Download "MRS Studies of Creatine Kinase Metabolism in Human Heart" in

MRS Studies of Creatine Kinase Metabolism in Human Heart
Paul A. Bottomley
Johns Hopkins University, Baltimore, MD, USA
The heart is the largest consumer of energy per gram of tissue and the creatine kinase (CK) reaction is its primary cellular energy reserve, providing
adenosine triphosphate (ATP) to fuel contraction by shuttling phosphocreatine (PCr) from the mitochondria. For 30 years the measurement of the
myocardial PCr/ATP ratio has been a common focus of human cardiac phosphorus (31 P) magnetic resonance spectroscopy (MRS). Today, our ability to
measure the absolute concentrations of PCr and ATP, the CK reaction rate and flux with 31 P MRS, and the total creatine pool (CR) using proton (1 H)
MRS, allows almost complete characterization of CK metabolism in the healthy and diseased human heart. The methods and limitations of human
cardiac MRS for measuring CK metabolism are reviewed herein, along with the application and findings in ‘normal’ physiological processes including
aging, diet, sedentary lifestyle, hypoxia, and obesity. The results from studies of patients with myocardial infarction, ischemia, dilated and hypertrophic
cardiomyopathies, valve disease, heart failure, diabetes, and other disorders are summarized, with the quantitative findings tabulated. The myocardial
PCr/ATP ratio is a sensitive indicator of cardiac energy reserve, existing in a meta-stable state that can be upset in many disorders. Reductions in
metabolite concentrations and the forward flux for delivering ATP are associated with myocardial infarction, congestive heart failure, cardiomyopathy,
and disease severity. Moreover, metrics of CK energy reserve and supply can independently predict long-term cardiovascular outcomes, and are now
being used to quantify the effect of pharmaceutical and lifestyle intervention on the heart’s energy budget.
Keywords: heart, magnetic resonance spectroscopy (MRS), energy metabolism, quantification, heart disease, cardiomyopathy, heart failure, aging,
diabetes, creatine kinase reaction
How to cite this article:
eMagRes, 2016, Vol 5: 1183–1202. DOI 10.1002/9780470034590.emrstm1488
Introduction
(CK reaction):
The heart is the largest consumer of energy per gram of tissue,
and disruptions in energy metabolism, including energy supply
and demand, are thought to play a central role in many common diseases of the heart.1,2 It was therefore inevitable that the
first human in vivo cardiac magnetic resonance spectroscopy
(MRS) in 19853,4 focused on those energy metabolites that
were shown by earlier animal studies5–8 (see also Cardiac MRS
Studies in Rodents and Other Animals) to be NMR detectable.
Human cardiac MRS required the development of high-field,
high-homogeneity magnet systems capable of accommodating
the torso.9 The performance advantages of such systems for
MRI, the sister technology of MRS,10 has so entwined the two
that human cardiac MRS is today inconceivable without MRI
guidance and gradient localization.
Human phosphorus (31 P) MRS can noninvasively detect
endogenous adenosine triphosphate (ATP), the energy that
fuels the cardiac pump via the ATP-ase reaction,1
ATP ⇐⇒ ADP + Pi + ⟨energy⟩.
(1)
Inorganic phosphate (Pi) is a by-product whose pHdependent chemical shift provides a measure of intracellular
pH.11,12 It can measure phosphocreatine (PCr),2–7 which serves
as the heart’s primary energy reserve. ATP is synthesized from
PCr and adenosine diphosphate (ADP) via the creatine kinase
Volume 5, 2016
PCr + ADP + H+k−1 ⇐⇒k1 ATP + Cr
(2)
releasing unphosphorylated creatine (Cr). It can even measure
the forward (k1 ) and reverse (k−1 ) reaction rate constants
(see Measuring Biochemical Reaction Rates In Vivo with
Magnetization Transfer).13 Switching to hydrogen (1 H) MRS
allows documentation of intra- and extra-myocellular lipid
fuel deposits in the myocardium,14,15 whose accumulation is
linked to cardiac dysfunction, obesity, and diabetes. 1 H MRS
can also be used to measure the total creatine pool (CR),
comprising PCr plus Cr.16–18 Thus, in combination, 31 P and
1 H MRS could almost completely characterize CK metabolism
and energy supply in the human heart.16 The depletion of CR
in heart failure17 and in myocardial infarction (MI),18 which
is routinely diagnosed by an elevated serum concentration of
myocardial CK enzyme, suggests its potential as an in situ 1 H
MRS metabolic marker. Meanwhile, significant reductions in
CK energy flux measured by 31 P MRS suggest a mechanistic
role for CK metabolism in the failing heart (see MRS in the
Failing Heart: From Mice to Humans)13,19 that can be directly
targeted for therapy.20
Access to the myocyte’s metabolic machinery for producing
ATP energy via oxidative phosphorylation (OXPHOS) using
carbon (13 C) MRS is limited by the low isotopic abundance
of 13 C, although it might be useful for seeing endogenous
lipids and perhaps glycogen in the human heart.21 However,
© 2016 John Wiley & Sons, Ltd.
1183
PA Bottomley
Myofibril
Cytosol
Mitochondria
OXPHOS
ATP
Cr
CK
ATP
Cr
CK
CK
k1
ADP + Pi
Energy sink
PCr
PCr
ADP + Pi
Energy source
myocytes.23,24
Figure 1. Illustration of the CK shuttle hypothesis in
ATP
generated by oxidative phosphorylation (OXPHOS) in the mitochondria
(right) phosphorylate Cr via the CK reaction, producing PCr. The process may continue in the cytosol until PCr arriving at the myofibrils (left),
combines with ADP via the CK reaction to recreate ATP to fuel an ensuing contraction. Unphosphorylated Cr disperses to the mitochondria for
rephosphorylation, resulting in a net transfer of biochemical energy from
mitochondria to myofibrils. The CK pseudo first-order reaction rate, k1 ,
measured by 31 P MRS saturation transfer,13 reflects the tissue average transfer rate (center)
recent approaches employing exogenous, hyperpolarized, 13 Cenriched metabolic substrates offer new minimally invasive
tools for probing OXPHOS ATP synthesis in future human
studies.22
Human 1 H MRS studies of cardiac lipids, 13 C MRS, hyperpolarization, and cardiac MRS in animals are reviewed elsewhere
in this volume (see Cardiac Lipids by 1 H MRS; Integration
of 13 C Isotopomer Methods and Hyperpolarization Provides
a Comprehensive Picture of Metabolism; Hyperpolarization
Methods for MRS; Cardiac MRS Studies in Rodents and Other
Animals). The present article focuses on CK metabolism, which
involves 31 P MRS except for the use of 1 H MRS to measure
CR.16–18 The importance of CK metabolism stems from its
putative role as an intracellular spatial energy shuttle and
temporal buffer to facilitate high-energy phosphate transfer – in the form of PCr – from the mitochondria where ATP
is produced by OXPHOS, to the myofibrils to fuel muscular
contraction (Figure 1).23,24 The by-product, Cr, ultimately
returns to the mitochondria for rephosphorylation. This role is
supported by observations that PCr begins to deplete almost
immediately to maintain ATP levels at the onset of critical
ischemia or hypoxia,1 and that the resting CK rate is many-fold
faster than OXPHOS,13 as would be required for a temporal
buffer. Accordingly, the ratio of reactant to product, PCr/ATP,
has long been considered a sensitive index of energy reserve,
and the PCr/ATP ratio is by far the most measured parameter
in cardiac MRS.
Since an earlier human cardiac MRS review in 2008,25 the
number of papers reporting PCr/ATP values in cohorts of
healthy controls has doubled to over 110 studies of more than
1600 volunteers. The NMR field strength (B0 ) for human cardiac studies has been extended from 1.5 to 7 T (Figure 2).26–28
Protocols for quantifying absolute concentrations of the CK
substrates, [PCr] and [ATP],29 the CK rate constant k1 and
1184
the flux of the ATP supply, k1 ⋅[PCr], have also been worked
out for human cardiac 31 P MRS at 3 T.30–32 The short take
in toto is that one or more of these parameters often show
profound differences in acute33 and chronic heart disease and
may independently predict cardiac events and outcomes.34,35
Cardiac MRS methods and results are now reviewed, starting
from the limitations imposed by the low signal-to-noise ratio
(SNR) of CK metabolites. The methods of metabolite detection,
localization, and artefacts are then summarized, and the ranges
of CK measures in the healthy and diseased human heart
presented.
Methods
Signal-to-noise Ratio (SNR)
SNR Relative to 1 H MRS. The SNR of a metabolite with an MRS
X
, per unit volume of ‘X-nuclei’
moiety bearing a number, Nml
with spin IX can be deduced from The Basics (Section 5;
Equations 32 and 33 therein). If a set of NX transients are
detected and averaged from a tissue volume VX using a receiver
with bandwidth BWX at an NMR frequency 𝜈 X , the SNR
relative to a 1 H NMR experiment (subscripts H) at the same
B0 is
√
X
𝜈X MX VX Nml
|B1X | NX RH BWH
SNRX
√
=
× 10(NFH −NFX )∕20
H
SNRH
𝜈H MH VH Nml
|B1H | NH RX BWX
(3)
Here the M, B1 , and R are, respectively, the nuclear magnetization per Tesla per unit volume, the magnitude of the circularly
polarized transverse RF magnetic field produced at the volume
by the detector excited with unit current, and its RF resistance
when loaded by the sample. The NFs are the MRI/MRS system noise figures at the two frequencies. Assuming that the
noise is sample dominant, and that the losses are inductive at
36 and derive from the same sample volume,
both
√ frequencies
√
RH ∕ RX = 𝜈H ∕𝜈X . But because
X 2
MX
Nml
𝜈X IX (IX + 1)
= H
MH
Nml 𝜈H2 IX (IH + 1)
(4)
for spin- 1∕2 nuclei, equation (3) reduces to
√
X
|B1X | NX BWH
𝜈X2 VX Nml
SNRX
√
= 2
H
SNRH
𝜈H VH Nml
|B1H | NH BWX
√
X
N
𝜈X2 VX Nml
√ X
= 2
H
𝜈H VH Nml NH
(5)
(6)
Equation (6) assumes that the detector coil efficiencies (B1 per
ampere) and receiver bandwidths are the same at the two frequencies. These expressions include the intrinsic difference in
NMR sensitivity of nucleus X versus 1 H (via the 𝜈 2 terms), but
ignore differences in the system NF, as well as NMR excitation
and relaxation effects.
Cardiac MRS of endogenous metabolism in humans is
limited by low metabolite SNR primarily due to low metabolite
X
concentrations, as reflected by the Nml
term above. In the
healthy heart, [ATP], [PCr], and [CR] are present at about 6,
10, and 30 μmol g−1 wet weight, respectively.25 This compares
© 2016 John Wiley & Sons, Ltd.
Volume 5, 2016
MRS Studies of Creatine Kinase Metabolism in Human Heart
3T
PCr
40
7T
γ-ATP
30
α-ATP
20
2,3-DPG
(b)
10
β-ATP
PDE
10
0
0
10
(a)
5
0
−5
−10
−15
ppm
(c)
Figure 2. Comparison of typical cardiac 31 P spectra at 3 T (blue) and 7 T (red) from a 5.6 ml voxel in the septum. (a) Spectra are apodized with an exponential filter matched to the PCr line width and scaled to illustrate the SNR advantage (y-axes offset for clarity; 2,3-DPG, blood 2,3-diphosphoglycerate; PD,
phosphodiesters). (b) Shows short-axis, and (c) four-chamber long-axis cine localizer images acquired at 7 T, and overlaid with the position of the septal
voxel marked in red. The yellow stripe shows the location of a saturation band used to minimize chest contamination. (Reproduced from Ref. 26. © John
Wiley & Sons, Ltd., 2014)
to about 43 mmol g−1 for tissue water (77% by weight), which
has a 1 H concentration of 2 × 43 = 86 mmol g−1 . For PCr and
X
H
ATP, Nml
∕Nml
are thus 1.2 × 10−4 and 7 × 10−5 , respectively.
After accounting for the lower sensitivity of 31 P [borne by the
quadratic frequency terms in equation (5)], the SNR of PCr
in a cardiac 31 P MRS experiment is about 1.9 × 10−5 times
lower than an equivalent 1 H NMR experiment on tissue water
(assuming VX /VH = BWH /BWX = B1X /B1H = 1) at the same B0 .
A 1 H MRS CR experiment performed on the 3.0 ppm moiety
with its 3 protons will have 45 times higher SNR than the 31 P
MRS PCr experiment. Nevertheless, CR SNR is still 9 × 10−4
times lower than tissue water. These enormous SNR deficits
must be made up somehow.
Negotiating the Low-SNR Experiment. Since increasing
the B0 by orders of magnitude cannot be contemplated
[and would preferentially advantage 1 H MRS due to the
quadratic frequency dependencies in equation (5)], other
sacrifices must be made, and spatial resolution is the
prime target. Choosing a 10 ml 31 P MRS voxel instead of
a 0.05 × 0.05 × 0.3 = 7.5 × 10−4 ml 1 H MRI pixel buys a factor
of about 13 000, bringing the shortfall to within a factor of
4 of tissue water, for example. It is only after this sacrifice
that increasing B0 or scan time (or number of averages, NX )
by a factor of 2 or 3 can have much practical effect on SNR.
Bandwidth is also an easy target for improving SNR, especially
relative to 1 H MRI whose bandwidths (BWH ) are often 16 kHz
or higher. For a cardiac MRS experiment, equation (5) shows
that setting BWX ∼ 1 kHz compared to 16 kHz can increase
SNR fourfold. NF should also not be overlooked. Non-1 H
MRS is often an afterthought in commercial MRI scanners:
the author has measured >2-fold losses in cardiac 31 P MRS
SNR due to insufficient preamplification or inadequate digital
Volume 5, 2016
noise suppression in two different leading brands of MRI/MRS
systems operating at 3 and 1.5 T (see The Basics and Magnetic
Resonance Spectroscopy Instrumentation on how to measure
system NF). Given the huge cost and siting requirements of
MRI/MRS magnets, the option of increasing B0 to obtain
higher SNR is really only justifiable after all the other factors in
equation (5) have been optimized.
At the scanner, the total examination time is typically limited by patient tolerance to about an hour. Allowing 20 min
for positioning, performing rudimentary MRI to guide MRS,
plus setting up MRI/MRS, means that the total number of
NX transients that can be acquired can consume at most,
about 40 min. To detect a 20% change in an MRS metric of
interest – say, PCr/ATP or [PCr] – requires an SNR > 5 per
data set. If multiple data sets are needed to determine that
metric within said 40 min acquisition window – say, four for
a CK reaction rate13 or a metabolite spin–lattice relaxation
(T1 ) measurement, or for reference scans for quantification29
or studies involving rest and exercise33 or pharmaceutical
intervention20 – then this SNR must often be realized in under
10 min. The spatial localization method and study protocol
must be tailored accordingly lest the study fail to achieve its
objectives.
Nuclear Overhauser Enhancement (nOe). The SNR of cardiac MRS performed on non-1 H nuclei such as 13 C and 31 P
can be improved by the combined effect of 1 H nOe (nuclear
overhauser enhancement) and 1 H-decoupling21,37–41 (see
The Basics). This technique requires the addition of an RF
excitation channel that provides an essentially continuous
low-level 1 H irradiation via associated 1 H excitation coils, and
electronics to isolate the 1 H nOe/decoupling power from the
31 P (or 13 C) receiver.21,38 At 1.5 T, the reported 31 P MRS nOe
© 2016 John Wiley & Sons, Ltd.
1185
PA Bottomley
for PCr, 𝛾-ATP and 𝛽-ATP in the human heart are, respectively, 𝜂(= the fractional increase in signal) = 0.61 ± 0.25 SD
(standard deviation), 0.6 ± 0.3, and 0.3 ± 0.2. Thus, SNR gains
of up to 1.6-fold are possible.38,40 Note that the nOe is likely B0
dependent. At 3 T, SNR gains from nOe of 13–43% are reported
(mean PCr gain, 1.23-fold, ±0.14; 𝛾-ATP, 1.27 ± 0.35; 𝛽-ATP,
1.14 ± 0.29).41
Cardiac MRS Detection Coils
Small ‘surface coils’ implanted directly about the heart in
intact animals7 or on isolated perfused hearts8 can be used
to study in vivo high-energy phosphate metabolism during
ischemia and monitor response to therapy. Indeed, all human
cardiac 31 P MRS performed since the first studies in the 1980s
have employed surface detection coils placed on the chest
closest to the anterior myocardium.3 Such spectra are typically
dominated by signals originating close to the coil due to its
nonuniform sensitivity profile (see Surface Coil NMR: Detection with Inhomogeneous Radiofrequency Field Antennas;
Surface and Other Local Coils for In Vivo Studies; Detection
Coils for MRS; Quantifying Metabolite Ratios and Concentrations by Non-1 H MRS). The nonuniform sensitivity also
affords surface coil-based detectors an optimum SNR when
the noise is sample dominant (which is typical for human torso
studies), by limiting the effective volume of tissue that can
contribute noise (see √
The Basics). This SNR advantage derives
primarily from the 1/ R dependence of equation (3).
