Download Position paper Discordant conclusions from HIV clinical trials — an

Antiviral Therapy 10:367–374
Position paper
Discordant conclusions from HIV clinical trials —
an evaluation of efficacy endpoints
Andrew Hill1* and Ralph DeMasi 2
1
Department of Pharmacology, University of Liverpool, Liverpool, UK
Trimeris, Morrisville, NC, USA
2
*Corresponding author: Tel: +44 207 223 2064; Fax: +44 151 794 5656; E-mail: [email protected]
The three main components of long-term efficacy for a
combination of antiretrovirals are: (i) the strength of the
antiviral effect, (ii) toxicity profile and (iii) patient
acceptability of the regimen. Intent-to-treat (ITT)
analysis, where discontinuations and switches are considered failures [ITT, switch equals failure (ITT/S=F)], is a
regulatory standard for analysing the efficacy of antiretrovirals. A review of all clinical trials published in FDA
product labels was conducted, including all clinical trials
of boosted protease inhibitor- or nucleoside reverse transcriptase inhibitor-based highly active antiretroviral
therapy in treatment-naive patients, and all clinical trials
of antiretrovirals in treatment-experienced patients.
Clinical trials where the results are presented in the standard ITT/S=F method were included. For randomized
clinical trials in treatment-naive patients, the majority of
treatment discontinuations have been either for toxicity
(32%) or patient refusal of treatment (41%), with only
27% of failure endpoints for virological reasons among
recent clinical trials in naive patients. Therefore, there is
the potential for the results from ITT/S=F analysis to be
driven by non-virological endpoints – a new treatment
can be classified as ‘more efficacious’ than control owing
to fewer discontinuations due to adverse events or
patient preference.
In order to understand the intrinsic potency of the antiretroviral regimen under study, ITT analysis needs to be
supplemented by standardized as-treated analyses,
excluding withdrawals for toxicity or other reasons. To
evaluate the efficacy of a treatment strategy or sequential treatment regimens, the ‘ITT, switch included’ (ITT/SI)
method: where changes from the initial randomized
treatment are not classified as treatment failure – can be
used. However, interpretation of clinical trials using ITT/SI
analysis is difficult and depends on the frequency of
treatment switching in the different arms of a trial.
Conclusions on efficacy from clinical trials can depend on
the primary analysis used; most commonly, treatments
could be significantly different by ITT/S=F analysis, but
then interpreted as equivalent using the ITT/SI or
as-treated methods.
Introduction
The standard of care for antiretroviral-naive patients
has become a triple combination of two nucleoside
reverse transcriptase inhibitors (NRTIs) plus either a
non-nucleoside reverse transcriptase inhibitor (NNRTI)
or boosted protease inhibitor (PI), and is termed highly
active antiretroviral therapy (HAART). More intensive
combinations are recommended for patients with drugresistant HIV [1–3]. The primary endpoint used in
most randomized HIV clinical trials is the percentage
of patients with HIV RNA suppression (below either
400 or 50 copies/ml) after 24–48 weeks of randomized
treatment. The HIV RNA assay is a direct and reproducible measure of viral replication [4] and can be used
to evaluate response in all treated patients; HIV RNA
© 2005 International Medical Press 1359-6535
suppression from antiretroviral treatment is correlated
with a lower risk of HIV disease progression [4]. In
addition, stronger HIV RNA suppression during
ongoing treatment may lower the risk of developing
drug resistance [5]. CD4 counts are normally included
as secondary endpoints.
For randomized clinical trials designed for antiretroviral drug approval, intent-to-treat (ITT) analysis is used
for comparing treatment groups. Patients who discontinue randomized treatment for any reason are classified
as treatment failures [6]. Classifying discontinuations
and switches as treatment failures has been termed
‘discontinuation and switch equals failure’, ‘switch
equals failure’, ‘non-completer equals failure’ or ‘ITT/d’.
367
A Hill & R DeMasi
This approach, hereafter termed ‘ITT, switch equals
failure’ (ITT/S=F), provides an estimate of the minimum
efficacy for the randomized treatment evaluated –
patients might show responses to second-line treatments,
but these are not included as successes in this type of
analysis. For drug approvals in North America, failure
endpoints are classified using standard definitions into
three main categories [6]: (i) virological failure – patients
whose HIV RNA was never suppressed below 400 (or
50) copies/ml or where HIV RNA levels rebounded
above this threshold after earlier suppression below the
threshold; (ii) discontinuation of randomized treatment
for safety reasons, including clinical adverse events,
lab abnormalities and death; and (iii) discontinuation
of randomized treatment for other reasons (for
example, withdrawal of consent, loss to follow-up,
non-adherence, pregnancy or protocol violation).
Table 1 shows the results from all recent registration
trials included in US drug labels for first-line treatment
with either NNRTI- or boosted PI-based HAART, with
endpoints analysed in a standard format for FDA drug
labelling. The failure endpoints are classified into three
categories as described above. For treatment-naive
patients, 27% of failure endpoints have been virological, with 32% due to treatment discontinuation for
adverse events and 41% due to discontinuation for
other reasons (Table 1). Table 2 shows similar results
for all trials included in drug labels for treatment of
antiretroviral-experienced patients: treatment failure
endpoints were 68% virological, 11% for toxicity and
21% for other reasons.
