anomalies were smaller or absent in patients who were treated with methylphenidate solely. Longitudinal studies, including pre-­‐treatment measurements and a stimulant-­‐naïve patient group, are needed to allow more conclusive interpretation of potential mechanisms leading to these volume reductions. 142 143 144 Chapter 8 MEMORY PERFORMANCE AND HIPPOCAMPUS
STRUCTURE AFTER INFREQUENT RECREATIONAL
STIMULANT USE IN YOUTH
Submitted as: Schweren LJS, Giedd J, Castro N, Bartsch H, Squeglia LM, Meloy MJ, Tapert S. Memory performance and hippocampus structure after infrequent recreational stimulant use in youth. 145 ABSTRACT
Stimulant use has been associated with memory problems and hippocampal changes in adult recreational users. Animal studies have shown that the developing brain may be especially vulnerable to these effects, but studies in human adolescents are scarce. Here, we selected three demographically matched groups of youth from a prospective neuroimaging study, based on their substance use patterns. Participants (N=98, MAGE=14.7) were stimulant-­‐naïve at baseline, and classified at ~5 year follow-­‐up as either nonuser controls (n=31), non-­‐stimulant users (e.g., marijuana/alcohol, n=33), or stimulant-­‐users (n=34). Neuropsychological testing and high-­‐resolution magnetic resonance imaging were collected at each time point. Across these ~5 years of late adolescence from baseline to follow-­‐up, performance on a verbal learning and memory test improved for non-­‐stimulant users (+5.9%, ptime=0.023) but not for stimulant users (-­‐1.3%, ptime=0.413; pgroup-­‐by-­‐time=0.032). Improvement for nonuser controls was not significant (+3.2%, ptime=0.181). Developmental trajectories of hippocampal volume were not different between groups. These findings suggest some very subtle yet detectable disadvantages for youth who used recreational stimulant drugs, even on a limit basis (i.e., one to two instances). 146 INTRODUCTION
A substantial proportion of recreational stimulant users (i.e., non-­‐dependent users of amphetamine, methamphetamine, 3,4-­‐methylenedioxy-­‐methamphetamine [MDMA or ecstasy], cocaine, and prescription stimulants including methylphenidate and d-­‐amphetamine) initiate stimulant use before the age of eighteen. Prevalence estimates of adolescent stimulant use range from 0.9 to 9.1 percent, with MDMA and cocaine use being most common in the U.S. (Frieden et al., 2014; Johnston et al., 2015; SAMHSA, 2014). Up to 65% of youngsters associate no risks with low-­‐frequency recreational stimulant use (Johnston et al., 2015), although the long-­‐term effects of adolescent recreational stimulant use on brain development are largely unknown. Stimulants are indirect monoaminergic agonists, modulating the availability of extracellular serotonin, norepinephrine, and dopamine in the brain. Monoaminergic receptors are abundantly expressed in the hippocampus, which receives afferent dopaminergic projections from the ventral tegmental area and serotonergic projections from the raphe nuclei. Animal studies have shown that chronic stimulant exposure results in impaired hippocampal neurogenesis, a process of stem cell renewal thought to be important in learning and memory (Eisch & Harburg, 2006). Chronic high-­‐dose stimulant exposure has been associated with a wide range of electrophysiological and molecular changes in the hippocampus, including reduced long-­‐term potentiation (Onaivi et al., 1996), reduced serotonergic and dopaminergic receptor binding (Armstrong & Noguchi, 2004), and memory impairments (Belcher et al., 2005) that may persist up to several months after exposure (van Nieuwenhuizen et al., 2010). Human studies have mostly focused on chronic cocaine and methamphetamine use in severely addicted adults. Brain changes including enlarged striata and widespread decreased grey matter volume, particularly in frontal, temporal, and hippocampal regions (Berman et al., 2008; Ersche et al., 2013), and impaired memory and learning have been reported in these patients (Potvin et al., 2014; Scott et al., 2007). Less is known about the effects of occasional illicit stimulant use. In an animal model of recreational stimulant use, single-­‐episode binge exposure to stimulants has induced serotonin depletion and memory deficits lasting up to several weeks (Biezonski & Meyer, 2010) or months (McGregor et al., 2003) after exposure. In human recreational users of MDMA (Murphy et al., 2012) and prescription stimulant misuse (Reske et al., 2010), deficits in verbal memory have been found. Neuroimaging studies, however, have yielded mixed results, partially due to methodological and participant group differences. Reduced frontal grey matter volume (Cowan et al., 2003; Daumann et al., 2011) and changes in striatal structure and function (De Win et al., 2008; Mackey et al., 2014) have been reported. Structural changes in the 147 hippocampus have not been found in whole-­‐brain studies (Cowan et al., 2003; Daumann et al., 2011; Mackey et al., 2014) nor when the hippocampus was specifically targeted as a region of interest (Koester et al., 2012). By contrast, recreational stimulant users did show reduced blood-­‐oxygen-­‐level dependent (BOLD)-­‐response in the left hippocampus and parahippocampal gyrus compared to nonusers during memory encoding (Becker et al., 2013; Roberts et al., 2009). In sum, memory deficits and reduced hippocampal activity during memory tasks have been seen in occasional stimulant users, but structural hippocampal changes have not been observed. The sparse literature regarding the effects of occasional stimulant use on hippocampal structure and memory has limitations. First, the majority of studies addressing memory performance, and all investigations of brain structure, in occasional stimulant users had a cross-­‐sectional design. Any alterations or deficiencies reported in this group may have existed prior to stimulant exposure, potentially indicating a risk factor for, rather than a consequence of, recreational stimulant use. Second, three out of four structural neuroimaging studies in occasional users utilized voxel-­‐based morphometry (VBM) techniques, optimized to detect volumetric changes in contiguous grey matter structures such as the cortex. Non-­‐
volumetric features, such as surface morphology of the hippocampus, may develop independent of volumetric changes, and may be more sensitive to subtle changes. Moreover, subcortical and limbic structures are composited of both grey and white matter. The absence of macroscopic grey matter changes does not preclude the presence of changes in white matter microstructure. In fact, a recent study found indications of loss of hippocampal myelin proteins and tissue integrity in patients with mild memory impairments, in the absence of volumetric changes (Granziera et al., 2015). Generalizability of stimulant effects in young adult occasional users to adolescents is uncertain. The mechanism by which stimulant abuse causes hippocampal damage in adult heavy users and memory deficits in adult occasional users likely applies to adolescents as well. During adolescence, however, the brain undergoes a series of complex developmental changes, rendering the adolescent brain especially vulnerable to external influences such as substance exposure and toxicity (Andersen & Navalta, 2011). Major modifications of brain architecture during adolescence include widespread cortical thinning (Giedd et al., 2009), and increasing myelination of cortico-­‐cortical connections (Uda et al., 2015). Hippocampus volume and entorhinal cortical thickness typically reach their peak before adolescence, after which the adolescence phase is characterized by relative stability (Krogsrud et al., 2014; Tamnes et al., 2014; Wierenga et al., 2014). By contrast, hippocampal white matter continues to develop throughout adolescence and into early adulthood, with 148 increasing fractional anisotropy (FA) and decreasing mean diffusivity (MD; Bava et al., 2010; Tamnes et al., 2010), both indicative of increased myelination. Ongoing development may sensitize the adolescent brain to lasting changes, i.e., may increase the severity of structural brain changes after stimulant use, and/or may increase the likelihood of structural changes after less frequent stimulant exposure. Such age-­‐
dependent effects have previously been reported in rats, that exhibited memory deficits and hippocampal shape changes after low-­‐dose stimulant exposure during adolescence, but not after exposure at later age (van der Marel et al., 2014). In humans, adolescent but not adult exposure to MDMA predicted 5-­‐HT transporter density in midbrain (Klomp et al., 2012). In this prospective longitudinal neuroimaging study, we investigated changes in hippocampus structure and verbal memory performance in adolescent recreational stimulant users. Changes in the stimulant users group were compared to typical developmental changes in a matched group of adolescents not using any substances, and to changes in a matched group of non-­‐stimulant substance users (e.g., users of alcohol, marijuana, and/or cigarette smokers). We hypothesized that adolescent stimulant users would present with subtle yet detectable signs of memory impairment and changes in hippocampal structure, i.e., hippocampal total volume reduction and/or localized surface changes over time and decreased hippocampal white matter integrity, compared to both control groups. METHODS
Procedure This study was part of an ongoing neuroimaging study of adolescents at familial risk for substance use problems. A total of 295 adolescents 12 to 14 years of age, who had had no or minimal exposure to substances, were recruited through local middle schools. The sample was enriched for adolescents with a family history of substance use disorders. Assessment included interviews and questionnaires regarding substance use history, background, family history, and mental health functioning, all obtained from the adolescent, a biological parent or legal guardian, and a close relative, as well as neuropsychological testing and a magnetic resonance imaging (MRI) session for the adolescent. After enrollment, participants were administered substance use interviews by phone every six months, and were invited for complete assessment including an MRI scan annually. All scans included in the current study were obtained between 2005 and 2015 on the same 3T scanner. The study protocol was approved by the University of California, San Diego, Human 149 Research Protections Program. Adolescents and their parents (while the adolescent was <18 years) signed informed assent and consent, respectively, at each assessment. Participants Exclusion criteria at study enrollment included the presence of any DSM-­‐IV axis I disorder, mental retardation or learning disabilities, a history of chronic medical illness or head trauma, prenatal alcohol or illicit drug exposure, and contra-­‐indication to MRI (for details, see Squeglia et al., 2015). Participants who had successfully completed at least two 3T structural MRI scans, reported no psychostimulant use prior to the first scan, and had no medical use of prescription psychostimulants at any time (e.g., for ADHD) were selected for this study. When more than two successful 3T measurements were available, one baseline and one follow-­‐up measurement were selected such that 1) baseline represented the earliest valid 3T measurement, 2) for participants reporting stimulant use, follow-­‐up represented the first follow-­‐up measurement after the last instance of stimulant use [to maximize the amount of stimulant exposure captured, while minimizing the delay between exposure and scan], and 3) for participants who did not report stimulant use, follow-­‐up duration optimally matched the average follow-­‐up duration across participants who did report stimulant use. Average follow-­‐up interval was 5.1 years (range: 0.8-­‐8.9 years). Of all eligible participants (n=199), 35 reported stimulant use between baseline and follow-­‐up (17.6%, ‘stimulant-­‐users’ or ‘STIM’). One participant reporting a pattern of regular rather than occasional stimulant use (>10 instances of stimulant use per year) was excluded. Of those participants reporting no stimulant use (i.e., controls), 71 reported no or minimal substance use between baseline and follow-­‐up (‘nonuser controls’ or NU-­‐CON, 35.7%, reporting ≤1 instance of being drunk, using marijuana or other drugs, or smoking cigarettes, per year), and 93 reported substance use, but not stimulant use (‘non-­‐stimulant controls’ or NS-­‐CON, 46.7%). Next, both control groups were optimally matched to the stimulant-­‐users group. Participants were one-­‐to-­‐one matched on baseline age (+/-­‐ 1SD), follow-­‐up duration (+/-­‐ 1SD), and sex, the latter criterion being dropped if necessary (n=2). Both control groups were group-­‐level matched to the STIM group regarding family history of substance use problems. Furthermore, the NS-­‐CON group was group-­‐matched to the STIM group regarding the use of non-­‐stimulant substances between baseline and follow-­‐up. The latter matching was only partially successful (Table 1). For one stimulant-­‐user, no match could be found in the NS-­‐CON group, and for three stimulant-­‐users no match could be found in the NU-­‐CON group. The final sample consisted of 34 participants in the STIM group, 33 participants in the NS-­‐CON group, and 31 participants in the NU-­‐
CON group. 150 151 51.6 14.5 4.7 1.2 0.0 0.0 0.0 0.0 0 -­‐ % with family history of SUD Baseline age Follow-­‐up interval in years # CVLT sessions (BL-­‐FU) # instances of alcohol use (before BL) # instances of substance use (before BL) Freq. of alcohol intoxication (times/year, BL-­‐FU) Freq. of marijuana use (times/year, BL-­‐FU) Freq. of tobacco use (cigarettes/year, BL-­‐FU) Days since last substance use (FU) -­‐ 0 0.0 0.0 0.0 0.0 1.2 1.8 1.8 6.2 326 28.1 13.7 4.5 4.0 1.8 5.3 14.8 63.6 63.6 M 5.8 991 50.7 16.1 15.8 13.1 1.4 1.8 1.7 SD NONUSER (n=31) 5.0 106 53.0 25.7 8.3 1.3 1.3 5.4 19.4 50.0 58.8 M 3.7 222 65.0 25.2 24.5 3.8 1.3 1.8 1.9 SD NONUSER (n=31) -­‐ -­‐ -­‐ -­‐ -­‐ -­‐ -­‐0.333 -­‐1.643 -­‐0.796 0.017 0.341 T / Chi2 STIM vs. NU-­‐CON 1.001 1.245 -­‐1.752 -­‐2.332* -­‐0.747 1.138 1.562 -­‐0.322 -­‐0.207 1.268 0.163 T / Chi2 STIM vs. NOSTIM 0.065 0.269 0.008 0.239 0.084 0.093 -­‐0.163 -­‐0.098 -­‐0.078 -­‐0.961 -­‐0.429 r / T Assoc. with stim use freq. BL baseline; FU follow-­‐up; NU-­‐CON nonuser control group; NS-­‐CON non-­‐stimulant control group; STIM stimulant users group; CVLT California Verbal Learning Test; SUD substance use disorder; * p<0.05 51.6 % male M SD NONUSER (n=31) TABLE 1. Demographic and substance use characteristics of nonuser controls, non-­‐stimulant controls, and stimulant users Substance use Self-­‐reported frequency of substance use was assessed with the Customary Drinking and Drug Use Record (CDDR; Brown et al., 1998). Reports from a second informant (e.g., parent, sibling, or friend) were collected for confirmation. The CDDR assesses the use of stimulants, as well as alcohol, marijuana, barbiturates, hallucinogens, inhalants, opiates, benzodiazepines, ketamine, gamma-­‐hydroxybutyric acid (GHB), phencyclidine (PCP, or ‘angel dust’), and cigarette smoking. Stimulant use could be reported on the following items: amphetamine-­‐type stimulants (ATS, “speed, crystal, meth, Ritalin, Adderall, stimulant pills, performance enhancing, ripped fuel, ephedrine, diet pills”), cocaine (“coke, blow, crack”), MDMA/ecstasy (“MDMA, MDA, E, X, Rolls, Molly”), and “misuse of prescription drugs.” Participants were classified as stimulant users if they reported ≥1 instance of ≥1 stimulant type between baseline and follow-­‐up. Annual stimulant use frequency was calculated as the total number of instances of stimulant use (of any type) between baseline and follow-­‐up, divided by follow-­‐up duration. Lifetime stimulant use frequency was calculated by dividing the lifetime number of stimulant use instances by age at follow-­‐up. Substance use frequencies were also calculated for each stimulant type separately, and for alcohol (instances of ‘being drunk’), marijuana, and cigarettes. Memory performance Verbal learning and memory performance was assessed using the California Verbal Learning Test for Children (CVLT-­‐C; Delis et al., 1994) in the first four years of the study and the CVLT-­‐II adult version (Delis et al., 2000) in later years. A list of common words belonging to word categories was presented to the participant five times. After the fifth trial, an interference list of new words was read once, and the participant was asked to recall items from the interference list. After a 20-­‐minute delay, during which testing continued with nonverbal tasks, a delayed recall of the original list was requested. The delayed recall memory score (CVLT-­‐LD) is defined as the percentage of words correctly stated after the 20-­‐minute interval. CVLT-­‐LD has been associated with variation in hippocampus volume in healthy populations, and is sensitive to changes in adolescent memory performance after substance exposure (Ashtari et al., 2011; Brown et al., 2000; Wright et al., 2015). MRI data acquisition All neuroimaging data reported here were collected with the same 3T General Electric (Milwaukee, Wisconsin) scanner with an 8-­‐channel phase-­‐array head coil, at 152 the University of California, San Diego, Keck Center for Functional MRI. Eight high-­‐
bandwidth receivers for ultrashort repetition times reduced signal distortion and signal dropout. Imaging included a sagittal high-­‐resolution 3D T1-­‐weighted structural acquisition (FOV=24 cm, 0.94×0.94×1 mm voxels, 176 slices, TR=20 ms, TE=4.8 ms; flip angle=12°). Diffusion-­‐weighted imaging was added later in the study; 66% (n=65) of participants received at follow-­‐up an axial diffusion-­‐weighted single-­‐shot dual spin echo acquisition (FOV = 24 cm, matrix size = 96x96; 53 interleaved slices, slice thickness = 2.5 mm; b=1000, 30 diffusion directions plus 2 b reference images, TR=12000, optimized TE). MRI analyses Left and right hippocampi were segmented from the original T1 images through the automated FMRIB integrated registration and segmentation tool pipeline (FIRST) of the FMRIB Software Library (FSL; Patenaude et al., 2011), with default settings. FIRST registers each scan to standard space (MNI152 with 1x1x1 mm resolution), and then integrates shape and intensity information into a Bayesian, probabilistic framework to accurately segment 15 subcortical structures, including the bilateral hippocampi. FIRST’s output consists of volumetric representation of the bilateral hippocampi, as well as a mesh representation of their surface. Images were visually inspected after each processing step (skull-­‐stripping, registration to standard space, and automated segmentation), and settings were manually adapted where needed. Surface meshes were filled to allow spatial statistics on the hippocampus surface shape, with global scaling to remove volumetric effects. Statistical modeling was performed in FSL Permutation Analysis of Linear Models (PALM; Winkler et al., 2014) with threshold-­‐free cluster enhancement (1000 permutations) for cluster inference, and with family-­‐wise error (FWE) adjustment for multiple testing. Diffusion scans (follow-­‐up only) were processed through an automated pipeline provided by the Multimodal Imaging Laboratory (MMIL) of the University of California, San Diego, integrating diffusion-­‐, T2-­‐, and high-­‐resolution T1-­‐weighted acquisitions. Diffusion-­‐weighted volumes were registered to the b reference volume by rigid body registration. Eddy current distortions were removed, and magnetic susceptibility artifacts were minimized using a nonlinear b -­‐unwarping method that enables highly accurate registration with the T1-­‐weighted images. Next, a diffusion tensor was fitted. Transformation matrices resulting from intra-­‐modal registrations (b to T1) were used to transform hippocampal segmentation masks to diffusion space. Average fractional anisotropy (FA) and mean diffusivity (MD) within these hippocampal masks were extracted. In addition, T1-­‐T2 ratio, indicative of myelin content (Glasser & van Essen, 2011), within the FreeSurfer hippocampal 0
0
0
0
153 segmentation mask was calculated as the average T1-­‐weighted voxel intensity (normalized to the corresponding bias-­‐field image to reduce artifacts), divided by the average T2-­‐weighted voxel intensity from the b diffusion image (adjusted for voxel intensity of cerebral spinal fluid in the same image, to reduce scanner-­‐ or upgrade-­‐
specific effects). Statistical testing In SPSS linear mixed effect models, we predicted memory performance and hippocampal volume from time (baseline vs. follow-­‐up), group (STIM vs. NS-­‐CON, or STIM vs. NU-­‐CON), and time-­‐by-­‐group-­‐interaction, while including baseline age, sex, and follow-­‐up duration as covariates. The effect of interest is captured in the time-­‐by-­‐
group interaction, testing whether memory performance/hippocampus volume changed differently over time for stimulant-­‐users compared to controls. For the vertex-­‐wise analysis of the hippocampal surface, the same longitudinal model was implemented in PALM. In the absence of diffusion measurements at baseline, follow-­‐
up hippocampal FA, MD, and T1-­‐T2 ratio were predicted from group (STIM vs. NS-­‐
CON, and STIM vs. NU-­‐CON), including the covariates sex and age at follow-­‐up. Second, we performed generalized additive modeling (GAM) regression analyses using the mgcv-­‐package in R (Wood, 2011), to assess associations between stimulant use frequency and each outcome measure (memory performance, hippocampus volume, FA, MD, and T1-­‐T2 ratio) within the STIM group. Memory performance and hippocampus volume at follow-­‐up were predicted from sex, baseline age, follow-­‐up duration, baseline memory performance/hippocampus volume, and frequency of alcohol intoxication, marijuana use, cigarette smoking, and stimulant use between baseline and follow-­‐up. The same model was again implemented in PALM to assess hippocampal surface changes associated with stimulant use frequency. The cross-­‐sectional outcomes FA, MD, and T1-­‐T2 were predicted from sex, age at follow-­‐
up, and lifetime frequency of alcohol intoxication, marijuana use, cigarette smoking, and stimulant use. Each of the substance use frequency predictors were positively skewed, with the majority of participants reporting infrequent use, hence the predictors were log-­‐transformed. In R, stimulant use frequency was added as a smooth regression term with no restrictions to the link function, to allow detection of non-­‐linear associations. For each significant stimulant effect, we evaluated the influence of potential confounders in a second step by adding the following additional covariates to the model: 1) an effect-­‐by-­‐sex interaction term, 2) an effect-­‐by-­‐age interaction term, 3) for contrasts not including the NU-­‐CON group: number of days since last substance 0
154 use instance, to account for potential acute substance effects, and 4) any demographic variable associated with stimulant use. For memory performance, alpha was divided by two (α=0.05/2=0.025; two case-­‐control contrasts) to adjust for multiple testing. For the hippocampus measures, alpha was divided by six (α=0.05/6=0.008; two times three outcome measures [memory performance, and hippocampus volume, shape, and microstructure]). Power to detect case-­‐control differences meeting this alpha-­‐level given our limited sample size was low, i.e., 0.67 for differences of large effect size (Cohen’s d=0.8) and 0.22 for differences of medium effect size (Cohen’s d=0.5). Preliminary outcomes meeting nominal significance are therefore also reported (α=0.05). TABLE 2. Cumulative lifetime stimulant use and stimulant use frequency between baseline and follow-­‐up, within the stimulant users group. N % Stimulants (any type) 34 100.0 MDMA 20 58.8 ATS 16 47.1 Cocaine 14 41.2 Lifetime use (instances) M SD Median Range 9.0 12.9 3.5 1-­‐60 3.5 3.5 2.0 1-­‐14 10.2 14.3 6.5 1-­‐60 5.2 3.0 1-­‐23 6.7 Frequency (instances/year) M SD Median Range 1.7 2.2 0.8 0.2-­‐9.6 0.7 0.8 0.4 0.1-­‐2.9 1.9 2.3 1.6 0.2-­‐9.6 1.0 1.2 0.4 0.1-­‐3.8 MDMA 3,4-­‐methylenedioxy-­‐methamphetamine, ATS amphetamine-­‐type stimulants RESULTS
Ninety-­‐eight participants (57 males, 58.1%) were included in three sex-­‐ and age-­‐matched groups (Table 1). Average age was 14.7 years at baseline and 19.8 years at follow-­‐up. Despite optimal matching, participants in the STIM group reported more frequent alcohol intoxication between baseline and follow-­‐up compared to the NS-­‐
CON group (25.7 instances/year vs. 13.7 instances/year). Frequency of marijuana use and cigarette smoking did not differ between the groups. Substance use other than stimulants, alcohol, marijuana, or nicotine included hallucinogens (n=24), opiates (n=8), benzodiazepines (n=7), inhalants (n=5), and ketamine (n=3). At follow-­‐up, 97 percent of participants in the combined STIM and NS-­‐CON group had used substances (either alcohol or drugs) within 30 days prior to study participation, with an average of 6 days since their most recent use. The STIM group included users of amphetamine-­‐type stimulants (n=16, 47.1%), cocaine (n=14, 41.2%), and MDMA (n=20, 58.8%), with 38% (n=13) of stimulant-­‐users reporting more than one stimulant type. The average frequency of stimulant use (any type) was 1.7 instances 155 per year (range: 0.2-­‐9.6), adding up to on average 9 lifetime instances (range: 1-­‐60; Table 2). Verbal memory performance The time-­‐by-­‐group interaction effect for stimulant users versus non-­‐
stimulant controls reached marginal significance (p=0.032; Figure 1). Whereas CVLT-­‐
LD memory performance improved in the NS-­‐CON group (M
=79.9%, M
=85.8%, β =0.424, pTIME=0.023), test performance was stable in the STIM group (M
=81.5%, M
=80.2%, β =-­‐0.145, pTIME=0.413). Change in performance for the NU-­‐CON group resembled that of the NS-­‐CON group but was not significant (M
=82.3%, M
=85.5%, β =0.251, p=0.181). The time-­‐by-­‐group interaction effect was not significant for stimulant users versus nonuser controls (p=0.127). At baseline, the STIM group performed similar to the NS-­‐CON group (p=0.587) and the NU-­‐CON group (p=0.295). At follow-­‐up the STIM group performed worse compared to the NU-­‐CON group (p=0.024) but not compared to the NS-­‐CON group (p=0.095). The difference in slope between stimulant users and non-­‐stimulant controls did not interact with sex (p=0.746) or age (p=0.173), and remained unchanged when accounting for potential acute substance use effects (pDAYS-­‐SINCE-­‐USE=0.389; pSTIMvs.NOSTIM=0.029) and frequency of alcohol intoxication (pALC-­‐FREQ=0.430; pSTIMvs.NOSTIM=0.033). Memory performance was not associated with stimulant use frequency between baseline and follow-­‐up within the stimulant-­‐users group (p=0.518). BASELINE
UP
FOLLOW-­‐
TIME
BASELINE
FOLLOW-­‐UP
TIME
BASELINE
FOLLOW-­‐UP
TIME
FIGURE 1. Change in CVLT delayed free recall performance (y-­‐axis: percentage correctly recalled) between baseline and follow-­‐up (x-­‐axis: age) in nonuser controls (blue), non-­‐stimulant controls (green), and stimulant users (red). Test performance is likely to improve as a result of exposure to the CVLT materials. Between baseline and follow-­‐up, participants performed the CVLT between 0 and 6 times (M=1.33, SD=1.26), with no difference between groups. The average delay between the last CVLT session (either baseline or an intermediate assessment) and the follow-­‐up session was 3.1 years (SD=1.6), with no difference between groups. 156 Eighty-­‐three percent of participants had been administered the children’s version at baseline and the adult version at follow-­‐up, with no difference between groups. Hippocampal volume Left and right hippocampal volumes changed little between baseline and follow-­‐up in all three groups. Changes in hippocampal volumes over time in the STIM group did not differ from those in the NU-­‐CON group (left: p=0.122; right: p=0.057) or the NS-­‐CON group (left: p=0.092; right: p=0.678). Across groups, changes over time were not significant (left: β =-­‐0.034, p=0.460; right: β =-­‐0.058, p=0.296), and there were no between-­‐group differences in hippocampus volume at baseline or follow-­‐up. Within the STIM group, stimulant use frequency was not associated with left or right hippocampus volume (left: p=0.121; right: p=0.415). TIME
TIME
FIGURE 2. Left: Right hippocampal region of significant (pFWE<0.05 in red, pFWE<0.01 in yellow) group-­‐by-­‐
time interaction. Right: surface expansion (y-­‐axis) by age (x-­‐axis) in the nonuser control group and the stimulant users group. Hippocampal surface Vertex-­‐wise analyses revealed no significant differences in hippocampus surface morphology between stimulant users and nonuser controls, or between stimulant users and non-­‐stimulant controls, at either baseline or follow-­‐up. The time-­‐
by-­‐group (STIM vs. NU-­‐CON) interaction was significant in a small cluster on the right lateral hippocampus surface (3.5% of surface voxels at p <0.05, 0.4% at p <0.01, Figure 2). Within this cluster, surface expansion over time was found in the NU-­‐CON group (β =0.493, p<0.001) but not in the STIM group (β =0.087, p=0.267). The effect was the same for male and female participants (p
=0.216), and for participants of different baseline ages (p
=0.495). Stimulant use frequency was not associated with surface expansion in the identified cluster (p=0.422), nor with surface changes in other vertices. Finally, there were no significant time-­‐by-­‐group FWE
TIME
FWE
TIME
TIME*GROUP*SEX
TIME*GROUP*AGE
157 interaction effects for stimulant users vs. non-­‐stimulant controls anywhere on the cortical surface. Hippocampal microstructure At follow-­‐up, there were no significant between-­‐group differences in hippocampal FA, MD, or T1-­‐T2 ratio between stimulant users and nonuser controls (left: pFA=0.324, pMD=0.824, pT1-­‐T2=0.194; right: pFA=0.121, pMD=0.822, pT1-­‐T2=0.657), or between stimulant users and non-­‐stimulant controls (left: pFA=0.395, pMD=0.212, pT1-­‐
T2=0.081; right: pFA=0.467, pMD=0.574, pT1-­‐T2=0.388). Within stimulant-­‐users, none of the microstructure parameters were associated with lifetime stimulant use frequency (left: pFA=0.862, pMD=0.970, pT1-­‐T2=0.241; right: pFA=0.857, pMD=0.483, pT1-­‐T2=0.168). DISCUSSION
We investigated changes in verbal memory performance and hippocampus structure in adolescent recreational stimulant users over a 5-­‐year period. We found a modest time-­‐by-­‐group interaction effect on CVLT delayed recall score: between baseline and follow-­‐up, memory test performance improved in the non-­‐stimulant control group, but not in the stimulant users group. Second, we found a small cluster on the right lateral hippocampal surface where stimulant users did not show typical surface expansion with increasing age. There was no effect of stimulant use on hippocampal volume or white matter microstructure. We had hypothesized that stimulant-­‐users would show subtle memory deficits compared to both control groups. In the NS-­‐CON group, delayed recall memory performance improved by 5.9% from baseline to follow-­‐up, in line with practice effects typically seen after repeated CVLT assessment (Woods et al., 2006). The NU-­‐CON group improved by 3.2% over time (non-­‐significant), such that memory performance at follow-­‐up was very similar for the NU-­‐CON and NS-­‐CON groups (85.5% and 85.8%, respectively). The stimulant users, despite having been exposed to the test materials to the same degree, did not benefit from practice. The effect of stimulant use on memory performance was not attributable to pre-­‐existing differences or acute substance effects. Memory deficits after recreational stimulant use had previously been shown in adults, who in most studies reported more than one-­‐hundred lifetime instances of stimulant use (Murphy et al., 2012), which is much higher compared to the average of nine and median of four lifetime instances of stimulant use in our adolescent sample. It is important to note that although memory test performance in the STIM group did not significantly deteriorate over time (-­‐
1.3%), the absence of improvement compared to controls was marginally significant. 158 However, there is the imminent risk of type 1 error, hence these findings are preliminary until replicated and should not be over-­‐interpreted. The effect of stimulant use on memory task performance occurred in the absence of changes in hippocampal volume, microstructure, or shape, with the exception of a small cluster on the right hippocampal surface. The absence of stimulant effects on hippocampus volume is in line with a cross-­‐sectional study in adult recreational users, in which hippocampus volume was not abnormal after infrequent (<5 lifetime instances of MDMA and/or <5 mg of amphetamine) or frequent stimulant use (>100 instances of MDMA and/or >50 mg of amphetamine; Koester et al., 2012). We had expected that microstructure/ diffusion parameters may be more sensitive compared to grey matter volume to detect subtle hippocampal changes, as had been shown in patients with mild memory impairment (Granziera et al., 2015), but we found no association between stimulant use and hippocampal microstructure either. Stimulant use may impact memory performance through changes in brain regions other than the hippocampus. In fact, it has been suggested that gradual memory improvement during adolescence may be facilitated by frontal rather than hippocampal changes (Sowell et al., 2001). The association between adolescent stimulant use, memory performance, and frontal brain changes would be an interesting target for future studies. We wish to emphasize that failure to detect hippocampal changes in the current sample, with low-­‐frequency stimulant use patterns and limited power to detect medium or small effects, does not necessarily imply an absence of hippocampal changes. In a small cluster on the right lateral hippocampus, a significant time-­‐by-­‐
group interaction effect indicated surface expansion in the NU-­‐CON group, but not in STIM group. Prior analyses in adult stimulant users found no significant surface changes after stimulant exposure (Bava et al., 2010). In rats, medial inward vertex displacement was found after adolescent exposure to stimulants, but not after adult exposure (van der Marel et al., 2015). To our best knowledge, adolescent hippocampal surface development has not yet been described. Across the adult lifespan, typical surface changes encompass contractions in medial regions and expansions in lateral regions (Voineskos et al., 2015), hence the absence of expansion in the stimulant users group could be interpreted as a deviation from normal development. However, we warrant cautious interpretation for several reasons. First, the region of significant effect is very small, spanning less than 1% of the hippocampal surface. The functional relevance of such a small alteration is likely very limited. In our sample, surface change within the cluster was not associated with memory test performance. Second, we had hypothesized that surface changes in the STIM group would deviate from those in both control groups. The NS-­‐CON group, however, showed no expansion over time within the cluster. We conclude that adolescent stimulant use could potentially 159 induce changes in a small cluster on the right hippocampal surface; however, the robustness and relevance of this effect requires further investigation. The current study presents several novelties. This is the first structural MRI study to investigate the effects of recreational stimulant use on the developing adolescent brain. With up to thirty percent of recreational users initiating stimulant use during adolescence (SAMHSA, 2014), and animal studies reporting amplification of negative stimulant effects when animals are exposed earlier in life (Van Der Marel et al., 2014 and 2015), investigations targeting human adolescents are urgently needed. The limited exposure to stimulants within our sample reflects the frequency of stimulant use in typical adolescent populations, ensuring the validity of our findings. In addition, the current study is the first with a prospective design, allowing for the first time a distinction between stimulant-­‐exposure effects and pre-­‐existing differences. There were limitations as well. First, our sample size provided limited power to detect small effects, and did not allow separate analyses per stimulant type. Differentiation between stimulants types was further limited as all stimulants were aggregated in one questionnaire category. As a result, the STIM group was likely a heterogeneous group, which may have further reduced statistical power. Second, the observational study design, although inevitable, may have introduced a susceptibility bias (e.g., stress may have simultaneously increased the likelihood of stimulant use and elicited memory problems). Furthermore, the use of self-­‐ and parent-­‐report substance use interviews, as well as the illicit origin of many substances, may compromise substance use data accuracy. Future studies employing biological markers for substance use (e.g., hair analysis) are needed. Finally, our focus on the hippocampus was firmly based in the literature yet precluded potential findings in other brain regions. The frontal cortex would be an interesting target for future studies. In sum, we found that changes in memory performance may occur after infrequent recreational stimulant use in adolescents. Furthermore, adolescent stimulant-­‐users may present with subtle changes in hippocampal surface morphology, but the functional significance of such changes needs further investigation. Replication in a larger sample is needed, for instance in the Adolescent Brain Cognitive Development study (ABCD). If replicated, these findings may also have implications for patients with ADHD, who are often prescribed stimulant treatment for extended periods of time, albeit typically at lower dose compared to recreational use. Declarative memory performance has received little attention in ADHD research, and subtle changes in stimulant-­‐treated patients may have gone undetected. 160 161 162 Chapter 9 NO LONG-TERM EFFECTS OF STIMULANT