The optimum SNR at a region of interest (ROI) lying at depth
√
d on the axis of a tuned loop detector, has a radius a = d/ 5.
Given approximate depths of the heart from the chest of
7–15 cm, this translates to an optimum detector comprised
of about 7–15 cm diameter surface coils.42,43 A multichannel
phased array comprised of such coils and arranged against the
chest adjacent to the heart, can yield, in practice, about 80% of
the ultimate intrinsic SNR (the maximum SNR that can be had
by any design assuming zero detector noise).42 This translates
to SNR gains of 6- to 10-fold compared to 1 H body coils.
This has not gone unnoticed in cardiac 1 H MRI where torso
arrays employing similar principles are routinely used. Cardiac
MRS that does not use similar technology may thus be further
SNR-disadvantaged relative to state-of-the-art cardiac 1 H MRI.
Bearing in mind that it is always possible to make√a lousy
detector of any geometry, in addition to (i) the a = d/ 5 rule,
other ‘rules of thumb’ for cardiac MRS detectors are that:
(ii)
(iii)
(iv)
(v)
1186
Noncircular coils (square, etc.), and ‘figure-8’ or ‘butterfly’ coils are less-efficient and will generally have lower
SNR than circular loops.
A ‘quadrature’ coil pair comprised of a figure-8 coil
plus circular loop will generally underperform a simple
phased array comprised of loop coils.
A multiturn (distributed capacitance) loop coil will generally outperform a single-turn (distributed capacitance)
loop coil, if they are both operating well below the coil’s
self resonance frequency.
Finally, because the heart is closer to the surface of the
chest when a subject is oriented prone compared to a
supine orientation, and due to the nonuniform sensitivity
of the optimized surface coil, a prone orientation can yield
2–3 times the SNR achieved by a supine orientation.43,44
Localization in Cardiac MRS
31 P
MRS. By itself, a surface coil placed on the chest is unsuitable for localizing MRS signals to a heart because the detected
spectrum will be dominated by signals from the intervening
superficial chest muscle. The first localized 31 P MRS of the
human heart was performed using depth resolved surface coil
spectroscopy (DRESS) with slice-selective excitation and MRI
gradients positioned using 1 H MRI guidance.3 Cardiac 31 P
MRS studies performed since 2008 have utilized the following
methods:
1. one-, two-, and three-dimensional (1-D, 2-D, and 3-D)
chemical shift imaging (CSI; see CSI and SENSE CSI and
Quantifying Metabolite Ratios and Concentrations by
Non-1 H MRS)33,45,46 ;
2. 1-D and 3-D image-selected in vivo spectroscopy (ISIS; see
Single-Voxel MR Spectroscopy)47 ;
3. 3-D CSI augmented at 3 T with 31 P saturation bands48 or
‘crusher’ gradients applied using a surface spoiling gradient coil at 7 T,28 to minimize contamination from chest
muscle and surrounding tissue;
4. 3-D CSI using the spatial localization with optimum pointspread function (SLOOP) method (see Accurate and Efficient Localized Spectroscopy from Anatomically Matched
Regions: SLOOP and Its Enhancements)40,49 ; and
5. spectroscopy with linear algebraic modeling (SLAM; see
Accelerated Spatially Encoded Spectroscopy of Arbitrarily Shaped Compartments Using Prior Knowledge and
Linear Algebraic Modeling).50
The CSI experiments are preferably ‘acquisition weighted’ by
collecting more signals with the lower order phase-encoding
gradients than with the higher order gradients in a graded
fashion,51 or by using a Fourier series window (FSW)52 strategy,
in order to optimize the point spread function (PSF) which
defines the spatial resolution of the technique. These methods
and the earlier DRESS (see Single-Voxel MR Spectroscopy)3
and rotating frame zeugmatography (RFZ; Localized MRS
Employing Radiofrequency Field (B1 ) Gradients)53 methods of cardiac MRS are detailed elsewhere and will not be
revisited here.
Often, hybrid methods that combine CSI with DRESS or ISIS
to encode one or more dimensions are used in cardiac 31 P MRS
to reduce the dimensionality or number of CSI phase-encoding
steps while retaining full 3-D gradient-mediated localization.
Common variants are as follows:
6. 2-D CSI with slice-selective excitation (DRESS) in the
third dimension39,45,54,55 ; and
7. 1-D CSI combined with 2-D ISIS.12,56,57
In method 6, the slice-selective gradient refocusing lobe is run
coincident with the phase-encoding CSI gradients. In method
7, the basic 1-D CSI sequence is preceded by two slice-selective
inversion pulses applied in the other two non-CSI dimensions,
and repeated with all four combinations of inversion pulses
© 2016 John Wiley & Sons, Ltd.
Volume 5, 2016
MRS Studies of Creatine Kinase Metabolism in Human Heart
applied and not applied. The resulting four signals are added
and subtracted for each phase encode in the manner of
2-D ISIS.
The hybrid 3-D ISIS and 3-D CSI methods yield substantially rectangular- or cubic-shaped voxels. SLOOP and SLAM
yield arbitrarily shaped voxels that conform to the segmented
portion of the heart.40,49,50 The 1-D methods alone, yield fuzzily
defined voxels whose shape and extent in the two dimensions
not explicitly defined by the gradients, are determined by
the intersection of the 1-D gradient-localized plane with the
detector coil’s sensitivity profile. This means that the larger the
detector coil or detector array is, the poorer the localization
in those dimensions. It also means that surface coil detectors
should be positioned with the ROI centered on the coil’s cylindrical axis and that the 1-D-localizing gradient be directed
along the cylindrical axis to localize a set of ‘sensitive disks’
parallel to the surface coil.
Indeed, the sensitivity profile of the detection system modulates the MRS signal acquired by every localization method
anyway. This preferentially weights the signal contributions in
every voxel to favor those that lie closest to, say, an optimum
7–15 cm diameter surface coil. The weighting is typically large
enough to warrant sensitivity corrections to measurements of
cardiac tissue volumes when they are used to quantify absolute
metabolite concentrations in voxels with dimensions >1 cm
(see Quantifying Metabolite Ratios and Concentrations by
Non-1 H MRS).29 In the case of the RFZ method which also
generates a set of 1-D-resolved spectra, the voxel shapes are
further modulated by the gradient contours of the B1 field of the
excitation coil (see Localized MRS Employing Radiofrequency
Field (B1 ) Gradients).53
1H
MRS. Cardiac 1 H MRS could use the same localization
methods as 31 P MRS, and these can work for studies of lipids
and water. However, multiple factors can confound the use of
CSI and ISIS for acquiring uncontaminated 1 H spectra of lowconcentration cardiac metabolites like CR. First, there is the
intense water resonance from tissue and moving blood, and,
second, the intense lipid resonance from pericardial fat and
the chest. Third, there is cardiac and breathing motion. Fourth,
intense signals can ‘bleed’ into cardiac voxels from adjacent
chest voxels due to an imperfect PSF for CSI, or to incomplete
signal cancellation in ISIS whose localization depends on
spectral subtraction. Fifth, B0 inhomogeneity across the CSI
field of view (FOV) combined with the much smaller chemical
shift dispersion of 1 H compared to 31 P can exacerbate all these
problems.
On the other hand, unlike 31 P where spin-echo methods are
avoided due to signal loss from J-coupling and spin–spin (T2 )
relaxation, 1 H moieties are generally blessed with longer T2 s.
This renders them amenable to spin-echo localization methods
that can avoid the excitation of large FOVs containing much of
the troublesome water and lipid signals. Thus, 1 H MRS studies
of human myocardial CR have to date have utilized singlevoxel point-resolved surface coil spectroscopy (PRESS)17,58,59
or stimulated-echo acquisition mode (STEAM)18 localization with water suppression, as detailed in Single-Voxel MR
Spectroscopy. These methods yield CR measurements from
about 3–9 ml voxels in scan times of <10 min at 1.5–3 T.
Volume 5, 2016
Known Artefacts of Cardiac MRS
Signal ‘Bleed’ and Partial Volume. The primary artefact in
cardiac 31 P MRS is the contamination of cardiac spectra from
extra-voxel signals due to imperfect localization. For CSI
methods, the artifact is commonly called ‘bleed’ and arises
from the combined effect of the sinc-function-shaped PSF
extending into adjacent voxels, and tissue heterogeneity within
each voxel. When the ‘center of mass’ of the signal in each
voxel coincides with the geometric center of the voxel, its PSF
is sampled perfectly at every zero crossing in adjacent voxels,
and there is no bleed artifact.60 However, when the signal
center of mass in the voxel is offset from the voxel’s geometric
center, its PSF in adjacent voxels is sampled at nonzero points
on the PSF’s side lobes.61 The nonzero signal alternately adds
or subtracts to the signal in adjacent voxels because the PSF
has both positive and negative lobes. The likelihood of such
artefacts increases with larger MRS voxel sizes because of the
increased probability of anatomic heterogeneity (it is not due to
CSI grid sizes that use eight or more phase-encoding steps60 ).
When the PCr distribution in a 31 P chest voxel is heterogeneous, the contamination will generally subtract from the PCr
in a bordering cardiac voxel, but add to the next deepest cardiac voxel.60 Because [PCr] in chest muscle is over twice that in
heart but [ATP] is the same, the effect is to alternately decrease
and increase the apparent PCr/ATP ratio with depth into the
heart. The amplitude of the modulation is exacerbated by the
higher sensitivity of a surface detector coil in the chest, which
amplifies the intensity of the muscle signal relative to the cardiac signal. On the distal side, the absence of PCr in ventricular
blood which does contain [ATP], albeit at a lower level than
muscle, reduces the apparent PCr/ATP ratio in voxels that intersect the ventricular chamber in proportion to the partial volume present in the voxel. The same is true of the nearby liver,
which also contains no PCr.
These problems are alleviated by minimizing the diameter
of the detector coil (see section titled ‘Cardiac MRS Detection
Coils’) – since larger coils increase the volume sensitivity to
extraneous sources; and/or by applying saturation pulses to the
chest28,48 ; and/or by using acquisition weighting to improve the
PSF by reducing the amplitude of the side lobes51,52 ; or through
the use of post-acquisition spatial filtering (e.g., applying a
Fermi filter) which does the same as acquisition weighting
but broadens voxel size; and/or by applying intra-voxel tissue
heterogeneity correction methods (see Quantifying Metabolite
Ratios and Concentrations by Non-1 H MRS).
Motion, Chemical Shift Displacement Artifact and Partial
Saturation. The primary artefacts that distinguish the application of PRESS and STEAM 1 H MRS to the heart, from their
application to the brain and other organs, are those arising
from cardiac motion and breathing, which are more intense
than anywhere else, and from the presence of intense lipid
resonances, as noted in section titled ‘1 H MRS’. 31 P MRS is
less susceptible to cardiac motion than 1 H MRS because the
voxels are larger, there are no interfering peaks whose intensity
is comparable to the fat or water 1 H peaks, and because spin
echoes are not used. However, protocols employing ISIS have
an extended exposure to motion artefacts because localization
© 2016 John Wiley & Sons, Ltd.
1187
PA Bottomley
requires the subtraction of large signal volumes acquired with
and without inversion over time periods of twice the pulse
sequence repetition period (TR), 4TR, and 8TR for 1-D, 2-D,
and 3-D ISIS, respectively. In both cardiac 31 P and 1 H MRS,
motion effects can be ameliorated using a prone orientation
and cardiac triggered acquisitions,17,18,33,44 or using a combined
cardiac-respiratory double-triggering strategy for 1 H MRS (see
Physiologic Motion: Dealing with Cardiac, Respiratory, and
Other Sporadic Motion in MRS).58,59
Both 1 H and 31 P MRS methods employing slice selective
excitation (DRESS, PRESS, STEAM) or selective inversion
(ISIS) in one or more dimensions suffer from chemical shift
displacement artefact in those dimensions (see Measuring
Metabolite Concentrations I: 1 H MRS; Single-Voxel MR
Spectroscopy). This problem may be alleviated by using
the strongest selective gradients and shortest RF pulses
allowed.
ISIS suffers from the added partial saturation effect imparted
by its inversion pulses (see Single-Voxel MR Spectroscopy) to
the 31 P metabolites, whose T1 s are somewhat longer than those
of 1 H metabolites. The use of longer TRs than typical of simple pulse-and-acquire approaches, may be needed to maintain
consistent saturation levels over the full ISIS cycle, albeit at a
cost to efficiency. A saturation effect may also arise from the
echo-producing 180∘ pulses in PRESS since the residual longitudinal magnetization is inverted, instead of being refocused, if
the initial pulse is not set at 90∘ .
Quantification
Ratios. The PCr/ATP ratio is measured from the ratio of the
31 P MRS peak areas of PCr and either the 𝛾-ATP or the 𝛽-ATP
resonances. Because the metabolite T1 s of PCr and ATP are different and long such that 31 P acquisitions are usually partially
saturated, it is standard practice to correct the PCr/ATP ratios
for partial saturation. This is done using either the known excitation flip angle at the heart and prior determined T1 values62 ,
or by directly measuring saturation factors from unlocalized
scans.44 Table 1 lists published relaxation times for CK metabolites in the healthy human heart. Metabolite T1 measurements
are unavailable for patients. However, an analysis of the dependence of saturation factors on chest muscle content in multiple patient populations suggests that saturation corrections for
myocardial PCr/ATP ratios do not vary significantly with disease state or chest muscle contamination at 1.5 T.67 Note, however, that the intrinsic T1 , T1int , a theoretical measure of the T1
in the absence of CK exchange, also does not differ significantly
in patients with heart failure at 3 T,32 while exchange rates do
vary.13
It is also standard practice to correct PCr/ATP ratios for
the distortion introduced by ventricular blood, by subtracting a signal amount equal to about 0.3 times the integrated
blood 2,3-diphosphoglycerate (DPG) peak, from the ATP peak
area.68 Although nOe is often not used in 31 P MRS37–39 , the
differences in 𝜂 for PCr and ATP are expected to distort the
PCr/ATP ratio, necessitating corrections if a physiological ratio
is being sought or if interlaboratory comparisons are being
made.38–41
1188
Table 1. Published relaxation times for CK metabolites in the normal
human heart
B0 (31 P) (T) T1 (PCr)
T1 (𝛾-ATP) T1 (𝛽-ATP) T1 (Pi)
References
1.5
1.5
1.5
2.0
3
3
4
7
3
3
3
4.2 ± 0.2a 2.6 ± 0.6a 2.24 ± 0.54a
—
4.3 ± 0.7
3.0 ± 0.5
—
—
4.2
1.7
—
—
4.2 ± 1
2.24 ± 0.63 2.46 ± 0.6 4.3 ± 2.4
3.8 ± 0.7
2.4 ± 1.1
—
—
5.8 ± 0.5
3.1 ± 0.6
—
—
5.3 ± 1.6
2.7 ± 0.6
—
—
3.1 ± 0.4 1.82 ± 0.09
—
—
—
—
—
7.4 ± 1.8b
—
—
—
8.2 ± 1.3c
—
—
—
8.4 ± 1.4d
B0 (1 H) (T)
T1 (CR)
T1 (water)
T2 (CR)
T2 (water)
Reference
1.48
1.21
135
33.1
64
1.5
62
63
64
62
65
66
56
26
31
32
32
Notes: Values are means ± SD; T1 in second, T2 in millisecond. 𝛾-ATP and
𝛽-ATP are the 𝛾- and 𝛽-phosphates of ATP.
a Reported literature average.62
b The ‘intrinsic T ’ of PCr, T
1
1int , calculated assuming no CK chemical
exchange (see Measuring Biochemical Reaction Rates In Vivo with Magnetization Transfer).
cT
1int for PCr corrected for spillover irradiation (uncorrected value
∼7.5 s).32
d The spillover-corrected T
1int for PCr in heart failure patients (P = 0.6 vs
healthy controls).32
Total myocardial creatine, CR, in 1 H MRS is usually
measured from its N-methyl 1 H resonance at 3.0 ppm, and
quantified relative to water in a nonsuppressed spectrum
acquired from the same voxel. The normal CR/water ratio
is about 10−3 .18 For standardization, the STEAM or PRESS
measurements of water ratios should be corrected for T2
attenuation as well as T1 saturation.16–18 In the absence of
formal T2 measurements, this can be done by extrapolating
CR measurements, recorded at long and short echo times,
exponentially back to time zero.18
Concentrations and Reaction Rates. Absolute concentration
measurements are far fewer, but all that is required after a
corrected ratio is obtained is a concentration reference plus
corrections for sensitivity and tissue volume differences when
the reference and cardiac voxels are at different locations.29,45,69
Tissue water contents are surprisingly stable and fairly well
known, enabling the water signal from a nonsuppressed 1 H
spectrum to serve as a concentration reference for both 1 H and
31 P MRS.16–18,64,69 External phosphate reference solutions are
often used as well.29
The reaction rates for the CK and ATPase reactions can be
measured by 31 P MRS saturation transfer protocols.13 Basically, a first reactant moiety, say PCr, is measured with and
without the reaction product moiety, 𝛾-ATP, saturated, and
the fractional change in amplitude of the first moiety due to
its chemical reaction, is recorded. An additional experiment is
then performed to determine the T1 of the first moiety while
saturation is applied to the second moiety, in order to convert
the fractional change to a rate. The product of the rate with
© 2016 John Wiley & Sons, Ltd.