For regulatory authorities granting marketing
approvals to new treatments based on efficacy and
safety in randomized clinical trials, the key issue is the
intrinsic potency and tolerability of the new experimental drugs under study. In this respect, discontinuation of a new experimental treatment for toxicity or
other reasons is a type of treatment failure and, if a
control treatment showed lower rates of these discontinuations, it might be favoured.
However, the ITT/S=F method has been criticized for
including a high proportion of non-virological
endpoints and for not examining the implications of
initial treatment failure with regards to response to
subsequent treatments [7]. It is very important to be
aware of the reason for any difference in overall ‘efficacy’ when the ITT/S=F analysis is used, as this may
not be the result of more potent and sustained effects of
the treatment regimen on virological response, but may
instead be dominated by improved toxicity profiles or
lower overall discontinuation rates (which may be for
subjective reasons in open-label trials). The three types
of endpoint will therefore be reviewed, with alternative
methods suggested for interpreting discordant results
from trial analyses.
Table 1. Summary of 48-week treatment failure endpoints from regulatory trials (data source: FDA drug labels) – antiretroviralnaive patients on HAART based on two NRTIs plus either an NNRTI or a boosted PI
Trial
Arm
n
%VF
%AE
%Other
Reference
DMP-006
FTC 301A
FTC 301A
EPV 201
EPV 201
Gilead 903
Gilead 903
CNA3021
CNA3021
APV3002
Abbott 863
BMS 034
ZDV/3TC/EFV
FTC/ddI/EFV
d4T/ddI/EFV
ZDV/3TCod/EFV
ZDV/3TCbid/EFV
TDF/3TC/EFV
d4T/3TC/EFV
ABCod/3TC/EFV
ABCbid/3TC/EFV
fAPV/r/ABC/3TC
LPV/r/d4T/3TC
EFV/ZDV/3TC
422
286
285
278
276
299
301
384
386
322
326
404
6
3
11
8
8
6
4
5
5
6
9
21
7
7
14
7
12
7
7
13
11
10
6
8
17
9
8
18
14
8
7
11
3
15
10
10
[32]
[33]
[33]
[34]
[34]
[35]
[35]
[36]
[36]
[37]
[38]
[39]
7.7
27
9.1
32
11.7
41
Unweighted mean
Percentage of total endpoints
%VF, percentage of randomized patients with virological failure at week 48 (HIV RNA >400 copies/ml). %AE, percentage of randomized patients discontinuing
randomized treatment for adverse events or death. %Other, percentage of randomized patients discontinuing randomized treatment for other reasons (including nonadherence, protocol violation, pregnancy, withdrawal of consent and loss to follow-up). HAART, highly active antiretroviral therapy; NRTI, nucleoside reverse
transcriptase inhibitor; NNRTI, non-nucleoside reverse transcriptase inhibitor; PI, protease inhibitor; 3TC, lamivudine; ABC, abacavir; ddI, didanosine; d4T, stavudine;
EFV, efavirenz; fAPV, fosamprenavir; FTC, emtricitabine; LPV/r, lopinavir/ritonavir; ZDV, zidovudine; od, once daily; bid, twice daily.
368
© 2005 International Medical Press
Efficacy endpoints in HIV clinical trials
Table 2. Summary of 48-week treatment failure endpoints from regulatory trials (data source: FDA drug labels) – antiretroviralexperienced patients on any approved HAART regimen
Trial
Arm
n
%VF
%AE
%Other
Reference
FTC 303
FTC 303
Abbott 888
Gilead 907
TORO 1/2
FTC/ZDV/NNRTI
3TC/ZDV/NNRTI
LPV/r/NVP/NRTI
NRTI/TDF
T-20 OBR
294
146
148
368
661
7
8
24
53
49
4
1
6
3
9
12
10
14
3
5
[33]
[33]
[38]
[38]
[40]
28.2
68
4.6
11
8.8
21
Unweighted mean
Percentage of total endpoints
%VF, percentage of randomized patients with virological failure at week 48 (HIV RNA >400 copies/ml). %AE, percentage of randomized patients discontinuing
randomized treatment for adverse events or death. %Other, percentage of randomized patients discontinuing randomized treatment for other reasons (including nonadherence, protocol violation, pregnancy, withdrawal of consent and loss to follow-up). NRTI, nucleoside reverse transcriptase inhibitor; NNRTI, non-nucleoside
reverse transcriptase inhibitor; 3TC, lamivudine; FTC, emtricitabine; LPV/r, lopinavir/ritonavir; NVP, nevirapine; TDF, tenofovir; ZDV, zidovudine.