TREATMENT ON ADHD SYMPTOMS, SOCIALEMOTIONAL FUNCTIONING, OR COGNITION
Submitted as: Schweren LJS, Hoekstra PJ, van Lieshout M, Oosterlaan J, Rommelse NNJ, Buitelaar JK, Franke B, Hartman CA. Long-­‐term effects of stimulant treatment on ADHD symptoms, social-­‐emotional functioning, and cognition. 163 ABSTRACT
Background: Methodological and ethical constraints have hampered research into lasting long-­‐term outcomes of stimulant treatment in individuals with attention-­‐
deficit/hyperactivity disorder (ADHD). Aims: To investigate whether stimulant treatment history predicts long-­‐term development of ADHD symptoms, social-­‐emotional functioning, or cognition, measured after medication wash-­‐out. Method: Outcomes were measured twice, six years apart, in two ADHD groups (stimulant-­‐treated vs. not stimulant-­‐treated between baseline and follow-­‐up), closely matched on baseline characteristics (n=148, 58% male, age=11.1). A matched healthy control group was included for reference. Results: All but two outcome measures (emotional problems and prosocial behavior) improved between baseline and follow-­‐up. For all outcomes, improvement over time was the same for participants who had received treatment and those who had not. Conclusions: Stimulant treatment is not associated with the long-­‐term developmental course of ADHD symptoms, social-­‐emotional functioning, or cognition. These findings are an important source to feed the scientific and public debate. 164 INTRODUCTION
Attention-­‐deficit/hyperactivity disorder (ADHD) is a prevalent and often persistent developmental disorder, characterized by age-­‐inappropriate and impairing levels of inattention and/or hyperactivity-­‐impulsivity. ADHD has been associated with a broad range of neurocognitive deficits, including impaired executive functioning (Willcutt et al., 2005), timing deficits (Noreika et al., 2013), and higher response time variability (Klein et al., 2006). In the majority of individuals with ADHD, stimulants acutely reduce symptoms (Swanson et al., 2001) and improve neurocognitive functioning (Coghill et al., 2014b). Concerns about potential harmful long-­‐term effects of stimulant treatment, as well as anticipation of potential lasting benefits of treatment have dominated the public and scientific debate. Adequately investigating long-­‐term treatment effects, especially in children, is methodologically and ethically challenging, hence evidence for either positive or negative long-­‐term outcomes of stimulant treatment is equivocal. In the Multimodal Treatment Study of ADHD (MTA), the largest controlled treatment study to date, the benefits of 14 months of stimulant treatment on a broad range of outcomes rapidly diminished in subsequent years (MTA Cooperative Group, 1999; Molina et al., 2009; Swanson et al., 2007a). In the MTA study, outcomes were assessed without a medication wash-­‐out phase, which impedes the distinction between lasting effects of prior treatment and acute effects of ongoing treatment. When rated while off-­‐medication, ADHD symptoms were found not to change with one year of stimulant treatment (Huang et al., 2012). Attention task performance and IQ did improve over the course of one year, but in the absence of a comparable non-­‐treated or healthy control group, these changes may reflect normal maturation (Tsai et al., 2013). Observational studies have reported higher ADHD persistence rates in stimulant-­‐treated patients compared to non-­‐treated patients (Biederman et al., 2012; van Lieshout et al., 2016a), while at the same time rates of comorbidity were found to be lower in treated patients (Biederman et al., 2009). Importantly, in these studies confounding-­‐by-­‐indication and self-­‐selection could not satisfactorily be addressed. Here, we applied stringent matching procedures to derive two comparable ADHD samples from a large prospective cohort study (i.e., stimulant-­‐treated and non-­‐stimulant-­‐treated) as well as a typically developing reference group. Outcomes were repeatedly measured over six years, always while participants were in their non-­‐medicated state. We investigated whether stimulant treatment between baseline and follow-­‐up predicted the developmental trajectory of ADHD symptoms, social-­‐emotional functioning, and/or cognition. 165 METHODS
Participants Participants were drawn from the prospective multi-­‐centre IMAGE-­‐
NeuroIMAGE cohort study (von Rhein et al., 2015a). The full cohort includes 751 children, adolescents, and young adults with ADHD from 590 families. At baseline, ADHD diagnosis was ascertained using the Strengths and Difficulties Questionnaire (van Widenfelt et al., 2003) (SDQ, >90th percentile on the hyperactivity subscale), the parent-­‐ and teacher-­‐rated Conners’ ADHD scales (CPRS and CTRS; T≥63 on the DSM inattentive or hyperactive/impulsive scale) (Conners et al., 1998a, 1998b) and the Parental Account of Children’s Symptoms interview (PACS; ≥6 symptoms, present in ≥2 situations and ≥1 symptom reported by the teacher) (Taylor, 1986). Participants with ≥6 symptoms but who did not fulfill all diagnostic criteria, were classified as subthreshold ADHD. At follow-­‐up, ADHD diagnosis in participants <18 years was ascertained again using the same CPRS and CTRS criteria, complemented with the Schedule for Affective Disorders and Schizophrenia for School-­‐Age Children interview (K-­‐SADS; ≥6 symptoms, present in ≥2 situations, causing impairment, and onset before age 12) (Kaufman et al., 1997). For participants ≥18 years, the self-­‐rated Conners’ scale (CAARS; Conners et al., 1999) was used instead of the teacher-­‐rated scale, and five symptoms were sufficient for diagnosis. Participants who scored T≥63 on either of the Conners’ scales or had sufficient symptoms, but did not fulfill all diagnostic criteria, were classified as subthreshold ADHD. Average follow-­‐up time was 5.9 years (SD=0.6), and the retention rate was high (77%). We applied the following inclusion criteria: (1) participation at baseline and follow-­‐up, (2) diagnosis of (subthreshold) ADHD at baseline and/or at follow-­‐up, (3) IQ>70 at baseline and follow-­‐up, and (4) no known genetic or neurological disorders. Eligible participants were split according to treatment between baseline and follow-­‐up into stimulant-­‐treated (n=337) and non-­‐stimulant-­‐treated participants (n=138). Stimulant treatment prior to baseline and treatment with non-­‐stimulant psychoactive medication was allowed in both groups. From the two ADHD groups we selected all participants who had a one-­‐to-­‐one match on gender, age (±<0.5 SD), and baseline number of ADHD symptoms (±<0.5 SD). This resulted in two comparable groups of 74 participants with ADHD each (Table 1). For reference, a gender-­‐ and age-­‐matched healthy control sample was drawn from the IMAGE-­‐NeuroIMAGE cohort as well, applying the same inclusion and matching criteria (except inclusion criterion two/symptom-­‐matching). In addition, control participants had no family-­‐history of any psychiatric disorder. All assessments took place at two sites in the Netherlands. Participants were asked to withhold use of 166 psychoactive drugs for 48 hours before each assessment. Informed consent was signed by all participants and their parents (only parents signed informed consent for participants < 12 years). Procedures were approved by the local ethical committee of each site. TABLE 1. Baseline characteristics of the two treatment groups. Treated Non-­‐treated M SD M SD Stat. p Gender=male N=43 58.1% N=43 58.1% 0.000 1.000 Age 11.14 3.29 11.00 3.23 0.066 0.798 Site=Amsterdam N=27 36.5% N=46 62.2% 9.759 0.002* IQ 99.93 10.47 103.55 10.77 3.605 0.060 Socio-­‐economic status 11.26 2.02 12.07 2.52 4.522 0.035* Follow-­‐up interval (years) 5.92 0.60 5.86 0.68 0.258 0.613 24.3% 31.335 <0.001* Treatment prior to baseline=yes ADHD type N=52 70.3% N=18 8.677 Unaffected N=6 8.1% N=7 9.5% Inattentive N=4 5.4% N=6 8.1% Hyperactive N=1 1.4% N=2 2.7% Combined N=55 74.3% N=39 52.7% Subthreshold N=8 10.8% N=20 27.0% Comorbid problems # 0.070 Anxiety/shyness 5.20 4.92 4.30 4.47 1.333 0.250 Perfectionism 3.85 4.24 3.55 3.55 0.214 0.644 Psychosomatic problems 3.45 3.33 2.80 3.16 1.445 0.231 Stat = Chi for categorical variables, student-­‐t for continuous variables. # scores on the anxiety/shyness scale, perfectionism scale, and psychosomatic problems scale of the parent-­‐ and teacher-­‐rated Conners’ questionnaires were used as a proxy of baseline comorbid problems. * = significant difference between treated and non-­‐treated participants. 2
167 Stimulant treatment Participants and parents provided written consent to request prescription records from their pharmacies. In addition, they reported lifetime history of psychoactive medication in a questionnaire at follow-­‐up measurement. Pharmacy data covering the baseline-­‐follow-­‐up interval were available for 91% of participants with ADHD (n=135). Participants were classified as stimulant-­‐treated if they had been prescribed any immediate or extended release methylphenidate preparations, or d-­‐amphetamine preparations, between baseline and follow-­‐up. When pharmacy transcripts were not available or incomplete (n=13), treatment history was derived from the questionnaire data. The questionnaire data was also used to determine stimulant treatment prior to baseline (“previously treated” or “stimulant-­‐naïve”) for all participants. Outcome measures Parent-­‐rated numbers of hyperactivity-­‐impulsivity and inattention symptoms were measured at baseline and follow-­‐up using the respective DSM subscales of the CPRS (range 0:27). For participants using medication, parents rated behavior in the participant’s non-­‐medicated state. Four indicators of social-­‐emotional functioning were derived from the SDQ for both time points: problems with emotion regulation, problems with peer relationships, conduct problems, and prosocial behavior (range 0-­‐
10). In addition, six cognitive tests were administered at both baseline and follow-­‐
up. Three tasks measured motor control: Baseline Speed, in which participants were required to press a key upon unpredictable appearance of a stimulus; Pursuit, where participants followed a randomly moving target with the cursor as precisely as possible; and Tracking, in which participants were required to trace an invisible midline between an inner and an outer circle as precisely as possible. Two tasks measured timing: Time Estimation, where participants were asked to reproduce the duration of visually presented stimuli of different lengths (4, 8, 12, 16, and 20 seconds); and Motor Timing, in which participants were instructed to produce 1-­‐
second intervals as accurately as possible. Working memory was assessed in the backwards condition of the Digit Span test (WISC-­‐III/WAIS-­‐III), in which participants had to reproduce an increasingly long sequence of numbers in reverse order. For details, see Supplementary Table 1. 168 Statistical analyses We used linear mixed effects models, predicting symptoms of hyperactivity/impulsivity and inattention, each of the four social-­‐emotional outcomes, and performance on each cognitive test from time (baseline or follow-­‐up), treatment (“stimulant-­‐treated” or “not stimulant-­‐treated” during the study phase), and time-­‐by-­‐
treatment-­‐interaction. The effect of interest is captured in the time-­‐by-­‐treatment interaction, which evaluated whether the outcome variables changed differently over time for the stimulant-­‐treated group compared to the non-­‐treated group. Baseline demographic/clinical between-­‐group differences were included as covariates, as was a random intercept per family to account for dependencies among siblings. Multiple testing was accounted for by Bonferroni adjustment: alpha was divided by two for ADHD symptoms (α=0.05/2=0.025), by four for social-­‐emotional outcomes (α=0.012), and by six for cognitive outcomes (α=0.008). Previous work by our group described changes over time in ADHD symptoms and cognitive functioning in participants with ADHD compared to typically developing participants (van Lieshout et al., 2016a; 2016b). Case-­‐control differences are thus not the focus of the current study. Rather, the matched control group was used as a reference group for normative developmental changes. For visualization of estimated marginal means of all groups (stimulant-­‐treated, not stimulant-­‐treated, and control), the models described above were re-­‐estimated across all participants with a fixed factor for group. Sensitivity analyses were performed to test the robustness of our findings. With a relatively short wash-­‐out time (48h), immediate withdrawal effects may have affected cognitive functioning in participants who received ongoing treatment at time of measurement. Therefore, analyses were repeated with an additional covariate encoding whether participants were actively being treated with stimulants within six months prior to assessment or not, and its interaction with the effect of interest (active treatment * time * treatment between baseline and follow-­‐up). Second, all analyses were repeated with baseline age as an additional predictor, to address the wide age-­‐range within our sample. Here, change over time in each outcome variable was predicted from age-­‐by-­‐treatment interaction, thus analyzing whether the effect of treatment on clinical/social-­‐emotional/cognitive changes over time was different for participants of different ages. RESULTS
Mean age of participants with ADHD was 11.1 years (SD=3.2) at baseline and 17.0 years (SD=3.3) at follow-­‐up. Fifty-­‐eight percent of participants was male. 169 Participants were diagnosed with ADHD or subthreshold ADHD at baseline (n=135, 91.2%) and/or at follow-­‐up (n=132, 89.2%). Most participants reached diagnostic criteria at both times (n=119, 80.41%). Fifteen participants (10.1%) with subthreshold ADHD never met criteria for full ADHD diagnosis. At baseline, the majority of participants had combined type ADHD (n=94, 63.5%), while at follow-­‐up the majority had either combined type (n=40, 27.0%) or inattentive type (n=51, 34.5%), with no differences between groups (Table 1). Within the stimulant-­‐treated group, average cumulative stimulant dose between baseline and follow-­‐up was 43336 mg, which equals 5.9 years of 20.1 mg per day. Forty participants (54.1%) had received active stimulant treatment within six months prior to follow-­‐up assessment; the other participants had ceased stimulant treatment earlier. Participants in the stimulant-­‐treated group had lower SES (p=0.035), were more likely to have received stimulant treatment prior to the initial assessment (Chi2=31.335, p=0.001), and more likely to have received atomoxetine treatment between baseline and follow-­‐up (nOVERALL=16, 10.8%; nTREATED=13, 17.6%; nNON-­‐TREATED=3, 4.1%; Chi2=6.862, p=0.009). There was a site effect for stimulant treatment as well (Chi2=9.759, p=0.002). Site, SES, and prior treatment were therefore added as covariates in all between-­‐group comparisons. At baseline, the two treatment groups did not differ from each other with regard to any of the clinical or cognitive outcome measures. There was a significant main effect of time on ADHD symptoms, as well as on two out of four social-­‐emotional outcome measures (Table 2). Across all participants with ADHD, symptoms of hyperactivity/impulsivity and inattention, peer problems, and conduct problems improved between baseline and follow-­‐up. There were no main effects of time on emotional problems or prosocial behavior. Improvement over time was also found for performance on all cognitive tasks: participants showed lower Baseline Speed variability, smaller deviations on the Tracking, Pursuit and Time Estimation tasks, and higher maximum Digit Span at follow-­‐up compared to baseline. There were no main effects of treatment group, and no time-­‐by-­‐treatment-­‐
group interaction effects on any of the outcome measures (Table 2, Figure 1 and 2). Thus, changes in ADHD symptoms, social-­‐emotional and cognitive functioning over time were the same for participants who received stimulant treatment between baseline and follow-­‐up and those who had not. Moreover, changes over time were the same for participants on active stimulant treatment at follow-­‐up assessment and those who were not, suggesting no confounding by withdrawal effects. Finally, there were no significant interactions with age, suggesting that treatment effects were similar at different ages. 170 TABLE 2. Baseline and follow-­‐up scores across treatment groups, and the effects of time, treatment, and time-­‐by-­‐treatment interaction. Baseline Follow-­‐up EMM SD EMM SD Hyper/imp symptoms 14.22 5.95 11.83 6.73 <0.001* 0.212 0.188 Inattention symptoms 12.28 6.15 7.38 5.55 <0.001* 0.557 0.054 Emotional problems 2.98 3.00 2.82 3.08 0.736 0.577 0.707 Prosocial behavior 7.15 2.08 7.38 2.19 0.351 0.280 0.142 Peer problems 2.82 2.12 2.19 1.98 0.003* 0.382 0.424 Conduct problems 3.09 2.00 2.43 1.83 0.002* 0.238 0.906 172.37 103.89 90.29 50.35 <0.001* 0.513 0.672 Pursuit (inaccuracy) 6.44 3.74 3.87 0.76 <0.001* 0.609 0.320 Tracking (inaccuracy) 2.85 1.81 1.34 0.94 <0.001* 0.798 0.175 203.11 95.10 148.83 51.48 <0.001* 0.449 0.341 Time Estimation (inaccuracy) 2.72 1.79 1.48 0.81 <0.001* 0.776 0.411 Digit Span 3.92 1.15 4.49 1.26 <0.001* 0.126 0.715 Baseline Speed variability Motor Timing (inaccuracy) p
TIME
p
TREATMENT
P
TIME*TREATMENT
EMM=estimated mean score across participants with ADHD, adjusted for stimulant treatment prior to baseline measurement, site, and SES. *=p<0.012 or p<0.008. DISCUSSION
Main findings We investigated developmental changes in a broad spectrum of outcomes, including social-­‐emotional and cognitive functioning, in stimulant-­‐treated and not stimulant-­‐treated individuals with ADHD who had been stringently matched on baseline characteristics and were non-­‐medicated at both assessments. We found no evidence for any (beneficial or adverse) stimulant treatment effects persisting after stimulant treatment had temporarily been ceased. ADHD symptoms, peer problems, conduct problems, and performance on tests of motor control, timing, and working memory improved over time, but improvement occurred irrespective of treatment. Even at a lenient threshold for statistical significance, stimulant treatment was not associated with any of the outcomes. 171 FIGURE 1. Change in ADHD symptoms and social-­‐emotional outcome measures over ~6 years, for stimulant-­‐treated (grey) and non-­‐treated (black) participants with ADHD, and control participants (grey dashed), matched on baseline age, gender, and ADHD symptoms. Baseline social-­‐emotional outcomes were not assessed for typically developing participants. The slopes of the two treatment groups did not differ for any outcome. FIGURE 2. Change in cognitive test performance over ~6 years, for stimulant-­‐treated (grey) and non-­‐
treated (black) participants with ADHD, and control participants (grey dashed), matched on baseline age, gender, and ADHD symptoms. The slopes of the two treatment groups did not differ for any outcome Interpretation and previous studies Our findings put into perspective previous studies reporting beneficial long-­‐
term effects of stimulant treatment. First, previous studies reporting long-­‐term beneficial treatment effects oftentimes assessed outcomes when patients were on active treatment (Abikoff et al., 2004; Charach et al., 2004). Their findings may thus 172 represent either lasting effects of prior treatment, transient effects of ongoing treatment, or a combination of both. Our findings, in conjunction with reports of better outcome during phases of active stimulant treatment (Chang et al., 2016; Lichtenstein et al., 2012), suggest that previously reported long-­‐term effects may be driven by ongoing transient effects rather than lasting effects. The absence of lasting treatment effects in our sample convenes with negative long-­‐term findings of the MTA study, that have previously been attributed to self-­‐selection during the observational phase (Molina et al., 2009; Swanson et al., 2007a). Our findings, however, underline the possibility that the theorized long-­‐term effects may in fact not occur. At the same time, we wish to emphasize that beneficial long-­‐term treatment effects have been found in outcomes that were not addressed here, such as the development of comorbid disorders later in life (Biederman et al., 2009). Second, our findings are in line with a previous report of improved attention task performance after a one-­‐year stimulant treatment episode even at drug-­‐free status (Huang et al., 2012), which, in the absence of a reference group, could indicate either lasting beneficial treatment effects or improved cognitive performance at older age. In the current study, changes over time were the same in the treated and non-­‐
treated groups, suggesting that improvement over time is not related to treatment. Third, several previous studies have reported more severe and/or more persistent ADHD in individuals who had received stimulant treatment during childhood, which could indicate either detrimental treatment effects or confounding-­‐
by-­‐indication (Biederman et al., 2012; Molina et al., 2009; van Lieshout et al., 2016a). The current findings, free of confounding-­‐by-­‐indication due to stringent matching procedures and accounting for baseline measurements, provide no evidence of detrimental treatment effects. Implications The current findings are an important source to feed the scientific and public debate about pharmacological treatment for ADHD that has focused on long-­‐term hazards and benefits. First, our findings emphasize that the course of ADHD symptoms and related outcomes are not altered by stimulant treatment. Previous work of our group showed that ADHD symptoms tend to decline but not disappear at later age (van Lieshout et al., 2016a). The current results add to these findings by showing that this conclusion holds for both stimulant-­‐treated and non-­‐treated individuals. Second, the absence of long-­‐term treatment effects on clinical and cognitive outcomes may guide the interpretation of findings of structural brain changes associated with stimulant treatment (or the absence thereof). The evidence for normalized brain structure in children with ADHD who had received long-­‐term 173 stimulant treatment is mixed (Schweren et al., 2015; Shaw et al., 2009; 2014). The absence of lasting treatment effects on a broad spectrum of clinical/behavioral outcomes emphasizes the importance of investigating behavioral correlates and clinical relevance of stimulant effects on the brain. Strengths and limitations This is the first longitudinal study investigating long-­‐term treatment effects that included a non-­‐treated ADHD and a typically developing sample and reported on a wide spectrum of clinical and cognitive outcomes. The average follow-­‐up time of almost six years allowed the detection of effects emerging at later age, and captured the late adolescent/early adulthood phase that is often characterized by both clinical and normative developmental changes, which we were able to tease apart. Our rigorous one-­‐to-­‐one matching procedure allowed firm conclusions. Finally, extensive diagnostic assessments resulted in a well-­‐characterized ADHD sample, and the availability of pharmacy records enabled highly reliable assessment of treatment history. The current study had limitations as well. Treatment allocation was not random. We were able to rule out confounding-­‐by-­‐indication for all measured baseline variables, but not for non-­‐measured potential between-­‐group differences. Especially functional impairment and comorbidity could not satisfactorily be addressed. Propensity score adjustment would have been valuable in this regard, but was not feasible with the available data. Confounding may also have occurred during the study phase, e.g. behavioral treatment (not assessed) may have been more common in one group compared to the other. Second, findings regarding clinical outcomes furthermore rely on reports by parents, who were not blind to the participant´s treatment history or status. Third, the current design did not allow full investigation of treatment timing, since participants had often initiated treatment prior to the baseline measurement and/or continued treatment after the follow-­‐up measurement. Treatment at different ages may be associated with different long-­‐term consequences, although in our sample we found no indications of such effects. 174 SUPPLEMENTAL INFORMATION
TABLE S1. Neurocognitive tasks. Task (aim) Description Performance measure n Baseline Speed (motor output in response to cue) A white square appeared unpredictably (500-­‐2500ms after response) on a screen, after which participants were required to press a key. Practiced and executed with the non-­‐preferred hand, thereafter with the preferred hand. Standard deviation of reaction times in ms averaged across both hands 78 (52.7%) Pursuit (motor control with continuous adaptation) Participants were required to ‘catch’ a randomly moving stimulus (asterisk, 10 mm/second) as precisely as possible by moving the cursor on top of the stimulus with the left hand. Mean absolute distance in mm between target and cursor 81 (54.7%) Tracking (motor control without continuous adaptation) With the left hand, participants traced an invisible midline between an inner and outer circle presented on the screen (radius 7.5 and 8.5 cm, respectively), counterclockwise and as quickly and precisely as possible. Mean absolute distance in mm between target (midline) and cursor 83 (56.1%) Digit Span (working memory) Participants were instructed to reproduce sequences of numbers, of increasing length, in reverse order. Maximum accurately reproduced sequence length 111 (75.0%) Time Estimation Stimuli (4, 8, 12, 16, 20 seconds) were randomly presented by a lightbulb. Participants were required to reproduce stimulus length by pressing a button. Absolute discrepancy between the response length and the stimulus length averaged across all 12-­‐second trials. 83 (56.1%) Motor Timing Participants were instructed to produce a 1-­‐
second interval after a tone, as accurately as possible. Visual feedback was given, indicating whether the response was correct, too short or too long (defined by a dynamic tracking algorithm). Median absolute deviation in ms from 1 second 88 (59.5%) n = number of participants with ADHD who completed the task at baseline and at follow-­‐up. 175 176 Chapter 10 GENERAL DISCUSSION
177 178 This thesis had two overall aims. Our first and foremost aim was to describe long-­‐term effects, or the absence thereof, of stimulant treatment on the developing brain in children, adolescents and young adults with ADHD. Our second aim was to advance our knowledge about mechanisms underlying long-­‐term effects of stimulant treatment in the brain. To these ends, we investigated associations between stimulant treatment history and various neural outcomes in a large cross-­‐sectional ADHD sample (NeuroIMAGE, see page 17 Box 1), with special focus on the frontal-­‐striatal system and on non-­‐volumetric brain parameters. Furthermore, we evaluated differences in patient characteristics and exposure patterns that may influence long-­‐
term effects, with at its extreme the investigation of an adolescent recreational stimulant user group without ADHD (Youth At Risk, see page 17 Box 2). In this final chapter, I first provide a summary of the findings in each of the chapters. Next, I describe how these findings relate to each other and to the existing literature, and how they provide answers to our two main research questions. After that, I discuss methodological strengths and limitations of our approach, potential clinical implications of our findings, and suggest directions for future studies. SUMMARY PER CHAPTER
At the start of this thesis, in chapter 2, we reviewed the MRI literature regarding short-­‐ and long-­‐term effects of stimulant treatment on brain structure and function. Evidence from structural MRI studies published up to 2011 suggested that those with ADHD who received long-­‐term stimulant treatment had structural white matter, anterior cingulate cortex, thalamus, and cerebellum indices in that were less different or not different at all from those of healthy controls compared to children with ADHD that were untreated. Moreover, preliminary evidence suggests that methylphenidate treatment may normalize the trajectory of cortical development in ADHD. Similarly, functional MRI studies had shown that case-­‐control differences in brain activation patterns during cognitive control, attention, and rest were attenuated after a single dose of methylphenidate, but evidence of long-­‐term effects on these activation patterns was missing. The effects of methylphenidate on the brain appeared highly specific and dependent on numerous factors, including biological factors (e.g., age), and the environment (e.g., task difficulty). In chapter 3, we investigated long-­‐term effects of stimulant treatment on cortical thickness. ADHD had previously been associated with widespread changes in cortical thickness, and these changes had been suggested to be reduced or even disappear at older age, but also after stimulant treatment. In our sample, we found that medial temporal cortical thickness in both hemispheres was reduced in adolescents/young adults with ADHD compared to their typically developing peers. 179 These differences were associated with symptoms of hyperactivity and prosocial behavior. However, they were independent of stimulant treatment history. In fact, stimulant treatment was not associated with cortical thickness in any brain region. Furthermore, thinner medial temporal cortex was present throughout adolescence and young adulthood, and age-­‐related changes in cortical thickness were the same for individuals with and without ADHD across the cortical mantle. Thus, we found no evidence of cortical normalization either with stimulant treatment or at older age. In the next chapter, chapter 4, we used diffusion tensor imaging (DTI) to study white matter structural connectivity. Previous DTI studies had revealed subtle abnormalities in white matter connectivity in individuals with ADHD, and stimulant treatment may in the long term reduce such abnormalities. We investigated associations between treatment history and white matter connectivity within five dopaminergic circuits (orbitofrontal-­‐striatal, orbitofrontal-­‐amygdalar, amygdalar-­‐
striatal, dorsolateral-­‐prefrontal-­‐striatal and medial-­‐prefrontal-­‐striatal). Individuals with ADHD presented with reduced fractional anisotropy in the orbitofrontal-­‐striatal pathway, indicative of impaired structural connectivity. In the same pathway, higher lifetime stimulant dose was associated with lower diffusivity, which may be indicative of improved structural connectivity at higher treatment levels. Next, in chapter 5, we evaluated associations between stimulant treatment history and brain activation patterns during performance of a monetary incentive delay task, known to activate the striatum by eliciting reward-­‐related dopaminergic neurotransmission. One previous study had reported indications of normalized striatal activation patterns in adult ADHD patients with a history of stimulant treatment. Here we found that, even though participants were in their non-­‐medicated state during scanning, stimulant treatment history predicted brain activation patterns during reward processing. Individuals who had initiated treatment early and had received a relatively high dose showed more activation in the supplementary motor area and dorsal anterior cingulate cortex, compared to individuals with later onset and lower dose treatment. These changes may indicate compensatory recruitment of brain regions for higher-­‐order integration of valence information in the intensely treated group. Contrary to our hypothesis, treatment history was not associated with (more normative) striatal activation patterns. In chapter 6, we investigated long-­‐term effects of treatment on striatum, hippocampus, and frontal cortex volumes, and how these may be moderated by age and/or by two common variants of dopaminergic genes (DAT1, encoding the presynaptic dopamine transporter; and DRD4, encoding the postsynaptic dopamine D4 receptor). Previous animal studies had shown exacerbated effects of methylphenidate (therapeutic dose) on brain structure when treatment was administered at younger age. Here, we found that striatum volume was not associated 180 with stimulant treatment history. In the frontal cortex and left hippocampus, however, DRD4 genotype and age predicted the effects of stimulant treatment. At younger age and lower treatment-­‐levels, but not at younger age and higher treatment levels, carriers of the DRD4 7R-­‐allele showed decreased frontal cortex volumes compared to controls. At older age, individuals with ADHD showed lower frontal volumes irrespective of genotype or treatment history. Left hippocampal volume was similar to controls at average treatment levels, and increased with treatment only in carriers of the DRD4 7R-­‐allele and at younger age. These findings may suggest that carriers of the DRD4 7R-­‐allele may at younger age be sensitive to cortical remodeling after stimulant treatment. Chapter 7 addressed combined stimulant and antipsychotic treatment, an off-­‐
label yet rather common augmentation strategy for the treatment of behavioral problems in ADHD. In animals, administration of antipsychotics had been shown to counteract the acute effects of stimulants in the striatum. Here, we compared volumes of brain regions involved in dopaminergic circuits between patients who had received combined treatment, patients who had received stimulants but not antipsychotics, and healthy controls. Patients in the combined treatment group, but not those in the stimulant-­‐only group, showed a reduction in total cortical volume and frontal cortex volume compared to healthy controls. Further, the combined treatment group, but not the stimulant-­‐only group, showed volume reduction in bilateral subthalamic nuclei and left thalamus. Structural abnormalities in the combined treatment group may have existed prior to treatment, perhaps in relation to behavioral problems or ADHD severity. However, our findings could also indicate that combined stimulant and antipsychotic treatment may result in volume reductions in the developing brain. In chapter 8 we took a sidestep and investigated the effects of non-­‐medical, recreational use of psychostimulants such as amphetamines or cocaine. Recreational stimulant exposure patterns are very different from typical ADHD treatment patterns. The neural consequences of recreational stimulant use had previously been studied in healthy adults, but not in adolescents. In the prospective, repeatedly scanned Youth At Risk sample, we evaluated the effect of incidental, high-­‐dose stimulant exposure on the development of hippocampus structure and memory performance. We found that learning/memory test performance changed differently over time for stimulant-­‐
exposed individuals compared to non-­‐exposed individuals. Whereas nonusers (fully abstinent) and non-­‐stimulant users (e.g., marijuana, but not stimulants) showed an upward trend in test performance over time which may reflect subtle learning effects, stimulant-­‐users showed a downward trend. No changes in hippocampal volume, microstructure, or surface morphology were found in the stimulant-­‐exposed group compared to both non-­‐exposed groups. 181 Finally, in chapter 9, we addressed long-­‐term effects of stimulant treatment in terms of clinical or behavioral outcomes longitudinally. In previous studies, adequately investigating such effects spanning multiple years had proven challenging, and findings had suggested clinical improvement but also worsening of symptoms after long-­‐term treatment. Here, we compared the development of ADHD symptoms, social-­‐emotional functioning, and test performance in three cognitive domains (all while participants were in their non-­‐medicated state), between stimulant-­‐treated and not stimulant-­‐treated individuals with ADHD that were closely matched on baseline characteristics. All but two outcome measures (emotional problems and prosocial behavior) improved over the follow-­‐up period of approximately six years. However, for all outcomes, improvement was the same for participants who had received treatment and those who had not, suggesting no long-­‐term treatment effects. WHAT ARE THE LONG-TERM EFFECTS OF STIMULANT TREATMENT ON THE DEVELOPING BRAIN?