Volume 5, 2016
MRS Studies of Creatine Kinase Metabolism in Human Heart
the concentration, for example, k1 ⋅[PCr], is the forward flux
through the reaction in micromole per gram second.
Protocol Specifics and Derivative Quantities. Detailed protocols, equations, and specific examples of quantification of 31 P
metabolite ratios, concentrations, and CK reaction rates in
the human heart are presented in Quantifying Metabolite
Ratios and Concentrations by Non-1 H MRS and Measuring
Biochemical Reaction Rates In Vivo with Magnetization
Transfer, to which the reader is referred. The formula for
determining intracellular pH from the chemical shift difference between Pi and PCr is in Cardiac MRS Studies in Rodents
and Other Animals, and the quantification of 1 H metabolites is
covered in Measuring Metabolite Concentrations I: 1 H MRS.
With measurements of PCr/ATP, PCr, CR, ATP, Pi, pH, k,
and CK flux in hand, a complete characterization of cardiac CK
metabolism is but a few short steps away. First, unphosphorylated creatine, [Cr] = [CR]–[PCr], can be derived from joint 31 P
and 1 H MRS measurements of PCr and CR.16 [ADP] can then
be estimated from the CK equilibrium equation:
[ADP] =
[ATP][Cr]
[PCr][H+ ]Keq
(7)
with [H+ ] derived from the pH and Keq (= 1.66 × 10 ±
9 l mol−1 ) as the equilibrium constant.13,70 The free energy of
ATP hydrolysis available to fuel cardiac contraction, −𝛥G∼ATP ,
is given by
𝛥G∼ATP = 𝛥G0 + RT ln([ADP][Pi]∕[ATP]) kJ mol−1
(8)
with 𝛥G0 (=−30.5 kJ mol−1 ) as the standard free energy
change, and RT is the universal gas constant times absolute
temperature.13,70
Results
Normal Values
Table 2 presents the average (±SD) of the mean values of
human cardiac CK metabolic parameters measured by 31 P and
1 H MRS published through 2015. These are unweighted by the
number of subjects or SDs listed in the source publications. The
mean myocardial PCr/ATP has risen slightly from 1.72 ± 0.26
in 200825 to 1.84. The individual study data are visualized in
Figure 3. The B0 at which the studies were conducted does not
significantly affect the mean PCr/ATP (Figure 2a) – a good
sign – and the scatter appears to decline at 3 and 4 T compared
to 1.5 T, although the 1.5 T data pool includes much older
data. However, the reported SDs accompanying the individual
mean values do not yet make a case that the increase in SNR
with B0 ,26,65,66 translates to more precise myocardial PCr/ATP
measurements overall (Figure 2b). Even so, direct comparative
measurements at 1.5, 3,65 and 7 T26 suggest otherwise, and
SNR gains at higher fields are often spent on improving spatial
resolution and/or speed [VX and NX in equation (3)] instead of
precision.56,141
The choice of localization method also does not appear
to systematically affect PCr/ATP or its variability overall
(Figure 2c). This does not mean that localization is not
important, but rather suggests that other factors – including
experimental and physiological variability – are dominant
sources of scatter. Regarding the former, about a half of the
studies have SDs ≤15% with a third of them below 10% overall.
This demonstrates that such levels of precision are in fact
commonly achievable, and suggests that experimental factors unrelated to B0 or localization method per se, represent
the primary sources of variability between studies of human
myocardial PCr/ATP ratios. PCr/ATP does not appear to vary
with location in the normal heart.56,106
Despite the added uncertainty in measuring reference concentrations and other factors needed to convert metabolite
ratios to absolute concentrations, cardiac [PCr] and [ATP]
measurements appear to be at least as precise as the PCr/ATP
measurements (Table 2). Although this may reflect better
control of experimental variables in this more quantitative
work, a caveat is that these studies remain limited to about
four independent research groups. The same is true of the [CR]
measurements, which at 28.0 ± 0.7 μmol g−1 appear tightest of
all. The tabulated values for [CR], the CK rate k1 , and the flux
each reflect the efforts of just two laboratories.
The dearth of measurements of Pi undoubtedly arises from
the difficulty in resolving it from blood DPG and other phosphomonoester (PM) resonances in the healthy heart. Although
1 H decoupling can reduce the line widths of DPG and PM signals and improve the spectral resolution of Pi, it is still undetectable in at least half of the normal subjects studied at 1.5 T.12
Moreover, Pi and pH appear ripe for characterization using the
Table 2. Literature average metabolic parameters measured in normal human heart by MRS
Variable
Mean ± SDa
Na
Source references
PCr/ATP
[PCr] (μmol g−1 wet)
[ATP] (μmol g−1 wet)
pHb
k1 (s−1 )
CK flux (μmol g−1 s−1 )
[CR] (μmol g−1 wet)
1.84 ± 0.27
10.14 ± 1.36
6.02 ± 1.00
7.13 ± 0.04
0.36 ± 0.06
3.35 ± 0.21
28.0 ± 0.7
1662
237
237
31
133
94
66
3, 4, 12, 13, 19, 26–29, 31, 33–35, 38, 39, 41, 45, 48, 49, 51–57, 63, 68, 69, 71–139 (114 cohorts)
13, 19, 29, 35, 45, 64, 69, 87, 95, 98, 110, 120, 126, 135, 140 (16 cohorts)
13, 19, 29, 35, 45, 64, 69, 87, 95, 98, 110, 120, 126, 135, 140 (16 cohorts)
12, 54, 70, 86
13, 19, 30–32, 35, 126, 140–142
13, 19, 31c, 35, 126, 140
17, 18, 64, 143, 144
Notes: Values are means ± SD of the average of the published mean values, unweighted by individual sample size or uncertainty. Excludes data known to be
uncorrected for saturation or nOe.
a Number of subjects – may include same subjects reported in repeat publications.
b Measurements from Pi at 1.5 T, which is undetectable about half the time.12
c Value based on a [PCr] determination scaled assuming [ATP] = 5.5 μmol g−1 wet wt.
Volume 5, 2016
© 2016 John Wiley & Sons, Ltd.
1189
PA Bottomley
PCr/ATP
60
2.5
SD (%)
2.3
2.3
50
2.1
1.9
2.1
1.9
40
1.7
1.7
1.5
30
1.5
0.9
10
0.7
0.7
B0 (T)
1
2
(a)
3
4
5
6
7
8
0
B0 (T)
0.5
1 2 3 4 5 6 7 8
(b)
(c)
3-DCSI, 2-DCSI+DRESS
0.9
1.1
1-D CSI+ 2-D ISIS
1.1
DRESS
20
Surface crusher
1-DCSI, RFZ
1-DCSI+DRESS
ISIS, SLOOP, FSW
1.3
1.3
0.5
PCr/ATP
2.5
Figure 3. Literature measurements of the cardiac PCr/ATP ratio in healthy controls by localized 31 P MRS. Each symbol represents a mean published value
for a study cohort, corrected for saturation and blood ATP contamination. (a) The PCr/ATP ratio as a function of B0 in Tesla. (b) The reported standard
deviation (SD) in the mean PCr/ATP measurements expressed as a percentage of the mean, plotted as a function of B0 . (c) PCr/ATP as a function of the
localization method used to measure it. The data from the three left-most columns utilized 1-D gradient localization (confined in the other two dimensions
only by a surface detector coil’s sensitivity profile). The 1-D CSI plus DRESS data were acquired with 2-D gradient localization. The right-most three columns
of data were acquired using full 3-D gradient-controlled MRS localization
better spectral resolution and SNR afforded at 7 T. Note that [Pi]
in whole human blood is only about 0.08 mM.11 This would be
imperceptible to 31 P MRS and could not contaminate myocardial Pi and pH measurements.84
From Table 2, we see that PCr constitutes 36% of the total
creatine pool in normal heart, hence [Cr] = 17.9 μmol g−1
wet. Thus, PCr’s fraction of the total creatine pool is a half of
that available to resting skeletal muscle (i.e., 34/45 mM = 76%
from Ref. 145; or 25.6/36.2 μmol g−1 = 71% from Refs 69,
146). This is consistent with a more active role of the Cr/PCr
system in shuttling energy from the mitochondria to fuel the
contracting myofibrils even at rest. ATP synthesis through
OXPHOS has been estimated from myocardial oxygen (O2 )
consumption, which is about 0.085 ml g−1 wet weight per
minute in healthy resting subjects.147,148 Assuming 3 mol of
ATP is produced per mole of O2 consumed yields a mean ATP
supply of 0.33 μmol g−1 s−1 via OXPHOS, which is consistent
with invasive measures of ∼0.43 μmol g−1 s−1 .149 This means
that the cardiac CK flux of 3.35 μmol g−1 s−1 (Table 2) would be
able to shuttle up to about 10 times the energy provided by aerobic metabolism at rest to support contraction, consistent with
a CK role as a spatiotemporal buffer.13 From equation (7), the
value of [ADP] is approximately 90 μmol l−1 . From equation (8)
then, 𝛥G∼ATP ≈ 60 kJ mol−1 for the normal heart.13,150
Physiological Variations
Cyclic Energy Demand and Cardiac Workload. All of the CK
metabolite values in Table 2 were measured at rest and/or
represent temporal averages. However, energy demand varies
throughout the cardiac cycle, arguably by at least threefold
between peak demand and relaxation.13 If the role of PCr and
1190
the CK reaction is as a rapid energy reserve, then an obvious
question is whether myocardial PCr/ATP varies cyclically
in the healthy heart at rest, or during exercise. Measurements of PCr/ATP triggered at different points in the cardiac
cycle in healthy subjects are few, but any metabolic variations other than those attributable to myocardial volume, are
undetectable.71 Indeed, a theoretical analysis of the CK enzymatic rate and Bloch equations employing empirical parameter
ranges recently concluded that (i) the maximum intracycle
changes in high-energy phosphate would be ≈0.4 mM, which
are at or below current detection levels and (ii), that saturation transfer 31 P MRS would be unable to detect intracyclic
fluctuations in k1 or CK flux.150
If CK metabolites did vary cyclically, even modestly, then
much larger changes would be expected with exercise when
energy demands would be much higher. PCr/ATP has been
studied with increased cardiac workloads while subjects lay
in the scanner using three types of stress test. First, an isometric exercise, performed, for example, with a hand-grip
dynamometer.33 Second, an aerobic exercise involving the
lifting of weights with the legs,75 and, third, stress induced by
pharmaceutical agents such as dobutamine.13,80 The increase in
cardiac workload as indexed by the heart-rate blood-pressure
product (HR × BP) in these protocols, is limited to: (i) about
40% with a hand-grip exercise involving exertion at 30% of
the subject’s maximum force33 ; (ii) about twofold with aerobic
exercise involving lifting multiple 2.5 kg weights74 ; and (iii)
up to fourfold using dobutamine.80,92,96 Compared to aerobic
exercising, the isometric and dobutamine protocols minimize
motion problems during acquisition.
Studies using isometric,33,85 aerobic,48,75,128 and dobutamine
stress13,80,138 at up to double the HR × BP did not produce
© 2016 John Wiley & Sons, Ltd.
Volume 5, 2016
MRS Studies of Creatine Kinase Metabolism in Human Heart
any significant changes in PCr/ATP. However, at a higher
stress of three to four times the resting HR × BP, significant
reductions of 14–21% (P < 0.001) in myocardial PCr/ATP were
observed92,99 even in athletes.96 It is interesting to note that
even though CK starts with a 10-fold buffering capacity for
shuttling OXPHOS energy at rest (see section titled ‘Normal
Values’), the CK flux might be expected to have difficulty meeting ATP demands at three to four times the resting HR × BP,
if peak intracycle energy demands are also taken into account.
That is of course, unless the CK flux rises to meet the increased
energy demand at the new stress level.
To test whether CK flux upregulates with cardiac workload, the CK rate and flux were measured in healthy subjects
during dobutamine stress using 31 P MRS saturation transfer
methods.13 No significant changes were observed in PCr/ATP,
[PCr], [ATP], k, or CK flux, at least for work-loads at an
average of double the resting rate.13 If CK flux does not change
with increasing workloads, then the CK ATP supply must be
limited. The observed decline in myocardial PCr/ATP at the
highest levels of stress92,96,99 would be a consequence of this
limit, which might even be a determinant of individual limits
to cardiac workload.
Cardiovascular Fitness. A number of studies have examined the effect of cardiovascular fitness on both resting and
stressed myocardial PCr/ATP in healthy volunteers. High-level
dobutamine stress at three- to four-fold HR × BP elicited
similar 14–17% decreases in myocardial PCr/ATP in both
elite cyclists and age-matched healthy controls, while their
resting PCr/ATP values were also the same (1.4 ± 0.2).96 A
more recent study, compared resting PCr/ATP in healthy
middle-aged (∼50 years old) volunteers as a function of their
maximum working capacity, MWC, as gauged by a stepped
exercise.55 Unlike the earlier high-level HR × BP study, this
study found that resting myocardial PCr/ATP correlated
with MWC and those with MWC > 230 W had significantly
higher PCr/ATP than those with MWC <200 W (1.93 ± 0.36
vs 1.59 ± 0.35; P < 0.001).55 On the other hand, a comparison of myocardial PCr/ATP ratios in elite track sprinters,
marathon runners, and young sedentary but lean men, published the same year, found no significant differences among
the three groups (PCr/ATP = 1.9–2.4), although athletes had
the highest mean PCr/ATP ratios.123 A follow-up study by
the same authors showed no significant differences between
young sedentary (25 ± 2 years; PCr/ATP = 1.94 ± 0.36) and
young (26 ± 5 years; PCr/ATP = 2.36 ± 0.36) and middle-aged
(47 ± 9 years; PCr/ATP = 2.19 ± 0.34) athletes, but sedentary
middle-aged (45 ± 6 years) subjects had significantly lower
cardiac PCr/ATP (1.83 ± 0.27) versus the three other young
and/or fit groups.127
Differences in myocardial PCr/ATP ratios among control
groups that are supposed to be comparable, combined with
observed changes that often fall within normal ranges published by others, represent major distractions for these studies.
However, if such differences are set aside as being attributable
to systematic factors of methodological origin, a picture that
emerges in toto is that most young healthy adults in their 20s
are endowed with a high cardiac reserve as indexed by a high
PCr/ATP ratio of, say, 1.8–2.0, regardless of their sedentary or
Volume 5, 2016
Table 3. Reported age and gender variations in PCr and ATP in normal
subjects
Age/gender
n
32 ± 3 years
60 ± 13 years
30 ± 6 years
53 ± 7 years
<40 years
>40 years
M
F
15
15
37
39
16
14
18
12
PCr/ATP
1.7 ± 0.3
1.6 ± 0.4
2.16 ± 0.36
1.83 ± 0.37†
1.9 ± 0.5
1.9 ± 0.4
1.9 ± 0.4
1.9 ± 0.6
[PCr]
(μmol g−1 )
13.5 ± 1.9
9.7 ± 2.5*
—
—
9.7 ± 2.4
7.7 ± 2.5‡
9.2 ± 2.4
8.0 ± 2.8
[ATP]
References
(μmol g−1 )
8.2 ± 1.4
6.4 ± 1.8
—
—
5.1 ± 1.0
4.1 ± 0.8*
4.9 ± 1.0
4.2 ± 0.9
95
95
39
39
120
120
120
120
Notes: Means ± SD as reported or calculated from standard errors; M, male;
F, female.