Virological failure endpoints
The trials most likely to change clinical practice are
those where a new treatment regimen has improved
rates of HIV RNA suppression over existing control
regimens. For treatment-naive patients, such trials
include the AIDS Clinical Trials Group (ACTG) 5095
trial, showing virological inferiority of zidovudine
(ZDV)/lamivudine (3TC)/abacavir (ABC) relative to a
combined treatment group of ZDV/3TC/efavirenz
(EFV) and ZDV/3TC/ABC/EFV in naive patients [8],
and the Abbott 863 trial, showing a virological benefit
of stavudine (d4T)/3TC/lopinavir/r over d4T/3TC/nelfinavir [5]. For treatment-experienced patients, examples
include the Gilead 907 trial, showing a benefit for addition of tenofovir (TDF) for patients with NRTI resistance [9], and the TORO trials, showing a virological
benefit for the addition of enfuvirtide (ENF) to optimized background therapy in triple-class experienced
patients [10,11]. Improved virological efficacy may be
caused by (i) improved or more consistent drug levels
despite occasional non-adherence [5], (ii) a simpler
regimen with a lower pill burden improving adherence
[12], (iii) synergy between the drugs in the combination
used [13] and (iv) activity against drug-resistant virus
[10,11] or other factors. For patients with pre-existing
drug resistance, continuation of treatment could still be
beneficial [14], possibly owing to poor replication
kinetics of drug-resistant virus [15]. There is still the
potential for major improvements in treatment efficacy
for those with resistance to the currently available
treatment classes, given the relatively high rates of virological failure in the active arms of the trials conducted
so far (Table 2).
For treatment-naive patients, the proportion of
patients with virological failure has become so low that
a very large trial would be required to demonstrate
higher rates of virological suppression in one treatment
Antiviral Therapy 10:3
arm versus another. A recent meta-analysis showed no
trend for an additional benefit of four-drug HAART
over conventional three-drug HAART combinations
[16]. New trials for antiretroviral-naive patients typically do not detect statistically significant differences in
virological endpoints versus the current standard of
care (HAART including either a boosted PI or
NNRTI); however there is the potential for improved
or at least different safety profiles for the experimental
antiretroviral. The benefits and risks of new drug
classes (for example, the CCR5 antagonists) are
unknown and clinical trials may need to include new
efficacy and safety endpoints, such as the emergence of
X4 tropic virus during treatment and its effect on clinical disease progression.
Treatment discontinuation for adverse
events
Patients can discontinue randomized treatment for a
range of different adverse events, with either detectable
or undetectable HIV RNA at the time of discontinuation. HIV drug toxicity is multifactorial, with a range
of adverse events occurring after different durations of
treatment [17], from the early onset of the ABC hypersensitivity reaction to longer-term development of
lipodystrophy, pancreatitis or CNS disturbances. The
severity of drug toxicity may not correlate with discontinuation of treatment – for example, in the
MaxCmin 1 trial, 66% of patients who withdrew from
randomized treatment for adverse events had grade 1
or 2 events (mild or moderate severity, ACTG classification) [18]. Gastrointestinal symptoms such as
nausea, diarrhoea, abdominal pain and vomiting have
been a common cause of treatment discontinuation in
several trials [5,19].
Table 3 shows examples of trials where the degree of
virological suppression was similar between the two
369
A Hill & R DeMasi
Table 3. Trials showing an apparent discordance in treatment effects between ITT/S=F and as-treated analysis
Trial
Treatment arms
n
ITT/S=F, %
AT, %
Reference
Gilead 934
BEST
MaxCmin 1
MaxCmin 2
CNAB3005
TDF/FTC/EFV vs ZDV/3TC/EFV
Continued IDV vs IDV/r, with NRTIs
SQV/r vs IDV/r, with NRTIs
SQV/r vs LPV/r, with NRTIs
ZDV/3TC/ABC vs ZDV/3TC/IDV
517
323
306
324
562
88 vs 80*
58 vs 74*
57 vs 46*
52 vs 60
51 vs 51
98 vs 97
92 vs 93
79 vs 77
75 vs 70
86 vs 94†
[29]
[31]
[18]
[41]
[19]
All trials show percentage of randomized patients with undetectable HIV RNA levels at 48 weeks, except for the Gilead 934 trial, where 24-week interim data are
shown. HIV RNA: 50 copies/ml endpoint for MaxCMin 1 and 2 trials, 400 copies/ml endpoint for the Gilead 934 and CNAB3005 trials and 500 copies/ml endpoint for
the BEST trial. ITT analyses include all patients randomized to treatment. Methodology for as-treated analyses is not standardized. *P<0.05; †P=0.05. ITT/S=F, intentto-treat, switch equals failure; AT, as-treated; BEST, BID Efficacy and Safety Trial; NRTI, nucleoside reverse transcriptase inhibitor; 3TC, lamivudine; ABC, abacavir; EFV,
efavirenz; FTC, emtricitabine; IDV/r, indinavir/ritonavir; LPV/r, lopinavir/ritonavir; SQV/r, saquinavir/ritonavir; TDF, tenofovir; ZDV, zidovudine.
treatment arms, but differences in toxicity endpoints
appeared to cause the difference using the ITT/S=F
method. Examples include the MaxCMin 1 trial, where
the number of virological failure endpoints was similar
in the two arms but more patients in the boosted indinavir (IDV) arm discontinued treatment for toxicity
reasons, leading to an advantage of the boosted
saquinavir (SQV) arm in the ITT/S=F analysis [18].
A problem with ITT/S=F analysis is that an
improved safety profile might compensate for inferior
virological potency, leading to an apparently similar
result. This effect was observed for the CNAB3005
trial of ZDV/3TC/ABC versus ZDV/3TC/indinavir, a
non-inferiority trial (Table 3), which had a high rate of
withdrawal from randomized treatment in both arms
[19]. In the ITT/S=F analysis, the two treatments were
found to be equivalent. However, the IDV arm showed
a significantly higher percentage of patients with HIV
RNA undetectability when as-treated analyses were
used (Table 3).