Neurotoxicity and harmful effects Concerns that stimulant treatment may adversely affect brain development are frequently voiced, especially in popular media that seem eager to emphasize similarities between ADHD medication and for instance hard drugs such as cocaine. Evidence for stimulant-­‐induced damage to dopaminergic nerve terminals stems almost exclusively from animal studies, in which stimulants are typically administered in binge-­‐like patterns at pharmacologically high dosages (Berman et al., 2009). However, two positron emission tomography (PET) radiotracer studies, allowing in vivo assessment of dopamine metabolism in humans, have also suggested long-­‐term effects of stimulant treatment in patients with ADHD. (Wang et al., 2013; Ludolph et al., 2008). MRI outcomes may indirectly reflect (correlates of) dopaminergic neurotransmission, but are far more distal. To date, MRI studies have reported no evidence of lasting detrimental treatment effects on basal ganglia volumes (e.g., Shaw et al., 2014; Semrud-­‐Clikeman et al., 2006), cortical thickness (e.g., Hoekzema et al., 2012; Shaw et al., 2013), or brain activation patterns (e.g., Pliszka et al., 2006; Schlochtermeier et al., 2011; Stoy et al., 2011). If anything, MRI findings to date more likely reflect neutral or even beneficial treatment effects, which will be discussed in the next section. In line with the existing MRI literature, our findings do not provide support for stimulant treatment negatively affecting brain development in children, adolescents, and young adults with ADHD. We found no indications that the development of cortical thickness may be impacted by stimulant treatment (chapter 3). Furthermore, stimulant treatment was not associated with alterations in striatal 182 volumes, nor with abnormal striatal activation patterns during reward (chapter 5 and 6). We also found no indication of detrimental changes in frontal-­‐striatal white matter structural connectivity after stimulant treatment (chapter 4). Finally, there was no indication that stimulant treatment resulted in an exacerbation of ADHD symptoms when medication was (temporarily) discontinued, as had previously been suggested (Wang et al., 2013), nor did long-­‐term stimulant treatment result in a decline in social-­‐emotional or neuropsychological functioning (chapter 9). Only in one instance did we find less normative brain structure in ADHD at higher treatment levels. In a specific subgroup of patients, namely those of young age and carrying the DRD4 7R-­‐allele, hippocampus volume increased with more stimulant exposure, deviating away from control levels (chapter 6). Larger volume of the hippocampus, a medial temporal lobe structure critically implicated in memory and learning, is not typically associated with negative outcomes. For example, memory impairments in severely addicted stimulant abusers are associated with hippocampal volume reduction/atrophy rather than with increased volumes (Berman et al., 2008). The dose-­‐response curve, with on its horizontal axis ‘stimulant dose’ and on its vertical axis ‘hippocampus volume’, could in theory follow an inverted U-­‐shape, such that a lower/clinical dose results in enlarged hippocampi while a higher/abusive dose results in hippocampal volume loss. A similar inverted U-­‐shape curve has previously been proposed to describe acute stimulant effects on cognition and brain activation patterns (Swanson et al., 2011). Acute stimulant-­‐induced improvement of declarative memory has been reported in patients with ADHD (Verster et al., 2010), as have hippocampal volume reductions in adults with a history of childhood stimulant treatment compared to those without (Onnink et al., 2014; Frodl et al., 2010). ADHD is not typically associated with declarative memory problems, and perhaps for that reason declarative memory performance was not among the cognitive domains assessed in our sample. Previous findings, in conjunction with our findings of subtle decline in memory performance after recreational (typically high-­‐dose) stimulant exposure (chapter 8) emphasize the need to further address potential lasting effects of stimulant treatment on the hippocampus and memory performance. At this point, we cannot exclude the possibility that enlarged hippocampi in young DRD4 7R-­‐
carriers could reflect damage to dopaminergic nerve terminals even at clinical dose. In short, we detected no harmful long-­‐term effects of stimulant treatment on frontal-­‐striatal grey or white matter structure, cortical thickness, reward-­‐related brain activation patterns, nor on any clinical, behavioral, or cognitive outcomes. In a subgroup of patients, high-­‐dose stimulant treatment was associated with enlarged hippocampus volume, but it is unclear whether such changes should be interpreted as beneficial or harmful, and whether they have functional/clinical implications. 183 Normalization and beneficial effects Contrary to the publicly endorsed concerns about harmful effects, the scientific community has in recent years reported mostly in favor of beneficial effects of long-­‐term treatment on the brain. More specifically, it has been suggested that stimulant treatment may reduce or normalize brain changes typically associated with ADHD. Previous MRI studies have indicated normalizing treatment effects on structural changes in for instance the basal ganglia (Frodl & Skokauskas, 2012; Nakao et al., 2011), lateral prefrontal cortex (Shaw et al., 2009), and anterior cingulate cortex (Semrud-­‐Clikeman et al., 2006), although there have also been negative findings (e.g., Onnink et al., 2014; Semrud-­‐Clikeman et al., 2014; Shaw et al., 2014). Sustainable neural modifications may result from repeated modulation of monoaminergic neurotransmission, or from repeated manifestation of age-­‐
appropriate behaviors (i.e., paying attention in class, making non-­‐impulsive decisions) increasing the strength of neural networks underlying these behaviors (Kasparek et al., 2015). One chapter in this thesis provides partial support for normalizing effects of stimulant treatment. In chapter 5, we found that frontal cortex volume was reduced in individuals with ADHD compared to their typically developing peers, and that in young patients carrying the DRD4 7R-­‐allele, higher levels of treatment were associated with increased frontal cortex volume. This may suggest treatment-­‐induced normalization of frontal cortex volumes in a specific subgroup of patients. Much more frequently, however, we concluded that normalization with stimulant treatment did not occur. We found decreased medial temporal cortical thickness (chapter 3), and lower frontal-­‐striatal fractional anisotropy (chapter 4), in both stimulant-­‐treated and stimulant-­‐naive participants. Striatal activation patterns during reward processing, that had previously been found to be altered in our ADHD sample, were not associated with treatment history either (chapter 5). Finally, although clinical and cognitive outcomes improved over time towards more normative levels, these improvements occurred in treated and non-­‐treated individuals alike (chapter 9). Rather than neural normalization, however, beneficial treatment effects may occur through compensatory strategies or processes. We found two indications of long-­‐term beneficial brain changes in stimulant-­‐treated patients with ADHD as compared to their typically developing peers, that were not also present at lower treatment levels or in stimulant-­‐naive patients. First, during reward outcome, individuals with a history of early-­‐onset and high-­‐dose stimulant treatment, but not those with late-­‐onset and low-­‐dose treatment, showed higher activation in brain regions of cognitive control (chapter 5). Increased cognitive control is likely beneficial to individuals with ADHD who tend to respond impulsively to reward, and may 184 compensate for deficient striatal activation patterns that were not normalized by stimulant treatment. Second, patients with ADHD presented with compromised structural connectivity in the orbitofrontal-­‐striatal pathway. In that same pathway, a negative association between cumulative stimulant dose and mean diffusivity indicated improved white matter structural connectivity at higher treatment levels, albeit through a different mechanism (mean diffusivity rather than fractional anisotropy; chapter 4). Potential mechanisms underlying such lasting treatment effects are further discussed below (see What mechanisms underlie effects of long-­‐
term stimulant treatment in the brain?) In short, we found little evidence that stimulant treatment may reduce or normalize brain changes typically associated with ADHD, except in the frontal cortex where treatment may normalize volume reductions in a specific subgroup of patients. However, beneficial treatment effects may have occurred through compensatory mechanisms without affecting/normalizing the initial deficit. The role of stimulant exposure patterns The distinction between previously treated and stimulant-­‐naive patients with ADHD is practical and often used, yet it may be an overly simplistic representation of complex stimulant treatment trajectories. Throughout this thesis, we went beyond this distinction and investigated whether different exposure patterns (e.g., younger or older age of treatment initiation, higher or lower daily dose, shorter or longer treatment duration, etc.) were associated with different long-­‐term outcomes, as had been suggested in animal studies (e.g., van der Marel et al., 2014 and 2015). The effects of incidental, high-­‐dose recreational stimulant exposure could be informative in this regard as well. For the studies in this thesis, detailed pharmacy transcripts provided detailed information regarding stimulant exposure patterns, such as age of treatment, dose, continuity, etc. Disentangling how these various exposure parameters contributed to stimulant-­‐induced brain changes in their unique and shared ways was challenging (see methodological considerations). All in all, when effects of treatment were found, they could typically not be attributed to a single treatment parameter. In chapter 6, where we reported a combined effect of age, genotype, and treatment on frontal cortex volumes, the treatment component in the model could be represented by either cumulative dose or treatment duration. For a similar combined effect in the hippocampus, the treatment component could be represented by cumulative dose, treatment duration, or age of treatment onset. It thus seems that the effects of these treatment indicators are shared rather than unique. Daily dose and recency of treatment appear to be of less importance. 185 Another indication that the different exposure parameters may have shared rather than unique effects was provided in chapter 5, where we performed a multivariate data-­‐driven classification of stimulant-­‐treated participants based on their lifetime exposure trajectories. In the optimal solution, the vast majority of stimulant-­‐treated participants was grouped in one of only two classes. Over 90% of stimulant-­‐treated participants could best be classified as either a) early treatment onset, long treatment duration, and high maximum and total dose, or b) late treatment onset, short treatment duration, and low maximum and total dose. Other exposure patterns (e.g., late onset and high dose) were highly uncommon. Thus, although we were able to compute various detailed exposure parameters from the pharmacy transcripts, within patients with ADHD these parameters were very highly correlated which hampered meaningful distinction between their potentially unique effects in the brain. Furthermore, although the distinction between stimulant-­‐treated and stimulant-­‐naive patients may indeed be overly simplistic, virtually all stimulant-­‐
users could be classified as either moderate-­‐users or intense-­‐users, indicating that differentiation between stimulant-­‐users is not necessarily very complex either. It is important that the prevalence and relevance of different exposure patterns are replicated in independent samples, preferably in samples based in different health care realities. Finally, there was the group of adolescent recreational stimulant users, who typically consumed stimulants in a binge-­‐like pattern (incidental, high-­‐dose exposure to cocaine or MDMA). Note that we did not directly compare stimulant exposure effects in recreational users to those in patients with ADHD. However, the effect of stimulant exposure on hippocampus volume was evaluated in both groups, and in both groups we found no exposure effect. High-­‐dose exposure did negatively affect declarative memory performance in the recreational users group. Unfortunately, declarative memory performance was not tested in the ADHD sample with low-­‐dose exposure, hence we are unable to conclude whether subtle memory changes occur specifically after high-­‐dose exposure patterns. Conclusion Taken together, we may reasonably conclude that long-­‐term effects of stimulant treatment on the developing brain in ADHD are subtle. Findings of long-­‐
term treatment effects were outnumbered by findings of no effects of treatment. We found no indications of harmful long-­‐term effects of stimulants on brain structure or function, although our findings of hippocampal changes do ask for further investigation. At the same time, we also found little evidence that brain changes typically associated with ADHD are reduced or even disappeared after stimulant 186 treatment, except in the frontal cortex where treatment may normalize volume reductions in a younger patients carrying the DRD4 7R-­‐allele. In two instances, however, we found that beneficial effects of long-­‐term treatment may have occurred through compensatory mechanisms. Importantly, even if stimulant treatment indeed caused either beneficial or harmful brain changes, such changes were not accompanied by better or worse outcomes in terms of ADHD symptoms, social-­‐
emotional functioning, or cognitive performance. In terms of exposure patterns, it appears that treatment duration, cumulative dose, and age of treatment onset have shared rather than unique contributions to long-­‐term brain changes. WHAT MECHANISMS UNDERLIE EFFECTS OF LONG-TERM STIMULANT TREATMENT IN THE BRAIN?
The second aim of this thesis was to advance our understanding of how long-­‐
term effects of stimulant treatment in the developing brain come about. Long-­‐term effects may be caused by the same mechanism that also underlies acute stimulant effects in the brain (i.e., blockade of dopaminergic autoreceptors in the striatum), but we propose that long-­‐term effects of treatment are likely the result of a different mechanism, on two grounds. First, previous PET studies have repeatedly shown that acute stimulant effects are localized within the striatum (e.g., Cherkasova et al., 2014), whereas here we found no evidence of treatment-­‐induced changes in striatal volume (chapter 6) or activation during reward (chapter 5). Rather, effects of long-­‐term treatment were localized in the frontal cortex, hippocampus, and supplementary motor area/dorsal anterior cingulate cortex. Second, if acute and long-­‐term effects would reflect the same mechanism, then a genetic variant predisposing acute stimulant effects (DAT1; Aarts et al., 2014, Kasparbauer et al., 2015) would be expected to predict the occurrence of long-­‐term changes as well. However, in our sample, DAT1 genotype was not predictive of long-­‐term changes in striatum volume (chapter 6), nor in frontal cortex or hippocampus volume (additional analyses, not shown). If different from the mechanism underlying acute stimulant effects, what could the process causing long-­‐term brain changes entail? Our findings of cortical remodeling after stimulant treatment in younger patients carrying the DRD4 7R-­‐allele may be especially informative in this regard. The DRD4 gene encodes postsynaptic dopamine D4 receptors, that play an important role in maintaining tonic dopamine levels and are crucial for the occurrence of neural plasticity (Asghari et al., 1995; Goto et al., 2010; Padmanabhan & Luna, 2013). Thus, the frontal cortex of young DRD4 7R-­‐
carriers may exhibit postsynaptic characteristics allowing for long-­‐term neural plasticity, i.e., structural adaptation of the brain, in the event of exposure to stimulants. Interestingly, DRD4 7R-­‐carriership has previously been associated with 187 better clinical outcome in older adolescents with ADHD of whom at least 78% had received stimulant treatment at some point in life (Shaw et al., 2007b). Moreover, DRD4 genotype and social environment together, but not DRD4 genotype alone, predicted prefrontal cortex activation during response inhibition (Richards et al., 2016). This may suggest a more general mechanism of DRD4 7R-­‐carriers being more sensitive to lasting brain changes when circumstances (such as stimulant treatment and/or a positive environment) are optimal. I do wish to emphasize that our findings of age-­‐ and genotype specific treatment effects await replication in an independent sample. Moreover, and more generally, our thoughts about underlying mechanisms remain highly speculative at this point, and require further investigation using other modalities such as PET or SPECT. METHODOLOGICAL CONSIDERATIONS
Treatment effects in an observational study design Any investigation of treatment or exposure effects spanning multiple years, including the current investigation, is inevitably naturalistic by design. An observational study design comes with limitations as well as opportunities. Associations between treatment and outcomes that are identified through observation may represent true effects of treatment, but could also reflect bias and/or confounding (Jepsen et al., 2004). We aimed to reduce information bias, occurring in observational studies when treatment itself is associated with the accuracy of reported outcomes (e.g., parent-­‐rated ADHD symptom scores while off-­‐medication may be less accurate after the child has consistently been using medication for some time), by the use of multiple informants. Furthermore, we attempted to statistically minimize the influence of confounding factors throughout this thesis, e.g., through matching and covariate adjustment. Nevertheless, bias and confounding cannot be ruled out. Especially confounding through unmeasured characteristics and confounding-­‐by-­‐indication / susceptibility bias (i.e., when treatment is preferentially prescribed to patients with worse prognosis, due to e.g., pretreatment severity) are difficult to address. Propensity score adjustment, i.e., statistical adjustment for the estimated propensity or likelihood that a given patient will receive treatment, could be valuable in this regard. Unfortunately, propensity estimates could not be calculated based on our data. Another drawback of observational study designs is their tendency to result in imbalanced exposure groups. Only a handful of participants within our ADHD sample were naïve to stimulant treatment, which is in accordance with high prescription rates in the Netherlands. In contrast, in the Youth At Risk sample, 188 recreational stimulant users were scarce which is in accordance with the relatively low incidence of recreational stimulant use amongst adolescents. Therefore, in both samples, attempts towards matching were often unsuccessful, or resulted in the exclusion of a substantial number of participants. Despite these limitations, observational cohort studies also have advantages. Results from observational studies are generally more consistent compared to those from randomized controlled trials, and concerns about susceptibility bias resulting in an overestimation of treatment effects have been invalidated in recent years (Concato et al., 2000; Concato, 2013). Moreover, large-­‐scale observational studies are especially suitable for the study of heterogeneous clinical populations such as patients with ADHD. Not only do such samples enhance generalizability of findings, but they also provide researchers with much-­‐needed statistical power to study within-­‐group variability. For example, the investigation of differential treatment effects at different ages, or at a wide range of doses, could not have been achieved in a randomized controlled trial. Effects of treatment in a cross-­‐sectional study design Both the IMAGE-­‐NeuroIMAGE and the Youth At Risk cohort are longitudinal by design. Unfortunately, however, several aspects of the NeuroIMAGE study design prevented optimal longitudinal investigation of stimulant treatment effects on the brain. First, the baseline assessment (IMAGE, approximately six years prior to NeuroIMAGE) did not include an MRI session. Therefore, at the level of brain outcomes, the available data were cross-­‐sectional. Cross-­‐sectional data are far from optimal for the investigation of developmental disorders, and perhaps even more problematic for the investigation of effects of treatment, as there is a substantial risk for confounding (Kraemer et al., 2000). Ideally, brain data is collected prior to initiation of treatment, to allow analysis of (and adjustment for) brain indices that at baseline distinguish between patients who will later receive stimulant treatment and those who will remain untreated. The cross-­‐sectional nature of our study should be kept in mind when interpreting associations with treatment and age. More conclusive interpretation of such effects can only be derived from longitudinal studies. Second, a substantial proportion of participants with ADHD had already received stimulant treatment prior to enrollment in the IMAGE study. Thus, for these participants, clinical baseline measurements did not represent their pre-­‐treatment state, and their pre-­‐
treatment status was not assessed. As a result, we were unable to statistically adjust for pre-­‐treatment characteristics such as symptom severity, which would have reduced the risk of susceptibility bias. Note that these two limitations do not apply to the analyses presented in chapter 8, as in the Youth At Risk sample all participants 189 underwent MRI scanning at baseline when they were drug-­‐naive. Ideally, longitudinal studies with repeated scanning should be prospective, i.e., starting prior to treatment onset. Fortunately, there is an advantage to the current cross-­‐sectional study as well. The NeuroIMAGE cohort stands out through the inclusion of older adolescents and young adults, which in a longitudinal design starting in childhood would be very time-­‐consuming and prone to dropout. The late adolescent/early adult phase is often characterized by marked clinical changes (e.g., reduction of hyperactivity symptoms) as well as normative developmental changes (e.g., improved executive functioning), yet has received very little attention in previous studies of ADHD. Specifically in terms of effects of treatment, the inclusion of older adolescents and young adults was an asset of our study as it allowed the investigation of stimulant effects occurring only after multiple years of treatment, and/or occurring long after treatment had been discontinued. The shared and unique effects of different treatment characteristics Naturally, characteristics of stimulant treatment history are highly correlated amongst each other. For instance, individuals who have started treatment at early age are likely to have longer treatment duration and higher lifetime cumulative dose during adolescence, when these parameters were calculated. Associations between treatment and age add to the complexity, as for example treatment duration but also daily dose tend to increase with age. Finally, treatment characteristics may also interact with each other and with age to affect brain development. For example, as was shown in animal studies (van der Marel et al., 2014 and 2015) and in chapter 6, a certain stimulant dose may cause lasting brain changes when administered at young age but not when administered at older age. Disentangling the influences of individual treatment parameters on brain development is very challenging, if not impossible. Nevertheless, throughout this thesis, we made several attempts to achieve such differentiation. In most of the empirical chapters, we initially evaluated associations between one encompassing stimulant treatment parameter on the one hand (i.e., age-­‐adjusted cumulative stimulant dose, or CSIADJ), and brain changes on the other. CSIADJ represented each of the other treatment indicators at least to some extent, while being independent of age at time of scan. We anticipated that CSIADJ was informative in itself, but would also serve as a proximate measure of other treatment parameters, which we would attempt to tease apart in second instance. As a benefit to this approach, the alpha level of statistical testing does not require adjustment for multiple comparisons, which is necessary when all treatment indicators would be tested one by one. However, as a disadvantage, those treatment parameters that are 190 less correlated with CSIADJ are less likely to have been detected. This shortcoming should be kept in mind when considering the unique and shared effects of stimulant treatment parameters (see The role of stimulant exposure patterns). We concluded that, when treatment effects were found, they could typically not be attributed to a single treatment parameter but rather represented an overall effect of cumulative dose, treatment duration, and onset age. Dose and recency of treatment appeared to be of less importance. Note that the lack of findings regarding dose and recency may also be reflective of lower correlations between these treatment parameters and CSIADJ. Detection of brain changes associated with treatment recency and daily dose may require univariate modeling of these parameters. CLINICAL IMPLICATIONS