* Probability that difference is not significant, P < 0.01 versus younger
group.
† P < 0.001 versus younger group.
‡ P < 0.05 versus younger group.
athletic lifestyles. However, with aging, a continued sedentary
lifestyle is associated with a declining cardiac PCr/ATP by the
time they reach about 50 years old, while maintenance of cardiovascular fitness through exercise can preserve myocardial
CK energy reserve at near-youthful levels.
Age and Gender. Indeed, several studies have reported some
variation of PCr/ATP ratios and/or [PCr] and [ATP] concentrations with age39,95,120 and gender,120 as summarized in
Table 3. This work suggests a trend of reduced metabolite
concentrations,95,120 and probably PCr/ATP,39 with age above
40 years, and perhaps in women versus men of similar age as
well.120 Such results are consistent with a decline in cardiac
CK energy reserve with age in cohorts of control patients that
include more sedentary individuals not selected for cardiovascular fitness. However, it is prudent to be cautious because
(i) there are systematic differences between the results in
Tables 2 and 3; (ii) there are systematic age- and gender-related
variations in chest composition that could affect the results if
chest contamination is present (see section titled ‘Signal ‘Bleed’
and Partial Volume’); and (iii) some laboratories that have
studied many subjects have not reported any significant age
dependence.
Diet and Obesity. Given that over a third of the adult American
population are considered obese as of 2012 and that twothirds are overweight, the inclusion in cardiac MRS studies of
‘normal volunteers’ who are overweight but do not otherwise
have exclusion criteria for heart disease, is perhaps inevitable.
The effect of diet on myocardial PCr/ATP was tested in a
crossover study of young men (age 22 ± 1 standard error, SE)
on whom 31 P MRS was performed before and after either a
5-day high-fat diet (75% of calories) or a standard 23% fat diet.
This was followed by a 2 week ‘washout’ period and crossover
to the alternate diet.134 The high-fat diet increased plasma free
fatty acids by 44%, and was associated with a 9% reduction
in cardiac PCr/ATP (2.0 ± 0.1 SE vs 2.3 ± 0.1 SE vs P < 0.01)
along with some cognitive impairment.
Cardiac 31 P MRS results from age-matched obese subjects (age, 44 ± 7 years; BMI = 39 ± SD kg m−2 ) with no
© 2016 John Wiley & Sons, Ltd.
1191
PA Bottomley
history of cardio- or peripheral-vascular disease, smoking,
hypertension (HT), or diabetes, were compared with those
from ‘normal-weight’ (age, 43 ± 10 years; body mass index,
BMI = 22 ± 2 SD kg m−2 ) subjects at rest and during dobutamine stress (with a 63 ± 16% HR increase).138 At rest,
diastolic function (peak filling rate) and cardiac PCr/ATP
were 15% lower in obese subjects than in the normal-weight
group (1.7 ± 0.4 vs 2.0 ± 0.3; P < 0.05), which decreased by an
additional 12% (to 1.5 ± 0.5; P = 0.03 vs rest) with dobutamine
stress. As in earlier studies, PCr/ATP was unaltered during
stress in the normal-weight group. A follow-up study by the
same authors compared obese patients (BMI = 34 ± 4 kg m−2 )
with ‘normal-weight’ subjects (BMI = 22 ± 2) before and after
a year in a supervised weight-loss program.151 At the initial
31 P MRS examination, the PCr/ATP ratio was 1.58 ± 0.47
compared to 2.03 ± 0.27 in normal-weight subjects (P = 0.002).
As a result of the weight-loss program, the obese subjects
shed 9 kg (to BMI = 29 ± 2), and cardiac PCr/ATP increased
by 24% to 1.96 ± 0.47 (P < 0.05), which was accompanied by
improvements in diastolic function.
Thus, both dietary fat and obesity are associated with
impaired cardiac CK energy reserve. In obese subjects with
diastolic dysfunction, PCr/ATP may be further reduced during
stress. This transient decrease in PCr/ATP with stress has to
be interpreted as evidence of ischemia, wherein the CK ATP
supply is unable to meet the increased workload. The good
news is that long-term weight loss may reverse this reduction
in myocardial CK energy reserve.
1 H MRS studies in obesity have focused not on CR but on
triglyceride elevations in the septum, which are reviewed elsewhere (see Cardiac Lipids by 1 H MRS).
Hypertension (HT). Hypertension, too, may affect myocardial
PCr/ATP ratios. Middle-aged hypertensive patients in whom
coronary disease, malignant HT, and diabetes were all ruled
out, have been studied at rest and during dobutamine stress
with a 2.2-fold increase in HR × BP. The resting PCr/ATP
was significantly lower than in healthy age-matched controls (1.2 ± 0.18 SD vs 1.39 ± 0.17, P < 0.05) and decreased
a further 21% with stress (to 0.95 ± 0.25, P < 0.01 vs rest).99
The metabolic changes were again associated with diastolic
dysfunction. However, another study showed no change in
resting myocardial PCr/ATP, [PCr], or [ATP] in hypertensive
patients who had a ∼50% increase in left ventricular (LV) mass
compared to controls, but otherwise had no symptoms of heart
failure.110
Hypoxia. Chronic exposure to hypobaric hypoxia may
represent another ‘normal’ variant for human myocardial
PCr/ATP. Six 20- to 30-year-old professional Sherpa trekker
guides who were native to an altitude of ∼3400 m in Nepal,
exhibited significant reductions in myocardial PCr/ATP
to 1.0 ± 0.37 SD as compared to control ‘lowlanders’ with
PCr/ATP = 1.76 ± 0.06.52 These results suggest that the heart
may adapt its energy sources and delivery over time in response
to chronic environmental stress.52
Yet even relatively short-term exposure to hypoxia may
reduce myocardial PCr/ATP. A 31 P MRS study of 14 trekkers,
age 38 ± 11 SD years, before and immediately after a 17-day
1192
trek to Mt. Everest base camp (5300 m) exemplifies an effect of
hypobaric hypoxia.137 Myocardial PCr/ATP was reduced 18%
from 2.05 ± 0.30 SD before, to 1.68 ± 0.30 (P = 0.003) after the
trek, accompanied by some alterations in diastolic function.
Cardiac metabolism and function returned to pre-trek levels
by 6 months. The effect of shorter normobaric hypoxic exposures were investigated by the same authors using a hypoxic
chamber with O2 maintained at an end-tidal expiration partial
pressure of 50–60 mmHg and 80% peripheral O2 saturation.136
After 20 h of exposure, young (age, 24 ± 7 SD years) healthy
volunteers exhibited 15% reductions in cardiac PCr/ATP (from
2.0 ± 0.3 to 1.7 ± 0.3; P < 0.01) and impaired diastolic function.
Thus, cardiac CK energy reserve and high-energy phosphate
balance is depressed in the presence of hypoxia, even at the relatively modest O2 reductions that can be tolerated by healthy
subjects over periods ranging from hours to lifetimes.
Disease
Myocardial Infarction (MI). Published 31 P MRS studies are
mixed on whether resting myocardial PCr/ATP ratios are
altered in MI. The data are summarized in Table 4. Most likely,
the mixed message arises from other conditions such as congestive heart failure (CHF) or cardiomyopathy (CM) associated
with post-MI remodeling. These would confound the PCr/ATP
measurements in chronic MI (see sections titled ‘Dilated
Cardiomyopathy (DCM) and Heart Failure’ and ‘Hypertrophy,
Valve Disease, and Heart Failure’), while playing lesser roles in
acute MI in which active ischemia was not present. Meanwhile,
in acute anterior MI, Pi may be elevated a week or so postonset,71 consistent with the post-MI time course of Pi seen in
canine studies after an acute decline in PCr and ATP.154
Table 4. Resting myocardial PCr/ATP ratios in patients with MI and/or
ischemia (ISCH)
MI/ISCH
MI
MI + ISCHa
MI
MIb
ISCHc
MIb
ISCHc
MI
MI
MI
MI
MI
ISCH
Averagee
Controls
Patients
References
1.6 ± 0.4
1.72 ± 0.15
1.95 ± 0.45
1.85 ± 0.25
1.85 ± 0.25
1.8 ± 1.03
1.8 ± 1.03
1.61 ± 0.18
1.72 ± 0.31
1.45 ± 0.29
1.87 ± 0.45
1.7 ± 0.40
1.45 ± 0.31
No change in patients
1.24 ± 0.3*
1.6 ± 0.19
0.94 ± 0.41
1.37 ± 0.57
1.51 ± 0.17
1.47 ± 0.38
0.6 ± 0.2d,†
1.74 ± 0.27
1.03 ± 0.39†
1.39 ± 0.23
1.34 ± 0.34‡
71
33
79
85
85
87
87
91
101
108
126
152
153
1.73 ± 0.16
Notes: Means ± SD, as reported or calculated from reported standard errors
(SE); ISCH = patients with ischemia.
a ISCH patients includes 6 with MI (P = 0.052 vs controls).
b Fixed defect on exercise 201 Tl imaging.
c Reversible defect on exercise 201 Tl imaging.
d Average for septal and anterior LV MI.
e Average of the studies listed (omitting duplicate entries).
* P < 0.05 in MI versus controls.
† P < 0.05 in MI versus uninvolved tissue.
‡ P < 0.005 versus controls
© 2016 John Wiley & Sons, Ltd.
Volume 5, 2016
MRS Studies of Creatine Kinase Metabolism in Human Heart
Indeed, biochemical analyses in animal heart models show
depletion of essentially all ATP and PCr within the first few
hours of an ischemic injury that results in cell death.1 Because
dead cells can contribute no high-energy phosphates, the
resting PCr/ATP ratios that are reported in MI must derive
from ‘metabolically normal’, ‘jeopardized’, and/or ischemic
myocardium adjacent to, or interspersed with the infarction.
Note that ‘metabolically normal’ need not mean functionally
normal. A study of 29 reperfused MI patients with dysfunction in the infarct area – ‘myocardial stunning’ – found no
significant changes in PCr/ATP from normal at 4 ± 2 days
post onset, with PCr/ATP remaining unchanged a month later
even though function had improved.91 Regardless, a preserved
PCr/ATP has to be interpreted as indicating the presence of
viable surviving myocardium in the MRS voxel.
The main MRS action in MI is the reduction in [PCr],
[ATP], and [CR] in the infarction itself, as summarized in
Table 5. Reductions of about 50% likely reflect the presence of
both infarcted and noninfarcted tissue in MRS voxels. There is
a significant negative correlation between cardiac ATP levels
and the size of perfusion deficits quantified by thallium 201 Tl
radionuclide imaging.78 In addition, [PCr] and [ATP] are
reduced in patients with fixed 201 Tl defects compared to those
with reversible defects.87 In the tabulated studies, uninvolved
myocardium without wall motion abnormalities at echocardiography had near-normal PCr, ATP, and CR concentrations.18,64
In addition, CK rate measurements measured by saturation
transfer also showed no change in k1 in MI (Table 5). Because
the measurements of k1 can only derive from the myocytes
that survive in the infarcted region, this result must reflect a
preserved CK metabolism in the surviving tissue, even though
the net CK flux given by the product, k1 ⋅[PCr], is halved due to
the loss of [PCr] in the infarct.126
Myocardial Ischemia. In patients with ischemic heart disease
involving stenosis of the anterior vessels, the resting anterior
myocardial PCr/ATP shows a declining trend (Table 4, shaded
entries), with CK metabolite concentrations in reversible
defects reaching statistical significance (Table 5).87 While
a number of other conditions are associated with reduced
metabolite levels at rest, the transient reduction in PCr/ATP
ratio that is elicited by increased cardiac workloads, indicates
an excess consumption of PCr that cannot be met by OXPHOS.
Thus, stress-induced changes in myocardial PCr/ATP could
serve as a specific metabolic hallmark of ischemia.33
As noted in the section titled ‘Cyclic Energy Demand and
Cardiac Workload’, in healthy subjects who are free of significant coronary disease, myocardial PCr/ATP is unaltered by
up to a ∼2-fold increase in cardiac workload,13,33,75,85,128,138
but a small reduction at three to four times the resting
workload92,96,99 may arise if the CK energy supply cannot
cope with demand. However, in patients with severe anterior
coronary stenosis and ischemia verified by echocardiography
or 201 Tl radionuclide imaging, a simple isometric handgrip
exercise that only elicited a 30–40% increase in HR × BP, was
able to produce a ∼40% reduction in myocardial PCr/ATP
compared to that at rest (Figure 4).33 After exercise, metabolite
ratios recovered to near-normal pre-exercise values. Patients
with nonischemic heart disease (CM or valve disease) who
Volume 5, 2016
Table 5. CK metabolite levels in patients with MI
[PCr]
(μmol g−1 wet weight)
Control
12.1 ± 4.3
11.9 ± 4.4
9.6 ± 1.1
[ATP]
(μmol g−1 wet weight)
Patients
Control
Patients
7.6 ± 3a, *
3.9 ± 2.2b,†
11.0 ± 4.4c
6.4 ± 1.8d, *
5.0 ± 2.4e, *
5.4 ± 1.2‡
7.7 ± 3
6.4 ± 3.2a
4.4 ± 1.5b, *
7.9 ± 4.3c
5.4 ± 2.2d
4.9 ± 2.5e
3.4 ± 1.1‡
[CR]
(μmol g−1 wet weight)
7.0 ± 1.8
5.5 ± 1.3
[CR]
(μmol g−1 wet weight)
Control
Patients
Control
Patients
28 ± 6
28.8 ± 3.0
9.8 ± 8.6§
28 ± 6
28.8 ± 3.0
28.8 ± 3.0
26 ± 11f
16.9 ± 7.1d, *
13.2 ± 5.4e, *
26.5 ± 4.3c
CK rate,
k1 (s−1 )
Control
0.33 ± 0.07
CK flux
(μmol g−1 wet weight s)
Patients
Control
Patients
0.31 ± 0.08
3.3 ± 0.8
1.7 ± 0.5‡
References
87
87
64
64
64
126
References
18
64
64
References
126
a Patients
with MI and reversible defects on 201 Tl scintigraphy.
with MI and fixed defects on 201 Tl scintigraphy.
c Normo-kinetic by echocardiography.
d Hypokinetic by echocardiography.
e Dyskinetic by echocardiography.
f Measurements from uninvolved tissue.
* P < 0.05 versus controls.
† P < 0.01 versus controls.
‡ P < 0.001 versus controls.
§ P < 0.0001 versus controls.
b Patients
were tested with the same exercise exhibited no PCr/ATP
changes. In addition, patients with ischemia who underwent
repeat stress testing after successful revascularization therapy
also showed no stress-induced PCr/ATP change, consistent
with a resolution of the metabolic abnormality following a successful clinical outcome.33 A similar 40% decrease in anterior
myocardial PCr/ATP was reported during hand-grip exercise
in patients with reversible anterior wall ischemia confirmed
by exercise 201 Tl radionuclide imaging.85 Those with fixed
201 Tl defects indicative of MI exhibited no exercise-induced
PCr/ATP changes.
These studies, and observations that dobutamine stresstesting at 1.6-fold resting HR × BP induced no significant
reductions in myocardial PCr/ATP in patients with DCM and
CHF,80 suggested that stress-induced changes in PCr/ATP
might well be specific to myocardial ischemia. However, as
noted in sections titled ‘Diet and Obesity’ and ‘Hypertension
(HT)’, 12% and 21% reductions in myocardial PCr/ATP have
also been observed with dobutamine stress in obese patients138
and those with HT,99 respectively, in the absence of independent evidence of ischemia. More recently, a 9% transient
reduction in cardiac PCr/ATP was reported in patients who
© 2016 John Wiley & Sons, Ltd.
1193
PA Bottomley
Exercise
Rest
Recovery
PCr
PCr
0.9
2.0
#16
endo.
Pi
1.6
#15
PCr
1.3
3.2
epi.