In other trials with later-onset toxicity endpoints,
there may be no difference in discontinuations for toxicity reasons between arms, with similar or equivalent
efficacy shown in the ITT/S=F analysis, but one arm is
still judged to have improved safety. One example is the
144 week analysis of the Gilead 903 non-inferiority
trial, which compared d4T/3TC/efavirenz (EFV) with
TDF/3TC/EFV in naive patients. The ITT/S=F and astreated analyses showed very similar responses for the
two arms throughout the trial, but with an improved
lipid and lipodystrophy profile for the TDF arm [20].
In other cases, the toxicity profiles of two arms may
not differ significantly but allow individualization for
certain patient groups – for example, the 2NN study
showed similar virological responses for first-line
HAART with either EFV or nevirapine (NVP),
suggesting a potential use of NVP for people where
EFV would be unsuitable, such as pregnant women
with low CD4 counts or those with a pre-disposition to
CNS toxicities [21]. For open-label trials, toxicity
370
endpoints need to be checked carefully as they can be
assessed subjectively.
Treatment discontinuation for other reasons
As shown in Table 1, ‘discontinuation for other
reasons’ is the most common endpoint in clinical trials
of antiretroviral-naive patients, more common than
either true virological failure or withdrawal for adverse
events. This category includes a range of different
reasons for withdrawal: non-adherence, withdrawal of
consent and loss to follow-up. Most of these endpoints
are subjective, which can create interpretational difficulties in open-label randomized trials.
A small proportion of patients are randomized to a
treatment but discontinue from the trial before dosing
starts. ‘ITT exposed’ analysis includes only patients
who received at least one dose of study medication. In
the 2NN trial, there was a significant difference in efficacy between twice daily NVP and EFV in naive
patients when only patients who received at least one
dose of study drug were included (P=0.03) [21]; the
treatment arms were not significantly different, however,
when patients randomized but not treated were included
in the analysis (classified as treatment failures). Even so,
there may be situations where an imbalance between
treatment arms in pre-randomization discontinuation
could point to an inherent bias or lack of perceived
equipoise in a clinical trial.
Once dosing is initiated, patients could discontinue
treatment with either detectable or undetectable HIV
RNA; patients and investigators often know the HIV
RNA results at the time of discontinuation. Patients
can potentially stop a treatment at the first signs of
virological failure, even if a protocol-defined virological endpoint has not yet been reached. There is an
overlap between the toxicity of a treatment and its
acceptability – in one survey of patients taking
HAART, those with adverse events were 12.8 times less
likely to have 95–100% adherence than those without
© 2005 International Medical Press
Efficacy endpoints in HIV clinical trials
adverse events [22]. Patients in most clinical trial
centres have other treatment options outside a clinical
trial. Patients can clearly withdraw their consent to
continue randomized treatment at any time if they
judge that improved treatment could be obtained
outside the trial.
A large meta-analysis showed that HAART combinations in studies of naive patients with lower pill
burdens tended to have superior virological responses
compared with those with higher pill burdens [12].
Patient surveys show that adherence may be improved
from combinations with lower pill counts, smaller pill
sizes, fewer daily doses, fewer requirements for
concomitant food and fluids, or room temperature
storage. For open-label randomized trials, balancing
the pill count and dosage frequency across treatment
arms minimizes the potential effect of adherence-driven
outcomes in the ITT/S=F analysis.
Discontinuation of treatment may also depend on the
cultural environment in which the trial is conducted.
Active recreational drug use is one of several predictors
of poor adherence to treatment and low rates of HIV
RNA suppression [23,24], so conducting trials in
centres with a high proportion of active recreational
drug users (or other markers of poor adherence) may
lead to higher rates of treatment discontinuation.
Adherence counselling may improve adherence and
consequently overall efficacy [25]. A recent prospective
trial of HAART with boosted SQV in Thailand [26]
showed a withdrawal rate of 1%, whereas this rate was
18% for an international trial of ZDV/3TC/EFV [27].
This low failure rate may also be related to the treatment options available outside clinical trials for people
in developing or middle-income countries.
One example of these apparently subjective treatment discontinuations influencing trial outcomes was a
meta-analysis of randomized open-label trials of
continued PI treatment versus a switch to NRTI combinations (typically ZDV/3TC/ABC) for patients with
virological suppression at baseline. At the time the
trials were conducted, the potential advantages of this
switching away from PI-based HAART to triple NRTI
combinations included improved lipid profiles, a lower
pill burden or the avoidance of PI-related toxicities. In
a meta-analysis of three randomized trials, significantly
more patients discontinued PI-based HAART than
ZDV/3TC/ABC. However, the risk of virological
failure was significantly higher for patients randomized
to ZDV/3TC/ABC [28].
remain on their randomized treatment. This method
excludes potentially high numbers of patients withdrawing for toxicity or other reasons. Inclusion of these
patients can interfere with the evaluation of virological
efficacy, and so as-treated analyses may have higher
sensitivity to detect differences in virological potency
between treatments. Table 3 shows examples of discordance between the ITT/S=F and as-treated analyses of
randomized clinical trials. While the ITT results
presented are all conducted by an ITT/S=F approach,
the as-treated analysis method is not stated in most of
the trials and may not include data from patients who
had previously withdrawn from the trials for virological failure.