I wish to discuss our findings in terms of potential implications for clinical practice. After all, concerns about long-­‐term consequences of stimulant treatment have dominated the public debate about such medications. Clinicians, parents, and patients, including those who dedicated their time to participate in our study, are longing for information that may aid their decision-­‐making. At the start of this section, however, I do want to emphasize once more that we were able to study correlates of long-­‐term stimulant treatment, that may or may not reflect long-­‐term consequences. Moreover, be advised that the brain is just one organ. For clinical decision-­‐making, long-­‐term risks and benefits for other organs, bodily functions, and other aspects of well-­‐being should be considered as well. Concerns that stimulant treatment may adversely affect brain development are frequently voiced. Quite contrarily, it has also been suggested that stimulant treatment may positively affect brain development. All in all, our findings provide little to no support for adverse long-­‐term treatment effects, while they are partially supportive of beneficial effects. At the neural level, potential subtle long-­‐term benefits may include 1) increased activation in cognitive control areas during reward processing; 2) improved orbitofrontal-­‐striatal structural connectivity; and 3) for a subgroup of patients, normalization of frontal cortex volume. Especially for clinicians, however, it is important to realize that even if certain aspects of brain structure or function improve after treatment, these changes may or may not ultimately benefit the child’s functioning. In chapter 9, we report a striking absence of lasting effects of treatment on ADHD symptoms, social-­‐emotional functioning, and cognitive functioning across various domains. Clinicians may be advised to communicate to parents and patients that stimulant treatment is likely to cause symptomatic relief, but is not expected to cause lasting improvement of ADHD pathology. 191 Clinicians may be interested to learn which patients benefit most from treatment, and which patients are at highest risk for harmful effects. In our studies, effects of treatment (if any) were the same for boys and girls. We found one indication that young patients possessing the DRD4 7R-­‐risk allele may be more sensitive to lasting treatment effects in the brain. However, even within this specific group, there are substantial differences between patients. At this moment, the prediction of lasting effects of treatment for an individual patient based on their age and genotype is far from likely. Clinicians may also be interested to learn what characterizes optimal stimulant treatment. Should treatment be initiated at early age? What is the optimal dose, and when should treatment be discontinued? Overall, when effects of treatment were found, they appeared to be more pronounced when lifetime total dose was higher, treatment duration was longer, and start age was earlier; however, not one of these characteristics could be identified as being the most important. Finally, I wish to draw the clinician’s attention to two preliminary findings in this thesis that, especially if replicated, could have clinical implications. First, our findings suggested that stimulant exposure could, in certain patients and/or at high dose, be associated with changes in hippocampus volume and declarative memory performance. Memory performance is not usually addressed during clinical assessments for ADHD. While awaiting further studies, clinicians should be vigilant of subtle changes in memory performance in stimulant-­‐treated patients, especially those who received a high dose at young age. Second, I wish to address prescription practices regarding combined stimulant and antipsychotic treatment. Combined stimulant and low-­‐dose antipsychotics treatment has been recommended, albeit off-­‐
label, for short-­‐term treatment of severe behavioral problems (Kutcher et al., 2004). In our sample, the suggested treatment duration of several weeks was exceeded en mass. Our findings indicating brain changes in this specific combined treatment group, suggest that antipsychotic treatment, especially for prolonged time, is at this point not commendable. FUTURE DIRECTIONS
In this final section, I suggest three major recommendations for future investigations into long-­‐term stimulant treatment effects in ADHD. First, evidently, there is a need for prospective longitudinal neuroimaging data, that includes a baseline assessment prior to treatment onset. In the two largest longitudinal ADHD cohorts to date (NIMH, and NeuroIMAGE) only a minority of participants were naive to stimulants at their initial MRI scan. Brain scans of stimulant-­‐naive patients will provide essential information about non-­‐treated ADHD that is now lacking. As the majority of these initially stimulant-­‐naive patients will at some point commence 192 treatment, targeted inclusion of additional stimulant-­‐naive participants during the follow-­‐up phase may be needed. I further suggest that prospective longitudinal studies, where possible, employ alternative neuroimaging techniques such as single photon emission computed tomography (SPECT) and pharmacological MRI (phMRI). (Semi-­‐) quantitative methods have previously suggested aggravation of case-­‐control differences in dopamine metabolism after long-­‐term treatment (Ludolph et al., 2008; Wang et al., 2013), which may not be measurable using conventional MRI. Commendable efforts have been made by another Netherlands-­‐based research group to employ such alternative neuroimaging methods (e.g., Bottelier et al., 2014; van der Marel et al., 2014 and 2015). The second recommendation inclines additional effort to investigate functional rather than structural long-­‐term correlates of stimulant treatment. Functional neuroimaging studies on this topic are still sparse, especially compared to the substantial number of structural MRI studies. In this thesis, we found that stimulant treatment history was associated with altered activation patterns in cognitive control areas during reward processing, in the absence of structural brain changes in this area. Changes in neural activation patterns, that last when stimulant treatment is temporarily ceased, may provide an intermediate level between long-­‐
term structural brain changes and acute behavioral or clinical treatment effects. Evaluating long-­‐term treatment effects on brain activation patterns would thus provide novel information regarding the long-­‐term risks and benefits of stimulant treatment. Finally, third, I recommend that future studies specifically address long-­‐term stimulant treatment effects on learning and memory performance, and on hippocampal structure and function. Perhaps as a result of their questionable link with ADHD symptoms, declarative memory and the hippocampus are not typically targeted in the study of ADHD. As an illustration, the Pubmed query ‘ADHD declarative memory’ yielded only 16 hits; a far cry from the 799 hits for ‘ADHD response inhibition’. Subtle changes in declarative memory performance after stimulant treatment, either detrimental or beneficial, could thus easily have gone undetected. By contrast, the hippocampus and learning/memory performance have traditionally been an important focus of (animal) research into stimulant-­‐induced neurotoxicity and addiction. Future studies would benefit from transdisciplinary efforts, in which expertise from the fields of neurotoxicity/addiction and child and adolescent psychiatry are combined. 193 194 REFERENCES
195 196 Aarts, E., Holstein, M. Van, Hoogman, M., Onnink, M., Kan, C., Franke, B., … Cools, R. (2014). Reward modulation of cognitive function in adult attention-­‐deficit/hyperactivity disorder: a pilot study on the role of striatal dopamine. Behavioural Pharmacology, 26(1-­‐2), 227–240. Abikoff, H., Hechtman, L., Klein, R., Weiss, G., Fleiss, K., Etcovitch, J., … Pollack, S. (2004). Symptomatic improvement in children with ADHD treated with long-­‐term methylphenidate and multimodal psychosocial treatment. Journal of the American Academy of Child and Adolescent Psychiatry, 43(7), 802–811. Almeida, L. G., Ricardo-­‐Garcell, J., Prado, H., Barajas, L., Fernández-­‐Bouzas, A., Avila, D., & Martínez, R. B. (2010). Reduced right frontal cortical thickness in children, adolescents and adults with ADHD and its correlation to clinical variables: a cross-­‐sectional study. Journal of Psychiatric Research, 44(16), 1214–1223. Almeida Montes, L. G., Prado Alcántara, H., Martínez García, R. B., De La Torre, L. B., Avila Acosta, D., & Duarte, M. G. (2013). Brain cortical thickness in ADHD: age, sex, and clinical correlations. Journal of Attention Disorders, 17(8), 641–654. American Psychiatric Association (2000). Diagnostic and statistical manual of mental disorders: DSM-­‐IV-­‐
TR. American Psychiatric Publishing Inc. American Psychiatric Association (2013). Diagnostic and Statistical Manual of Mental Disorders (5th ed.). Arlington, VA: American Psychiatric Publishing Inc. Amico, F., Stauber, J., Koutsouleris, N., & Frodl, T. (2011). Anterior cingulate cortex gray matter abnormalities in adults with attention-­‐deficit/hyperactivity disorder: a voxel-­‐based morphometry study. Psychiatry Research, 191(1), 31–35. An, L., Cao, X. H., Cao, Q. J., Sun, L., Yang, L., Zou, Q. H., … Wang, Y. F. (2013). Methylphenidate normalizes resting-­‐state brain dysfunction in boys with attention-­‐deficit/hyperactivity disorder. Neuropsychopharmacology, 38(7), 1287–1295. Andersen, S. L. (2005). Stimulants and the developing brain. Trends in Pharmacological Sciences, 26(5), 237–243. Andersen, S. L., & Navalta, C. P. (2011). Annual research review: New frontiers in developmental neuropharmacology: Can long-­‐term therapeutic effects of drugs be optimized through carefully timed early intervention? Journal of Child Psychology and Psychiatry and Allied Disciplines, 52(4), 476–503. Anderson, C. M., Polcari, A., Lowen, S. B., Renshaw, P. F., & Teicher, M. H. (2002). Effects of methylphenidate on functional magnetic resonance relaxometry of the cerebellar vermis in boys with ADHD. American Journal of Psychiatry, 159(8), 1322–1328. Antshel, K., Zhang-­‐James, Y., & Faraone, S. (2013). The comorbidity of ADHD and autism spectrum disorder. Expert Review of Neurotherapeutics, 13(10), 1117–1128. Armstrong, B. D., & Noguchi, K. K. (2004). The neurotoxic effects of 3,4-­‐methylenedioxymethamphetamine (MDMA) and methamphetamine on serotonin, dopamine, and GABA-­‐ergic terminals: An in-­‐vitro autoradiographic study in rats. NeuroToxicology, 25(6), 905–914. Arnold, L. E., Abikoff, H. B., Cantwell, D. P., Conners, C. K., Elliott, G. R., Greenhill, L. L., … Wells, K. C. (1997). NIMH collaborative multimodal treatment study of children with ADHD (MTA): Design, methodology, and protocol evolution. Archives of General Psychiatry, 54(9), 865-­‐870. Arnsten, A. F. T. (2006). Therapeutic actions in ADHD. Neuropharmacology, 31, 2376–2383. Asghari, V., Sanyal, S., Buchwaldt, S., Paterson, A, Jovanovic, V., & van Tol, H. H. (1995). Modulation of intracellular cyclic AMP levels by different human dopamine D4 receptor variants. Journal of Neurochemistry, 65(3), 1157–1165. Ashburner, J., & Friston, K. (2000). Voxel-­‐based morphomethy -­‐ the methods. NeuroImage, 11(6), 805–821. Ashburner, J., & Friston, K. J. (2005). Unified segmentation. NeuroImage, 26(3), 839–851. 197 Ashtari, M., Avants, B., Cyckowski, L., Cervellione, K. L., Roofeh, D., Cook, P., … Kumra, S. (2011). Medial temporal structures and memory functions in adolescents with heavy cannabis use. Journal of Psychiatric Research, 45(8), 1055–1066. Ashtari, M., Kumra, S., Bhaskar, S. L., Clarke, T., Thaden, E., Cervellione, K. L., … Ardekani, B. A. (2005). Attention-­‐deficit/hyperactivity disorder: a preliminary diffusion tensor imaging study. Biological Psychiatry, 57(5), 448–455. Bachmann, C. J., Lempp, T., Glaeske, G., & Hoffmann, F. (2014). Antipsychotic prescription in children and adolescents: an analysis of data from a German statutory health insurance company from 2005 to 2012. Deutsches Ärzteblatt International, 111(3), 25–34. Barbaresi, W. J., Katusic, S. K., Colligan, R. C., Weaver, A. L., Leibson, C. L., & Jacobsen, S. J. (2006). Long-­‐term stimulant medication treatment of attention-­‐deficit/hyperactivity disorder: Results from a population-­‐based study. Journal of Developmental and Behavioral Pediatrics, 27, 10. Batty, M. J., Liddle, E. B., Pitiot, A., Toro, R., Groom, M. J., Scerif, G., … Hollis, C. (2010). Cortical gray matter in attention-­‐deficit/hyperactivity disorder: A structural magnetic resonance imaging study. Journal of the American Academy of Child and Adolescent Psychiatry, 49(3), 229–238. Bava, S., Thayer, R., Jacobus, J., Ward, M., Jernigan, T. L., & Tapert, S. F. (2010). Longitudinal characterization of white matter maturation during adolescence. Brain Research, 1327, 38–46. Becker, B., Wagner, D., Koester, P., Bender, K., Kabbasch, C., Gouzoulis-­‐Mayfrank, E., & Daumann, J. (2013). Memory-­‐related hippocampal functioning in ecstasy and amphetamine users: A prospective fMRI study. Psychopharmacology, 225(4), 923–934. Bédard, A. C., Schulz, K. P., Cook, E. H., Fan, J., Clerkin, S. M., Ivanov, I., … Newcorn, J. H. (2010). Dopamine transporter gene variation modulates activation of striatum in youth with ADHD. NeuroImage, 53(3), 935–942. Belcher, A. M., O’Dell, S. J., & Marshall, J. F. (2005). Impaired object recognition memory following methamphetamine, but not p-­‐chloroamphetamine-­‐ or d-­‐amphetamine-­‐induced neurotoxicity. Neuropsychopharmacology, 30(11), 2026–2034. Bellgrove, M. A, Johnson, K. A, Barry, E., Mulligan, A., Hawi, Z., Gill, M., … Chambers, C. D. (2009). Dopaminergic haplotype as a predictor of spatial inattention in children with attention-­‐
deficit/hyperactivity disorder. Archives of General Psychiatry, 66(10), 1135–1142. Benjamini, E., & Hochberg, Y. (1995). Controlling the False Discovery Rate: A practical and powerful approach to multiple testing. Journal of the Royal Statistical Society, 57(1), 289–300. Berman, S., Kuczenski, R., McCracken, J., & London, E. D. (2009). Potential adverse effects of amphetamine treatment on brain and behavior: a review. Molecular Psychiatry, 14, 123–142. Berman, S., O’Neill, J., Fears, S., Bartzokis, G., & London, E. D. (2008). Abuse of amphetamines and structural abnormalities in the brain. Annals of the New York Academy of Sciences, 1141, 195–220. Biederman, J., Monuteaux, M. C., Spencer, T., Wilens, T. E., & Faraone, S. V. (2009). Do stimulants protect against psychiatric disorders in youth with ADHD? A 10-­‐year follow-­‐up study. Pediatrics, 124(1), 71–78. Biederman, J., Petty, C. R., O’Connor, K. B., Hyder, L. L., & Faraone, S. V. (2012). Predictors of persistence in girls with attention-­‐deficit/hyperactivity disorder: results from an 11-­‐year controlled follow-­‐up study. Acta Psychiatrica Scandinavica, 125(2), 147–156. Biezonski, D. K., & Meyer, J. S. (2010). Effects of 3,4-­‐methylenedioxymethamphetamine (MDMA) on serotonin transporter and vesicular monoamine transporter 2 protein and gene expression in rats: Implications for MDMA neurotoxicity. Journal of Neurochemistry, 112(4), 951–962. Bledsoe, J., Semrud-­‐Clikeman, M., & Pliszka, S. R. (2009). A magnetic resonance imaging study of the cerebellar vermis in chronically treated and treatment-­‐naïve children with attention-­‐
deficit/hyperactivity disorder combined type. Biological Psychiatry, 65(7), 620–624. Blumenthal, J. D., Zijdenbos, A., Molloy, E., & Giedd, J. N. (2002). Motion artifact in magnetic resonance imaging: implications for automated analysis. NeuroImage, 16(1), 89–92. 198 Bottelier, M. A, Schouw, M. L. J., Klomp, A., Tamminga, H. G. H., Schrantee, A. G. M., Bouziane, C., … Reneman, L. (2014). The effects of Psychotropic drugs On Developing brain (ePOD) study: methods and design. BMC Psychiatry, 14, 48. Brookes, K. J., Mill, J., Guindalini, C., Curran, S., Xu, X., Knight, J., … Asherson, P. (2006). A common haplotype of the dopamine transporter gene associated with attention-­‐deficit/hyperactivity disorder and interacting with maternal use of alcohol during pregnancy. Archives of General Psychiatry, 63(1), 74–81. Brown, S. A, Tapert, S. F., Granholm, E., & Delis, D. C. (2000). Neurocognitive functioning of adolescents: effects of protracted alcohol use. Alcoholism, Clinical and Experimental Research, 24(2), 164–171. Brown, S., Myers, M., Lippke, L., Tapert, S., Steward, D., & Vik, P. (1998). Psychometric evaluation of the Customary Drinking and Drug Use Record (CDDR): a measure of adolescent alcohol and drug involvement. Journal of Studies on Alcohol, 59(4), 427–438. Buitelaar, J. K., & Coghill, D. R. (2013). Brain imaging: closing the gap between basic research and clinical application is urgently needed. European Child and Adolescent Psychiatry, 22(12), 715–717. Buitelaar, J., & Medori, R. (2010). Treating attention-­‐deficit/hyperactivity disorder beyond symptom control alone in children and adolescents: a review of the potential benefits of long-­‐acting stimulants. European Child and Adolescent Psychiatry, 19(4), 325–340. Bush, G., Spencer, T. J., Holmes, J., Shin, L. M., Valera, E. M., Seidman, L. J., … Biederman, J. (2008). Functional magnetic resonance imaging of methylphenidate and placebo in attention-­‐deficit/hyperactivity disorder during the multi-­‐source interference task. Archives of General Psychiatry, 65(1), 102–114. Bush, G., Valera, E. M., & Seidman, L. J. (2005). Functional neuroimaging of attention-­‐deficit/hyperactivity disorder: a review and suggested future directions. Biological Psychiatry, 57(11), 1273–1284. Bussing, R., Grudnik, J., Mason, D., Wasiak, M., & Leonard, C. (2002). ADHD and conduct disorder : An MRI study in a community sample. World Journal of Biological Psychiatry, 3(4), 6–10. Canese, R., Adriani, W., Marco, E. M., De Pasquale, F., Lorenzini, P., De Luca, N., … Laviola, G. (2009). Peculiar response to methylphenidate in adolescent compared to adult rats: A phMRI study. Psychopharmacology, 203(1), 143–153. Cao, Q., Sun, L., Gong, G., Lv, Y., Cao, X., Shuai, L., … Wang, Y. (2010). The macrostructural and microstructural abnormalities of corpus callosum in children with attention-­‐deficit/hyperactivity disorder: a combined morphometric and diffusion tensor MRI study. Brain Research, 1310, 172–
180. Carmona, S., Hoekzema, E., Ramos-­‐Quiroga, J. A., Richarte, V., Canals, C., Bosch, R., … Vilarroya, O. (2012). Response inhibition and reward anticipation in medication-­‐naïve adults with attention-­‐
deficit/hyperactivity disorder: A within-­‐subject case-­‐control neuroimaging study. Human Brain Mapping, 33(10), 2350–2361. Carmona, S., Vilarroya, O., Bielsa, A., Tremols, V., Soliva, J. C., Rovira, M., … Bulbena, A. (2005). Global and regional gray matter reductions in ADHD: A voxel-­‐based morphometric study. Neuroscience Letters, 389(2), 88–93. Casey, B. J., Epstein, J. N., Buhle, J., Liston, C., Davidson, M. C., Tonev, S. T., … Glover, G. (2007). Frontostriatal connectivity and its role in cognitive control in parent-­‐child dyads with ADHD. American Journal of Psychiatry, 164(11), 1729–1736. Castellanos, F. X., Lee, P. P., Sharp, W., Jeffries, N. O., Greenstein, D. K., Clasen, L. S., … Rapoport, J. L. (2002). Developmental trajectories of brain volume abnormalities in children and adolescents with attention-­‐deficit/hyperactivity disorder. Journal of the American Medical Association, 288(14), 1740–1748. Chan, E., Fogler, J. M., & Hammerness, P. G. (2016). Treatment of attention-­‐deficit/hyperactivity disorder in adolescents. Journal of the American Medical Association, 315(18), 1997. 199 Chang, Z., D’Onofrio, B. M., Quinn, P. D., Lichtenstein, P., & Larsson, H. (2016). Medication for attention-­‐
deficit/hyperactivity disorder and risk for depression: A nationwide longitudinal cohort study. Biological Psychiatry (ePub) Charach, A., Ickowicz, A., & Schachar, R. (2004). Stimulant treatment over five years: adherence, effectiveness, and adverse effects. Journal of the American Academy of Child and Adolescent Psychiatry, 43(5), 559–567. Cheng, J. T., & Li, J. S. (2013). Intra-­‐orbitofrontal cortex injection of haloperidol removes the beneficial effect of methylphenidate on reversal learning of spontaneously hypertensive rats in an attentional set-­‐
shifting task. Behavioural Brain Research, 239(1), 148–154. Cherkasova, M. V., Faridi, N., Casey, K. F., O’Driscoll, G. A., Hechtman, L., Joober, R., … Benkelfat, C. (2014). Amphetamine-­‐induced dopamine release and neurocognitive function in treatment-­‐naive adults with ADHD. Neuropsychopharmacology, 39(6), 1498–1507. Coghill, D. R., Seth, S., & Matthews, K. (2014a). A comprehensive assessment of memory, delay aversion, timing, inhibition, decision making and variability in attention-­‐deficit/hyperactivity disorder: advancing beyond the three-­‐pathway models. Psychological Medicine, 44(9), 1989-­‐2001. Coghill, D. R., Seth, S., Pedroso, S., Usala, T., Currie, J., & Gagliano, A. (2014b). Effects of methylphenidate on cognitive functions in children and adolescents with attention-­‐deficit/hyperactivity disorder: evidence from a systematic review and a meta-­‐analysis. Biological Psychiatry, 76(8), 603–615. Cohen, J. (1992). A power primer. Quantitative Methods in Psychology, 112(1), 155–159. Concato, J. (2013). Study design and “evidence” in patient-­‐oriented research. American Journal of Respiratory and Critical Care Medicine, 187(11), 1167–1172. Concato, J., Shah, N., & Horwitz, R. I. (2000). Randomized, controlled trials, observational studies, and the hierarchy of research designs. The New England Journal of Medicine, 342(25), 1887–1892. Conners, C. (2002). Forty years of methylphenidate treatment in attention-­‐deficit/hyperactivity disorder. Journal of Attention Disorders, 6(suppl 1), S17–30. Conners, C. K., Erhardt, D., & Sparrow, A. P. (1999). Conner’s Adult ADHD Rating Scales: CAARS. North Tonawanda, NY: Multi-­‐Health Systems. Conners, C. K., Sitarenios, G., Parker, J. D. A., & Epstein, J. N. (1998a). Revision and re-­‐standardization of the Conners’ Teacher Rating Scale (CTRS-­‐R): factor structure, reliability, and criterion validity. Journal of Abnormal Child Psychology, 26(4), 279–291. Conners, C. K., Sitarenios, G., Parker, J. D. A., & Epstein, J. N. (1998b). The revised Conners’ Parent Rating Scale (CPRS-­‐R): factor structure, reliability, and criterion validity. Journal of Abnormal Child Psychology, 26(4), 257–268. Connor, D. F., Steeber, J., & McBurnett, K. (2010). A review of attention-­‐deficit/hyperactivity disorder complicated by symptoms of oppositional defiant disorder or conduct disorder. Journal of Developmental and Behavioral Pediatrics, 31(5), 427–440. Contini, V., Rovaris, D. L., Victor, M. M., Grevet, E. H., Rohde, L. A, & Bau, C. H. D. (2013). Pharmacogenetics of response to methylphenidate in adult patients with attention-­‐deficit/hyperactivity disorder (ADHD): a systematic review. European Neuropsychopharmacology, 23(6), 555–560. Contini, V., Victor, M. M., Bertuzzi, G. P., Salgado, C. A. I., Picon, F. A, Grevet, E. H., … Bau, C. H. D. (2012). No significant association between genetic variants in 7 candidate genes and response to methylphenidate treatment in adult patients with ADHD. Journal of Clinical Psychopharmacology, 32(6), 820–823. Copeland, W. E., Adair, C. E., Smetanin, P., Stiff, D., Briante, C., Colman, I., … Angold, A. (2013). Diagnostic transitions from childhood to adolescence to early adulthood. Journal of Child Psychology and Psychiatry, and Allied Disciplines, 54(7), 791–799. Cortese, S. (2012). The neurobiology and genetics of attention-­‐deficit/hyperactivity disorder (ADHD): What every clinician should know. European Journal of Paediatric Neurology, 16(5), 422–433. 200 Cowan, R. L., Lyoo, I. K., Sung, S. M., Ahn, K. H., Kim, M. J., Hwang, J., … Renshaw, P. F. (2003). Reduced cortical gray matter density in human MDMA (ecstasy) users: A voxel-­‐based morphometry study. Drug and Alcohol Dependence, 72(3), 225–235. Cubillo, A., Halari, R., Giampietro, V., Taylor, E., & Rubia, K. (2011). Fronto-­‐striatal underactivation during interference inhibition and attention allocation in grown up children with attention-­‐
deficit/hyperactivity disorder and persistent symptoms. Psychiatry Research -­‐ Neuroimaging, 193(1), 17–27. Cubillo, A., Halari, R., Smith, A., Taylor, E., & Rubia, K. (2012a). A review of fronto-­‐striatal and fronto-­‐cortical brain abnormalities in children and adults with attention-­‐deficit/hyperactivity disorder (ADHD) and new evidence for dysfunction in adults with ADHD during motivation and attention. Cortex, 48(2), 194–215. Cubillo, A., Smith, A. B., Barrett, N., Giampietro, V., Brammer, M. J., Simmons, A., & Rubia, K. (2012b). Shared and drug-­‐specific effects of atomoxetine and methylphenidate on inhibitory brain dysfunction in medication-­‐naive ADHD Boys. Cerebral Cortex, 24(1), 174-­‐185. Dale, A. M., Fischl, B., & Sereno, M. I. (1999). Cortical surface-­‐based analysis. I. Segmentation and surface reconstruction. NeuroImage, 9(2), 179–194. Dalsgaard, S., Mortensen, P. B., Frydenberg, M., & Thomsen, P. H. (2014). ADHD, stimulant treatment in childhood and subsequent substance abuse in adulthood -­‐ a naturalistic long-­‐term follow-­‐up study. Addictive Behaviors, 39(1), 325–328. Dalsgaard, S., Nielsen, H. S., & Simonsen, M. (2013). Five-­‐fold increase in national prevalence rates of attention-­‐deficit/hyperactivity disorder medications for children and adolescents with autism spectrum disorder, attention-­‐deficit/hyperactivity disorder, and other psychiatric disorders: A Danish register-­‐based study. Journal of Child and Adolescent Psychopharmacology, 23(7), 432–
439. Daumann, J., Koester, P., Becker, B., Wagner, D., Imperati, D., Gouzoulis-­‐Mayfrank, E., & Tittgemeyer, M. (2011). Medial prefrontal gray matter volume reductions in users of amphetamine-­‐type stimulants revealed by combined tract-­‐based spatial statistics and voxel-­‐based morphometry. NeuroImage, 54(2), 794–801. De Luis-­‐García, R., Cabús-­‐Piñol, G., Imaz-­‐Roncero, C., Argibay-­‐Quiñones, D., Barrio-­‐Arranz, G., Aja-­‐Fernández, S., & Alberola-­‐López, C. (2015). Attention-­‐deficit/hyperactivity disorder and medication with stimulants in young children: A DTI study. Progress in Neuro-­‐Psychopharmacology and Biological Psychiatry, 57, 176–184. De Win, M. M. L., Jager, G., Booij, J., Reneman, L., Schilt, T., Lavini, C., … van den Brink, W. (2008). Sustained effects of ecstasy on the human brain: A prospective neuroimaging study in novel users. Brain, 131(11), 2936–2945. De Zeeuw, P., Mandl, R. C. W., Hulshoff Pol, H. E., van Engeland, H., & Durston, S. (2012). Decreased frontostriatal microstructural organization in attention-­‐deficit/hyperactivity disorder. Human Brain Mapping, 33(8), 1941–1951. Defagot, M. C., Malchiodi, E. L., Villar, M. J., & Antonelli, M. C. (1997). Distribution of D4 dopamine receptor in rat brain with sequence-­‐specific antibodies. Molecular Brain Research, 45(1), 1–12. Del Campo, N., Chamberlain, S. R., Sahakian, B. J., & Robbins, T. W. (2011). The roles of dopamine and noradrenaline in the pathophysiology and treatment of attention-­‐deficit/hyperactivity disorder. Biological Psychiatry, 69(12), e145–e157. Delis, D., Kramer, J., Kaplan, A., & Ober, B. (1994). Manual for the California Verbal Learning Test -­‐ Children’s version. San Antonio, TX: Psychological Corporation. Delis, D., Kramer, J., Kaplan, A., & Ober, B. (2000). California Verbal Learning Test -­‐ 2nd edition (CVLT-­‐II) manual. San Antonio, TX: The Psychological Corporation. 201 Dennis, M., Francis, D. J., Cirino, P. T., Schachar, R., Barnes, M. A., & Fletcher, J. M. (2009). Why IQ is not a covariate in cognitive studies of neurodevelopmental disorders. Journal of the International Neuropsychological Society, 15(3), 331–343. Depue, B. E., Burgess, G. C., Bidwell, L. C., Willcutt, E. G., & Banich, M. T. (2010). Behavioral performance predicts grey matter reductions in the right inferior frontal gyrus in young adults with combined type ADHD. Psychiatry Research -­‐ Neuroimaging, 182(3), 231–237. Desikan, R. S., Ségonne, F., Fischl, B., Quinn, B., Dickerson, B., Blacker, D., … Killiany, R. (2006). An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest. NeuroImage, 31(3), 968–980. Di Martino, A., Scheres, A., Margulies, D. S., Kelly, S. M. C., Uddin, L. Q., Shehzad, Z., … Milham, M. P. (2008). Functional connectivity of human striatum: A resting state fMRI study. Cerebral Cortex, 18(12), 2735–2747. Dickstein, D. P., Bannon, K., Castellanos, F. X., & Milham, M. P. (2006). The neural correlates of attention-­‐