Chest
muscle
3.2
#14
ATP
(a)
(b)
0
20
(c)
0
20
(d)
0
20 ppm
Figure 4. Hand-grip exercise stress testing of a patient with myocardial ischemia performed at 1.5 T.33 (a) Axial 1 H surface coil image of the chest and
anterior LV of a patient with an occluded right coronary and 70–90% stenosis of the left anterior coronary artery. The image is annotated with the locations
of three 1 cm thick coronal slices localized in the 31 P MRS exam (SE, septum; RV, right ventricle; ENDO, endocardial; EPI, epicardial; ST, sternum; REF,
reference vial in the 31 P coil). (b–d) Cardiac-gated 1-D CSI surface coil 31 P spectra from the three slices (b) at rest, and during (c), an 8 min isometric
hand-grip exercise, and (d) a recovery acquisition commencing 2 min post exercise. The PCr/ATP decreases from 2.0 to 0.9, and 1.6 to 1.3 in the endo- and
epicardial slices respectively, consistent with ischemia. The PCr/ATP ratio is unchanged in the chest (slice #14)
had HCM of a genetic etiology but no evidence of coronary
artery disease, during aerobic exercise testing at a 1.7-fold
increase in HR × BP workload (see also section titled ‘Hypertrophy, Valve Disease, and Heart Failure’).48 This work suggests
that factors other than frank vessel disease may also lead to
transient reductions in energy reserve – that is, an ischemic
response – in obesity, hypertensive, and hypertrophic disease.
Two additional studies have used isometric hand-grip
exercise to measure myocardial ischemia. One studied
‘women’s ischemic syndrome’ involving chest pain – also
in the absence of significant coronary vessel disease –reporting
that some women had cardiac PCr/ATP ratios during stress
that were two SDs below the mean of control subjects.155
However, the overall variations in PCr/ATP were higher than
in controls and exhibited both increases and decreases during
stress. The second study was a double-blinded application of
stress 31 P MRS to test the efficacy of an experimental antiischemic pharmaceutical that reduces the binding affinity of
oxygen to hemoglobin, thereby potentially increasing oxygen
availability.153 While the stress produced a 31% PCr/ATP
reduction during control studies, the lesser 20% decline seen
with anti-ischemic therapy in the same subjects was not
significant and the outcome was inconclusive.
Because LV dysfunction and heart failure are common
prognoses for ischemic events, another study examined the
effect of trimetazidine (TMZ) on resting cardiac PCr/ATP post
ischemia. TMZ is an anti-ischemic therapy that shifts energy
utilization from free fatty acids to glucose.156 A double-blinded
90-day crossover protocol with placebo or TMZ was implemented in patients with CHF but not active ischemia, acute
MI, or anterior lesions at locations where the 31 P MRS was
performed. It was found that TMZ reduced the symptom-based
New York Heart Association (NYHA) CHF class, improved
LV ejection fraction (EF), and increased myocardial PCr/ATP
from an abnormally low 1.35 ± 0.33 to 1.8 ± 0.5 (P = 0.03).
1194
Dilated Cardiomyopathy (DCM) and Heart Failure. Initial 31 P
MRS studies of DCM in the early 1990s revealed significant68
and nonsignificant46,77 reductions in myocardial PCr/ATP. An
elevated phosphodiester (PD) peak,46,77 and an elevated Pi/PCr
Table 6. Myocardial PCr/ATP in patients with dilated cardiomyopathy
(DCM)
Controls
1.54 ± 0.11
1.8 ± 0.21
1.65 ± 0.26
2.09 ± 0.44
1.95 ± 0.45
1.86 ± 0.17
2.02 ± 0.41
1.94 ± 0.60
1.75 ± 0.25
2.07 ± 0.17
n.s.
n.s.
1.81 ± 0.49
n.s.
1.86 ± 0.17c
DCM
NYHA
References
1.51 ± 0.29
1.46 ± 0.31*
1.52 ± 0.58
1.88 ± 0.4
1.78 ± 0.51
1.44 ± 0.52†
1.94 ± 0.43
1.63 ± 0.24a
1.54 ± 0.48†
1.63 ± 0.43b
1.26 ± 0.29
1.31 ± 0.38*
1.63 ± 0.33*
1.54 ± 0.34
1.58 ± 0.41
1.59 ± 0.48c
1.5 ± 0.4
1.56 ± 0.16d, *
I–III
(CHF)
II–III
n.s.
II–IV
≥III
<III
I–III
II–III
I–III
n.s.
III–IV
I–II
n.s.
II–III
I–III
n.s.
77
68
12
81
79
80
89
34
107
122
40
20
35
27
Average
Notes: Values are means ± SD, as reported or calculated from reported data
or standard errors (SEs) excluding repeat entries or data uncorrected for
saturation when known; NYHA, New York Heart Association classification
for congestive heart failure (CHF); n.s., not specified.
a No change with dobutamine stress at a 1.6-fold increase in HR × BP.80
b PCr/ATP < 1.6 predicted mortality at 2.5 years (P < 0.02).34
c Patients had DCM or left ventricular hypertrophy (LVH).35
d Average of the studies listed.
* P < 0.001 versus controls.
† P < 0.05 versus controls.
© 2016 John Wiley & Sons, Ltd.
Volume 5, 2016
MRS Studies of Creatine Kinase Metabolism in Human Heart
Table 7. CK metabolite concentrations and flux rates in patients with cardiomyopathy
[PCr] (μmol g−1 wet weight)
DCM + CHFa
DCM
DCM + CHFc
LVH + CHFc,d
LVH + CHF
LVH no CHF
CM + CHFc,f
CM + CHFa
LVH
HCMg
HCMh
[ATP] (μmol g−1 wet weight)
Control
Patients
Control
Patients
10.1 ± 1.3
—
8.8 ± 1.3
—
9.4 ± 1.1
—
—
9.75 ± 1.33
9.7 ± 2.5
9.4 ± 1.2
8.8 ± 2.6
8.3 ± 2.6*
6.2 ± 1.5b
4.3 ± 1.2†
6.3 ± 1.5*
7.2 ± 3.7
6.1 ± 2.0†
7.6 ± 1.8
7.39 ± 2.5‡
6.1 ± 2.2†
7.1 ± 2.3†
6.1 ± 1.9e,†
5.7 ± 1.3
—
5.7 ± 1.0
—
5.5 ± 1.3
—
—
5.7 ± 1.3
6.4 ± 1.8
5.8 ± 1.2
4.6 ± 1.0
5.2 ± 1.3
4.1 ± 0.8b
3.7 ± 0.5*
4.9 ± 0.9
5.0 ± 1.1
4.7 ± 1.3
4.9 ± 1.3
4.8 ± 1.4*
4.1 ± 1.3*
5.0 ± 0.8
3.9 ± 1.5*
[CR] (μmol g−1 wet weight)
DCM + CHFc
DCM + CHFa
HCM + CHFa
CM + CHFa
[CR] (μmol g−1 wet weight)
Control
Patients
Control
Patients
27.6 ± 4.1
27.6 ± 4.1
—
27.1 ± 3.2
15.0 ± 5.4‡
—
—
—
—
—
—
—
—
16.1 ± 4.5*
22.6 ± 8.1*
16.5 ± 6‡
CK rate, k1 (s−1 )
DCM + CHFa
CM + CHFc,f
LVH no CHF
LVH + CHF
HCMg
CM + CHFa
CK flux (μmol g−1 wt s)
Control
Patients
Control
Patients
0.32 ± 0.07
0.21 ± 0.07‡
3.2 ± 0.9
1.6 ± 0.6‡
2.07 ± 1.27
2.2 ± 0.7*
1.1 ± 0.4‡
2.0 ± 1.4‡
1.75 ± 0.96‡
0.32 ± 0.06
0.38 ± 0.07
0.38 ± 0.08
0.28 ± 0.13
0.36 ± 0.04
0.17 ± 0.06‡
0.28 ± 0.15*
0.24 ± 0.12‡
3.1 ± 0.8
3.6 ± 0.9
3.6 ± 0.8
References
13
40
110
110
19
19
20
35
95
140
135
References
143
17
17
144
References
13
20
19
19
140
35
Notes: LVH, left ventricular hypertrophy; DCM, nonischemic dilated cardiomyopathy; CM, nonischemic cardiomyopathy; CHF, congestive heart failure;
HCM, hypertrophic cardiomyopathy. Control values are listed only once per reference.
a Patients with NYHA class I–IV CHF.
b Control study for patients undergoing exercise training: training improved LV function but [PCr] and [ATP] were unchanged.
c Patients with NYHA class II–III CHF.
d Patients with LVH due to valve disease.
e Enzyme replacement therapy increased [PCr] to 7.1 ± 1.5 μmol g−1 (P = 0.012).
f Control study for double-blind evaluation of acute allopurinol treatment: allopurinol increased [PCr] by ∼11% and CK flux by 39%.20
g HCM due to a familial myosin heavy chain Arg403Gln mutation.
h HCM due to Fabry disease.
* P < 0.05 versus controls.
† P < 0.01 versus controls.
‡ P < 0.001 versus controls.
in several patients12 were also reported. PD was subsequently
found to correlate with blood DPG signal, and attributed to
blood contamination, exacerbated by the thin ventricular wall
in these patients (see section titled ‘Ratios’).12 The published
PCr/ATP ratios for DCM are summarized in Table 6. The
preponderance of studies suggests that myocardial PCr/ATP
is reduced on average by nearly 20% in DCM, an amount that
does not always reach statistical significance. The PCr/ATP
ratios generally correlate only weakly with functional and morphological indices of disease severity such as EF and/or fractional shortening. However, significant correlations between an
increasing severity of NYHA class and a decreasing PCr/ATP
have been reported.79,110,122 In addition, PCr/ATP has been
observed to recover in patients whose NYHA class improves
Volume 5, 2016
due to drug therapy.79 Conversely, a significantly higher risk
of cardiovascular death within 2.5 years was associated with
those who had myocardial PCr/ATP < 1.6, compared to those
with PCr/ATP > 1.6 (mortality 40% vs 11%, P < 0.02).34
Because the relative reduction in PCr represents a drop
in myocardial high-energy phosphate reserve, these findings
support the old hypothesis that the failing heart is energy
starved.2 The picture is reinforced by quantitative 31 P and 1 H
MRS measurements of myocardial [PCr], [ATP], and [CR] in
patients with CM and CHF, summarized in Table 7. Reductions
in both [ATP] and [PCr] mean that measurements of PCr/ATP
ratios often underestimate the loss in high-energy phosphate
reserves.110 Importantly, [CR] also declines with increasing
NYHA severity.144 Whether the fraction of the total creatine
© 2016 John Wiley & Sons, Ltd.
1195
PA Bottomley
pool that is phosphorylated is preserved – which is critical
for CK reaction kinetics in equations (2) and (7) – is as yet
unknown, but is at least within the grasp of MRS.16,64 As is
evident from Table 7, for all of these CM studies, the reductions
in the creatine pool are striking. Mindlessly averaging the tabulated results as a whole reveals a 30% reduction in [PCr] from
9.4 to 6.6 μmol g−1 (P < 0.001) and a 40% reduction in [CR]
from 27 to 18 μmol g−1 , which presumably all has something
to do with a 20% loss in [ATP] (P = 0.001).
Measurements of CK reaction kinetics in these patients
reveal that even more fuel has been taken from the fire. On
top of the reductions in [PCr] as substrate, the CK reaction
rate is also reduced by about a third (Table 7, lower panel).
Consequently, the CK ATP supply at rest in those exhibiting
symptoms of CHF is approximately half that in healthy subjects
(Figure 5).13 Given variable intracyclic energy demands, this
loss in CK ATP supply could be enough to compromise energy
supply, leading to contractile dysfunction and symptoms during periods of moderate exercise, by limiting the heart’s ability
to work (see section titled ‘Cyclic Energy Demand and Cardiac
Workload’).13
So what mechanism could slow the CK chemical reaction
rate in CHF patients? Focus shifts to the possibility that CK
enzyme is somehow being damaged, for example, by reactive
oxygen species (ROS) generated from increased xanthine oxidase (XO) activity, a terminal reaction in purine degradation. If
so, an XO inhibitor could reduce ROS, thereby increasing CK
activity and flux. A double-blinded saturation transfer study of
allopurinol, a ROS inhibitor used for treating gout, was recently
CK flux (μmol−1 (g. wt. s) )
5
4
Rest
200%
Stress
CHF
LVH
HCM
MI
DCM
LVH
3
*
2
†
†
1
0
n = 16
n=6
n = 11
n=9
n =15
†
†
n = 17 n = 10
Normal
Figure 5. The forward CK flux (micromole per gram wet weight per second) in healthy controls and patients with heart disease as measured by
saturation transfer 31 P MRS at 1.5 T. Healthy subjects at rest and with dobutamine stress at a 200% in workload (HR × BP) have the highest flux (left
two columns).13 Patients with pressure-overload LVH19 and anterior MI126
(third and fourth columns) were not in failure and had a normal CK rate
constant, k1 , but a reduced [PCr], which resulted in a significantly lower
CK flux – although only locally in the case of the MI patients. Patients with
familial HCM140 had both significantly reduced CK rates and a lower [PCr],
as well as symptoms of CHF (NYHA class I–III, but mostly I). DCM13
and LVH19 patients in the two right-most columns had CHF. These data
link a decline in CK ATP supply with heart failure (n, number of subjects;
*P < 0.01 and ** P < 0.001 vs healthy controls)
1196
reported in patients with CHF. Acute administration of allopurinol increased cardiac PCr/ATP and [PCr] by about 10%
(P < 0.02), but the mean CK flux – the CK ATP supply – went up
by nearly 40% (P < 0.007).20 This provides direct evidence that
cardiac energy delivery can be pharmaceutically augmented in
the failing heart. The CK reaction may thus be a viable target
for heart failure therapy, and 31 P MRS is a suitable – if not
unique – noninvasive tool for evaluating such approaches (see
MRS in the Failing Heart: From Mice to Humans).
Hypertrophy, Valve Disease, and Heart Failure. The first localized 31 P MRS study of HCM appeared in the late 1980s.72 Studying cardiac hypertrophy with MRS has the practical advantage
that the MRS signal from the thickened ventricle is less likely
to be contaminated by chamber blood whose lack of PCr could
systematically reduce PCr/ATP and [PCr] measurements (see
section titled ‘Ratios’). Common underlying causes of HCM
include familial and genetic factors, while HT and valve disease are common causes of left ventricular hypertrophy (LVH).
As in DCM, CHF is a common outcome, and from an MRS
standpoint, the CK metabolism of HCM and LVH are indistinguishable from DCM. Published myocardial PCr/ATP values
for nonischemic hypertrophic disease to date are summarized
in Table 8: the concentrations and fluxes are included in Table 7.
Overall, the 31 P MRS data suggest that myocardial PCr/ATP
is reduced by about 23% in hypertrophic disease, including
HCM of various genetic origins, LVH, and underlying HT
and valve disease. Myocardial PCr/ATP tends to decrease with
the presence and severity of symptoms of CHF,46,53,93,97,110
and/or the degree of severity of underlying valve disease,97,121
or dysfunction in HT (Table 8).121 Yet reduced PCr/ATP is
also seen in hypertrophy that is not explicitly linked to CHF
or severity.19,46,94,140 Indeed, a study of LVH patients who
had comparable cardiac function and morphology reported
that PCr/ATP was the same in those with and those without
symptoms of CHF: in fact, mean [PCr] was lowest in those
LVH patients who were not in heart failure, though the difference vs. CHF was not significant19 (Table 7). So high-energy
phosphate concentrations – mostly PCr – are reduced in
hypertrophy.19,95,110,135,140 As if to preserve the relative fractions of phosphorylated and unphosphorylated creatine, [CR]
is down by a similar fraction as [PCr] in the only reported
study (Table 7).17 While [CR] reductions were less in HCM
than in DCM in that study,17 this may reflect the more severe
disease present in the DCM cohort.
A couple of studies have reported elevated Pi in hypertrophic
disease.12,94 As in healthy subjects (see section titled ‘Normal
Values’) given the potential for blood DPG and PM contamination, the dearth of Pi data likely reflects quantification
difficulties rather than these findings being anomalous, a situation that may be resolved at higher B0 . As also noted in section
titled ‘Myocardial Ischemia’, a recent study of 40-year-olds with
genetic HCM found not only a lower resting cardiac PCr/ATP
ratio (1.71 ± 0.35 vs 2.14 ± 0.35 in controls, P < 0.0001), but
that PCr/ATP declined further (to 1.56 ± 0.29; P = 0.02) when
cardiac workload was increased 1.7-fold.48 These results, which
imply a CK response in HCM analogous to that seen with
transient ischemia due to coronary artery disease, differ from
previous, more-limited studies of isometric handgrip exercise
© 2016 John Wiley & Sons, Ltd.