Any apparent discordance between results from ITT
and as-treated analyses should be explained. When a
treatment is superior to control in both ITT and astreated analyses, the explanation is simple: higher virological efficacy. When a treatment is superior in the
ITT/S=F but not the as-treated analysis, the explanation
may be improved safety and tolerability – this was seen
for the 24-week interim analysis of a comparative trial
of TDF/emitricabine/EFV versus ZDV/3TC/EFV [29],
shown in Table 3. In these cases, the reason for the
tolerability advantage should be identified, but this
may not always be possible.
A more standardized definition of as-treated analysis
is needed. In particular, patients who withdraw from a
trial with detectable HIV RNA levels should be classified as failures in as-treated analyses, by using either a
Kaplan–Meier approach or using the last observation
carried forward approach. Excluding patients with
prior virological failure from long-term as-treated
analysis could overestimate efficacy owing to a
‘survivor effect’, where only those patients with
continued virological suppression remain on their
randomized treatment. For clinical trials of treatmentexperienced patients where a minority of patients
achieve HIV RNA undetectability, the continuous
measure of log10 reduction in HIV RNA (and/or the
proportion of patients with at least a 1 log10 reduction)
provides higher statistical power to detect differences
between treatments than rates of undetectability. To
further investigate antiretroviral potency, subgroup
analyses can be conducted to analyse antiviral reductions for patients with the highest HIV RNA levels at
baseline (for example, over 100 000 copies/ml for trials
of naive patients). Alternatively, measures of adherence
could be used to restrict analysis to those known to be
receiving randomized medication.
As-treated analysis
The ‘ITT switch included’ approach
An as-treated analysis is also known as ‘on treatment’,
‘per protocol’ or ‘observed data’ analysis. As-treated
analysis includes only the data from patients who
Antiviral Therapy 10:3
For clinical trials where patients are randomized to a
planned series of treatments (including cross-over
371
A Hill & R DeMasi
designs) or different treatment strategies, the primary
endpoint could then be the long-term success of the
combined treatments received, rather than the success
of the initial randomized treatment in isolation.
Analyses which examine the overall outcome of a
sequence of treatments are often called ‘ITT switch
included’ (ITT/SI). For example, in the ACTG 384 trial,
patients were randomized to different pre-set sequences
of first-line and second-line NRTI/PI or NRTI/NNRTI
regimens – the primary protocol endpoint was the time
to failure of the second-line treatment [30]. ITT/SI
analysis could potentially detect effects of resistance
acquired during first-line treatment on response to
second-line options. This method could also help in
assessing long-term toxicities of an initial treatment if
these persist or worsen during subsequent treatments.
ITT/SI analyses will always yield HIV RNA efficacy
results that are at least as great as the corresponding
results using ITT/S=F, since responses on second-line
treatments can also be included as successes, depending
on the level of viraemia, whereas these are always
considered failures using ITT/S=F analysis.
However, conclusions from clinical trials analysed
using the ITT/SI method can differ considerably from
those in an ITT/S=F analysis. For example, in the BEST
trial [31] [a randomized comparison of continued IDV
versus a switch to IDV/ritonavir (RTV)] the treatment
arms differed significantly in the ITT/S=F analysis, but
were equivalent in the ITT/SI approach. More patients
discontinued IDV/RTV treatment but could be successfully salvaged with second-line options by week 48
(Table 4). Another example is the MaxCmin 1 trial,
where SQV/RTV showed significant benefits over
IDV/RTV in the ITT/S=F analysis, but not in the ITT/SI
approach – patients failing the IDV arm again could be
successfully salvaged with second-line options [18].
Since the ITT/S=F and ITT/SI methods are attempting
to answer different research questions, both have
validity in different situations. A conclusion from
ITT/S=F is only relevant for the initial randomized treatment and takes no account of future treatment options.
A conclusion from ITT/SI analysis is only relevant for
an overall sequence of treatments received and does not
distinguish the efficacy of the initial randomized treatment from responses on second-line options.
Given the above examples, there is significant potential for misinterpretation of trial results if the analytical
method being presented is not widely understood.
When presenting results from a clinical trial, ITT
analyses need to be clearly labelled, for example as
either ITT/S=F or ITT/SI, in order to convey the correct
interpretation. In any case, even if a trial is analysed
using the ITT/S=F method, there is still the need for
long-term follow-up of all patients on subsequent treatment regimens. Long-term follow-up after withdrawal
from randomized treatment could help to determine
the consequences of cumulative drug toxicities and
implications of drug resistance for second-line treatment options.
Conclusions and recommendations
ITT/S=F analysis is the current gold standard for analysis
and interpretation of HIV clinical trials for regulatory
approval. However, an advantage of a new treatment
over control in ITT/S=F analysis could be driven by
improved tolerability or patient acceptance, rather than
an intrinsic increase in antiretroviral potency. In addition, a high withdrawal rate in a trial analysed by
ITT/S=F could lead to ‘equivalence by default’.