deficit/hyperactivity disorder: an ALE meta-­‐analysis. Journal of Child Psychology and Psychiatry, 47(10), 1061–1062. Dodds, C. M., Müller, U., Clark, L., van Loon, A., Cools, R., & Robbins, T. W. (2008). Methylphenidate has differential effects on blood oxygenation level-­‐dependent signal related to cognitive subprocesses of reversal learning. The Journal of Neuroscience, 28(23), 5976–5982. Doyle, A. (2006). Executive functions in attention-­‐deficit/hyperactivity disorder. The Journal of Clinical Psychiatry, 67(suppl 8), 21–26. Doyle, A. E., Willcutt, E. G., Seidman, L. J., Biederman, J., Chouinard, V. A., Silva, J., & Faraone, S. V. (2005). Attention-­‐deficit/hyperactivity disorder endophenotypes. Biological Psychiatry, 57(11), 1324–
1335. Duerden, E. G., Tannock, R., & Dockstader, C. (2012). Altered cortical morphology in sensorimotor processing regions in adolescents and adults with attention-­‐deficit/hyperactivity disorder. Brain Research, 1445, 82–91. Durston, S., de Zeeuw, P., & Staal, W. G. (2009). Imaging genetics in ADHD: A focus on cognitive control. Neuroscience and Biobehavioral Reviews, 33(5), 674–689. Durston, S., Fossella, J. A, Casey, B. J., Hulshoff Pol, H. E., Galvan, A, Schnack, H. G., … van Engeland, H. (2005). Differential effects of DRD4 and DAT1 genotype on fronto-­‐striatal gray matter volumes in a sample of subjects with attention-­‐deficit/hyperactivity disorder, their unaffected siblings, and controls. Molecular Psychiatry, 10(7), 678–685. Durston, S., Fossella, J. A, Mulder, M. J., Casey, B. J., Ziermans, T. B., Vessaz, M. N., & van Engeland, H. (2008). Dopamine transporter genotype conveys familial risk of attention-­‐deficit/hyperactivity disorder through striatal activation. Journal of the American Academy of Child and Adolescent Psychiatry, 47(1), 61–67. Dwan, K., Altman, D. G., Arnaiz, J. A., Bloom, J., Chan, A. W., Cronin, E., … Williamson, P. R. (2008). Systematic review of the empirical evidence of study publication bias and outcome reporting bias. PLoS One, 11(6), e1001666 Eisch, A. J., & Harburg, G. C. (2006). Opiates, psychostimulants, and adult hippocampal neurogenesis: Insights for addiction and stem cell biology. Hippocampus, 16(3), 271–286. Epstein, J. N., Casey, B. J., Tonev, S. T., Davidson, M. C., Reiss, A. L., Garrett, A., … Spicer, J. (2007). ADHD-­‐ and medication-­‐related brain activation effects in concordantly affected parent-­‐child dyads with ADHD. Journal of Child Psychology and Psychiatry, 48(9), 899–913. Epstein, T., Na, P., & Weiser, M. (2014). Immediate-­‐release methylphenidate for attention-­‐
deficit/hyperactivity disorder (ADHD) in adults (Review). The Cochrane Library, (9), 1–75. Ersche, K. D., Williams, G. B., Robbins, T. W., & Bullmore, E. T. (2013). Meta-­‐analysis of structural brain abnormalities associated with stimulant drug dependence and neuroimaging of addiction vulnerability and resilience. Current Opinion in Neurobiology, 23(4), 615–624. 202 Faber, A., Kalverdijk, L. J., de Jong-­‐van den Berg, L. T. W., Hugtenburg, J. G., Minderaa, R. B., & Tobi, H. (2010). Co-­‐morbidity and patterns of care in stimulant-­‐treated children with ADHD in the Netherlands. European Child and Adolescent Psychiatry, 19(2), 159–166. Faraone, S. V, Biederman, J., & Mick, E. (2006). The age-­‐dependent decline of attention-­‐deficit/hyperactivity disorder: a meta-­‐analysis of follow-­‐up studies. Psychological Medicine, 36(2), 159–165. Faraone, S. V, Bonvicini, C., & Scassellati, C. (2014). Biomarkers in the diagnosis of ADHD -­‐ promising directions. Current Psychiatry Reports, 16(11), 497. Faraone, S. V., & Buitelaar, J. (2010). Comparing the efficacy of stimulants for ADHD in children and adolescents using meta-­‐analysis. European Child and Adolescent Psychiatry, 19(4), 353–364. Fernández-­‐Jaén, A., López-­‐Martín, S., Albert, J., Fernández-­‐Mayoralas, D. M., Fernández-­‐Perrone, A. L., Tapia, D. Q., & Calleja-­‐Pérez, B. (2014). Cortical thinning of temporal pole and orbitofrontal cortex in medication-­‐naïve children and adolescents with ADHD. Psychiatry Research, 224(1), 8–13. Fischl, B., & Dale, A. M. (2000). Measuring the thickness of the human cerebral cortex from magnetic resonance images. Proceedings of the National Academy of Sciences of the United States of America, 97(20), 11050–11055. Fischl, B., Salat, D. H., van der Kouwe, A., Makris, N., Ségonne, F., Quinn, B., & Dale, A. M. (2004). Sequence-­‐
independent segmentation of magnetic resonance images. NeuroImage, 23 suppl 1, s69–s84. Fischl, B., Sereno, M. I., & Dale, A. M. (1999). Cortical surface-­‐based analysis II: Inflation, flattening, and a surface-­‐based coordinate system. NeuroImage, 9(2), 195–207. Francx, W., Zwiers, M. P., Mennes, M., Oosterlaan, J., Heslenfeld, D., Hoekstra, P. J., … Buitelaar, J. K. (2015). White matter microstructure and developmental improvement of hyperactive/impulsive symptoms in attention-­‐deficit/hyperactivity disorder. Journal of Child Psychology and Psychiatry, and Allied Disciplines, 56(12), 1289-­‐1297. Franke, B., Vasquez, A. A., Johansson, S., Hoogman, M., Romanos, J., Boreatti-­‐Hümmer, A., … Reif, A. (2010). Multicenter analysis of the SLC6A3/DAT1 VNTR haplotype in persistent ADHD suggests differential involvement of the gene in childhood and persistent ADHD. Neuropsychopharmacology, 35(3), 656–664. Frazier, J. A, Giedd, J. N., Kaysen, D., Albus, K., Hamburger, S., Alaghband-­‐Rad, J., … Rapoport, J. L. (1996). Childhood-­‐onset schizophrenia: Brain MRI rescan after 2 years of clozapine maintenance treatment. American Journal of Psychiatry, 153(4), 564–566. Frieden, T., Jaffe, H., Cono, J., Richards, C., & Iademarco, M. (2014). Youth Risk Behavior Surveillance -­‐ US2013. Atlanta, United States: Center for Surveillance, Epidemiology, and Laboratory Services, Centers for Disease Control and Prevention (CDC), U.S. Department of Health and Human Services Frodl, T., & Skokauskas, N. (2012). Meta-­‐analysis of structural MRI studies in children and adults with attention-­‐deficit/hyperactivity disorder indicates treatment effects. Acta Psychiatrica Scandinavica, 125(2), 114–126. Frodl, T., Stauber, J., Schaaff, N., Koutsouleris, N., Scheuerecker, J., Ewers, M., … Meisenzahl, E. (2010). Amygdala reduction in patients with ADHD compared with major depression and healthy volunteers. Acta Psychiatrica Scandinavica, 121(2), 111–118. Froehlich, T. E., Epstein, J. N., Nick, T. G., Melguizo Castro, M. S., Stein, M. A., Brinkman, W. B., … Kahn, R. S. (2011). Pharmacogenetic predictors of methylphenidate dose-­‐response in attention-­‐
deficit/hyperactivity disorder. Journal of the American Academy of Child and Adolescent Psychiatry, 50(11), 1129–1139. Fusar-­‐Poli, P., Rubia, K., Rossi, G., Sartori, G., & Balottin, U. (2012). Striatal dopamine transporter alterations in ADHD: Pathophysiology or adaptation to psychostimulants? A meta-­‐analysis. American Journal of Psychiatry, 169(3), 264–272. Garbe, E., Mikolajczyk, R. T., Banaschewski, T., Petermann, U., Petermann, F., Kraut, A. A, & Langner, I. (2012). Drug treatment patterns of attention-­‐deficit/hyperactivity disorder in children and 203 adolescents in Germany: results from a large population-­‐based cohort study. Journal of Child and Adolescent Psychopharmacology, 22(6), 452–458. Giedd, J. N., Lalonde, F. M., Celano, M. J., White, S. L., Wallace, G. L., Lee, N. R., & Lenroot, R. K. (2009). Anatomical brain magnetic resonance imaging of typically developing children and adolescents. Journal of the American Academy of Child and Adolescent Psychiatry, 48(5), 465–470. Gilsbach, S., Neufang, S., Scherag, S., Vloet, T. D., Fink, G. R., Herpertz-­‐Dahlmann, B., & Konrad, K. (2012). Effects of the DRD4 genotype on neural networks associated with executive functions in children and adolescents. Developmental Cognitive Neuroscience, 2(4), 417–427. Glasser, M., & van Essen, D. (2011). Mapping human cortical areas in vivo based on myelin content as revealed by T1-­‐ and T2-­‐weighted MRI. Journal of Neuroscience, 31(32), 11597–11616. Goto, Y., Yang, C. R., & Otani, S. (2010). Functional and dysfunctional synaptic plasticity in prefrontal cortex: Roles in psychiatric disorders. Biological Psychiatry, 67(3), 199–207. Granziera, C., Daducci, A., Donati, A., Bonnier, G., Romascano, D., Roche, A., … Krueger, G. (2015). A multi-­‐
contrast MRI study of microstructural brain damage in patients with mild cognitive impairment. NeuroImage Clinical, 8, 631–619. Greenhill, L. L., Findling, R. L., & Swanson, J. M. (2002). A double-­‐blind, placebo-­‐controlled study of modified-­‐release methylphenidate in children with attention-­‐deficit/hyperactivity disorder. Pediatrics, 109(3), E39 Greven, C. U., Bralten, J., Mennes, M., O’Dwyer, L., van Hulzen, K. J. E., Rommelse, N., … Buitelaar, J. K. (2015). Developmentally stable whole-­‐brain volume reductions and developmentally sensitive caudate and putamen volume alterations in those with attention-­‐deficit/hyperactivity disorder and their unaffected siblings. JAMA Psychiatry, 72(5), 490–499. Groenman, A. P., Oosterlaan, J., Rommelse, N. N. J., Franke, B., Greven, C. U., Hoekstra, P. J., … Faraone, S. V. (2013). Stimulant treatment for attention-­‐deficit/hyperactivity disorder and risk of developing substance use disorder. The British Journal of Psychiatry, 203(2), 112–119. Groenman, A., Schweren, L., Weeda, W., Luman, M., Hartman, C., Hoekstra, P., … Oosterlaan, J. (submitted). Predicting substance use disorder and nicotine dependence in ADHD: the crucial role of stimulant medication profiles. Haber, S. N., & Knutson, B. (2010). The reward circuit: linking primate anatomy and human imaging. Neuropsychopharmacology, 35(1), 4–26. Hamilton, L. S., Levitt, J. G., O’Neill, J., Alger, J. R., Luders, E., Phillips, O. R., … Narr, K. L. (2008). Reduced white matter integrity in attention-­‐deficit/hyperactivity disorder. Neuroreport, 19(17), 1705–
1708. Han, X., Jovicich, J., Salat, D., van der Kouwe, A., Quinn, B., Czanner, S., … Fischl, B. (2006). Reliability of MRI-­‐
derived measurements of human cerebral cortical thickness: the effects of field strength, scanner upgrade and manufacturer. NeuroImage, 32(1), 180–194. Harrison, J. N., Cluxton-­‐Keller, F., & Gross, D. (2012). Antipsychotic medication prescribing trends in children and adolescents. Journal of Pediatric Health Care, 26(2), 139–145. Hart, H., Radua, J., Nakao, T., Mataix-­‐Cols, D., & Rubia, K. (2013). Meta-­‐analysis of functional magnetic resonance imaging studies of inhibition and attention in attention-­‐deficit/hyperactivity disorder: exploring task-­‐specific, stimulant medication, and age effects. JAMA Psychiatry, 70(2), 185–198. Hartman, C. A., Luteijn, E., Serra, M., & Minderaa, R. (2006). Refinement of the Children’s Social Behavior Questionnaire (CSBQ): an instrument that describes the diverse problems seen in milder forms of PDD. Journal of Autism and Developmental Disorders, 36(3), 325–342. Hawrylycz, M. J., Lein, E. S., Guillozet-­‐Bongaarts, A. L., … Jones, A. R. (2012). An anatomically comprehensive atlas of the adult human transcriptome. Nature, 489(7416), 391-­‐399. The Allen Human Brain Atlas, available from: http://human.brain-­‐map.org Heal, D. J., Cheetham, S. C., & Smith, S. L. (2009). The neuropharmacology of ADHD drugs in vivo: insights on efficacy and safety. Neuropharmacology, 57(7-­‐8), 608–618. 204 Heinz, A., Goldman, D., Jones, D. W., Palmour, R., Hommer, D., Gorey, J. G., … Weinberger, D. R. (2000). Genotype influences in vivo dopamine transporter availability in human striatum. Neuropsychopharmacology, 22(2), 133–139. Hoekzema, E., Carmona, S., Ramos-­‐Quiroga, J. A., Richarte Fernández, V., Picado, M., Bosch, R., … Vilarroya, O. (2012). Laminar thickness alterations in the fronto-­‐parietal cortical mantle of patients with attention-­‐deficit/hyperactivity disorder. PloS One, 7(12), e48286. Honey, G., & Bullmore, E. (2004). Human pharmacological MRI. Trends in Pharmacological Sciences, 25(7), 366–374. Huang, Y. S., Wang, L. J., & Chen, C. K. (2012). Long-­‐term neurocognitive effects of methylphenidate in patients with attention/deficit-­‐hyperactivity disorder, even at drug-­‐free status. BMC Psychiatry, 12(1), 194. Huang-­‐Pollock, C. L., Karalunas, S. L., Tam, H., & Moore, A. N. (2012). Evaluating vigilance deficits in ADHD: a meta-­‐analysis of CPT performance. Journal of Abnormal Psychology, 121(2), 360–371. IBM. (2011). IBM SPSS Statistics for Windows. Armonk: NY: IBM Corp. Ivanov, I., Bansal, R., Hao, X., Zhu, H., Miller, L., Sanchez-­‐pena, J., … Peterson, B. S. (2010). Morphological abnormalities of the thalamus in youths with attention-­‐deficit/hyperactivity disorder. American Journal of Psychiatry, 167(4), 397–408. Ivanov, I., Liu, X., Clerkin, S., Schulz, K., Fan, J., Friston, K., … Newcorn, J. H. (2014). Methylphenidate and brain activity in a reward/conflict paradigm: Role of the insula in task performance. European Neuropsychopharmacology, 24(6), 897–906. Jay, T. M. (2003). Dopamine: A potential substrate for synaptic plasticity and memory mechanisms. Progress in Neurobiology, 69(6), 375–390. Jensen, P. S., Arnold, L. E., Swanson, J. M., Vitiello, B., Abikoff, H. B., Greenhill, L. L., … Hur, K. (2007). 3-­‐year follow-­‐up of the NIMH MTA study. Journal of the American Academy of Child and Adolescent Psychiatry, 46(8), 989–1002. Jepsen, P., Johnsen, S. P., Gillman, M. W., & Sørensen, H. T. (2004). Interpretation of observational studies. Heart, 90(8), 956–960. Johnston, L., O’Malley, P., Miech, R., Bachman, J., & Schulenberg, J. (2015). Monitoring the Future national survey results on drug use, 1975-­‐2015: Overview, key findings on adolescent drug use. University of Michigan: Ann Arbor: Institute for Social Research. Jolles, D. D., van Buchem, M. A., Crone, E. A., & Rombouts, S. A. (2011). A comprehensive study of whole-­‐
brain functional connectivity in children and young adults. Cerebral Cortex, 21(2), 385–391. Jones, D. K., Knösche, T. R., & Turner, R. (2013). White matter integrity, fiber count, and other fallacies: the do’s and don'ts of diffusion MRI. NeuroImage, 73, 239–254. Kasparbauer, A., Rujescu, D., Riedel, M., Pogarell, O., Costa, A., Meindl, T., … Ettinger, U. (2015). Methylphenidate effects on brain activity as a function of SLC6A3 genotype and striatal dopamine transporter availability. Neuropsychopharmacology, 40(3), 736–745. Kasparek, T., Theiner, P., & Filova, A. (2013). Neurobiology of ADHD from childhood to adulthood: Findings of imaging methods. Journal of Attention Disorders, 20(10), 1–13. Kasper, L. J., Alderson, R. M., & Hudec, K. L. (2012). Moderators of working memory deficits in children with attention-­‐deficit/hyperactivity disorder (ADHD): A meta-­‐analytic review. Clinical Psychology Review, 32(7), 605–617. Kaufman, J., Birmaher, B., Brent, D., Rao, U., Flynn, C., Moreci, P., … Ryan, N. (1997). Schedule for Affective Disorders and schizophrenia for School-­‐age children -­‐ Present and Lifetime version (K-­‐SADS-­‐PL): initial reliability and validity data. Journal of the American Academy of Child and Adolescent Psychiatry, 36(7), 980–988. Kebir, O., & Joober, R. (2011). Neuropsychological endophenotypes in attention-­‐deficit/hyperactivity disorder: A review of genetic association studies. European Archives of Psychiatry and Clinical Neuroscience, 261(8), 583–594. 205 Klein, C., Wendling, K., Huettner, P., Ruder, H., & Peper, M. (2006). Intra-­‐subject variability in attention-­‐
deficit/hyperactivity disorder. Biological Psychiatry, 60(10), 1088–1097. Klomp, A., den Hollander, B., de Bruin, K., Booij, J., & Reneman, L. (2012). The effects of ecstasy (MDMA) on brain serotonin transporters are dependent on age-­‐of-­‐first exposure in recreational users and animals. PLoS One, 7(10), e47524. Knutson, B., Bjork, J. M., Fong, G. W., Hommer, D., Mattay, V. S., & Weinberger, D. R. (2004). Amphetamine modulates human incentive processing. Neuron, 43(2), 261–269. Kobel, M., Bechtel, N., Specht, K., Klarhöfer, M., Weber, P., Scheffler, K., … Penner, I. K. (2010). Structural and functional imaging approaches in attention-­‐deficit/hyperactivity disorder: does the temporal lobe play a key role? Psychiatry Research, 183(3), 230–236. Kobel, M., Bechtel, N., Weber, P., Specht, K., Klarhöfer, M., Scheffler, K., … Penner, I. K. (2009). Effects of methylphenidate on working memory functioning in children with attention-­‐deficit/hyperactivity disorder. European Journal of Paediatric Neurology, 13(6), 516–523. Koester, P., Tittgemeyer, M., Wagner, D., Becker, B., Gouzoulis-­‐Mayfrank, E., & Daumann, J. (2012). Cortical thinning in amphetamine-­‐type stimulant users. Neuroscience, 221, 182–192. Konrad, A., Dielentheis, T., El Masri, D., Bayerl, M., Fehr, C., Gesierich, T., … Winterer, G. (2010). Disturbed structural connectivity is related to inattention and impulsivity in adult attention deficit hyperactivity disorder. European Journal of Neuroscience, 31(5), 912–919. Konrad, K., & Eickhoff, S. B. (2010). Is the ADHD brain wired differently? A review on structural and functional connectivity in attention-­‐deficit/hyperactivity disorder. Human Brain Mapping, 31(6), 904–916. Konrad, K., Neufang, S., Fink, G. R., & Herpertz-­‐Dahlmann, B. (2007). Long-­‐term effects of methylphenidate on neural networks associated with executive attention in children with ADHD: results from a longitudinal functional MRI study. Journal of the American Academy of Child and Adolescent Psychiatry, 46(12), 1633–1641. Kooij, J. S., Boonstra, A. M., Vermeulen, S. H., Heister, A. G., Burger, H., Buitelaar, J. K., & Franke, B. (2008). Response to methylphenidate in adults with ADHD is associated with a polymorphism in SLC6A3 (DAT1). American Journal of Medical Genetics, Part B: Neuropsychiatric Genetics, 147(2), 201–208. Kraemer, H. C., Yesavage, J. A., Taylor, J. L., & Kupfer, D. (2000). How can we learn about developmental processes from cross-­‐sectional studies, or can we? American Journal of Psychiatry, 157(2), 163–
171. Krebs, R. M., Heipertz, D., Schuetze, H., & Duzel, E. (2011). Novelty increases the mesolimbic functional connectivity of the substantia nigra/ventral tegmental area (SN/VTA) during reward anticipation: Evidence from high-­‐resolution fMRI. NeuroImage, 58(2), 647–655. Krogsrud, S., Tamnes, C., Fjell, A., Amlien, I., Grydeland, H., Sulutvedt, U., … Walhovd, K. (2014). Development of hippocampal subfield volumes from 4 to 22 years. Human Brain Mapping, 35(11), 5646–5657. Kuczenski, R., & Segal, D. S. (1997). Effects of methylphenidate on extracellular dopamine, serotonin, and norepinephrine: comparison with amphetamine. Journal of Neurochemistry, 68(5), 2031–2037. Kuriyan, A. B., Pelham, W. E., Molina, B. S. G., Waschbusch, D. A., Sibley, M. H., & Gnagy, E. M. (2014). Concordance between parent and physician medication histories for children and adolescents with attention-­‐deficit/hyperactivity disorder. Journal of Child and Adolescent Psychopharmacology, 24(5), 269–274. Kutcher, S., Aman, M. G., Brooks, S. J., Buitelaar, J. K., van Daalen, E., Fegert, J. M., … Tyano, S. (2004). International consensus statement on attention-­‐deficit/hyperactivity disorder (ADHD) and disruptive behaviour disorders (DBDs): clinical implications and treatment practice suggestions. European Neuropsychopharmacology, 4(1), 11–28. Langevin, L. M., MacMaster, F. P., & Dewey, D. (2014). Distinct patterns of cortical thinning in concurrent motor and attention disorders. Developmental Medicine and Child Neurology, 57(3), 257-­‐264. 206 Langley, K., Fowler, T., Ford, T., Thapar, A. K., van den Bree, M., Harold, G., … Thapar, A. (2010). Adolescent clinical outcomes for young people with attention-­‐deficit hyperactivity disorder. The British Journal of Psychiatry, 196(3), 235–240. Lee, Y. S., Han, D. H., Lee, J. H., & Choi, T. Y. (2010). The effects of methylphenidate on neural substrates associated with interference suppression in children with ADHD: A preliminary study using event related fMRI. Psychiatry Investigation, 7(1), 49–54. Leh, S. E., Ptito, A., Chakravarty, M. M., & Strafella, A. P. (2007). Fronto-­‐striatal connections in the human brain: A probabilistic diffusion tractography study. Neuroscience Letters, 419(2), 113–118. Li, Q., Sun, J., Guo, L., Zang, Y., Feng, Z., Huang, X., … Gong, Q. (2010). Increased fractional anisotropy in white matter of the right frontal region in children with attention-­‐deficit/hyperactivity disorder: a diffusion tensor imaging study. Neuro Endocrinology Letters, 31(6), 747–753. Lichtenstein, P., Halldner, L., Zetterqvist, J., Sjölander, A., Serlachius, E., Fazel, S., … Larsson, H. (2012). Medication for attention-­‐deficit/hyperactivity disorder and criminality. New England Journal of Medicine, 367(21), 2006–2014. Liddle, E. B., Hollis, C., Batty, M. J., Groom, M. J., Totman, J. J., Liotti, M., … Liddle, P. F. (2011). Task-­‐related default mode network modulation and inhibitory control in ADHD: Effects of motivation and methylphenidate. Journal of Child Psychology and Psychiatry, and Allied Disciplines, 52(7), 761–
771. Linton, D., Barr, A. M., Honer, W. G., & Procyshyn, R. M. (2013). Antipsychotic and psychostimulant drug combination therapy in attention-­‐deficit/hyperactivity and disruptive behavior disorders: a systematic review of efficacy and tolerability. Current Psychiatry Reports, 15(5), 355. Loo, S. K., Specter, E., Smolen, A., Hopfer, C., Teale, P. D., & Reite, M. L. (2003). Functional effects of the DAT1 polymorphism on EEG measures in ADHD. Journal of the American Academy of Child and Adolescent Psychiatry, 42(8), 986–993. Ludolph, A. G., Kassubek, J., Schmeck, K., Glaser, C., Wunderlich, A., Buck, A. K., … Mottaghy, F. M. (2008). Dopaminergic dysfunction in attention-­‐deficit/hyperactivity disorder (ADHD), differences between pharmacologically treated and never treated young adults: a 3,4-­‐dihdroxy-­‐6-­‐[18F]fluorophenyl-­‐l-­‐
alanine PET study. NeuroImage, 41(3), 718–727. Luman, M., Oosterlaan, J., & Sergeant, J. A. (2005). The impact of reinforcement contingencies on AD/HD: A review and theoretical appraisal. Clinical Psychology Review, 25(2), 183–213. Lynch, W. J., Nicholson, K. L., Dance, M. E., Morgan, R. W., & Foley, P. L. (2010). Animal models of substance abuse and addiction: Implications for science, animal welfare, and society. Comparative Medicine, 60(3), 177–188. Mackey, S., Stewart, J. L., Connolly, C. G., Tapert, S. F., & Paulus, M. P. (2014). A voxel-­‐based morphometry study of young occasional users of amphetamine-­‐type stimulants and cocaine. Drug and Alcohol Dependence, 135(1), 104–111. Mannuzza, S., Klein, R. G., Truong, N. L., Moulton, J. L., Roizen, E. R., Howell, K. H., & Castellanos, F. X. (2008). Age of methylphenidate treatment initiation in children with ADHD and later substance abuse: Prospective follow-­‐up into adulthood. American Journal of Psychiatry, 165(5), 604–609. Martins, M., Reinke, A., Petronilho, F., Gomes, K., Dal-­‐Pizzol, F., & Quevedo, J. (2006). Methylphenidate treatment induces oxidative stress in young rat brain. Brain Research, 1078(1), 189–197. Mattai, A., Chavez, A., Greenstein, D., Clasen, L., Bakalar, J., Stidd, R., … Gogtay, N. (2010). Effects of clozapine and olanzapine on cortical thickness in childhood-­‐onset schizophrenia. Schizophrenia Research, 116(1), 44–48. McAlonan, G. M., Cheung, V., Chua, S. E., Oosterlaan, J., Hung, S., Tang, C., … Leung, P. W. L. (2009). Age-­‐
related grey matter volume correlates of response inhibition and shifting in attention-­‐
deficit/hyperactivity disorder. The British Journal of Psychiatry, 194(2), 123–129. 207 McCarthy, S., Wilton, L., Murray, M. L., Hodgkins, P., Asherson, P., & Wong, I. C. K. (2012). Persistence of pharmacological treatment into adulthood, in UK primary care, for ADHD patients who started treatment in childhood or adolescence. BMC Psychiatry, 12(1), 219. McGauran, N., Wieseler, B., Kreis, J., Schüler, Y. B., Kölsch, H., & Kaiser, T. (2010). Reporting bias in medical research -­‐ a narrative review. Trials, 11, 37. McGough, J. J., McCracken, J. T., Loo, S. K., Manganiello, M., Leung, M. C., Tietjens, J. R., … Sugar, C. A. (2009). A candidate gene analysis of methylphenidate response in attention-­‐deficit/hyperactivity disorder. Journal of the American Academy of Child Adolescent Psychiatry, 48(12), 1155–1164. McGregor, I. S., Clemens, K. J., van der Plasse, G., Li, K. M., Hunt, G. E., Chen, F., & Lawrence, A. J. (2003). Increased anxiety 3 months after brief exposure to MDMA (‘ecstasy’) in rats: association with altered 5-­‐HT transporter and receptor density. Neuropsychopharmacology, 28, 1472–1484. Meijer, W. M., Faber, A., van den Ban, E., & Tobi, H. (2009). Current issues around the pharmacotherapy of ADHD in children and adults. Pharmacy World and Science, 31(5), 509–516. Meinzer, M. C., Pettit, J. W., & Viswesvaran, C. (2014). The co-­‐occurrence of attention-­‐deficit/hyperactivity disorder and unipolar depression in children and adolescents: A meta-­‐analytic review. Clinical Psychology Review, 34(8), 595–607. Menon, V. (2015). Salience Network. Brain Mapping (Vol. 2). Elsevier Inc. Miller, G., & Chapman, J. (2001). Misunderstanding analysis of covariance. Journal of Abnormal Psychology, 110(1), 40–48. Minzenberg, M. (2012). Pharmacological MRI approaches to understanding mechanisms of drug action. Current Topics in Behavioral Neuroscience, 11, 365-­‐388. Molina, B. S. G., Hinshaw, S. P., Swanson, J. M., Arnold, L. E., Vitiello, B., Jensen, P. S., … Houck, P. R. (2009). The MTA at 8 years: prospective follow-­‐up of children treated for combined-­‐type ADHD in a multisite study. Journal of the American Academy of Child and Adolescent Psychiatry, 48(5), 484–
500. Monuteaux, M. C., Seidman, L. J., Faraone, S. V., Makris, N., Spencer, T., Valera, E., … Biederman, J. (2008). A preliminary study of dopamine D4 receptor genotype and structural brain alterations in adults with ADHD. American Journal of Medical Genetics, Part B: Neuropsychiatric Genetics, 147(8), 1436–1441. Moskvina, V., & Schmidt, K. M. (2008). On multiple-­‐testing correction in genome-­‐wide association studies. Genetic Epidemiology, 32(6), 567–573. MTA Cooperative Group (1999). A 14-­‐month randomized clinical trial of treatment strategies for attention-­‐