Volume 5, 2016
MRS Studies of Creatine Kinase Metabolism in Human Heart
Table 8. Literature myocardial PCr/ATP ratios for patients with hypertrophy (LVH, HCM), including associated valve disease (VAL) and heart failure
(CHF)
Cause/CHF
LVH-mild
LVH-severe
VAL-CHF
VAL-non-CHF
VAL
HCM
HCM
LVH-HT, VAL
HCM
HCM
VAL, NYHA III
VAL, NYHA I-II
AS, CHF
AI, CHF
VAL, NYHA I-III
HCM
HCM
HCM-familial
HCM
MR-severe, NYHA I-III
MR-dyspnea
MR, NYHA I-III
MR-mild, NYHA I
MR-moderate, NYHA I-II
HT
LVH, VAL
AS, LVH
HCM, FRD
HT, non-CHF
AS, NYHA II-III
NYHA II, CM, AS
NYHA III, CM, AS
HCM-genetic origin
LVH, HT, CHF
LVH, HT, non-CHF
HT
HT
LVH, NYHA II-IIIe
HCM
HCMf, NYHA I-IV
HCM-genetic origin
HCM
Averagea
Controls
1.55 ± 0.2
0.89 ± 0.30
1.5 ± 0.2
—
1.76 ± 0.22
—
2.09 ± 0.44
—
1.65 ± 0.26
1.71 ± 0.32
2.02 ± 0.41
—
—
—
—
2.15 ± 0.51
2.46 ± 0.53
2.46 ± 0.53
1.6 ± 0.4
1.61 ± 0.3
—
—
—
—
1.39 ± 0.17
1.65 ± 0.21
1.43 ± 0.14
2.39 ± 0.13
1.59 ± 0.33
—
—
—
2.44 ± 0.3
1.9 ± 0.3
—
2.07 ± 0.17
—
2.14 ± 0.63
2.41 ± 0.3
1.92 ± 0.5
1.7 ± 0.3
2.14 ± 0.35
1.91 ± 0.35
Patients
0.9 ± 0.2
0.78 ± 0.16
0.8 ± 0.29
1.1 ± 0.32*
1.56 ± 0.15
1.34 ± 0.25
1.73 ± 0.23
1.43 ± 0.36*
1.89 ± 0.2
1.32 ± 0.29†
1.07 ± 0.44‡
1.51 ± 0.09* SE
1.86 ± 0.18 SE
1.55 ± 0.43‡
1.77 ± 0.36
1.64 ± 0.42‡
1.93 ± 0.57
1.98 ± 0.37†
1.81 ± 0.28†
1.6 ± 0.6
1.29 ± 0.3†
1.21 ± 0.24‡
1.29 ± 0.29‡
1.73 ± 0.17§
1.49 ± 0.18||
1.2 ± 0.18†
0.8 ± 0.25*
1.24 ± 0.17‡,b
1.47 ± 0.55*
No change
1.3 ± 0.2
1.36 ± 0.22
1.21 ± 0.3†
1.7 ± 0.43*
1.3 ± 0.5‡
1.3 ± 0.3‡
1.65 ± 0.25*,c
1.43 ± 0.21*,d
1.57 ± 0.53‡
2.18 ± 0.41†
1.68 ± 0.43
1.4 ± 0.4
1.71 ± 0.35*,g
1.47 ± 0.31*
References
72
46
46
53
53
67
67
81
81
12
83
93
93
93
93
93
94
54
103
95
97
97
97
97
97
99
102
57
109
110
110
110
110
115
19
19
121
121
125
130
135
140
48
Notes: Means ± SD as reported or calculated from SE where possible or specified SE where not; HT, hypertension; AS, aortic valve stenosis; AI, aortic
incompetence; MR, mitral regurgitation; FRD, Friedreich ataxia. Control values are listed only once per reference.
a Unweighted average of means, excluding Ref. 46 (not saturation corrected).
b PCr/ATP improved to 1.47 ± 0.14 (P < 0.05) after aortic valve replacement.
c Patients with diastolic dysfunction only.
d Patients with both diastolic and systolic dysfunction.
e Patients with preserved ejection fraction.
f Due to Fabry disease.
g Exercise that increased HR × BP by 70% reduced PCr/ATP further to 1.56 ± 0.29 (P = 0.02 vs rest).
* P < 0.001 versus controls.
† P < 0.05 versus controls.
‡ P < 0.01 versus controls.
§ P < 0.01 versus severe.
|| P < 0.05 versus mild MR.
Volume 5, 2016
© 2016 John Wiley & Sons, Ltd.
1197
PA Bottomley
in nonischemic patients with valve disease, HCM, or DCM,33
and with DCM patients undergoing dobutamine stress,80
although the diagnoses and levels of stress differ.
All told, the work in hypertrophic disease suggests a compromised CK energy reserve regardless of the underlying
cause. Disease severity, including but not limited to heart
failure, appears to exacerbate a deterioration in CK energy
metabolism. Measurements of CK reaction kinetics, k1 and
CK flux, might offer better specificity for CHF. LVH patients
with CHF have lower k1 , while the k1 in LVH patients with no
CHF but comparable cardiac function and morphology was
shown to be the same as in healthy controls (Table 7).19 The
combined effect of the reduction in k and reduced [PCr] means
that the CK ATP supply or flux, k1 ⋅[PCr], can be reduced
to as little as a third of that in healthy controls (1.1 ± 0.4 vs
3.1 ± 0.8 μmol g−1 s−1 ; P < 0.001).19 It is difficult to see how
a threefold hit to cardiac ATP supply would not negatively
impact cardiac energetics and function.
Indeed, a 31 P MRS study of 58 patients with a mix of nonischemic LVH or DCM, and with NYHA class I–IV heart failure
symptoms, found that myocardial CK flux was an independent
predictor of all-cause and cardiac death, CHF hospitalization,
cardiac transplantation, and ventricular-assist device placement at 5 years.35 The other independent predictors were
NYHA class, EF, and African-American race.
Other Specific Disorders. A number of other specific diseases
have been studied by cardiac 31 P MRS including congenital cardiomegaly in an infant;4 muscular dystrophy, cardiac beri–beri
and amyloidosis;81 and diseases associated with progressive
systemic sclerosis.86 All these showed reductions in myocardial PCr/ATP ratios compared to healthy subjects. Similarly,
patients with untreated familiar hypercholesterolemia are
reported as having reduced myocardial PCr/ATP compared to
statin-treated patients (1.78 ± 0.34 vs 2.15 ± 0.26, P < 0.001)
and healthy controls (2.04 ± 0.26, P < 0.01).116 PCr/ATP is
lower in patients with hereditary hemochromatosis as well
(1.6 ± 0.41 vs 1.93 ± 0.36 in healthy controls; P = 0.004).117
31
P MRS studies of type I diabetes mellitus (T1DB) show
a modest decrease in myocardial PCr/ATP (to 1.9 ± 0.4 vs
2.15 ± 0.3, P < 0.05) in patients with no other symptoms of
cardiac disease.111 More significant PCr/ATP reductions that
were not correlated with perfusion reserve as measured by
stress MRI, were reported in another study of asymptomatic
T1DB patients (1.5–1.6 vs 2.1 in controls; P < 0.0001). In
T1DB patients with uremia and diastolic dysfunction, an even
lower PCr/ATP has been reported (1.36 ± 0.4 vs 1.91 ± 0.18;
P < 0.01).119
Two studies of patients who had type II diabetes (T2DB)
but no coronary disease or systolic dysfunction, again showed
a reduced PCr/ATP (1.47 ± 0.28 vs 1.88 ± 0.34, P < 0.01;113
and 1.5 ± 0.11 vs 2.3 ± 0.12; P < 0.001).114 On the other hand,
another study found no change in PCr/ATP in T2DB patients
who had no cardiovascular disease or impaired perfusion
CHF.129
Heart Transplant Patients. The management of patients with
transplanted hearts requires the timely detection of graft rejection to assess whether augmented immunosuppressive therapy
1198
Table 9. Summary of MRS findings in common heart diseases through
2015
Disorder
MI acute
MI chronic
ISCH rest
ISCH exercise
LVH, HCM
DCM
CHF + CM
Valve disease
XPLANT
T1DB
T2DB
PCr/ATP
Pi
–
↓
–
↓
↓
↓
↓
↓
↓
↓
↓
↑
↑
↑
– or ↑
[PCr] [ATP]
↓
↓
↓
↓
↓
↓
↓
↓
k
CK flux [CR]
–
– or ↓ – or ↓
– or ↓
↓
– or ↓
↓
– or ↓
↓
↓
↓
↓
↓
↓
↓
Notes: XPLANT, transplanted heart (post operation); ↓, decrease versus
normal; ↑, increase versus normal; –, unchanged versus normal; – or ↓ (↑),
unchanged or reduced (increased) versus normal. Blank entries are not yet
determined.
is warranted. The standard of practice requires that endomyocardial biopsies be acquired at regularly scheduled cardiac
catheterization procedures, from which histological evidence
of myocyte necrosis is read. Clearly, a noninvasive method
of assessing rejection would be most welcome. The idea that
changes in myocardial metabolite ratios might predict histological rejection in human heart transplants arose from animal
31 P MRS studies of nonimmunosuppressed allografts, in which
metabolic changes and histological evidence for acute rejection
in the first week or so post-transplantation were seen.157–159
A first paper applying 31 P MRS to patients with heart
transplants studied up to 5.5 years post-transplantation did
find significantly lower resting anterior myocardial PCr/ATP
ratios compared to normal controls (1.57 ± 0.5 vs 1.93 ± 0.21;
P < 0.01).76 However, 31 P MRS agreed with histological evidence of necrosis that distinguished mild from moderate
rejection (which would trigger augmented therapy), in only
about 60–70% of examinations. Consequently, it was concluded
that 31 P MRS did not precisely predict significant histological
rejection in many transplant patients.76 This finding was confirmed in two other studies.104,160 One of these followed up
13 patients serially for 13–294 days post-transplantation and
found a significant correlation between improving PCr/ATP
and time post-transplantation, independent of rejection.160
That PCr/ATP ratios are imprecise predictors of histological
rejection likely reflects fundamental differences between histological and metabolic indices. Thus, myocyte necrosis, which is
key to the histological evaluation, cannot directly cause altered
PCr/ATP ratios because dead cells can contribute no highenergy phosphates. The correlation between PCr/ATP and time
post-transplantation160 suggests that the low PCr/ATP acutely
following transplantation is associated with the transplant procedure itself, with factors that include graft harvesting, storage,
duration of hypoxia, and so on. Metabolic recovery requires an
extended period of months, which are typically punctuated by
episodes of acute rejection. The use of 31 P MRS to assess the viability of excised donor hearts at transplantation is reviewed elsewhere (see Assessing Cardiac Transplant Viability with MRS).
© 2016 John Wiley & Sons, Ltd.
Volume 5, 2016
MRS Studies of Creatine Kinase Metabolism in Human Heart
Conclusions
31 P
1H
and
MRS can almost completely characterize CK
metabolism in the healthy and diseased human heart with
measurements of [PCr], [ATP], [CR], k1 , and CK flux. The
‘almost’ is because not all measures have been obtained from all
major diseases; [PCr] and [CR] have not both been measured
on the same subjects to quantify [Cr] unambiguously; [Pi] and
pH are not reliably quantified; nor is [ADP]. The measurements are made possible in patient examinations primarily by
sacrificing spatial resolution. Accordingly, focal disease – MI
and ischemia – have taken a back seat to studies of diseases
that affect the heart globally – CMs, CHF, obesity, diabetes, and
so on.
The vast majority of published work has focused on just a
single variable, the myocardial PCr/ATP ratio. It is clear from
Tables 2–8 that quantification is an ongoing challenge: the
normal control values are listed because the inter-laboratory
variations (e.g., 1.4–2.5 in Table 8) are often comparable to the
intralaboratory variations being attributed to various diseases.
Establishing reproducibility for measurements performed at
the same site, and between different sites, is essential for any
multicenter trial and for establishing normal ranges of values
that can be relevant to individual patients. As noted in section
titled ‘Normal Values’, about one-third of studies are able to
achieve SDs that are about 10% of the mean, so the rest of
the range of variation between 1.4 and 2.5 in control subjects
must relate to extraneous factors. A larger variation in the
values of k1 and flux is expected due to the accumulation of
factors required for those measurements: for example, when k1
is measured from four different spectra, plus a concentration
measurement for flux.13
Nevertheless, it is now evident that there are physiological
variations as well. Before in vivo MRS came along, some
assumed that the CK reaction in healthy hearts would behave
just like the CK reaction in skeletal muscle: as energy demand
increased, PCr would go down to supply more ATP, and
Pi would rise. After MRS showed that this did not happen,
the assumption then was that PCr/ATP was fixed: regulated
somehow to a single value whose scatter primarily reflected an
inability to measure it accurately. While the latter is true, intralaboratory comparisons of different cohorts of healthy subjects
with various conditions that are not normally designated
as heart disease (as reviewed in section titled ‘Physiological
Variations’) now indicate that there are variations in resting
myocardial PCr/ATP with age, cardiovascular fitness, hypertension, exposure to hypoxia, obesity, and diet that might be
considered ‘preclinical’. In patients, the presence of ischemic
disease, heart failure, CM, diabetes, or having had an MI or
a transplanted heart appear as even stronger risk factors for
a reduced CK energy reserve as indexed by a lower myocardial PCr/ATP ratio (Table 9). All these point not to a fixed
myocardial PCr/ATP ratio but to a meta-stable operating value
that may be eroded and compounded by one or more of the
aforementioned factors. The erosion probably commences with
a decline in the creatine pool, which noninvasive 1 H MRS is
ideally suited to document.
When CK ATP flux was measured in the human heart
with 31 P MRS, two of the possible outcomes could well have
Volume 5, 2016
rendered the CK reaction irrelevant to cardiac energetics. First,
if the rate was either much smaller than that of OXPHOS
or, secondly, if it far exceeded OXPHOS, then the changes
in creatine substrate levels, flux, or PCr/ATP being observed
could not have a relevant effect on energy supply in the context
of ongoing energy demands. Instead, the measured CK flux
fell right in a range of about 7–10 times OXPHOS, and does
not appear to vary much with cardiac workload in the healthy
heart. This means that the reductions in flux and/or ATP
supply seen in patients with ischemia, CM, or heart failure
could be enough during stress to consume some of the PCr
reserve and ultimately limit cardiac work.13,33 While we do not
know yet whether limited CK reserve limits physical activity,
we do know that compromised CK metabolism is a long-term
predictor of poor cardiovascular outcomes.34,35 Given that
predicting bad things is less rewarding than fixing them, it is
most encouraging to see cardiac 31 P MRS metrics being used
to quantify the status of the cardiovascular energy reserve in
order to evaluate the benefits of both pharmaceutical20,153,156
and lifestyle interventions.40,127,134,137,151
Acknowledgments
I thank RG Weiss for years of fruitful collaboration and discussions and the National Institutes of Health for past support.
Biographical Sketch
Paul A. Bottomley. BSc (Hon), 1975, PhD, 1978, Physics, University of Nottingham, United Kingdom. Research Associate, Johns
Hopkins University, Baltimore, 1978–1980. Physicist, G. E. Research
and Development Center, 1980–1994. Currently Russell H Morgan
Professor and Director of the Division of MR Research of the Russell
H Morgan Department of Radiology and Radiological Sciences, Johns
Hopkins University. Fellow and Gold Medal recipient of the Society
of Magnetic Resonance in Medicine, 1989; Coolidge Fellowship and
medal, G.E. Company, 1990; Gold, Silver, and Bronze patent medals,
G.E. Company; Gold Medal, American Roentgen Ray Society 2015;
Member National Academy of Inventors; over 40 issued patents, about
180 peer-reviewed papers, 24 book chapters, 13 editorials, and over
225 published abstracts. Research specialties: in vivo NMR, MRI,
tissue relaxation times, localized NMR spectroscopy, human cardiac
NMR spectroscopy, interventional MRI, MRI safety.