As-treated analysis is a useful cross-check on the ITT
result, which might include only a minority of true
virological endpoints. As-treated analyses need to be
better standardized – one approach is to include all
patients who have withdrawn with detectable HIV
RNA levels as failures and to exclude only patients
who withdrew with undetectable HIV RNA levels.
For any clinical trial, further follow-up after withdrawal from randomized treatment is important to
evaluate the implications for long-term toxicity and
cross resistance.
The ITT/SI method may help to evaluate trials
comparing pre-set sequences of treatments or treatment
strategies. The statistical power of this method is
Table 4. The BEST Trial – percentage of randomized patients with HIV RNA <500 copies/ml at week 48 using different analyses [31]
Treatment arm/result
IDV
IDV/r
P value
Sample size, n
ITT/S=F method, HIV RNA <500 copies/ml, %
ITT/SI method, HIV RNA <500 copies/ml, %
As-treated analysis, HIV RNA <500 copies/ml, %
Discontinuation for adverse events, n
Virological failure, n
161
74
86
92
20
7
162
58
88
93
47
6
P=0.003
P=NS
P=NS
P<0.001
P=NS
ITT/S=F, ITT switch equals failure; ITT/SI, ITT switch included; NS, not significant; IDV/r, indinavir/ritonavir.
372
© 2005 International Medical Press
Efficacy endpoints in HIV clinical trials
limited where second-line treatments are not well
defined. The ITT/SI method is not suitable as the
primary analysis of new experimental treatments, since
failure of the initial experimental treatment could be
hidden by successful salvage using approved secondline options.
ITT analyses should be more clearly labelled,
showing whether the ‘switch equals failure’ or ‘switch
included’ method is being used. The two methods have
very different meanings and need to be interpreted
correctly.
Disclosures
Andrew Hill has received honoraria from Roche and
Tibotec. Ralph DeMasi is an employee of Trimeris, Inc.
References
1.
Yeni PG, Hammer SM, Hirsch MS, Saag MS, Schechter M,
Carpenter CC, Fischl MA, Gatell JM, Gazzard BG,
Jacobsen DM, Katzenstein DA, Montaner JS, Richman
DD, Schooley RT, Thompson MA, Vella S & Volberding
PA. Treatment for adult HIV infection: 2004 recommendations of the International AIDS Society-USA Panel. Journal
of the American Medical Association 2004; 292:251–265.
2.
World Health Organisation. Scaling up Antiretroviral
Therapy in Resource-Limited Settings: Treatment
Guidelines for a Public Health Approach. 2004.
3.
Pozniak A, Gazzard B, Anderson J, Babiker A, Churchill
D, Collins S, Fisher M, Johnson M, Khoo S, Leen C,
Loveday C, Moyle G, Nelson M, Peter B, Phillips A, Pillay
D, Wilkins E, Williams I & Youle M; BHIVA Writing
Committee BHIVA Executive Committee. British HIV
Association (BHIVA) guidelines for the treatment of HIVinfected adults with antiretroviral therapy. HIV Medicine
2003; 4(Suppl 1):1–41.
4.
Murray JS, Elashoff MR, Iacono-Connors LC, Cvetkovich
TA & Struble KA. The use of plasma HIV RNA as a study
endpoint in efficacy trials of antiretroviral drugs. AIDS
1999; 13:797–804.
5.
Walmsley S, Bernstein B, King M, Arribas J, Beall G,
Ruane P, Johnson M, Johnson D, Lalonde R, Japour A,
Brun S & Sun E; M98-863 Study Team. Lopinavir-ritonavir versus nelfinavir for the initial treatment of HIV
infection. New England Journal of Medicine 2002;
346:2039–2046.
6.
Food and Drug Administration. Guidance for Industry:
Antiretroviral Drugs Using Plasma HIV RNA
Measurements – Clinical Considerations for Accelerated
and Traditional Approval. US Department of Health and
Human Services, Center for Drug Evaluation and Research
(CDER), Clinical Medical. October 2002.
http://www.fda.gov/cder/guidance/index.htm
7.
8.
9.
Phillips A & Walker S. Drug switching and virologic based
endpoints in trials of antiretroviral drugs for HIV infection. AIDS 2004; 18:365–370.
Gulick RM, Ribaudo HJ, Shikuma CM, Lustgarten S,
Squires KE, Meyer WA 3rd, Acosta EP, Schackman BR,
Pilcher CD, Murphy RL, Maher WE, Witt MD, Reichman
RC, Snyder S, Klingman KL & Kuritzkes DR; AIDS
Clinical Trials Group Study A5095 Team. Triple-nucleoside regimens versus efavirenz-containing regimens for the
initial treatment of HIV-1 infection. New England Journal
of Medicine 2004; 350:1850–1861.
Squires K, Pozniak AL, Pierone G Jr, Steinhart CR, Berger
D, Bellos NC, Becker SL, Wulfsohn M, Miller MD, Toole
JJ, Coakley DF & Cheng A; Study 907 Team. Tenofovir
disoproxil fumarate in nucleoside-resistant HIV-1 infec-
Antiviral Therapy 10:3
tion: a randomized trial. Annals of Internal Medicine 2003;
139:313–320.