deficit/hyperactivity disorder. Archives of General Psychiatry, 56(12), 1073–1086. MTA Cooperative Group (2004). National Institute of Mental Health Multimodal Treatment study of ADHD follow-­‐up: 24-­‐month outcomes of treatment strategies for attention-­‐deficit/hyperactivity disorder. Pediatrics, 113(4), 754–761. Müller, U. C., Asherson, P., Banaschewski, T., Buitelaar, J. K., Ebstein, R. P., Eisenberg, J., … Steinhausen, H. C. (2011). The impact of study design and diagnostic approach in a large multi-­‐centre ADHD study. Part 1: ADHD symptom patterns. BMC Psychiatry, 11, 54. Mulligan, R. C., Kristjansson, S. D., Reiersen, A. M., Parra, A. S., & Anokhin, A. P. (2014). Neural correlates of inhibitory control and functional genetic variation in the dopamine D4 receptor gene. Neuropsychologia, 62, 306–318. Murphy, P., Bruno, R., Ryland, I., Wareing, M., Fisk, J., Montgomery, C., & Hilton, J. (2012). The effects of ‘ecstasy’ (MDMA) on visuospatial memory performance: findings from a systematic review with meta-­‐analyses. Human Psychopharmacology, 27, 113–138. Nakao, T., Radua, J., Rubia, K., & Mataix-­‐Cols, D. (2011). Gray matter volume abnormalities in ADHD: voxel-­‐
based meta-­‐analysis exploring the effects of age and stimulant medication. American Journal of Psychiatry, 168(11), 1154–1163. 208 Narr, K. L., Woods, R. P., Lin, J., Kim, J., Phillips, O. R., Del’Homme, M., … Levitt, J. G. (2009). Widespread cortical thinning is a robust anatomical marker for attention-­‐deficit/hyperactivity disorder. Journal of the American Academy of Child and Adolescent Psychiatry, 48(10), 1014–1022. NHS. (2008). NICE clinical guideline 72. Manchester. Nijmeijer, J. S., Hoekstra, P. J., Minderaa, R. B., Buitelaar, J. K., Altink, M. E., Buschgens, C. J., … A, H. C. (2009). PDD symptoms in ADHD, an independent familial trait? Journal of Abnormal Child Psychology, 37(3), 443–453. Noreika, V., Falter, C. M., & Rubia, K. (2013). Timing deficits in attention-­‐deficit/hyperactivity disorder (ADHD): evidence from neurocognitive and neuroimaging studies. Neuropsychologia, 51(2), 235–
266. O’Dwyer, L., Tanner, C., van Dongen, E. V, Greven, C. U., Bralten, J., Zwiers, M. P., … Buitelaar, J. K. (2014). Brain volumetric correlates of autism spectrum disorder symptoms in attention-­‐
deficit/hyperactivity disorder. PloS One, 9(6), e101130. Obeso, J. A., Rodriguez-­‐Oroz, M. C., Benitez-­‐Temino, B., Blesa, F. J., Guridi, J., Marin, C., & Rodriguez, M. (2008). Functional organization of the basal ganglia: therapeutic implications for Parkinson’s disease. Movement Disorders, 23 suppl., s548–s559. Onaivi, E. S., Bishop-­‐Robinson, C., Motley, E. D., Chakrabarti, A., & Chirwa, S. S. (1996). Neurobiological actions of cocaine in the hippocampus. Annals of the New York Academy of Sciences, 801(615), 76–
94. Onnink, A. M. H., Zwiers, M. P., Hoogman, M., Mostert, J. C., Kan, C. C., Buitelaar, J., & Franke, B. (2014). Brain alterations in adult ADHD: Effects of gender, treatment and comorbid depression. European Neuropsychopharmacology, 24(3), 397-­‐409. Onnink, A. M. H., Franke, B., van Hulzen, K., Zwiers, M. P., Mostert, J. C., Schene, A. H., … Hoogman, M. (2016). Enlarged striatal volume in adults with ADHD carrying the 9-­‐6 haplotype of the dopamine transporter gene DAT1. Journal of Neural Transmission (ePub). Padmanabhan, A., & Luna, B. (2013). Developmental imaging genetics: Linking dopamine function to adolescent behavior. Brain and Cognition, 89, 27–38. Paloyelis, Y., Mehta, M., Faraone, S., Asherson, P., & Kuntsi, J. (2012). Striatal sensitivity during reward processing in attention deficit hyperactivity disorder (ADHD). Journal of the American Academy of Child and Adolescent Psychiatry, 51(7), 997–1003. Panagiotopoulos, C., Ronsley, R., Elbe, D., Davidson, J., & Smith, D. H. (2010). First do no harm: promoting an evidence-­‐based approach to atypical antipsychotic use in children and adolescents. Journal of the Canadian Academy of Child and Adolescent Psychiatry, 19(2), 124–137. Patenaude, B., Smith, S. M., Kennedy, D. N., & Jenkinson, M. (2011). A Bayesian model of shape and appearance for subcortical brain segmentation. NeuroImage, 56(3), 907–922. Pavuluri, M. N., Yang, S., Kamineni, K., Passarotti, A. M., Srinivasan, G., Harral, E. M., … Zhou, X. J. (2009). Diffusion tensor imaging study of white matter fiber tracts in pediatric bipolar disorder and attention-­‐deficit/hyperactivity disorder. Biological Psychiatry, 65(7), 586–593. Peterson, B. S., Potenza, M. N., Wang, Z., Zhu, H., Martin, A., Marsh, R., … Yu, S. (2009). An fMRI study of the effects of psychostimulants on default-­‐mode processing during stroop task performance in youths with ADHD. American Journal of Psychiatry, 166(11), 1286–1294. Peterson, D. J., Ryan, M., Rimrodt, S. L., Cutting, L. E., Denckla, M. B., Kaufmann, W. E., & Mahone, E. M. (2011). Increased regional fractional anisotropy in highly screened attention-­‐deficit/hyperactivity disorder (ADHD). Journal of Child Neurology, 26(10), 1296–1302. Pliśzka, S. R. (2000). Patterns of psychiatric comorbidity with attention-­‐deficit/hyperactivity disorder. Child and Adolescent Clinics of North America, 9(3), 525–540. Pliszka, S. R., Glahn, D. C., Semrud-­‐Clikeman, M., Franklin, C., Perez, R., Xiong, J., & Liotti, M. (2006). Neuroimaging of inhibitory control areas in children with attention-­‐deficit/hyperactivity disorder 209 who were treatment naive or in long-­‐term treatment. American Journal of Psychiatry, 163(6), 1052–1060. Pliszka, S. R., Liotti, M., Bailey, B. Y., Perez, R., Glahn, D., & Semrud-­‐Clikeman, M. (2007). Electrophysiological effects of stimulant treatment on inhibitory control in children with attention-­‐deficit/hyperactivity disorder. Journal of Child and Adolescent Psychopharmacology, 17(3), 356–366. Plomp, E., van Engeland, H., & Durston, S. (2009). Understanding genes, environment and their interaction in attention-­‐deficit/hyperactivity disorder: Is there a role for neuroimaging? Neuroscience, 164(1), 230–240. Polanczyk, G., de Lima, M. S., Horta, B. L., Biederman, J., & Rohde, L. A. (2007). The worldwide prevalence of ADHD: A systematic review and meta-­‐regression analysis. American Journal of Psychiatry, 164(6), 942–948. Posner, J., Maia, T. V., Fair, D., Peterson, B. S., Sonuga-­‐Barke, E. J., & Nagel, B. J. (2011). The attenuation of dysfunctional emotional processing with stimulant medication: An fMRI study of adolescents with ADHD. Psychiatry Research -­‐ Neuroimaging, 193(3), 151–160. Posner, J., Siciliano, F., Wang, Z., Liu, J., Sonuga-­‐Barke, E., & Greenhill, L. (2014). A multimodal MRI study of the hippocampus in medication-­‐naive children with ADHD: What connects ADHD and depression? Psychiatry Research, 224(2), 112-­‐118. Potvin, S., Stavro, K., Rizkallah, É., & Pelletier, J. (2014). Cocaine and cognition: a systematic quantitative review. Journal of Addiction Medicine, 8(5), 368–376. Powers, R. L., Marks, D. J., Miller, C. J., Newcorn, J. H., & Halperin, J. M. (2008). Stimulant treatment in children with attention-­‐deficit/hyperactivity disorder moderates adolescent academic outcome. Journal of Child and Adolescent Psychopharmacology, 18(5), 449–459. Prehn-­‐Kristensen, A., Krauel, K., Hinrichs, H., Fischer, J., Malecki, U., Schuetze, H., … Baving, L. (2011). Methylphenidate does not improve interference control during a working memory task in young patients with attention-­‐deficit/hyperactivity disorder. Brain Research, 1388, 56–68. Proal, E., Reiss, P. T., Klein, R. G., Mannuzza, S., Gotimer, K., Ramos-­‐Olazagasti, M. A., … Castellanos, F. X. (2011). Brain gray matter deficits at 33-­‐year follow-­‐up in adults with attention-­‐
deficit/hyperactivity disorder established in childhood. Archives of General Psychiatry, 68(11), 1122–1134. Pruim, R. H. R., Mennes, M., Buitelaar, J. K., & Beckmann, C. F. (2015). Evaluation of ICA-­‐AROMA and alternative strategies for motion artifact removal in resting state fMRI. NeuroImage, 112, 278–287. Pruim, R. H. R., Mennes, M., van Rooij, D., Llera, A., Buitelaar, J. K., & Beckmann, C. F. (2015). ICA-­‐AROMA: A robust ICA-­‐based strategy for removing motion artifacts from fMRI data. NeuroImage, 112, 267–
277. Rapoport, J. L., & Gogtay, N. (2008). Brain neuroplasticity in healthy, hyperactive and psychotic children: insights from neuroimaging. Neuropsychopharmacology, 33(1), 181–197. Ray, N. J., Brittain, J. S., Holland, P., Joundi, R. A., Stein, J. F., Aziz, T. Z., & Jenkinson, N. (2012). The role of the subthalamic nucleus in response inhibition: Evidence from local field potential recordings in the human subthalamic nucleus. NeuroImage, 60(1), 271–278. Raznahan, A., Shaw, P., Lalonde, F., Stockman, M., Wallace, G. L., Greenstein, D., … Giedd, J. N. (2011). How does your cortex grow? Journal of Neuroscience, 31(19), 7174–7177. Reske, M., Eidt, C., Delis, D., & Paulus, M. (2010). Non-­‐dependent stimulant users of cocaine and prescription amphetamines show verbal learning and memory deficits. Biological Psychiatry, 68(8), 762–769. Richards, J., Arias Vásquez, A., Franke, B., Hoekstra, P., Heslenfeld, D., Oosterlaan, J., … Hartman, C. (2016). Developmentally sensitive interaction effects of genes and the social environment on total and subcortical brain volumes. PloS One (ePub) Roberts, G. M. P., Nestor, L., & Garavan, H. (2009). Learning and memory deficits in ecstasy users and their neural correlates during a face-­‐learning task. Brain Research, 1292, 71–81. 210 Robinson, T. E., & Kolb, B. (2004). Structural plasticity associated with exposure to drugs of abuse. Neuropharmacology, 47(suppl. 1), 33–46. Rohde, L. A., Roman, T., Szobot, C., Cunha, R. D., Hutz, M. H., & Biederman, J. (2003). Dopamine transporter gene, response to methylphenidate and cerebral blood flow in attention-­‐deficit/hyperactivity disorder: a pilot study. Synapse, 48(2), 87–89. Rommelse, N. N., Altink, M. E., Martin, N. C., Buschgens, C. J., Buitelaar, J. K., Sergeant, J., & Oosterlaan, J. (2008). Neuropsychological measures probably facilitate heritability research of ADHD. Archives of Clinical Neuropsychology, 23(5), 579–591. Rubia, K., Alegria, A. a., Cubillo, A. I., Smith, A. B., Brammer, M. J., & Radua, J. (2014). Effects of stimulants on brain function in attention-­‐deficit/hyperactivity disorder: A systematic review and meta-­‐analysis. Biological Psychiatry, 76(8), 616–628. Rubia, K., Halari, R., Cubillo, A., Smith, A. B., Mohammad, A. M., Brammer, M., & Taylor, E. (2011a). Methylphenidate normalizes fronto-­‐striatal underactivation during interference inhibition in medication-­‐nave boys with attention-­‐deficit/hyperactivity disorder. Neuropsychopharmacology, 36(8), 1575–1586. Rubia, K., Halari, R., Cubillo, A., Mohammad, A. M., Brammer, M., & Taylor, E. (2009). Methylphenidate normalises activation and functional connectivity deficits in attention and motivation networks in medication-­‐naïve children with ADHD during a rewarded continuous performance task. Neuropharmacology, 57(7-­‐8), 640–652. Rubia, K., Halari, R., Mohammad, A.-­‐M., Taylor, E., & Brammer, M. (2011b). Methylphenidate normalizes frontocingulate underactivation during error processing in attention-­‐deficit/hyperactivity disorder. Biological Psychiatry, 70(3), 255–262. Rubinov, M., & Sporns, O. (2011). Weight-­‐conserving characterization of complex functional brain networks. NeuroImage, 56(4), 2068–2079. Russell, V., Villiers, A. De, Sagvolden, T., Lamm, M., & Taljaard, J. (1998). Differences between electrically-­‐ , ritalin-­‐ and D -­‐amphetamine-­‐stimulated release of [3H]dopamine from brain slices suggest impaired vesicular storage of dopamine in an animal model of attention-­‐deficit/hyperactivity disorder. Behavioral Brain Research, 94(1), 163-­‐171. SAMHSA Substance Abuse and Mental Health Services Administration (2014). Results from the 2013 national survey on drug use and health: Summary of national findings. Rockville, MD: Substance Abuse and Mental Health Service Administration, NSDUH Series H-­‐48. Schaid, D. J., Rowland, C. M., Tines, D. E., Jacobson, R. M., & Poland, G. A. (2002). Score tests for association between traits and haplotypes when linkage phase is ambiguous. American Journal of Human Genetics, 70(2), 425–434. Schlochtermeier, L., Stoy, M., Schlagenhauf, F., Wrase, J., Park, S. Q., Friedel, E., … Ströhle, A. (2011). Childhood methylphenidate treatment of ADHD and response to affective stimuli. European Neuropsychopharmacology, 21(8), 646–654. Schoenfelder, E. N., Faraone, S. V, & Kollins, S. H. (2014). Stimulant treatment of ADHD and cigarette smoking: A meta-­‐analysis. Pediatrics, 133(6), 1070–1080. Schweren, L. J. S., de Zeeuw, P., & Durston, S. (2013). MR imaging of the effects of methylphenidate on brain structure and function in attention-­‐deficit/hyperactivity disorder. European Neuropsychopharmacology, 23(10), 1151–1164. Schweren, L. J. S., Hartman, C. A., Heslenfeld, D. J., van der Meer, D., Franke, B., Oosterlaan, J., … Hoekstra, P. J. (2015a). Thinner medial temporal cortex in adolescents with attention-­‐deficit/hyperactivity disorder and the effects of stimulants. Journal of the American Academy of Child and Adolescent Psychiatry, 54(8), 660–667. Schweren, L. J. S., Hartman, C. A., Zwiers, M. P., Heslenfeld, D. J., van der Meer, D., Franke, B., … Hoekstra, P. J. (2015b). Combined stimulant and antipsychotic treatment in adolescents with attention-­‐
211 deficit/hyperactivity disorder: a cross-­‐sectional observational structural MRI study. European Child and Adolescent Psychiatry, 24(8), 959–968. Scott, J. C., Woods, S. P., Matt, G. E., Meyer, R. A., Heaton, R. K., Atkinson, J. H., & Grant, I. (2007). Neurocognitive effects of methamphetamine: A critical review and meta-­‐analysis. Neuropsychology Review, 17(3), 275–297. Seidman, L. J., Valera, E. M., & Makris, N. (2005). Structural brain imaging of attention-­‐deficit/hyperactivity disorder. Biological Psychiatry, 57(11), 1263–1272. Semrud-­‐Clikeman, M., Pliszka, S. R., Bledsoe, J., & Lancaster, J. (2014). Volumetric MRI differences in treatment naive and chronically treated adolescents with ADHD-­‐combined type. Journal of Attention Disorders, 18(6), 511-­‐520 Semrud-­‐Clikeman, M., Pliśzka, S. R., Lancaster, J., & Liotti, M. (2006). Volumetric MRI differences in treatment-­‐naïve vs. chronically treated children with ADHD. Neurology, 67(6), 1023–1027. Shaffer, D., Gould, M. S., Brasic, J., Ambrosini, P., Fisher, P., Bird, H., & Aluwahlia, S. (1983). A Children’s Global Assessment Scale (CGAS). Archives of General Psychiatry, 40(11), 1228-­‐12231. Shafritz, K. M., Marchione, K. E., Gore, J. C., Shaywitz, S. E., & Shaywitz, B. A. (2004). The effects of methylphenidate on neural systems of attention in attention-­‐deficit/hyperactivity disorder. American Journal of Psychiatry, 161(11), 1990–1997. Shang, C. Y., Wu, Y. H., Gau, S. S., & Tseng, W. Y. (2013). Disturbed microstructural integrity of the frontostriatal fiber pathways and executive dysfunction in children with attention-­‐
deficit/hyperactivity disorder. Psychological Medicine, 43(5), 1093–1107. Shaw, P., de Rossi, P., Watson, B., Wharton, A., Greenstein, D., Raznahan, A., … Chakravarty, M. M. (2014). Mapping the development of the basal ganglia in children with attention-­‐deficit/hyperactivity disorder. Journal of the American Academy of Child and Adolescent Psychiatry, 53(7), 780–789. Shaw, P., Eckstrand, K., Sharp, W., Blumenthal, J., Lerch, J. P., Greenstein, D., … Rapoport, J. L. (2007a). Attention-­‐deficit/hyperactivity disorder is characterized by a delay in cortical maturation. Proceedings of the National Academy of Sciences of the United States of America, 104(49), 1–6. Shaw, P., Gilliam, M., Liverpool, M., Weddle, C., Malek, M., Sharp, W., … Giedd, J. (2011). Cortical development in typically developing children with symptoms of hyperactivity and impulsivity: Support for a dimensional view of attention-­‐deficit/hyperactivity disorder. American Journal of Psychiatry, 168(2), 143–151. Shaw, P., Gornick, M., Lerch, J., Addington, A., Seal, J., Greenstein, D., … Rapoport, J. L. (2007b). Polymorphisms of the dopamine D4 receptor, clinical outcome, and cortical structure in attention-­‐
deficit/hyperactivity disorder. Archives of General Psychiatry, 64(8), 921–931. Shaw, P., Lerch, J., Greenstein, D., Sharp, W., Clasen, L., Evans, A., … Rapoport, J. (2006). Longitudinal mapping of cortical thickness and clinical outcome in children and adolescents with attention-­‐
deficit/hyperactivity disorder. Archives of General Psychiatry, 63(5), 540–549. Shaw, P., Malek, M., Watson, B., Greenstein, D., de Rossi, P., & Sharp, W. (2013). Trajectories of cerebral cortical development in childhood and adolescence and adult attention-­‐deficit/hyperactivity disorder. Biological Psychiatry, 74(8), 599–606. Shaw, P., Malek, M., Watson, B., Sharp, W., Evans, A., & Greenstein, D. (2012). Development of cortical surface area and gyrification in attention-­‐deficit/hyperactivity disorder. Biological Psychiatry, 72(3), 191–197. Shaw, P., Sharp, W. S., Morrison, M., Eckstrand, K., Greenstein, D. K., Clasen, L. S., … Rapoport, J. L. (2009). Psychostimulant treatment and the developing cortex in attention-­‐deficit/hyperactivity disorder. American Journal of Psychiatry, 166(1), 58–63. Sheridan, M. A., Hinshaw, S. P., & D’Esposito, M. (2010). The effect of stimulant medication on prefrontal connectivity during working memory in attention-­‐deficit/hyperactivity disorder. Journal of Attention Disorders, 14(1), 69–78. 212 Shima, K., & Tanji, J. (1998). Role for cingulate motor area cells in voluntary movement selection based on reward. Science, 282(5392), 1335–1338. Shook, D., Brady, C., Lee, P. S., Kenealy, L., Murphy, E. R., Gaillard, W. D., … Vaidya, C. J. (2011). Effect of dopamine transporter genotype on caudate volume in childhood ADHD and controls. American Journal of Medical Genetics, Part B: Neuropsychiatric Genetics, 156(1), 28–35. Šidàk, Z.. (1967). Retangular confidence regions for the means of multivariate normal distributions. Journal of the American Statistical Association, 62(318), 626–633. Silk, T. J., Vance, A., Rinehart, N., Bradshaw, J. L., & Cunnington, R. (2009a). Structural development of the basal ganglia in attention/deficit-­‐hyperactivity disorder: a diffusion tensor imaging study. Psychiatry Research, 172(3), 220–225. Silk, T. J., Vance, A., Rinehart, N., Bradshaw, J. L., & Cunnington, R. (2009b). White-­‐matter abnormalities in attention-­‐deficit/hyperactivity disorder: a diffusion tensor imaging study. Human Brain Mapping, 30(9), 2757–2765. Silverman, S. L. (2009). From randomized controlled trials to observational studies. The American Journal of Medicine, 122(2), 114–120. Simon, V., Czobor, P., Balint, S., Meszaros, A., & Bitter, I. (2009). Prevalence and correlates of adult attention-­‐
deficit/hyperactivity disorder: meta-­‐analysis. The British Journal of Psychiatry, 194(3), 204–211. Smith, S. M., Jenkinson, M., Woolrich, M. W., Beckmann, C. F., Behrens, T. E. J., Johansen-­‐Berg, H., … Matthews, P. M. (2004). Advances in functional and structural MR image analysis and implementation as FSL. NeuroImage, 23(suppl. 1), 208–219. Sobel, L. J., Bansal, R., Maia, T. V, Sanchez, J., Mazzone, L., Durkin, K., … Peterson, B. S. (2010). Basal ganglia surface morphology and the effects of stimulant medications in youth with attention-­‐
deficit/hyperactivity disorder. American Journal of Psychiatry, 167(8), 977–986. Sonuga-­‐Barke, E., Bitsakou, P., & Thompson, M. (2010). Beyond the dual pathway model: Evidence for the dissociation of timing, inhibitory, and delay-­‐related impairments in attention-­‐deficit/hyperactivity disorder. Journal of the American Academy of Child and Adolescent Psychiatry, 49(4), 345–355. Sowell, E., Delis, D., Stiles, J., & Jernigan, T. (2001). Improved memory functioning and frontal lobe maturation between childhood and adolescence: A structural MRI study. Journal of the International Neuropsychological Society, 7(3), 312–322. Spencer, T., Biederman, J., Wilens, T., Harding, M., O’Donnell, D., & Griffin, S. (1996). Pharmacotherapy of attention-­‐deficit hyperactivity disorder across the life cycle. Journal of the American Academy of Child and Adolescent Psychiatry, 35(4), 409–432. Spencer, T. J., Brown, A., Seidman, L. J., Valera, E. M., Makris, N., Lomedico, A., … Biederman, J. (2013). Effect of psychostimulants on brain structure and function in ADHD: a qualitative literature review of magnetic resonance imaging-­‐based neuroimaging studies. The Journal of Clinical Psychiatry, 74(9), 902–917. Squeglia, L. M., Tapert, S. F., Sullivan, E. V., Jacobus, J., Meloy, M. J., Rohlfing, T., & Pfefferbaum, A. (2015). Brain development in heavy-­‐drinking adolescents. The American Journal of Psychiatry, 172(6), 531–542. Storebø, O. J., Simonsen, E., & Gluud, C. (2016). Methylphenidate for attention-­‐deficit/hyperactivity disorder in children and ddolescents. JAMA, 315(18), 2009-­‐2010. Stoy, M., Schlagenhauf, F., Schlochtermeier, L., Wrase, J., Knutson, B., Lehmkuhl, U., … Ströhle, A. (2011). Reward processing in male adults with childhood ADHD -­‐ a comparison between drug-­‐naïve and methylphenidate-­‐treated subjects. Psychopharmacology, 215(3), 467–481. Suskauer, S. J., Simmonds, D. J., Fotedar, S., Blankner, J. G., Pekar, J. J., Denckla, M. B., & Mostofsky, S. H. (2008). Functional magnetic resonance imaging evidence for abnormalities in response selection in attention-­‐deficit/hyperactivity disorder: differences in activation associated with response inhibition but not habitual motor response. Journal of Cognitive Neuroscience, 20(3), 478-­‐493. 213 Swanson, J., Baler, R. D., & Volkow, N. D. (2011). Understanding the effects of stimulant medications on cognition in individuals with attention-­‐deficit/hyperactivity disorder: a decade of progress. Neuropsychopharmacology, 36(1), 207–226. Swanson, J. M., Hinshaw, S. P., Arnold, L. E., Gibbons, R. D., Marcus, S., Hur, K., … Wigal, T. (2007a). Secondary evaluations of MTA 36-­‐month outcomes: propensity score and growth mixture model analyses. Journal of the American Academy of Child and Adolescent Psychiatry, 46(8), 1003–1014. Swanson, J. M., Kinsbourne, M., Nigg, J., Lanphear, B., Stefanatos, G. A., Volkow, N., … Wadhwa, P. D. (2007b). Etiologic subtypes of attention-­‐deficit/hyperactivity disorder: brain imaging, molecular genetic and environmental factors and the dopamine hypothesis. Neuropsychology Review, 17(1), 39–59. Swanson, J. M., Kraemer, H. C., Hinshaw, S. P., Arnold, L. E., Conners, C. K., Abikoff, H. B., … Wu, M. (2001). Clinical relevance of the primary findings of the MTA: Success rates based on severity of ADHD and ODD symptoms at the end of treatment. Journal of the American Academy of Child and Adolescent Psychiatry, 40(2), 168–179. Swanson, J., McBurnett, K., Wigal, T., Pfiffner, L., Lerner, M., Williams, L., … Fisher, T. (1993). Effect of stimulant medication on children with attention-­‐deficit disorder: a ‘review of reviews’. Exceptional Children, 154–162. Tamnes, C. K., Østby, Y., Fjell, A. M., Westlye, L. T., Due-­‐Tønnessen, P., & Walhovd, K. B. (2010). Brain maturation in adolescence and young adulthood: Regional age-­‐related changes in cortical thickness and white matter volume and microstructure. Cerebral Cortex, 20(3), 534–548. Tamnes, C. K., Walhovd, K. B., Engvig, A., Grydeland, H., Krogsrud, S. K., Østby, Y., … Fjell, A. M. (2014). Regional hippocampal volumes and development predict learning and memory. Developmental Neuroscience, 36(3-­‐4), 161–174. Teicher, M. H., Anderson, C. M., Polcari, A., Glod, C. A., Maas, L. C., & Renshaw, P. F. (2000). Functional deficits in basal ganglia of children with attention-­‐deficit/hyperactivity disorder shown with functional magnetic resonance imaging relaxometry. Nature Medicine, 6(4), 470–473. ter Huurne, N., Fallon, S. J., van Schouwenburg, M., van der Schaaf, M., Buitelaar, J., Jensen, O., & Cools, R. (2015). Methylphenidate alters selective attention by amplifying salience. Psychopharmacology, 232(23), 4317-­‐43123. Thapar, A., & Cooper, M. (2016). Attention-­‐deficit/hyperactivity disorder. Lancet, 387, 1240–1250. Thissen, A. J. A. M., Bralten, J., Rommelse, N. N. J., Arias-­‐Vasquez, A., Greven, C. U., Heslenfeld, D., … Buitelaar, J. K. (2015). The role of age in association analyses of ADHD and related neurocognitive functioning: A proof of concept for dopaminergic and serotonergic genes. American Journal of Medical Genetics Part B: Neuropsychiatric Genetics, 168(6), 471–479. Thomason, M. E., & Thompson, P. M. (2011). Diffusion imaging, white matter, and psychopathology. Annual Review of Clinical Psychology, 7, 63–85. Tomasi, D., Volkow, N., Wang, G., Wang, R., Telang, F., Caparelli, E., … Fowler, J. (2011). Methylphenidate enhances brain activation and deactivation responses to visual attention and working memory tasks in healthy controls. NeuroImage, 54(4), 3101–3110. Trip, A. M., Visser, S. T., Kalverdijk, L. J., & de Jong-­‐van den Berg, L. T. W. (2009). Large increase of the use of psychostimulants among youth in the Netherlands between 1996 and 2006. British Journal of Clinical Pharmacology, 67(4), 466–468. Tsai, C. S., Huang, Y. S., Wu, C. L., Hwang, F. M., Young, K. B., Tsai, M. H., & Chu, S. M. (2013). Long-­‐term effects of stimulants on neurocognitive performance of Taiwanese children with attention-­‐
deficit/hyperactivity disorder. BMC Psychiatry, 13, 330. Tseng, K. Y., & O’Donnell, P. (2007). Dopamine modulation of prefrontal cortical interneurons changes during adolescence. Cerebral Cortex, 17(5), 1235–1240. Uda, S., Matsui, M., Tanaka, C., Uematsu, A., Miura, K., Kawana, I., & Noguchi, K. (2015). Normal development of human brain white matter from infancy to early adulthood: A diffusion tensor imaging study. Developmental Neuroscience, 37(2), 182–194. 214 Ullsperger, M., & von Cramon, D. Y. (2003). Error monitoring using external feedback: specific roles of the habenular complex, the reward system, and the cingulate motor area revealed by functional magnetic resonance imaging. The Journal of Neuroscience, 23(10), 4308–4314. Vaidya, C. J., Austin, G., Kirkorian, G., Ridlehuber, H. W., Desmond, J. E., Glover, G. H., & Gabrieli, J. D. E. (1998). Selective effects of methylphenidate in attention-­‐deficit/hyperactivity disorder: a functional magnetic resonance study. Proceedings of the National Academy of Sciences of the United States of America, 95(24), 14494–14499. Valera, E. M., Brown, A., Biederman, J., Faraone, S. V, Makris, N., Monuteaux, M. C., … Seidman, L. J. (2010). Sex differences in the functional neuroanatomy of working memory in adults with ADHD. American Journal of Psychiatry, 167(1), 86–94. Valera, E. M., Faraone, S. V, Murray, K. E., & Seidman, L. J. (2007). Meta-­‐analysis of structural imaging findings in attention-­‐deficit/hyperactivity disorder. Biological Psychiatry, 61(12), 1361–1369. van de Loo-­‐Neus, G. H. H., Rommelse, N., & Buitelaar, J. K. (2011). To stop or not to stop? How long should medication treatment of attention-­‐deficit/hyperactivity disorder be extended? European Neuropsychopharmacology, 21(8), 584–599. van den Ban, E. F., Souverein, P. C., van Engeland, H., Swaab, H., Egberts, T. C. G., & Heerdink, E. R. (2015). Differences in ADHD medication usage patterns in children and adolescents from different cultural backgrounds in the Netherlands. Social Psychiatry and Psychiatric Epidemiology, 50(7), 1153–