Related Articles
Cardiac MRS Studies in Rodents and Other Animals; MRS in
the Failing Heart: From Mice to Humans; Cardiac Lipids by 1 H
MRS; Assessing Cardiac Transplant Viability with MRS; Measuring Biochemical Reaction Rates In Vivo with Magnetization
Transfer; Quantifying Metabolite Ratios and Concentrations by
Non-1 H MRS
References
1. R. B. Jennings and K. A. Reimer, Am. J. Pathol., 1981, 102, 241.
2. S. Neubauer, N. Engl. J. Med., 2007, 356, 1140.
3. P. A. Bottomley, Science, 1985, 229, 769.
4. J. R. Whitman, B. Chance, H. Bode, J. Maris, J. Haselgrove, R. Kelley, B. J.
Clark, and A. H. Harken, J. Am. Coll. Cardiol., 1985, 5, 745.
© 2016 John Wiley & Sons, Ltd.
1199
PA Bottomley
5. W. E. Jacobus, G. J. Taylor, D. P. Hollis, and R. L. Nunnally, Nature, 1977, 263,
756.
34. S. Neubauer, M. Horn, M. Cramer, K. Harre, J. B. Newell, W. Peters, T. Pabst,
G. Ertl, D. Hahn, J. S. Ingwall, and K. Kochsiek, Circulation, 1997, 96, 2190.
6. P. B. Garlick, G. K. Radda, P. J. Seeley, and B. Chance, Biochem. Biophys. Res.
Commun., 1977, 74, 1256.
35. P. A. Bottomley, G. S. Panjrath, S. Lai, G. A. Hirsch, K. Wu, S. S. Najjar, A.
Steinberg, G. Gerstenblith, and R. G. Weiss, Sci. Trans. Med., 2013, 5, 215re3.
7. T. H. Grove, J. J. H. Ackerman, G. K. Radda, and P. J. Bore, Proc. Natl. Acad.
Sci. U. S. A., 1980, 77, 299.
36. D. I. Hoult and P. C. Lauterbur, J. Magn. Reson., 1979, 34, 425.
8. R. L. Nunnally and P. A. Bottomley, Science, 1981, 211, 177.
9. P. A. Bottomley, H. R. Hart, W. A. Edelstein, J. F. Schenck, L. S. Smith, W. M.
Leue, O. M. Mueller, and R. W. Redington, Lancet, 1983, 322, 273.
10. P. A. Bottomley, H. R. Hart, W. A. Edelstein, J. F. Schenck, L. S. Smith, W. M.
Leue, O. M. Mueller, and R. W. Redington, Radiology, 1984, 150, 441.
11. D. G. Gadian, Nuclear Magnetic Resonance and its Applications to Living
Systems, Oxford University Press: Oxford, 1982, 30.
12. A.de Roos, J. Doornbos, P. R. Luyten, L. J. M. P. Oosterwaal, E. E.van der Wall,
and J. A.den Hollander, J. Magn. Reson. Imaging, 1992, 2, 711.
13. R. G. Weiss, G. Gerstenblith, and P. A. Bottomley, Proc. Natl. Acad Sci. U. S.
A., 2005, 102, 808.
14. J. A.den Hollander, W. T. Evanochko, and G. M. Pohost, Magn. Reson. Med.,
1994, 32, 175.
15. L. S. Szczepaniak, R. L. Dobbins, G. J. Metzger, G. Sartoni-D’Ambrosia,
D. Arbique, W. Vongpatanasin, R. Unger, and R. G. Victor, Magn. Reson.
Med., 2003, 49, 417.
37. P. R. Luyten, G. Bruntink, F. M. Sloff, J. I. Vermeulen, J. I.van der Heijden, J.
A.den Hollander, and A. Heerschap, NMR Biomed., 1989, 1, 177.
38. P. A. Bottomley and C. J. Hardy, Magn. Reson. Med., 1992, 24, 384.
39. M. F. H. Schocke, B. Metzler, C. Wolf, P. Steinboeck, C. Kremser, O. Pachinger,
W. Jaschke, and P. Lukas, Magn. Reson. Imaging, 2003, 21, 553.
40. M. Beer, D. Wagner, J. Myers, J. Sandstede, H. Köstler, D. Hahn, S. Neubauer,
and P. Dubach, J. Am. Coll. Cardiol., 2008, 51, 1883.
41. D. J. Tyler, Y. Emmanuel, L. E. Cochlin, L. E. Hudsmith, C. J. Holloway,
S. Neubauer, K. Clarke, and M. D. Robson, NMR Biomed., 2009, 22, 405.
42. P. A. Bottomley, C. H. Lugo-Olivieri, and R. Giaquinto, Magn. Reson. Med.,
1997, 37, 591.
43. C. J. Hardy, P. A. Bottomley, K. W. Rohling, and P. B. Roemer, Magn. Reson.
Med., 1992, 28, 54.
44. P. A. Bottomley and C. J. Hardy, Phil. Trans. R. Soc. Lond. A, 1990, 333, 531.
45. P. A. Bottomley, C. J. Hardy, and P. B. Roemer, Magn. Reson. Med., 1990, 14,
425.
16. P. A. Bottomley and R. G. Weiss, Radiology, 2001, 219, 411.
46. S. Schaefer, J. R. Gober, G. G. Schwartz, D. B. Twieg, M. W. Weiner, and
B. Massie, Am. J. Cardiol., 1990, 65, 1154.
17. I. Nakae, K. Mitsunami, T. Omura, T. Yabe, T. Tsutamoto, S. Matsuo, M. Takahashi, S. Morikawa, T. Inubushi, Y. Nakamura, M. Kinoshita, and M. Horie, J.
Am. Coll. Cardiol., 2003, 42, 1587.
47. S. Schaefer, J. Gober, M. Valenza, G. S. Karczmar, G. B. Matson, S. A. Camacho,
E. H. Botvinick, B. Massie, and M. W. Weiner, J. Am. Coll. Cardiol., 1988, 12,
1149.
18. P. A. Bottomley and R. G. Weiss, Lancet, 1998, 351, 714.
19. C. S. Smith, P. A. Bottomley, S. P. Schulman, G. G. Gerstenblith, and R. G.
Weiss, Circulation, 2006, 114, 1151.
48. S. Dass, L. E. Cochlin, J. J. Suttie, C. J. Holloway, O. J. Rider, L. Carden, D. J.
Tyler, T. D. Karamitsos, K. Clarke, S. Neubauer, and H. Watkins, Eur. Heart. J.,
2015, 36, 1547.
20. G. A. Hirsch, P. A. Bottomley, G. Gerstenblith, and R. G. Weiss, J. Am. Coll.
Cardiol., 2012, 59, 802.
49. R. Löffler, R. Sauter, H. Kolem, A. Haase, and M.von Kienlin, J. Magn. Reson.,
1998, 134, 287.
21. P. A. Bottomley, C. J. Hardy, P. B. Roemer, and O. M. Mueller, Magn. Reson.
Med., 1989, 12, 348.
50. Y. Zhang, R. E. Gabr, M. Schär, R. G. Weiss, and P. A. Bottomley, J. Magn.
Reson., 2012, 218, 66. ol., 1988, 12, 1449.
22. M. A. Schroeder, K. Clarke, S. Neubauer, and D. J. Tyler, Circulation, 2011,
124, 1580.
23. T. Wallimann, Curr. Biol., 1994, 4, 42.
24. P. P. Dzeja and A. Terzic, J. Exp. Biol., 2003, 206, 2039.
25. P. A. Bottomley, in Encyclopedia of Magnetic Resonance, eds R. K. Harris and R. E. Wasylishen, John Wiley & Sons, Ltd: Chichester, 2009. DOI:
10.1002/9780470034590.emrstm0345.pub2.
26. C. T. Rodgers, W. T. Clarke, C. Snyder, J. T. Vaughan, S. Neubauer, and M. D.
Robson, Magn. Reson. Med., 2014, 72, 304.
27. V. Stoll, W. T. Clarke, E. Levelt, S. G. Myerson, M. D. Robson, S. Neubauer,
and C. Rodgers, J. Cardiovasc. Magn. Reson., 2015, 17, 249.
28. B. Schaller, W. T. Clarke, S. Neubauer, M. D. Robson, and C. T. Rodgers, Magn.
Reson. Med., 2015. DOI: 10.1002/mrm.25755.
29. A. E. M. El-Sharkawy, R. E. Gabr, M. Schar, R. G. Weiss, and P. A. Bottomley,
NMR Biomed., 2013, 26, 1363.
51. R. Pohmann and M.von Kienlin, Magn. Reson. Med., 2001, 45, 817.
52. P. W. Hochachka, C. M. Clark, J. E. Holden, C. Stanley, K. Ugurbil, and R. S.
Menon, Proc. Natl. Acad. Sci. U. S. A., 1996, 93, 1215.
53. M. A. Conway, J. Allis, R. Ouwerkerk, T. Niioka, B. Rajagopalan, and G. K.
Radda, Lancet, 1991, 338, 973.
54. W. I. Jung, L. Sieverding, J. Breuer, T. Hoess, S. Widmaier, O. Schmidt,
M. Bunse, F.van Erckelens, J. Apitz, O. Lutz, and G. J. Dietze, Circulation,
1998, 97, 2536.
55. G. Klug, R. H. Zwick, M. Frick, C. Wolf, M. F. H. Schocke, E. Conci, W. Jaschke,
O. Pachinger, and B. Metzler, Int. J. Sports Med., 2007, 28, 667.
56. R. S. Menon, K. Hendrich, X. Hu, and K. Ugurbil, Magn. Reson. Med., 1992,
26, 368.
57. H. P. Beyerbacht, H. J. Lamb, A.van der Laarse, H. W. Vliegen, F. Leujes, M. G.
Hazekamp, A.de Roos, and E. E.van der Wall, Radiology, 2001, 219, 637.
58. J. Felblinger, B. Jung, J. Slotboom, C. Boesch, and R. Kreis, Magn. Reson.
Med., 1999, 42, 903.
30. M. Schar, A. E. M. El-Sharkawy, R. G. Weiss, and P. A. Bottomley, Magn.
Reson. Med., 2010, 63, 1493.
59. M. Schär, S. Kozerke, and P. Boesiger, Magn. Reson. Med., 2004, 51, 1091.
31. A. Bashir and R. Gropler, NMR Biomed., 2014, 27, 663.
60. P. A. Bottomley, C. J. Hardy, P. B. Roemer, and R. G. Weiss, NMR Biomed.,
1989, 2, 284.
32. M. Schär, R. E. Gabr, A. M. El-Sharkawy, A. Steinberg, P. A. Bottomley, and R.
G. Weiss, J. Cardiovasc. Magn. Reson., 2015, 17, 70. DOI: 10.1186/s12968015-0175-4.
33. R. G. Weiss, P. A. Bottomley, C. J. Hardy, and G. Gerstenblith, N. Engl. J. Med.,
1990, 323, 1593.
1200
61. Y. Zhang, J. Zhou, and P. A. Bottomley, Magn. Reson. Med., 2015, 74, 320.
62. P. A. Bottomley and R. Ouwerkerk, Magn. Reson. Med., 1994, 32, 137.
63. J. O. Van Dobbenburgh, C. Lekkerkerk, R.de Beer, and C. J. A. Van Echteld,
NMR Biomed., 1994, 7, 218.
© 2016 John Wiley & Sons, Ltd.
Volume 5, 2016
MRS Studies of Creatine Kinase Metabolism in Human Heart
64. I. Nakae, K. Mitsunami, T. Yabe, T. Inubushi, S. Morikawa, S. Matsuo, T. Koh,
and M. Horie, J. Cardiovasc. Magn. Reson., 2004, 6, 685.
92. H. J. Lamb, H. P. Beyerbacht, R. Ouwerkerk, J. Doornbos, B. M. Pluim, E. E.van
der Wall, A.van der Laarse, and A.de Roos, Circulation, 1997, 96, 2969.
65. D. J. Tyler, L. E. Hudsmith, K. Clarke, S. Neubauer, and M. D. Robson, NMR
Biomed., 2008, 21, 793.
93. S. Neubauer, M. Horn, T. Pabst, K. Harre, H. Stromer, G. Bertsch, J. Sandstede,
G. Ertl, D. Hahn, and K. Kochsiek, J. Invest. Med., 1997, 45, 453.
66. A. E. M. El-Sharkawy, M. Schär, R. Ouwerkerk, R. G. Weiss, and P. A. Bottomley, Magn. Reson. Med., 2009, 61, 785.
94. L. Sieverding, W. I. Jung, J. Breuer, S. Widmaier, A. Staubert, F.van Erckelens,
O. Schmidt, M. Bunse, T. Hoess, O. Lutz, G. J. Dietze, and J. Apitz, Am. J.
Cardiol., 1997, 80, 34A.
67. P. A. Bottomley, C. J. Hardy, and R. G. Weiss, J. Magn. Reson., 1991, 95, 341.
68. C. J. Hardy, R. G. Weiss, P. A. Bottomley, and G. Gerstenblith, Am. Heart J.,
1991, 122, 795.
69. P. A. Bottomley, E. A. Atalar, and R. G. Weiss, Magn. Reson. Med., 1996, 35,
664.
70. C. Gibbs, J. Mol. Cell. Cardiol., 1985, 17, 727.
71. P. A. Bottomley, R. J. Herfkens, L. S. Smith, and T. M. Bashore, Radiology,
1987, 165, 703.
95. M. Okada, K. Mitsunami, T. Inubushi, and M. Kinoshita, Magn. Reson. Med.,
1998, 39, 772.
96. B. M. Pluim, H. J. Lamb, H. W. M. Kayser, F. Leujes, H. P. Beyerbacht, A. H.
Zwinderman, A.van der Laarse, H. W. Vliegen, A.de Roos, and E. E.van der
Wall, Circulation, 1998, 97, 666.
97. M. A. Conway, P. A. Bottomley, R. Ouwerkerk, G. K. Radda, and B.
Rajagopalan, Circulation, 1998, 97, 1716.
72. B. Rajagopalan, M. J. Blackledge, W. J. McKenna, N. Bolas, and G. K. Radda,
Ann. N. Y. Acad. Sci., 1987, 508, 321.
98. M. Meininger, W. Landschutz, M. Beer, T. Seyfarth, M. Horn, T. Pabst, A.
Haase, D. Hahn, S. Neubauer, and M.von Kienlin, Magn. Reson. Med., 1999,
41, 657.
73. M. A. Conway, J. D. Bristow, M. J. Blackledge, B. Rajagopalan, and G. K.
Radda, Lancet, 1988, 332, 692.
99. H. J. Lamb, H. P. Beyerbacht, A.van der Laarse, B. C. Stoel, J. Doornbos,
E. E.van der Wall, and A.de Roos, Circulation, 1999, 99, 2261.
74. T. M. Grist, J. B. Kneeland, W. R. Rilling, A. Jesmanowicz, W. Froncisz, and
J. S. Hyde, Radiology, 1989, 170, 357.
100. J. O. Van Dobbenburgh, M. C. H. De Groot, N. De Jonge, C. Klopping, J. R.
Lahpor, S. R. Woolley, E. O. R.de Medina, and C. J. A. Van Echteld, NMR
Biomed., 1999, 12, 515.
75. M. A. Conway, J. D. Bristow, M. J. Blackledge, B. Rajagopalan, and G. K.
Radda, Br. Heart J., 1991, 65, 25.
76. P. A. Bottomley, R. G. Weiss, C. J. Hardy, and W. A. Baumgartner, Radiology,
1991, 181, 67.
101. M. Beer, J. Sandstede, W. Landschutz, M. Viehrig, K. Harre, M. Horn, M.
Meininger, T. Pabst, W. Kenn, A. Haase, M.von Kienlin, S. Neubauer, and
D. Hahn, Eur. Radiol., 2000, 10, 1323.
77. W. Auffermann, W. M. Chew, C. L. Wolfe, N. J. Tavares, W. W. Parmley, R. C.
Semelka, T. Donnelly, K. Chatterjee, and C. B. Higgins, Radiology, 1991, 179,
253.
102. M. Beer, M. Viehrig, T. Seyfarth, J. Sandstede, C. Lipke, T. Pabst, W. Kenn, K.
Harre, M. Horn, W. Landschutz, M.von Kienlin, S. Neubauer, and D. Hahn,
Radiologie, 2000, 40, 162.
78. K. Mitsunami, M. Okada, T. Inoue, M. Hachisuka, M. Kinoshita, and T.
Inubushi, Jpn. Circ. J., 1992, 56, 614.
103. W. I. Jung, T. Hoess, M. Bunse, S. Widmaier, L. Sieverding, J. Breuer, J. Apitz,
O. Schmidt, F.van Erckelens, G. J. Dietze, and O. Lutz, Circulation, 2000, 101,
e121.