10. Lalezari JP, Henry K, O’Hearn M, Montaner JS, Piliero PJ,
Trottier B, Walmsley S, Cohen C, Kuritzkes DR, Eron JJ Jr,
Chung J, DeMasi R, Donatacci L, Drobnes C, Delehanty J
& Salgo M; TORO 1 Study Group. Enfuvirtide, an HIV-1
fusion inhibitor, for drug-resistant HIV infection in North
and South America. New England Journal of Medicine
2003; 348:2175–2185.
11. Lazzarin A, Clotet B, Cooper D, Reynes J, Arasteh K,
Nelson M, Katlama C, Stellbrink HJ, Delfraissy JF, Lange
J, Huson L, DeMasi R, Wat C, Delehanty J, Drobnes C &
Salgo M; TORO 2 Study Group. Efficacy of enfuvirtide in
patients infected with drug-resistant human immunodeficiency virus type 1 in Europe and Australia. New England
Journal of Medicine 2003; 348:2186–2195.
12. Bartlett J, DeMasi R, Quinn J, Moxham C & Rousseau F.
Overview of the effectiveness of triple combination therapy
in antiretroviral-naive HIV-1 infected adults. AIDS 2001;
15:1369–1377.
13. Snyder S, D’Argenio DZ, Weislow O, Bilello JA & Drusano
GL. The triple combination indinavir-zidovudine-lamivudine
is highly synergistic. Antimicrobial Agents &
Chemotherapy 2000; 44:1051–1058.
14. Ledergerber B, Lundgren JD, Walker AS, Sabin C, Justice
A, Reiss P, Mussini C, Wit F, d’Arminio Monforte A,
Weber R, Fusco G, Staszewski S, Law M, Hogg R, Lampe
F, Gill MJ, Castelli F & Phillips AN; PLATO
Collaboration. Predictors of trend in CD4-positive T cell
count and mortality among HIV-1 infected individuals with
virological failure to all three antiretroviral-drug classes.
Lancet 2004; 364:51–62.
15. Sune C, Brennan L, Stover DR & Klimkait T. Effect of
polymorphisms on the replicative capacity of protease
inhibitor-resistant HIV-1 variants under drug pressure.
Clinical Microbiology & Infection 2004; 10:119–126.
16. Moyle G & Hill A. A meta-analysis of the virological efficacy of regimens containing four versus three active
antiretroviral agents as initial therapy for HIV-1 infection.
15th International AIDS Conference. 11–16 July 2004,
Bangkok, Thailand. Abstract TuPeB4519.
17. Fellay J, Boubaker K, Ledergerber B, Bernasconi E, Furrer
H, Battegay M, Hirschel B, Vernazza P, Francioli P, Greub
G, Flepp M & Telenti A; Swiss HIV Cohort Study.
Prevalence of adverse events associated with potent antiretroviral treatment: Swiss HIV cohort study. Lancet 2001;
358:1322–1327.
18. Dragsted UB, Gerstoft J, Pedersen C, Peters B, Duran A,
Obel N, Castagna A, Cahn P, Clumeck N, Bruun JN,
Benetucci J, Hill A, Cassetti I, Vernazza P, Youle M, Fox Z
& Lundgren JD; MaxCmin1 Trial Group. Randomized
trial to evaluate indinavir/ritonavir versus saquinavir/ritonavir in human immunodeficiency virus type 1 infected
patients: the MaxCmin 1 trial. Journal of Infectious
Diseases 2003; 188:635–642.
19. Staszewski S, Morales-Ramirez J, Tashima KT, Rachlis A,
Skiest D, Stanford J, Stryker R, Johnson P, Labriola DF,
Farina D, Manion DJ & Ruiz NM. Efavirenz plus zidovudine and lamivudine, efavirenz plus indinavir, and indinavir
plus zidovudine and lamivudine in the treatment of HIV-1
infection in adults. Study 006 Team. New England Journal
of Medicine 1999; 341:1865–1873.
20. Gallant JE, Staszewski S, Pozniak AL, DeJesus E, Suleiman
JM, Miller MD, Coakley DF, Lu B, Toole JJ & Cheng AK;
903 Study Group. Efficacy and safety of tenofovir DF vs
stavudine in combination therapy in antiretroviral-naive
patients: a 3-year randomized trial. Journal of the
American Medical Association 2004; 292:191–201.
21. van Leth F, Phanuphak P, Ruxrungtham K, Baraldi E,
Miller S, Gazzard B, Cahn P, Lalloo UG, van der
Westhuizen IP, Malan DR, Johnson MA, Santos BR,
Mulcahy F, Wood R, Levi GC, Reboredo G, Squires K,
Cassetti I, Petit D, Raffi F, Katlama C, Murphy RL,
Horban A, Dam JP, Hassink E, van Leeuwen R, Robinson
P, Wit FW & Lange JM; 2NN Study Team. Comparison of
first-line antiretroviral therapy with regimens including
373
A Hill & R DeMasi
nevirapine, efavirenz, or both drugs, plus stavudine and
lamivudine: a randomized open-label trial, the 2NN Study.
Lancet 2004; 363:1253–1263.
23. Arnsten JH, Demas PA, Grant RW, Gourevitch MN,
Farzadegan H, Howard AA & Schoenbaum EE. Impact of
active drug use on antiretroviral therapy adherence and
viral suppression in HIV-infected drug users. Journal of
General Internal Medicine 2002; 17:377–381.