1162. van den Ban, E., Souverein, P. C., Swaab, H., van Engeland, H., Egberts, T. C. G., & Heerdink, E. R. (2010). Less discontinuation of ADHD drug use since the availability of long-­‐acting ADHD medication in children, adolescents and adults under the age of 45 years in the Netherlands. Attention Deficit and Hyperactivity Disorders, 2(4), 213–220. van der Marel, K., Bouet, V., Meerhoff, G. F., Freret, T., Boulouard, M., Dauphin, F., … Reneman, L. (2015). Effects of long-­‐term methylphenidate treatment in adolescent and adult rats on hippocampal shape, functional connectivity and adult neurogenesis. Neuroscience, 309, 243-­‐258. van der Marel, K., Klomp, A., Meerhoff, G. F., Schipper, P., Lucassen, P. J., Homberg, J. R., … Reneman, L. (2014). Long-­‐term oral methylphenidate treatment in adolescent and adult rats: differential effects on brain morphology and function. Neuropsychopharmacology, 39(2), 263–273. van der Meer, D., Hoekstra, P. J., Zwiers, M., Mennes, M., Schweren, L. J. S., Franke, B., … Hartman, C. A. (2015). Brain correlates of the interaction between 5-­‐HTTLPR and psychosocial stress mediating attention-­‐deficit/hyperactivity disorder severity. American Journal of Psychiatry, 172(8), 768-­‐775. van Ewijk, H., Groenman, A. P., Zwiers, M. P., Heslenfeld, D. J., Faraone, S. V, Hartman, C. A., … Oosterlaan, J. (2015). Smoking and the developing brain: altered white matter microstructure in attention-­‐
deficit/hyperactivity disorder and healthy controls. Human Brain Mapping, 36(3), 1180–1189. van Ewijk, H., Heslenfeld, D. J., Luman, M., Rommelse, N. N., Hartman, C. A., Hoekstra, P., … Oosterlaan, J. (2014a). Visuospatial working memory in ADHD patients, unaffected siblings, and healthy controls. Journal of Attention Disorders, 18(4), 369–378. van Ewijk, H., Heslenfeld, D. J., Zwiers, M. P., Buitelaar, J. K., & Oosterlaan, J. (2012). Diffusion tensor imaging in attention-­‐deficit/hyperactivity disorder: a systematic review and meta-­‐analysis. Neuroscience and Biobehavioral Reviews, 36(4), 1093–1106. van Ewijk, H., Heslenfeld, D. J., Zwiers, M. P., Faraone, S. V., Luman, M., Hartman, C. A., … Oosterlaan, J. (2014b). Different mechanisms of white matter abnormalities in attention-­‐deficit/hyperactivity disorder: A diffusion tensor imaging study. Journal of the American Academy of Child and Adolescent Psychiatry, 53(7), 790–799. van Hulst, B. M., de Zeeuw, P., & Durston, S. (2014). Distinct neuropsychological profiles within ADHD: a latent class analysis of cognitive control, reward sensitivity and timing. Psychological Medicine, 45(4), 735-­‐745. 215 van Lieshout, M., Luman, M., Schweren, L. J. S., Twisk, J., Faraone, S., Heslenfeld, D., … Oosterlaan, J. (in preparation). Neurocognitive development as predictor for behavioral outcome in children with ADHD and their unaffected siblings: are we looking at a dead end? van Lieshout, M., Luman, M., Twisk, J. W. R., van Ewijk, H., Groenman, A. P., Thissen, A. J. A. M., … Oosterlaan, J. (2016). A 6-­‐year follow-­‐up of a large European cohort of children with attention-­‐
deficit/hyperactivity disorder-­‐combined subtype: outcomes in late adolescence and young adulthood. European Child and Adolescent Psychiatry (ePub) van Nieuwenhuijzen, P. S., Long, L. E., Hunt, G. E., Arnold, J. C., & McGregor, I. S. (2010). Residual social, memory and oxytocin-­‐related changes in rats following repeated exposure to gamma-­‐
hydroxybutyrate (GHB), 3,4-­‐methylenedioxymethamphetamine (MDMA) or their combination. Psychopharmacology, 212(4), 663–674. van Rooij, D., Hartman, C. A., Mennes, M., Oosterlaan, J., Franke, B., Rommelse, N., … Hoekstra, P. J. (2015a). Altered neural connectivity during response inhibition in adolescents with attention-­‐
deficit/hyperactivity disorder and their unaffected siblings. NeuroImage Clinical, 7, 325–35. van Rooij, D., Hoekstra, P.J., Mennes, M., von Rhein, D., Thissen, A.J., Heslenfeld, D., … Hartman, C.A. (2015b). Distinguishing adolescents with ADHD from their unaffected siblings and healthy comparison subjects by neural activation patterns during response inhibition. American Journal of Psychiatry, 172(7), 674-­‐683. van Widenfelt, B. M., Goedhart, A. W., Treffers, P. D. A., & Goodman, R. (2003). Dutch version of the Strengths and Difficulties Questionnaire (SDQ). European Child and Adolescent Psychiatry, 12(6), 281–289. Vassena, E., Cobbaert, S., Andres, M., Fias, W., & Verguts, T. (2015). Unsigned value prediction-­‐error modulates the motor system in absence of choice. NeuroImage, 122, 73–79. Verster, J. C., Bekker, E. M., Kooij, J. J. S., Buitelaar, J. K., Verbaten, M. N., Volkerts, E. R., & Olivier, B. (2010). Methylphenidate significantly improves declarative memory functioning of adults with ADHD. Psychopharmacology, 212(2), 277–281. Villemonteix, T., de Brito, S. A., Kavec, M., Balériaux, D., Metens, T., Slama, H., … Massat, I. (2015). Grey matter volumes in treatment naïve vs. chronically treated children with attention-­‐
deficit/hyperactivity disorder: a combined approach. European Neuropsychopharmacology, 25(8), 1118-­‐1127. Voineskos, A., Winterburn, J., Felsky, D., Pipitone, J., Rajji, T., Mulsant, B., & Chakravarty, M. (2015). Hippocampal (subfield) volume and shape in relation to cognitive performance across the adult lifespan. Human Brain Mapping, 36(8), 3020–3037. Volkow, N. D., Wang, G., Fowler, J. S., Logan, J., Gerasimov, M., Maynard, L., … Franceschi, D. (2001). Therapeutic doses of oral methylphenidate significantly increase extracellular dopamine in the human brain. Journal of Neuroscience, 21(2), RC121. Volkow, N. D., Wang, G. J., Fowler, J. S., Logan, J., Angrist, B., Hitzemann, R., … Pappas, N. (1997). Effects of methylphenidate on regional brain glucose metabolism in humans: relationship to dopamine D2 receptors. American Journal of Psychiatry, 154(1), 50–55. Volkow, N. D., Wang, G. J., Fowler, J. S., Telang, F., Maynard, L., Logan, J., … Swanson, J. M. (2004). Evidence that methylphenidate enhances the saliency of a mathematical task by increasing dopamine in the human brain. American Journal of Psychiatry, 161(7), 1173–1180. Volkow, N. D., Wang, G. J., Newcorn, J. H., Kollins, S. H., Wigal, T. L., Telang, F., … Swanson, J. M. (2011). Motivation deficit in ADHD is associated with dysfunction of the dopamine reward pathway. Molecular Psychiatry, 16(11), 1147–1154. Volkow, N. D., Wang, G. J., Tomasi, D., Kollins, S. H., Wigal, T. L., Newcorn, J. H., … Swanson, J. M. (2012). Methylphenidate-­‐elicited dopamine increases in ventral striatum are associated with long-­‐term symptom improvement in adults with attention-­‐deficit/hyperactivity disorder. The Journal of Neuroscience, 32(3), 841–849. 216 von Rhein, D., Mennes, M., van Ewijk, H., Groenman, A. P., Zwiers, M. P., Oosterlaan, J., … Buitelaar, J. (2015a). The NeuroIMAGE study: a prospective phenotypic, cognitive, genetic and MRI study in children with attention-­‐deficit/hyperactivity disorder. Design and descriptives. European Child and Adolescent Psychiatry, 24(3), 265-­‐281. von Rhein, D., Cools, R., Zwiers, M. P., van der Schaaf, M., Franke, B., Luman, M., … Buitelaar, J. (2015b). Increased neural responses to reward in adolescents and young adults with attention-­‐
deficit/hyperactivity disorder and their unaffected siblings. Journal of the American Academy of Child and Adolescent Psychiatry, 54(5), 394–402. Wang, G. J., Volkow, N. D., Wigal, T., Kollins, S. H., Newcorn, J. H., Telang, F., … Swanson, J. M. (2013). Long-­‐
term stimulant treatment affects brain dopamine transporter level in patients with attention-­‐
deficit/hyperactive disorder. PLoS One, 8(5), e63023. Wechsler (2000). WAIS-­‐III Nederlandstalige bewerking. Technische handleiding. London, UK: The Psychological Corporation. Wechsler (2002). WISC-­‐III Handleiding. London, UK: The Psychological Corporation. Wierenga, L., Langen, M., Ambrosino, S., van Dijk, S., Oranje, B., & Durston, S. (2014). Typical development of basal ganglia, hippocampus, amygdala and cerebellum from age 7 to 24. NeuroImage, 96, 67–72. Wilens, T., McBurnett, K., Bukstein, O., McGough, J., Greenhill, L., Lerner, M., … Lynch, J. (2006). Multisite controlled study of OROS methylphenidate in the treatment of adolescents with attention-­‐
deficit/hyperactivity disorder. Archives of Pediatric and Adolescent Medicine, 160, 82–90. Willcutt, E. G., Doyle, A. E., Nigg, J. T., Faraone, S. V, & Pennington, B. F. (2005). Validity of the executive function theory of attention-­‐deficit/hyperactivity disorder: a meta-­‐analytic review. Biological Psychiatry, 57(11), 1336–1346. Winkler, A. M., Ridgway, G. R., Webster, M. A., Smith, S. M., & Nichols, T. E. (2014). Permutation inference for the general linear model. NeuroImage, 92, 381–397. Winstanley, C. A., Eagle, D. M., & Robbins, T. W. (2006). Behavioral models of impulsivity in relation to ADHD: Translation between clinical and preclinical studies. Clinical Psychology Review, 26(4), 379–395. Wong, C. G., & Stevens, M. C. (2012). The effects of stimulant medication on working memory functional connectivity in attention-­‐deficit/hyperactivity disorder. Biological Psychiatry, 71(5), 458–466. Wong, I. C. K., Asherson, P., Bilbow, A., Clifford, S., Coghill, D., Desoysa, R., … Taylor, E. (2009). Cessation of attention-­‐deficit/hyperactivity disorder drugs in the young (CADDY) – a pharmaco-­‐epidemiological and qualitative study. Health Technology Assessment, 13(50), 1-­‐120 Wood, A., & Neale, M. (2010). Twin studies and their implications for molecular genetic studies: Endophenotypes integrate quantitative and molecular genetics in ADHD research. Journal of the American Academy of Child and Adolescent Psychiatry, 49(9), 1–14. Wood, S. N. (2011). Fast stable restricted maximum likelihood and marginal likelihood estimation of semiparametric generalized linear models. Journal of the Royal Statistical Society. Series B: Statistical Methodology, 73(1), 3–36. Woods, S. P., Delis, D. C., Scott, J. C., Kramer, J. H., & Holdnack, J. A. (2006). The California Verbal Learning Test -­‐ second edition: Test-­‐retest reliability, practice effects, and reliable change indices for the standard and alternate forms. Archives of Clinical Neuropsychology, 21(5), 413–420. Wright, N. E., Strong, J. A., Gilbart, E. R., Shollenbarger, S. G., & Lisdahl, K. M. (2015). 5-­‐HTTLPR genotype moderates the effects of past ecstasy use on verbal memory performance in adolescent and emerging adults: A pilot study. PLoS One, 10(7), e0134708. Wu, Y. H., Gau, S. S. F., Lo, Y. C., & Tseng, W. Y. I. (2014). White matter tract integrity of frontostriatal circuit in attention-­‐deficit/hyperactivity disorder: association with attention performance and symptoms. Human Brain Mapping, 35(1), 199–212. 217 Xia, S., Li, X., Kimball, A. E., Kelly, M. S., Lesser, I., & Branch, C. (2012). Thalamic shape and connectivity abnormalities in children with attention-­‐deficit/hyperactivity disorder. Psychiatry Research, 204(2-­‐3), 161–167. Yang, X. R., Carrey, N., Bernier, D., & Macmaster, F. P. (2015). Cortical thickness in young treatment-­‐naive children with ADHD. Journal of Attention Disorders, 19(11), 925-­‐930. Yang, Z., Kelly, C., Castellanos, F. X., Leon, T., Milham, M. P., & Adler, L. a. (2016). Neural correlates of symptom improvement following stimulant treatment in adults with attention-­‐
deficit/hyperactivity disorder. Journal of Child and Adolescent Psychopharmacology (ePub) Yanofski, J. (2010). The dopamine dilemma: using stimulants and antipsychotics concurrently. Psychiatry (Edgemont), 7(6), 18–23. Zang, Y. F., Jin, Z., Weng, X. C., Zhang, L., Zeng, Y. W., Yang, L., … Faraone, S. V. (2005). Functional MRI in attention-­‐deficit/hyperactivity disorder: evidence for hypofrontality. Brain and Development, 27(8), 544–550. Zimmer, L. (2009). Positron emission tomography neuroimaging for a better understanding of the biology of ADHD. Neuropharmacology, 57(7-­‐8), 601–607. Zwiers, M. P. (2010). Patching cardiac and head motion artefacts in diffusion-­‐weighted images. NeuroImage, 53(2), 565–575. 218 219 220 NEDERLANDSTALIGE SAMENVATTING
221 222 In het voorjaar van 2016 zendt KRO-­‐NCRV de Monitor haar tweede uitzending uit in een programmareeks genaamd ‘de ADHD epidemie’. Centraal staat de vraag of de toename in het gebruik van ADHD medicatie medisch gerechtvaardigd is. Na de eerste aankondiging van deze serie ontvangt de redactie naar eigen zeggen honderden berichten van bezorgde ouders, leraren en psychiaters. De onrust over ADHD medicatie wordt in de media aangewakkerd door suggestieve berichtgeving over de overeenkomsten met harddrugs als cocaïne en ecstasy. Terecht vraagt men zich af: willen we onze kinderen blootstellen aan een middel waarvan we het lange-­‐
termijn effect op de hersenontwikkeling niet kennen? In haar antwoord op Kamervragen naar aanleiding van de ADHD epidemie refereert de Minister van Gezondheid, Welzijn en Sport aan Nederlands onderzoek naar de lange-­‐termijn effecten van ADHD medicatie op het zich ontwikkelende brein. In dit proefschrift presenteer ik de resultaten van dat onderzoek. ADHD EN HET BREIN
ADHD (attention-­‐deficit/hyperactivity disorder) is een veelvoorkomende ontwikkelingsstoornis die wordt gekenmerkt door aandachtsproblemen en/of impulsiviteit en hyperactiviteit. Op neuropsychologische taken, met name taken die een beroep doen op zelfregulatie, planning, werkgeheugen en timing, scoren kinderen en volwassenen met ADHD vaak slechter dan hun leeftijdsgenoten zonder ADHD. Een deel van de patiënten met ADHD kampt daarnaast met comorbide problemen, zoals depressieve klachten, gedragsproblemen of moeilijkheden in sociale relaties. Hoewel ADHD veelal wordt gediagnosticeerd in de kindertijd, zetten de problemen zich vaak voort in de adolescentie en volwassenheid. De exacte oorzaak van ADHD is niet bekend. Waarschijnlijk leveren veel verschillende factoren, inclusief genetische en omgevingsfactoren, elk een kleine bijdrage aan het ontstaan en het beloop van de stoornis. ADHD wordt in verband gebracht met veranderingen in het dopaminesysteem. De neurotransmitter dopamine fungeert als chemische boodschapper in de verbindingen tussen het striatum (een onder de hersenschors gelegen grijze stof kern) en de frontale hersengebieden. Deze frontostriatale verbindingen zijn belangrijk voor motorische aansturing, maar ook voor hogere cognitieve functies als zelfregulatie, planning, risico-­‐taxatie en het nemen van beslissingen. Neuroimaging studies laten subtiele afwijkingen zien in frontostriatale gebieden en verbindingen van mensen met ADHD. Ook in andere hersensystemen (bv. het noradrenaline systeem) en andere hersengebieden (bv. parietaalschors, cerebellum) worden bij mensen met ADHD subtiele afwijkingen gevonden. 223 BEHANDELING MET STIMULANTIA
Naast cognitieve therapie en/of oudertraining wordt ADHD vaak behandeld met medicatie. Stimulantia zoals methylfenidaat (beter bekend als Ritalin® of Concerta®) zorgen in de meerderheid van patiënten met ADHD voor een afname van symptomen en geven slechts in een kleine minderheid bijwerkingen. Behandeling met stimulantia begint vaak na de diagnose op jonge leeftijd, en wordt in sommige gevallen gecontinueerd tot in de adolescentie. De acute effecten van stimulantia in het brein zijn vaak onderzocht. Binnen een uur na inname verhoogt een enkele dosis methylfenidaat dopaminerge neurotransmissie in het striatum. Functionele magnetic resonance imaging (fMRI) meet de indirecte gevolgen van deze chemische verandering op de hersenactiviteit. De resultaten van de meeste fMRI studies suggereren dat methylfenidaat de activatiepatronen in het brein van patiënten met ADHD normaliseert: kinderen, jongeren en volwassenen met ADHD laten tijdens het maken van bijvoorbeeld aandachts-­‐ of plannings-­‐taken afwijkende activatiepatronen zien; na inname van een enkele dosis methylfenidaat zijn deze afwijkingen minder groot of zelfs verdwenen. Over de lange-­‐termijn gevolgen van stimulantia op het brein is veel minder bekend. Verschillende mogelijkheden zijn denkbaar. Herhaaldelijk ‘gepast gedrag’ (bv. opletten in de klas, of het maken van niet-­‐impulsieve beslissingen) zou kunnen leiden tot een versterking van de onderliggende neurale netwerken. Mogelijk raakt gepast gedrag daardoor steeds meer ingesleten, en is medicatie op den duur zelfs niet meer nodig. Een andere mogelijkheid, echter, is dat het herhaaldelijk toedienen van dopaminerge medicatie een toxisch effect heeft op het brein. Dopaminerge zenuwcellen zouden blijvend beschadigd kunnen raken. Voor beide hypothesen ontbreekt op dit moment doorslaggevend bewijs. DIT PROEFSCHRIFT
Het eerste doel van dit promotieonderzoek was het beschrijven van lange-­‐
termijn effecten, of de afwezigheid daarvan, van ADHD medicatie op het brein van kinderen, jongeren en jongvolwassenen met ADHD. Het tweede doel was het vergaren van kennis over de mechanismen die ten grondslag liggen aan lange-­‐termijn effecten van behandeling. In een grote groep kinderen, jongeren en jongvolwassenen met ADHD (NeuroIMAGE, zie BLOK 1) onderzochten we of en hoe verschillende parameters in het brein gerelateerd waren aan medicatiegebruik. We onderzochten ook of verschillen tussen patiënten en verschillen in blootstellingspatronen van invloed waren op lange-­‐termijn effecten. Daarnaast onderzochten we recreatief gebruik van stimulantia in een groep jongeren zonder ADHD (Youth At Risk, zie BLOK 2). Eerst geef 224 ik een samenvatting per hoofdstuk. Daarna volgt een beknopte beantwoording van de twee bovengenoemde onderzoeksvragen, en een uiteenzetting van de beperkingen van observationeel en cross-­‐sectioneel onderzoek zoals het huidige onderzoek. Tot slot bespreek ik de betekenis van de belangrijkste onderzoeksresultaten voor de klinische praktijk. BLOK 1. NeuroIMAGE De NeuroIMAGE studie vond tussen 2009 en 2012 plaats in drie Nederlandse centra (Nijmegen, Amsterdam en Groningen). Het doel van de studie was het lange-­‐termijn beloop van ADHD te beschrijven bij de Nederlandse deelnemers van het International Multicenter ADHD Genetics project (IMAGE; 2003-­‐2006). Meer dan 1000 kinderen, jongeren en jongvolwassenen uit bijna 500 families namen deel aan NeuroIMAGE (70.3% ADHD, 56% jongens, gemiddelde leeftijd = 17.0 jaar oud). Deelname omvatte onder meer diagnostische interviews met ouder en kind, gedragsvragenlijsten, neuropsychologische tests, structurele en functionele MRI, genotypering en het in kaart brengen van de complete medicatiegeschiedenis met behulp van apotheekgegevens. Op dit moment vindt in Nijmegen de derde follow-­‐up meting plaats (NeuroIMAGE2). Voor meer informatie over de NeuroIMAGE studie, zie www.neuroimage.nl en von Rhein et al., 2015. Hoofdstuk 3, 4, 5, 6, 7 en 9 zijn gebaseerd op het NeuroIMAGE cohort. SAMENVATTING PER HOOFDSTUK
In HOOFDSTUK 2 worden bevindingen beschreven uit eerder onderzoek naar korte-­‐ en lange-­‐termijn effecten van ADHD medicatie op het brein. In de beschikbare literatuur wordt het beeld geschetst dat behandeling met stimulantia de ontwikkeling van het brein mogelijk normaliseert. Functionele neuroimaging studies demonstreren dat het toedienen van een enkele dosis methylfenidaat resulteert in acute normalisatie van afwijkende activatiepatronen in het brein. Over lange-­‐termijn effecten van behandeling op het brein is minder bekend, hoewel aanwijzingen worden gevonden voor lange-­‐termijn normalisatie van hersenstructuur: kinderen die niet zijn behandeld met stimulantia laten in verschillende hersengebieden afwijkingen zien, die niet of minder aanwezig zijn bij behandelde kinderen met ADHD. 225 BLOK 2. Youth At Risk De Youth At Risk (YAR) studie is een prospectief longitudinaal onderzoek onder jongeren met een verhoogd risico op problematisch middelengebruik. De studie startte in 2002 in San Diego, California. Bijna 300 gezonde jongens en meisjes in de leeftijd van twaalf tot veertien jaar, die nog niet of nauwelijks waren blootgesteld aan alcohol en drugs, werden geïncludeerd. Ongeveer de helft van deze kinderen was afkomstig uit een gezin waarin problematisch middelengebruik voorkwam, wat werd beschouwd als risicofactor. De eerste meting bestond uit interviews over het gebruik van alcohol en drugs, gedragsvragenlijsten, neuropsychologische tests en structurele en functionele MRI. De middelengebruik interviews worden elke zes maanden herhaald, en deelnemers worden tot op heden elk jaar uitgenodigd voor een volledige follow-­‐up meting inclusief MRI sessie. Hoofdstuk 8 is gebaseerd op het YAR cohort. In HOOFDSTUK 3 onderzoeken we het lange-­‐termijn effect van stimulantia op de dikte van de hersenschors. Eerder werd gevonden dat ADHD gepaard gaat met een algehele verdunning van de hersenschors, die mogelijk minder prominent is op latere leeftijd en/of na behandeling met stimulantia. In de NeuroIMAGE groep vinden we bij jongeren en jongvolwassenen met ADHD een dunnere mediaal-­‐temporale schors in vergelijking met leeftijdsgenoten zonder ADHD. Deze verdunning is aanwezig op verschillende leeftijden en is niet gerelateerd aan behandeling met stimulantia. We vinden dus geen bewijs dat leeftijd en/of stimulantia een normaliserend effect hebben op de ontwikkeling van corticale dikte bij jongeren en jongvolwassenen met ADHD. In HOOFDSTUK 4 beschrijven we het verband tussen het gebruik van ADHD medicatie enerzijds en witte stof structurele connectiviteit (gemeten met diffusion tensor imaging, of DTI) anderzijds. Afwijkingen in structurele connectiviteit bij mensen met ADHD waren al eerder gevonden, maar het effect van medicatie op deze afwijkingen was onbekend. We onderzoeken structurele connectiviteit in vijf dopaminerge verbindingen tussen frontale, striatale en limbische hersengebieden. We vinden dat fractionele anisotropie in orbitofrontaal-­‐striatale verbindingen lager is in patiënten met ADHD ten opzichte van gezonde controles, wat wijst op verminderde structurele connectiviteit. In dezelfde witte-­‐stof banen is stimulantiagebruik geassocieerd met lagere diffusiviteit, wat kan wijzen op verbeterde structurele connectiviteit na behandeling met stimulantia. In HOOFDSTUK 5 onderzoeken we breinactivatie tijdens het uitvoeren van een beloningstaak. In een eerdere studie vertoonden NeuroIMAGE-­‐deelnemers met ADHD 226 afwijkende activatie in het striatum. Echter, in een groep volwassenen met ADHD die tijdens hun jeugd waren behandeld met stimulantia werden eerder geen afwijkingen gevonden. In hoofdstuk 5 beschrijven we dat medicatiegeschiedenis inderdaad voorspellend is voor hersenactivatie tijdens het ontvangen van beloning, maar niet in het striatum. Deelnemers die veel medicatie hebben gekregen (vroege startleeftijd, hoge dosis) vertonen meer activatie in de supplementary motor cortex en dorsal anterior cingulate cortex, vergeleken met deelnemers die minder intensief zijn behandeld (latere startleeftijd, lagere dosis). Mogelijk compenseert de inzet van deze ‘hogere-­‐orde’ hersengebieden de afwijkende activatie in het striatum in de intensief-­‐
behandelde groep. In HOOFDSTUK 6 beschrijven we het lange-­‐termijn effect van medicatie op het volume van het striatum, de frontaalschors en de hippocampus, en hoe deze wordt beïnvloed door leeftijd en/of genetische aanleg. Dierstudies suggereerden dat stimulantia op jongere leeftijd een groter effect zouden hebben op de structuur van het brein dan op latere leeftijd. In dit hoofdstuk beschrijven we dat DRD4 genotype, leeftijd en medicatiegeschiedenis samen voorspellend zijn voor het volume van de frontaalschors en linker hippocampus. Het frontaalschorsvolume van jonge dragers van het DRD4 7R-­‐allel met ADHD is vergelijkbaar met dat van gezonde controles, echter alleen bij een hoge dosis stimulantia; bij een lagere dosis is de frontaalschors kleiner. Bij oudere deelnemers met ADHD vinden we een kleinere frontaalschors, onafhankelijk van genotype of behandelgeschiedenis. Het volume van de hippocampus is niet afwijkend bij deelnemers met ADHD, echter in jonge dragers van het DRD4 7R-­‐allel is het volume van de hippocampus wel groter bij meer medicatiegebruik. Deze resultaten wijzen erop dat jonge dragers van het DRD4 7R-­‐
allel mogelijk gevoeliger zijn voor structurele veranderingen in de cortex na behandeling met stimulantia. HOOFDSTUK 7 betreft een studie naar gecombineerde behandeling met stimulantia en atypische antipsychotica, een veelvoorkomende off-­‐label behandeling van gedragsproblemen bij ADHD. Dierstudies hebben aangetoond dat antipsychotica het acute effect van stimulantia op dopamine neurotransmissie in het striatum kan opheffen. Wij vergelijken volumina van dopaminerge hersengebieden tussen drie groepen: patiënten die zijn behandeld met stimulantia en antipsychotica, patiënten die zijn behandeld met enkel stimulantia en gezonde controles. Vergeleken met gezonde controles hebben patiënten in de gecombineerde behandelgroep een lager totaal brein volume en lagere volumina van de frontaalschors, thalamus, en subthalamische kernen. Patiënten die uitsluitend zijn behandeld met stimulantia hebben deze afwijkingen niet. Afwijkingen in de antipsychotica groep waren mogelijk al aanwezig vóór de behandeling en kunnen hebben bijgedragen aan het inzetten van antipsychotica. Echter, de bevindingen kunnen er ook op wijzen dat de combinatie 227 van stimulantia en atypische antipsychotica resulteert in structurele veranderingen in het brein. In HOOFDSTUK 8 maken we een zijsprong en onderzoeken het effect van niet-­‐
medisch, recreatief stimulantiagebruik (bv. amfetamine, cocaïne). Het patroon van recreatieve blootstelling verschilt sterk van het typische blootstellingspatroon bij de behandeling van ADHD. Wij bestuderen het effect van incidentele blootstelling aan hoog-­‐gedoseerde stimulantia op de ontwikkeling van de hippocampus en het geheugen, in de prospectieve Youth At Risk groep. We vinden dat geheugenprestatie van jonge stimulantiagebruikers een afwijkend verloop over tijd liet zien. Jongeren die geen middelen of enkel niet-­‐stimulerende middelen gebruiken vertonen over de tijd een stabiele geheugenfunctie of lichte verbetering. Stimulantiagebruikers daarentegen laten een negatieve trend zien. We vinden geen veranderingen in de structuur van de hippocampus bij de stimulantiagebruikers vergeleken met beide andere groepen. Tot slot onderzoeken we in HOOFDSTUK 9 de lange-­‐termijn effecten van stimulantia op gedragsmatige en klinische uitkomsten. Eerdere studies waren slechts beperkt in staat geweest effecten over meerdere jaren adequaat te onderzoeken en hadden aanwijzingen gevonden voor zowel klinische verbetering als verslechtering na langdurig medicatiegebruik. Hier vergelijken we de ontwikkeling van ADHD symptomen, sociaal-­‐emotioneel functioneren en testprestatie op drie cognitieve domeinen (steeds gemeten zonder medicatie) tussen twee zorgvuldig gematchte groepen van behandelde en niet-­‐behandelde patiënten met ADHD. Met uitzondering van emotionele problemen en pro-­‐sociaal gedrag verbeteren alle uitkomstmaten over een periode van zes jaar. Echter, patiënten die tijdens deze periode wel en niet zijn behandeld met stimulantia verbeteren evenveel. We concluderen dat behandeling met stimulantia geen lange-­‐termijn effect heeft op de ontwikkeling van ADHD symptomen, sociaal-­‐emotioneel functioneren of cognitie. VRAAG 1: WAT IS HET LANGE-TERMIJN EFFECT VAN ADHD MEDICATIE OP HET ZICH ONTWIKKELENDE
BREIN?