79. S. Neubauer, T. Krahe, R. Schindler, M. Horn, H. Hillenbrand, C. Entzeroth,
H. Mader, E. P. Kromer, G. A. J. Riegger, K. Lackner, and G. Ertl, Circulation,
1992, 86, 1810.
80. S. Schaefer, G. G. Schwartz, S. K. Steinman, D. J. Meyerhoff, B. M. Massie,
and W. M. Weiner, Magn. Reson. Med., 1992, 25, 260.
81. Y. Masuda, Y. Tateno, H. Ikehira, T. Hashimoto, F. Shishido, M. Sekiya, Y.
Imazeki, H. Imai, S. Watanabe, and Y. Inagaki, Jpn. Circ. J., 1992, 56, 620.
82. S. Neubauer, T. Krahe, R. Schindler, H. Hillenbrand, C. Entzeroth, M. Horn, W.
R. Bauer, T. Stephan, K. Lackner, A. Haase, and G. Ertl, Magn. Reson. Med.,
1992, 26, 300.
83. H. Sakuma, K. Takeda, T. Tagami, T. Nakagawa, S. Okamoto, T. Konishi, and
T. Nakano, Am. Heart J., 1993, 125, 1323.
84. H. Sakuma, S. J. Nelson, D. B. Vigneron, J. Hartiala, and C. B. Higgins, Magn.
Reson. Med., 1993, 29, 688.
85. T. Yabe, K. Mitsunami, M. Okada, S. Morikawa, T. Inubushi, and M. Kinoshita,
Circulation, 1994, 89, 1709.
86. J. Doornbos, P. R. Luyten, M. Janssen, M. Wasser, and A. De Roos, J. Magn.
Reson. Imaging, 1994, 4, 165.
87. T. Yabe, K. Mitsunami, T. Inubushi, and M. Kinoshita, Circulation, 1995, 92,
15.
88. H. P. Hetherington, D. J. E. Luney, J. T. Vaughan, J. W. Pan, S. L. Ponder, O.
Tschendel, D. B. Twieg, and G. M. Pohost, Magn. Reson. Med., 1995, 33, 427.
89. S. Neubauer, M. Horn, T. Pabst, M. Godde, D. Lubke, B. Jilling, D. Hahn, and
G. Ertl, Eur. Heart J., 1995, 16, 115.
90. H. J. Lamb, J. Doornbos, J. A.den Hollander, P. R. Luyten, H. P. Beyerbacht,
E. E.van der Wall, and A.de Roos, NMR Biomed., 1996, 9, 217.
91. R. Kalil-Filho, C. P.de Albuquerque, R. G. Weiss, A. Mocelim, G. Bellotti,
G. Cerri, and F. Pileggi, J. Am. Coll. Cardiol., 1997, 30, 1228.
Volume 5, 2016
104. S. D. Buchthal, T. O. Noureuil, J. A.den Hollander, R. C. Bourge, J. K. Kirklin, C.
R. Katholi, J. B. Caulfield, G. M. Pohost, and W. T. Evanochko, J. Cardiovasc.
Magn. Reson., 2000, 2, 51.
105. J. G. Crilley, E. A. Boehm, B. Rajagopalan, A. M. Blamire, P. Styles, F. Muntoni,
D. Hilton-Jones, and K. Clarke, J. Am. Coll. Cardiol., 2000, 36, 1953.
106. H. Kostler, M. Beer, W. Landschutz, S. Buchner, J. Sandstede, T. Pabst, W.
Kenn, S. Neubauer, M.von Kienlin, and D. Hahn, ROFO, 2001, 173, 1093.
107. M.von Kienlin, M. Beer, A. Greiser, D. Hahn, K. Harre, H. Kostler, W. Landschutz, T. Pabst, J. Sandstede, and S. Neubauer, J. Magn. Reson. Imaging,
2001, 13, 521.
108. M. Beer, S. Buchner, J. Sandstede, M. Viehrig, C. Lipke, A. Krug, H. Kostler,
T. Pabst, W. Kenn, W. Landschutz, M.von Kienlin, K. Harre, S. Neubauer, and
D. Hahn, Magn. Reson. Mater. Phys., 2001, 13, 70.
109. R. Lodi, B. Rajagopalana, A. M. Blamire, J. M. Cooper, C. H. Davies, J. L.
Bradley, P. Styles, and A. H. V. Schapira, Cardiovasc. Res., 2001, 52, 111.
110. M. Beer, T. Seyfarth, J. Sandstede, W. Landschutz, C. Lipke, H. Kostler, M.von
Kienlin, K. Harre, D. Hahn, and S. Neubauer, J. Am. Coll. Cardiol., 2002, 40,
1267.
111. B. Metzler, M. F. H. Schocke, P. Steinboeck, C. Wolf, W. Judmaier, M. Lechleitner, P. Lukas, and O. Pachinger, J. Cardiovasc. Magn. Reson., 2002, 4, 493.
112. W. T. Evanochko, S. D. Buchthal, J. A.den Hollander, C. R. Katholi, R. C. Bourge,
R. L. Benza, J. K. Kirklin, and G. M. Pohost, J. Heart Lung Transplant., 2002,
21, 522.
113. M. Diamant, H. J. Lamb, Y. Groeneveld, E. L. Endert, J. W. A. Smit, J. J. Bax, J.
A. Romijn, A.de Roos, and J. K. Radder, J. Am. Coll. Cardiol., 2003, 42, 328.
114. M. Scheuermann-Freestone, P. L. Madsen, D. Manners, A. M. Blamire, R. E.
Buckingham, P. Styles, G. K. Radda, S. Neubauer, and K. Clarke, Circulation,
2003, 107, 3040.
© 2016 John Wiley & Sons, Ltd.
1201
PA Bottomley
115. J. G. Crilley, E. A. Boehm, E. Blair, B. Rajagopalan, A. M. Blamire, P. Styles, W. J.
McKenna, I. Ostman-Smith, K. Clarke, and H. Watkins, J. Am. Coll. Cardiol.,
2003, 41, 1776.
116. M. F. H. Schocke, M. Martinek, C. Kremser, C. Wolf, P. Steinboeck, M. Lechleitner, W. Jaschke, O. Pachinger, and B. Metzler, J. Cardiovasc. Magn. Reson.,
2003, 5, 595.
117. M. F. H. Schocke, H. Zoller, W. Vogel, C. Wolf, C. Kremser, P. Steinboeck, G.
Poelzl, O. Pachinger, W. R. Jaschke, and B. Metzler, Magn. Reson. Imaging,
2004, 22, 515.
118. T. Caus, F. Kober, A. Mouly-Bandini, A. Riberi, D. R. Metras, P. J. Cozzone, and
M. Bernard, Eur. J. Cardiothorac. Surg., 2005, 28, 576.
119. G. Perseghin, P. Fiorina, F. De Cobelli, P. Scifo, A. Esposito, T. Canu, M. Danna,
C. Gremizzi, A. Secchi, L. Luzi, and A. Del Maschio, J. Am. Coll. Cardiol., 2005,
46, 1085.
135. W. Machann, F. Breunig, F. Weidemann, J. Sandstede, D. Hahn, H. Kostler, S.
Neubauer, C. Wanner, and M. Beer, Eur. J. Heart Fail., 2011, 13, 278.
136. C. J. Holloway, L. E. Cochlin, I. Codreanu, E. Bloch, M. Fatemian, C. Szmigielski, H. Atherton, L. Heather, J. Francis, S. Neubauer, P. Robbins, H. E. Montgomery, and K. Clarke, FASEB J., 2011, 25, 3130.
137. C. J. Holloway, H. E. Montgomery, A. J. Murray, L. E. Cochlin, I. Codreanu, N.
Hopwood, A. W. Johnson, O. J. Rider, D. Z. H. Levett, D. J. Tyler, J. M. Francis,
S. Neubauer, M. P. W. Grocott, and K. Clarke, FASEB J., 2011, 25, 792.
138. O. J. Rider, J. M. Francis, M. K. Ali, C. Holloway, T. Pegg, M. D. Robson, D.
Tyler, J. Byrne, K. Clarke, and S. Neubauer, Circulation, 2012, 125, 1511.
139. O. Geier, A. M. Weng, A. Toepell, D. Hahn, M. Spindler, M. Beer, and H. Köstler,
Z. Med. Phys., 2014, 24, 49.
140. M. R. Abraham, P. A. Bottomley, V. L. Dimaano, A. Pinheiro, A. Steinberg, T.
A. Traill, T. P. Abraham, and R. G. Weiss, Am. J. Cardiol., 2013, 112, 861.
120. H. Kostler, W. Landschutz, S. Koeppe, T. Seyfarth, C. Lipke, J. Sandstede, M.
Spindler, M.von Kienlin, D. Hahn, and M. Beer, Magn. Reson. Med., 2006, 56,
907.
142. R. E. Gabr, R. G. Weiss, and P. A. Bottomley, J. Magn. Reson., 2008, 191, 248.
121. J. P. Heyne, R. Rzanny, A. Hansch, U. Leder, J. R. Reichenbach, and W. A.
Kaiser, Eur. Radiol., 2006, 16, 1796.
143. I. Nakae, K. Mitsunami, S. Matsuo, T. Matsumoto, S. Morikawa, T. Inubushi,
T. Koh, and M. Horie, Magn. Reson. Med. Sci., 2004, 3, 19.
122. A. Hansch, R. Rzanny, J. P. Heyne, U. Leder, J. R. Reichenbach, and W. A.
Kaiser, Eur. Radiol., 2005, 15, 319.
144. I. Nakae, K. Mitsunami, S. Matsuo, T. Inubushi, S. Morikawa, T. Tsutamoto, T.
Koh, and M. Horie, Circ. J., 2005, 69, 711.
123. G. Perseghin, F. De Cobelli, A. Esposito, G. Lattuada, I. Terruzzi, A. La Torre,
E. Belloni, T. Canu, P. Scifo, A. Del Maschio, L. Luzi, and G. Alberti, Am. Heart.
J., 2007, 154, 937.
145. G. J. Kemp, M. Meyerspeer, and E. Moser, NMR Biomed., 2007, 20, 555.
124. G. Perseghin, G. Lattuada, F. De Cobelli, A. Esposito, E. Belloni, G. Ntali,
F. Ragogna, T. Canu, P. Scifo, A. Del Maschio, and L. Luzi, Hepatology, 2008,
47, 51.
125. T. T. Phan, K. Abozguia, G. N. Shivu, G. Mahadevan, I. Ahmed, L. Williams,
G. Dwivedi, K. Patel, P. Steendijk, H. Ashrafian, A. Henning, and M. Frenneaux,
J. Am. Coll. Cardiol., 2009, 54, 402.
126. P. A. Bottomley, K. C. Wu, G. Gerstenblith, S. P. Schulman, A. Steinberg, and
R. G. Weiss, Circulation, 2009, 119, 1918.
127. G. Perseghin, F. De Cobelli, A. Esposito, E. Belloni, G. Lattuada, T. Canu, P. L.
Invernizzi, F. Ragogna, A. La Torre, P. Scifo, G. Alberti, A. Del Maschio, and L.
Luzi, Heart, 2009, 95, 630.
128. L. E. Hudsmith, D. J. Tyler, Y. Emmanuel, S. E. Petersen, J. M. Francis, H.
Watkins, K. Clarke, M. D. Robson, and S. Neubauer, Int. J. Cardiovasc. Imaging, 2009, 25, 819.
129. L. J. Rijzewijk, R. W.van der Meer, H. J. Lamb, H. W. A. M.de Jong, M. Lubberink, J. A. Romijn, J. J. Bax, A.de Roos, J. W. Twisk, R. J. Heine, A. A.
Lammertsma, J. W. A. Smit, and M. Diamant, J. Am. Coll. Cardiol., 2009, 54,
1524.
130. A. Esposito, F. De Cobelli, G. Perseghin, M. Pieroni, E. Belloni, R. Mellone, T.
Canu, F. Gentinetta, P. Scifo, C. Chimenti, A. Frustaci, L. Luzi, A. Maseri, and
A. Del Maschio, Heart, 2009, 95, 228.
131. G. N. Shivu, T. T. Phan, K. Abozguia, I. Ahmed, A. Wagenmakers, A. Henning,
P. Narendran, M. Stevens, and M. Frenneaux, Circulation, 2010, 121, 1209.
132. G. N. Shivu, K. Abozguia, T. T. Phan, I. Ahmed, A. Henning, and M. Frenneaux,
Eur. J. Radiol., 2010, 73, 255.
133. D. E. J. Jones, K. Hollingsworth, G. Fattakhova, G. MacGowan, R. Taylor, A.
Blamire, and J. L. Newton, Am. J. Physiol. Gastrointest. Liver Physiol., 2010,
298, G764.
134. C. J. Holloway, L. E. Cochlin, Y. Emmanuel, A. Murray, I. Codreanu, L. M.
Edwards, C. Szmigielski, D. J. Tyler, N. S. Knight, B. K. Saxby, B. Lambert, C.
Thompson, S. Neubauer, and K. Clarke, Am. J. Clin. Nutr., 2011, 93, 748.
1202
141. P. A. Bottomley and C. J. Hardy, J. Magn. Reson., 1992, 99, 443.
146. P. A. Bottomley, Y. H. Lee, and R. G. Weiss, Radiology, 1997, 204, 403.
147. K. T. Sun, L. A. Yeatman, D. B. Buxton, K. Chen, J. A. Johnson, S.-C. Huang,
K. F. Kofoed, S. Weismueller, J. Czernin, M. E. Phelps, et al., J. Nucl. Med.,
1998, 39, 272.
148. B. L. Gerber, W. Wijns, J. J. Vanoverschelde, G. R. Heyndrickx, B. DeBruyne,
J. Bartunek, and J. A. Melin, J. Am. Coll. Cardiol., 1999, 34, 1939.
149. A. C. Cave, J. S. Ingwall, J. Friedrich, R. Liao, K. W. Saupe, C. S. Apstein, and
F. R. Eberli, Circulation, 2000, 101, 2090.
150. K. Weiss, P. A. Bottomley, and R. G. Weiss, NMR Biomed., 2015, 28, 694.
151. O. J. Rider, J. M. Francis, D. Tyler, J. Byrne, K. Clarke, and S. Neubauer, Int. J.
Cardiovasc. Imaging, 2013, 29, 1043.
152. M. Beer, M. Spindler, J. J. W. Sandstede, H. Remmert, S. Beer, H. Kostler, and
D. Hahn, J. Magn. Reson. Imaging, 2004, 20, 798.
153. S. S. Najjar, P. A. Bottomley, S. P. Schulman, M. M. Waldron, R. P. Steffen,
G. Gerstenblith, and R. G. Weiss, J. Cardiovasc. Magn. Reson., 2005, 7, 1.
154. P. A. Bottomley, L. S. Smith, S. Brazzamano, L. W. Hedlund, R. W. Redington,
and R. J. Herfkens, Magn. Reson. Med., 1987, 5, 129.
155. S. D. Buchthal, J. A. Den Hollander, C. N. B. Merz, W. J. Rogers, C. J. Pepine,
N. Reichek, B. L. Sharaf, S. Reis, S. F. Kelsey, and G. M. Pohost, N. Engl. J.
Med., 2000, 342, 829.
156. G. Fragasso, G. Perseghin, F.de Cobelli, A. Esposito, A. Palloshi, G. Lattuada,
P. Scifo, G. Calori, A.del Maschio, and A. Margonato, Eur. Heart J., 2006, 27,
942.
157. R. C. Canby, W. T. Evanochko, L. V. Barrett, J. K. Kirklin, D. C. McGiffen, T. T.
Sakai, M. E. Brown, R. E. Foster, R. C. Reeves, and G. M. Pohost, J. Am. Coll.
Cardiol., 1987, 9, 1067.
158. C. E. Haug, J. L. Shapiro, L. Chan, and R. Weil, Transplantation, 1987, 44,
175.
159. C. D. Fraser, V. P. Chacko, W. E. Jacobus, G. M. Hutchins, J. Glickson, B. A.
Reitz, and W. A. Baumgartner, Transplantation, 1989, 48, 1068.
160. J. O. Van Dobbenburgh, J. R. Lahpor, S. R. Woolley, N. de Jonge, C. Klopping,
and C. J. Van Echteld, Circulation, 1996, 94, 2831.
© 2016 John Wiley & Sons, Ltd.
Volume 5, 2016