24. Kleeberger CA, Buechner J, Palella F, Detels R, Riddler S,
Godfrey R & Jacobson LP. Changes in adherence to highly
active antiretroviral therapy medications in the Multicenter
AIDS Cohort Study. AIDS 2004; 18:683–688.
25. Kleeberger CA, Buechner J, Palella F, Detels R, Riddler S,
Godfrey R & Jacobson LP. Enhancing adherence to combination antiretroviral therapy in non-adherent HIV-positive
men. AIDS Care 2000; 12:399–404.
Agents & Chemotherapy. 30 October–2 November 2004,
Washington, DC, USA. Abstract H-1137c.
30. Robbins GK, De Gruttola V, Shafer RW, Smeaton LM,
Snyder SW, Pettinelli C, Dube MP, Fischl MA, Pollard RB,
Delapenha R, Gedeon L, van der Horst C, Murphy RL,
Becker MI, D’Aquila RT, Vella S, Merigan TC & Hirsch MS;
AIDS Clinical Trials Group 384 Team. Comparison of
sequential three-drug regimens as initial therapy for HIV-1
infection. New England Journal of Medicine 2003;
349:2293–2303.
31. Arnaiz JA, Mallolas J, Podzamczer D, Gerstoft J, Lundgren
JD, Cahn P, Fatkenheuer G, D’Arminio-Monforte A, Casiro
A, Reiss P, Burger DM, Stek M & Gatell JM; BEST Study
Team. Continued indinavir versus switching to
indinavir/ritonavir in HIV-infected patients with suppressed
viral load. AIDS 2003; 17:831–840.
26. Ananworanich J, Ruxrungtham K, Siangphoe U,
Chetchotisakd P, Prasithsirikul W, Kiertiburanakul S,
Munsakul W, Klinbuayam W, Tansuphasawadikul S,
Nuesch R, Ubolyam S, Yakasem S, Hill A, Cooper D &
Hirschel B. A prospective cohort study of efficacy and
safety of 2 NRTIs plus once-daily ritonavir boosted
saquinavir hard gel capsule (SQV-HGC/R) at 24 weeks.
15th International AIDS Conference. 11–16 July 2004,
Bangkok, Thailand. Abstract TuPeB4469.
32. Bristol-Myers Squibb. Efavirenz (Sustiva) US Prescribing
Information. September 2004. www.sustiva.com
27. DeJesus E, McCarty D, Farthing CF, Shortino DD,
Grinsztejn B, Thomas DA, Schrader SR, Castillo SA,
Sension MG, Gough K & Madison SJ; EPV20001
International Study Team. Once-daily versus twice-daily
lamivudine, in combination with zidovudine and efavirenz,
for the treatment of antiretroviral-naive adults with HIV
infection: a randomized equivalence trial. Clinical
Infectious Diseases 2004; 39:411–418.
36. GlaxoSmithKilne. Abacavir (Ziagen) US Prescribing
Information. August 2004. www.ziagen.com
37. GlaxoSmithKilne/Vertex. FosAmprenavir (Lexiva) US
Prescribing Information. May 2004. www.lexiva.com
28. Bucher HC, Kofler A, Nuesch R, Young J, Battegay M &
Opravil M. Meta-analysis of randomized controlled trials
of simplified versus continued protease inhibitor-based antiretroviral therapy in HIV-1-infected patients. AIDS 2003;
17:2451–2459.
40. Roche Pharmaceuticals/Trimeris. Enfuvirtide (Fuzeon) US
Prescribing Information. March 2003.
29. Gazzard B, DeJesus E, Campo R, Jemsek J, Gallant JE,
Arribas JR, Pozniak AL, Lu B, Miller MD, Enejosa J &
Cheng A. The combination of tenofovir DF (TDF), emtricitabine (FTC) and efavirenz (EFV) has significantly greater
response vs fixed dose zidovudine/lamivudine(CBV) and
EFV in antiretroviral naive patients: a 24 week preliminary
analysis. 44th Interscience Conference on Antimicrobial
33. Gilead Sciences, Inc. Emitricabine (Emitriva, FTC) US
Prescribing Information. July 2003. www.emtriva.com
34. GlaxoSmithKline. Lamivudine (Epivir) US Prescribing
Information. May 2004. www.epivir.com
35. Gilead Sciences, Inc. Tenofovir (Viread) US Prescribing
Information. June 2004. www.viread.com
38. Abbott Laboratories. Lopinavir/ritoanvir (Kaletra) US
Prescribing Information. August 2004. www.kaletra.com
39. Bristol-Myers Squibb. Atazanavir (Reyataz) US Prescribing
Information. Revised July 2004. www.reyataz.com
41. Youle M, Gerstoft J, Fox Z, Losso M, Jayaweera DT,
Rieger A, Brunn JN, Castagna A, Walmsley S, Hill A,
Dragsted UB & Lundgren JD. The final 48 week analysis of
a phase IV, randomized, open-label, multicentre trial to evaluate safety and efficacy of lopinavir/ritonavir (400/100 mg
BID) versus saquinavir/ritonavir (1000/100 mg BID): the
MaxCmin 2 trial. 9th European AIDS Conference. 25–29
October 2003, Warsaw, Poland. Abstract F11/3.
Received 3 December 2004, accepted 7 March 2005
374
© 2005 International Medical Press