Neurotoxiciteit en schadelijke effecten De bevindingen in dit proefschrift wijzen niet op schadelijke effecten van stimulantia op de ontwikkeling van het brein. Medicatiegebruik leidt niet tot veranderingen in de dikte van de hersenschors, het volume van het striatum of activatiepatronen in het striatum tijdens beloning. We vinden ook geen aanwijzingen voor beschadiging van de frontostriatale witte stof connectiviteit na gebruik van ADHD medicatie. Bovendien vinden we dat kinderen met ADHD die medicatie 228 gebruiken dezelfde ontwikkeling laten zien in hun gedrag als kinderen met ADHD die geen medicatie gebruiken. In één hoofdstuk vinden we een structurele verandering in het brein die mogelijk zou kunnen wijzen op een negatief gevolg van medicatiegebruik. In een specifieke subgroep, namelijk jong behandelde kinderen die drager zijn van het DRD4 7R-­‐allel, staat meer blootstelling aan stimulantia in verband met een groter volume van de hippocampus. De hippocampus is belangrijk voor leren en geheugen. Echter, geheugenproblemen zoals die bijvoorbeeld worden gezien bij ernstig verslaafde stimulantiagebruikers gaan gepaard met een kleiner, niet groter, volume van de hippocampus. Lange-­‐termijn geheugenfunctie wordt in de NeuroIMAGE sample helaas niet gemeten. Onze bevinding dat incidentele recreatieve stimulantiagebruikers een subtiele verslechtering van geheugenfunctie laten zien benadrukt het belang van vervolgonderzoek. Normalisatie en positieve effecten Resultaten van eerder MRI onderzoek suggereerden dat ADHD medicatie op lange termijn de structurele veranderingen in het brein van kinderen met ADHD (ten opzichte van leeftijdsgenoten zonder ADHD) kunnen verminderen of zelfs kunnen doen verdwijnen. In slechts één hoofdstuk onderschrijven onze bevindingen, ten dele, een dergelijk normaliserend effect van ADHD medicatie. We vinden dat de frontaalschors kleiner is bij kinderen, jongeren en jongvolwassenen met ADHD vergeleken met leeftijdsgenoten zonder ADHD. Bij jonge kinderen met het DRD4 7R-­‐
allel die stimulantia gebruiken is de frontaalschors niet afwijkend klein. Echter, veel vaker concluderen we dat de verwachte normalisatie na behandeling niet optreedt. We vinden vermindering van mediaal-­‐temporale corticale dikte bij zowel behandelde als onbehandelde jongeren met ADHD. Activatie in het striatum, waarvan eerder was aangetoond dat deze afwijkend is bij onze ADHD groep, staat bovendien niet in verband met medicatiegeschiedenis. Tot slot vinden we dat op gedragsniveau verbeteringen tijdens de adolescentie evenzeer optreden bij kinderen die stimulantia gebruiken, als bij kinderen die dat niet doen. Positieve effecten van stimulantia kunnen zich ook manifesteren in compensatoire processen, in plaats van in vermindering van initiële verschillen tussen mensen met en zonder ADHD. In twee hoofdstukken vinden we aanwijzingen voor lange-­‐termijn compensatie in het brein na behandeling met stimulantia. Patiënten met een geschiedenis van vroeg begonnen en hoog-­‐gedoseerde behandeling, maar niet patiënten met een geschiedenis van laat begonnen en laag-­‐
gedoseerde behandeling, vertonen tijdens beloning activatie in hersengebieden die belangrijk zijn voor cognitieve controle. Voor patiënten met ADHD, die vaak geneigd 229 zijn impulsief te reageren op beloning, is meer cognitieve controle waarschijnlijk positief. Ten tweede vinden we een verminderde structurele connectiviteit in orbitofrontale-­‐striatale witte stof banen in patiënten met ADHD ten opzichte van controles. In diezelfde banen vinden we, met een andere indicator voor connectiviteit (diffusiviteit in plaats van fractionele anisotropie), een verbeterde connectiviteit na meer blootstelling aan stimulantia. Conclusie Lange-­‐termijn effecten van behandeling met stimulantia op het zich ontwikkelende brein met ADHD zijn subtiel. Veel vaker dan wél, vinden we géén verband tussen medicatie en breinstructuur. We vinden geen aanwijzingen voor schadelijke effecten, hoewel het verband tussen stimulantiagebruik en hippocampusvolume verder onderzoek behoeft. Ook vinden we weinig bewijs dat langdurig medicatiegebruik de ontwikkeling van het brein normaliseert, met uitzondering van de frontaalschors waar stimulantia mogelijk een positief lange-­‐
termijn effect hebben bij jongere dragers van het DRD4 7R-­‐allel. In twee gevallen vinden we mogelijke aanwijzingen voor compensatoire lange-­‐termijn effecten, namelijk in de vorm van activatie van cognitieve controle gebieden tijdens beloning en verbeterde connectiviteit. We vinden geen lange-­‐termijn positieve dan wel negatieve gevolgen van medicatiegebruik op de ontwikkeling van ADHD symptomen of sociaal-­‐emotioneel of cognitief functioneren. VRAAG 2: WELKE MECHANISMEN LIGGEN TEN GRONDSLAG AAN DE LANGE-TERMIJN EFFECTEN VAN
ADHD MEDICATIE IN HET BREIN?
Het tweede doel van dit onderzoek was het beschrijven van neurobiologische mechanismen die ten grondslag kunnen liggen aan lange-­‐termijn medicatie-­‐effecten. Onze resultaten suggereren dat lange-­‐termijn effecten van ADHD-­‐medicatie in het brein niet op dezelfde manier tot stand komen als acute medicatie effecten (nl. door blokkade van dopaminerge autoreceptoren in het striatum), maar door (een) ander(e) mechanisme(n). Immers, we vinden lange-­‐termijn veranderingen in de frontaalschors, hippocampus en supplementary motor area, maar niet in het striatum. Bovendien vinden we dat het DAT1 gen, dat zoals gebleken uit eerder onderzoek het acute effect van stimulantia modereert, niet van invloed is op lange-­‐termijn effecten van ADHD-­‐medicatie in het brein. Hoe ziet dit alternatieve mechanisme eruit? Met name de genetische bevindingen (lange-­‐termijn corticale veranderingen bij jonge dragers van het DRD4 7R-­‐allel) zijn hier informatief. Neurale plasticiteit verwijst naar het optreden van 230 structurele veranderingen in het brein als gevolg van activatie. Het DRD4 gen codeert postsynaptische dopamine receptoren, die van cruciaal belang zijn voor neurale plasticiteit. Mogelijk hebben postsynaptische neuronen in de cortex van jonge dragers van het DRD4 7R-­‐allel specifieke kenmerken die neurale plasticiteit bevorderen, waardoor de kans groter is dat het effect van medicatie ‘beklijft’. Eerder werd in een groep behandelde kinderen met ADHD gevonden dat het beloop van ADHD symptomen gunstiger was in dragers van het DRD4 7R-­‐allel vergeleken met niet-­‐
dragers van het allel. Bovendien werd in de NeuroIMAGE groep gevonden dat de combinatie van DRD4 genotype en een positieve sociale omgeving voorspellend is voor activiteit in de frontaalschors tijdens een respons inhibitie taak. Dit wijst mogelijk op een meer algemeen mechanisme waardoor dragers van het DRD4 7R-­‐allel in gunstige omstandigheden (bv. door behandeling met ADHD medicatie en/of een positieve sociale omgeving) meer kans hebben op blijvende veranderingen in het brein. Onze bevindingen komen overeen met een dergelijk mechanisme; echter voor direct bewijs zijn PET/SPECT imaging of dierstudies noodzakelijk. METHODOLOGISCHE OVERWEGINGEN
Elke studie naar lange-­‐termijn effecten van medicatie is noodzakelijkerwijs observationeel; immers, het willekeurig en bovendien geblindeerd toewijzen van een behandeling (of juist géén behandeling) die meerdere jaren duurt stuit op zowel ethische als praktische bezwaren. Ook het huidige onderzoek is daarom observationeel. Een verband tussen blootstelling enerzijds en lange-­‐termijn uitkomsten anderzijds dat wordt gevonden in observationeel onderzoek kan duiden op een werkelijk behandeleffect, maar kan ook het gevolg zijn van bias of verstorende variabelen. Ondanks pogingen om zulke verstoring zoveel mogelijk te beperken, kunnen ze niet worden uitgesloten. Een observationeel design resulteert bovendien vaak in ongelijke behandelgroepen. In het huidige onderzoek is slechts een handjevol deelnemers met ADHD volledig vrij van medicatie, waardoor we deze groep minder goed hebben kunnen onderzoeken. Observationeel onderzoek biedt echter ook kansen. Juist in heterogene groepen zoals patiënten met ADHD zijn grootschalige observationele studies waardevol: ze schetsen een representatief beeld, en maken onderzoek mogelijk naar individuele verschillen tussen patiënten. Naast observationeel zijn de neuroimaging studies in het NeuroIMAGE cohort (maar niet de studie in het Youth At Risk sample, of de studie naar het effect van ADHD medicatie op gedrag) ook cross-­‐sectioneel. Cross-­‐sectionele data zijn verre van ideaal voor onderzoek naar ontwikkelingsstoornissen en/of behandeleffecten. Voorzichtigheid is daarom geboden bij de interpretatie van bevindingen met betrekking tot leeftijd en medicatiegebruik. Het is van belang dat onze resultaten 231 worden gerepliceerd in longitudinale studies. Tegelijkertijd biedt het cross-­‐sectionele design de mogelijkheid om ADHD te onderzoeken tijdens de adolescentie en jongvolwassenheid; een fase die vaak gekenmerkt wordt door grote klinische veranderingen. Zo kunnen ook medicatie effecten die pas na meerdere jaren optreden worden onderzocht. Vervolgonderzoek wat op dit moment plaatsvindt (NeuroIMAGE2, zie BLOK 1) biedt hiervoor verdere mogelijkheden. DEZE BEVINDINGEN IN DE KLINISCHE PRAKTIJK
Lange-­‐termijn gevolgen van het gebruik van ADHD medicatie zijn in de eerste plaats relevant voor kinderen, jongeren en jongvolwassenen met ADHD, waaronder zij die vrijwillig hebben deelgenomen aan het NeuroIMAGE onderzoek. Patiënten, hun ouders en behandelaars hebben behoefte aan betrouwbare informatie die ze kan helpen in beslissingen rondom medicatiegebruik. Hoewel het huidig onderzoek in deze behoefte tracht te voorzien, wil ik benadrukken dat onze bevindingen lange-­‐
termijn correlaten van medicatiegebruik weergeven, die mogelijk maar niet noodzakelijkerwijs overeenkomen met lange-­‐termijn gevolgen van medicatiegebruik. Bovendien werd enkel het brein onderzocht; lange-­‐termijn gevolgen voor andere organen en lichaamsfuncties zijn buiten beschouwing gelaten. Alles tezamen wijzen de resultaten van het huidige onderzoek niet op schadelijke lange-­‐termijn effecten van het gebruik van ADHD medicatie. Anderzijds vinden we ondersteuning voor enkele positieve lange-­‐termijn effecten. Op brein-­‐
niveau bestaan positieve effecten mogelijk uit 1) toegenomen activatie van cognitieve controle gebieden tijdens activatie van het beloningssysteem; 2) verbeterde orbitofrontale-­‐striatale structurele connectiviteit; en 3) voor een subgroep van patiënten, normalisatie van frontaalschorsvolume. Echter, we rapporteren tevens een opvallende afwezigheid van lange-­‐termijn effecten op gedragsniveau. Lange-­‐termijn veranderingen in het brein lijken zich dus niet te vertalen naar beter functioneren. Het is voor clinici van belang met jongeren en hun ouders te bespreken dat ADHD medicatie waarschijnlijk de symptomen van ADHD zal verminderen, maar op lange-­‐
termijn niet zal leiden tot een betere uitkomst. Clinici vragen zich mogelijk af welke patiënten het meeste baat hebben bij ADHD medicatie en bij welke patiënten het risico op schadelijke gevolgen het grootst is. In geen van de onze studies vinden we verschillen tussen jongens en meisjes. Jonge patiënten met het DRD4 7R-­‐allel zijn mogelijk gevoeliger voor lange-­‐termijn effecten van medicatie in het brein. Echter, zelfs binnen deze specifieke groep vinden we grote verschillen tussen patiënten. We concluderen daarom dat het voorspellen van lange-­‐
termijn effecten op basis van genetische aanleg en leeftijd op dit moment niet mogelijk is. Een tweede vraag uit de klinische praktijk betreft het optimaliseren van 232 behandeling. Moet behandeling op jonge leeftijd worden gestart en wat is de optimale dosis? Wanneer behandeling met ADHD medicatie in onze studies gepaard gaat met veranderingen in het brein, zijn deze veranderingen groter wanneer de behandelduur langer, de totale dosis hoger, en de startleeftijd jonger is. Echter we vonden niet dat één van deze kenmerken belangrijker is dan de andere kenmerken. Tot slot wil ik twee bevindingen toelichten die mogelijk, na replicatie, van belang zijn voor de klinische praktijk. Ten eerste vinden we aanwijzingen dat het gebruik van stimulantia, bij bepaalde patiënten en/of bij een hoge dosis, gepaard kan gaan met veranderingen in de hippocampus en in het lange-­‐termijn geheugen. Lange-­‐
termijn geheugenfunctie wordt in de klinische praktijk van ADHD vaak niet expliciet onderzocht. In afwachting van verder onderzoek raden we clinici aan bedacht te zijn op subtiele veranderingen in geheugenfunctie, met name bij patiënten die op jonge leeftijd zijn behandeld met een hoge dosis stimulantia. De tweede bevinding betreft het voorschrijven van antipsychotica. Antipsychotica zijn geïndiceerd als kortdurende behandeling van ernstige gedragsproblemen. In de huidige onderzoeksgroep wordt de voorgeschreven behandelduur van enkele weken en mass overschreden. Structurele veranderingen in het brein van patiënten die zijn behandeld met antipsychotica geven, in afwachting van verder onderzoek, aanleiding tot terughoudendheid in het voorschrijven van deze medicatie aan kinderen met ADHD. 233 234 ACKNOWLEDGEMENTS / DANKWOORD
235 236 Allereerst wil ik alle kinderen bedanken die belangeloos meedoen aan wetenschappelijk onderzoek. Het is niet niks om in een MRI scanner te gaan, testjes te doen, en je door een dik pak vragenlijsten heen te worstelen. Ook alle ouders, leerkrachten, artsen en apothekers die hun medewerking hebben verleend, veel dank. Het NeuroIMAGE team op afstand: Jan, Jaap, Barbara, Dirk, professor Faraone, Marcel, Maarten, Nanda, Alejandro, Marjolein, Corina, Larry, Jill Fractalkine, Hanneke, Jennifer, Annabeth, Winke, Marloes, Daan, Marianne, Daniel, Siri, Janita, Andrieke, Dennis en Sophie. Speciale dank aan de promovendi die een groot deel van hun tijd hebben besteed aan het verzamelen van data, waardoor ik dat niet meer hoefde te doen. Ook bijzonder veel dank voor alle co-­‐auteurs op mijn (onze!) papers. Collega’s en ex-­‐collega’s bij Accare: Andrea, Anne, Anne-­‐Flore, Anne-­‐Marie, Annelies, Annemiek, Arianne, Barbara, Djûke, Eke, Ellen, Florianne, Halewijn, Hyun, Jessica, Judith, Julie, Lianne, Lotte, Marco, Marian, Marieke, Mariken, Marjan, Mark-­‐Peter, Marlien, Mascha, Monika, Natalie, Neeltje, Renee, Rianne, Sanne, Simone, Thaira, Vera en Yvonne (wat zijn jullie met véél!). Bijzonder veel dank gaat naar Florianne, Mascha, en Marlien, voor alle honderden kinderen, ouders, en apotheken die jullie hebben gebeld, gemaild, geschreven, en gefaxt. Ook speciale dank voor Vera, Sanne en Jessica, voor alle halve-­‐uurtjes, drie-­‐kwartiertjes, fruitmomentjes en lief-­‐en-­‐leedjes. I was extremely fortunate to spend four months of my PhD in San Diego. Thank you Susan Tapert and Jay Giedd for the very warm welcome in your research group at the University of California, San Diego. Many thanks also to Norma Castro, Tam Nguyen, Lara Wierenga, Lindsay Squeglia, Hauke Bartsch, MJ Meloy, Ty Brumback, and Joanna Jacobus for your invaluable help. Er zijn een aantal mensen en plekken die elk op hun eigen manier hebben bijgedragen aan mijn wetenschappelijke carrière. Jeroen Geurts en Hanneke Hulst: met jullie is dit alles begonnen, en ik ben heel blij dat dat is gebeurd; Odile van den Heuvel: dank voor je vertrouwen en waardevolle lessen; Sarah Durston, Patrick de Zeeuw, en NICHE-­‐lab: mijn jaar in Utrecht was heerlijk en (ook) voor deze promotie heel waardevol; Jochen Seitz: jij inspireert, motiveert, lacht, én luistert. Dank daarvoor. Catharina: wat heb ik van jou veel geleerd; zonder jouw statistiek-­‐ én sollicitatie-­‐
lessen had ik de baan in Cambridge nooit bemachtigd. Jouw vriendelijke, ongefilterde, soms warrige en soms boze woorden stuurden me altijd in de juiste richting. Ik heb heel erg genoten van onze samenwerking. Pieter: bedankt voor je vertrouwen, en voor de ruimte die je me steeds liet om mijn eigen koers te varen. Ik heb bewondering voor je gedrevenheid, die tijdens de vier jaar van mijn PhD resulteerde in een indrukwekkende toestroom van nieuwe collega’s. Veel dank! Tot slot, mijn lieve familie, vrienden, vriendinnen. Pap, mam, Hilde, en Teunis: ik voel me gezegend met zoveel zorgzame ouders. De wetenschap brengt me steeds verder van huis, maar ik voel me nu dichter bij jullie dan ooit. Lieve broer en zus, para-­‐ en partynimf: Oh, wat ben ik trots op jullie! Lieve Denise, paranimf Natalie, Meike, Tessa, Mirte, en Lisanne: wat ben ik dankbaar voor jullie vriendschap. Jullie zijn dapper en sterk, en zo belangrijk voor mij. 237 238 CURRICULUM VITAE
239 240 Lizanne Schweren was born in Geldrop, the Netherlands on December 28th 1986. She attended highschool (VWO, or pre-­‐university secondary education) at the Varendonck College in Asten and the ROC ter AA in Helmond, from which she graduated in 2003. After highschool, Lizanne enrolled in a school-­‐based intercultural exchange program to live for one year with a local family in Mae Chan, Thailand. Upon her return, she studied neuropsychology at the Radboud University in Nijmegen, and obtained her Bachelor’s degree and Honours Program certificate. She was accepted in the Master Neuroscience program, a biomedically-­‐oriented Research Master’s program of the VU University in Amsterdam. Here, she successfully completed two research internships: the first at the MS Center where she studied functional MRI correlates of cognitive decline in patients with multiple sclerosis, and the second at the Department of Psychiatry where she performed a multi-­‐center mega-­‐analysis of structural MRI scans of patients with obsessive compulsive disorder. Her Master’s thesis, conducted in collaboration with the University Medical Center Utrecht, addressed the effects of stimulant treatment on the developing brain in ADHD. The results of both internships and the Master’s thesis were published in peer-­‐reviewed journals. Lizanne graduated cum laude in 2011. Throughout her studies, she worked as a teaching assistant at the Radboud University and as a research assistant at the VU University. After graduating, Lizanne joined the Neuroimaging in Childhood (NICHE) lab at the University Medical Center Utrecht as a research assistant, before starting her PhD project in the summer of 2012. During her PhD, supervised by Prof. Pieter Hoekstra and Dr. Catharina Hartman at the University of Groningen, she investigated the long-­‐term effects of stimulant treatment on the developing brain, using various structural and functional neuroimaging modalities. As part of her studies, she collaborated closely with Prof. Jay Giedd and Prof. Susan Tapert at the University of California, San Diego, to investigate adolescent recreational stimulant use. From June 2016, Lizanne works as a postdoctoral Research Associate in the Neuroscience in Psychiatry Network at the Department of Developmental Psychiatry, University of Cambridge, UK. For more information, updates, and blogs, see www.lizanneschweren.com 241 SCHWEREN LJS, Hartman CA, Heslenfeld DJ, Groenman AP, Franke B, Oosterlaan J, Buitelaar JK, Hoekstra PJ (2016). Age and DRD4 genotype moderate associations between stimulant treatment history and cortex structure in ADHD. J Am Acad Child Adolesc Psychiatry, 55(10), 877-­‐85. SCHWEREN LJS, Hartman CA, Heslenfeld DJ, van der Meer D, Franke B, Oosterlaan J, Buitelaar JK, Faraone SV, Hoekstra PJ (2015). Thinner medial temporal cortex in adolescents with attention-­‐deficit/hyperactivity disorder and the effects of stimulants. J Am Acad Child Adolesc Psychiatry, 54(8), 660-­‐7. SCHWEREN LJS, Hartman CA, Zwiers MP, Heslenfeld DJ, Franke B, Oosterlaan J, Buitelaar JK, Hoekstra PJ (2016). Stimulant treatment history predicts frontal-­‐striatal structural connectivity in adolescents with attention-­‐deficit/hyperactivity disorder. Eur Neuropsychopharmacol, 26(4), 674-­‐83. SCHWEREN LJS, Groenman AP, von Rhein D, Weeda W, Faraone SF, Luman M, van Ewijk H, Heslenfeld DJ, Franke B, Buitelaar JK, Oosterlaan J, Hoekstra PJ, Hartman CA. (2016). Stimulant treatment trajectories are associated with neural reward processing in attention-­‐deficit/hyperactivity disorder. J Clin Psychiatry, in press. SCHWEREN LJS, Hartman CA, Zwiers MP, Heslenfeld DJ, van der Meer D, Franke B, Oosterlaan J, Buitelaar JK, Hoekstra PJ (2015). Combined stimulant and antipsychotic treatment in adolescents with attention-­‐deficit/hyperactivity disorder: a cross-­‐sectional observational structural MRI study. Eur Child Adolesc Psychiatry, 24(8), 959-­‐68. SCHWEREN LJS, de Zeeuw P, Durston S (2013). MR imaging of the effects of methylphenidate on brain structure and function in attention-­‐
deficit/hyperactivity disorder. Eur Neuropsychopharmacol, 23(10), 1151-­‐64. SCHWEREN LJS, Hoekstra PJ, van Lieshout M, Rommelse NNJ, Franke B, Oosterlaan J, Buitelaar JK, Hartman CA. No long-­‐term effects of stimulant treatment on ADHD symptoms, social-­‐emotional, or cognitive functioning. Submitted SCHWEREN LJS, Giedd J, Castro N, Bartsch H, Squeglia LM, Meloy MJ, Tapert S. Memory performance and hippocampus structure after low-­‐frequency recreational stimulant use in adolescents. Submitted 242 Greven CU, Bralten J, Mennes M, O'Dwyer L, van Hulzen K, Rommelse N, SCHWEREN LJS, Hoekstra PJ, Hartman CA, Heslenfeld D, Oosterlaan J, Faraone SV, Franke B, Zwiers MP, Arias-­‐Vasquez A, Buitelaar JK (2015). Developmentally stable whole-­‐brain volume reductions and developmentally sensitive caudate and putamen volume alterations in those with attention-­‐deficit/hyperactivity disorder and their unaffected siblings. JAMA psychiatry, 72(5), 490-­‐9. Van der Meer D, Hoekstra PJ, Zwiers M, Mennes M, SCHWEREN LJS, Franke B, Heslenfeld DJ, Oosterlaan J, Faraone SV, Buitelaar JK, Hartman CA (2015). Brain correlates of the interaction between 5-­‐HTTLPR and psychosocial stress mediating attention-­‐deficit/hyperactivity disorder severity. Am J Psychiatry, 172(8), 768-­‐
75. De Wit SJ, Alonso P, SCHWEREN LJS, Mataix-­‐Cols D, Lochner C, Menchón JM, [the OBIC-­‐
consortium, and] van den Heuvel OA (2014). Multicenter voxel-­‐based morphometry mega-­‐analysis of structural brain scans in obsessive-­‐compulsive disorder. Am J Psychiatry, 171(3), 340-­‐9. Hulst HE, Schoonheim MM, Roosendaal SD, Popescu V, SCHWEREN LJS, van der Werf YD, Visser LH, Polman CH, Barkhof F, Geurts JJ (2012). Functional adaptive changes within the hippocampal memory system of patients with multiple sclerosis. Hum Brain Mapp, 33(10), 2268-­‐80. Mudra S, Völker U, SCHWEREN LJS, Wessing I, Seitz J (2015). YICAP/ECAP international young investigators paper and grant writing workshop. Eur Child Adolesc Psychiatry, 24(2), 247-­‐8. Forde JN, Ronan L, Zwiers M, SCHWEREN LJS, Alexander-­‐Bloch AF, Franke B, Faraone SV, Oosterlaan J, Heslenfeld DJ, Hartman CA, Buitelaar JK, Hoekstra PJ. Healthy cortical development through adolescence and early adulthood. Submitted Hoogman M, Bralten J, Hibar, Mennes M, Zwiers M, SCHWEREN LJS, [the ADHD-­‐ENIGMA consortium, and] Franke F. Subcortical brain volume differences between participants with ADHD and healthy individuals across the lifespan: an ENIGMA collaboration. Submitted 243