European Heart Journal (1996) 17, 1158-1166

Review Article

Phosphorus magnetic resonance spectroscopy of thehuman heart: current status and clinical implications

H. P. Beyerbacht*, H. W. Vliegen*, H. J. Lambf, J. Doornbosf, A. de Roost,A. van der Laarse* and E. E. van der Wall*

Departments of * Cardiology and ^Diagnostic Radiology, University Hospital Leiden, The Netherlands

Introduction

Magnetic resonance spectroscopy is a non-invasive tech-nique which, by means of a spectrum, can be used toreveal the presence of certain molecules in a specificarea. Magnetic resonance spectroscopy uses a strongexternal magnetic field and radiofrequency pulses todetect the magnetic resonance signal of nuclei whichpossess magnetic spin. Several biologically interestingatomic nuclei are suitable for magnetic resonance spec-troscopy, such as hydrogen ('H), carbon (13C), sodium(23Na), and phosphorus (3IP). For non-invasive studiesof the heart in vivo, 31P magnetic resonance spectros-copy has been used extensively because it providesinformation about the high-energy phosphate metab-olism of the heart. The first cardiac 31P magnetic res-onance spectroscopy experiments were conducted byGadian et al.m in 1976, using isolated Langendorff-perfused animal hearts. In 1985 Bottomley121 reportedthe first localized human cardiac 3IP spectra obtained inhealthy volunteers. Since then, the development of 3IPmagnetic resonance spectroscopy for studies in men hasgrown steadily, but has been hampered by several tech-nical difficulties. An important limitation of in vivo 31Pmagnetic resonance spectroscopy is the relatively lowsensitivity of the technique resulting in long acquisitiontimes and a low spatial resolution. Consequently, 31Pmagnetic resonance spectroscopy currently has nowell-defined application in clinical cardiology. Futuredevelopments, such as the application of stronger mag-netic field strengths, are expected to provide higher

Key Words: Magnetic resonance spectroscopy, phosphorus-31,myocardia! ischaemia, cardiomyopathy, left ventricular hyper-trophy, heart transplant.

Revision submitted and accepted 21 November 1995.

Correspondence: Ernst E. van der Wall, MD, FESC, Uni-versity Hospital Leiden, Department of Cardiology, C5-P25,Rijnsburgerweg 10, 2333 AA Leiden, The Netherlands.

spatial resolution. The promise of non-invasive insightinto the metabolic status of the heart seems so powerful,especially when combined with magnetic resonanceimaging, that the question: 'when will this diagnostictool be available in clinical cardiology?', is more promi-nent than 'whether 31P magnetic resonance spectroscopywill become available in clinical cardiology'.

The present report provides an overview of thestatus, prospects and possible clinical implications ofcardiac 31P magnetic resonance spectroscopy. Results ofmajor human studies conducted to evaluate myocardialischaemia, heart transplantation, cardiomyopathy andleft ventricular hypertrophy will be discussed.

Heart muscle metabolismAdenosine triphosphate (ATP) serves as a high-energyintermediate for maintaining cellular homeostasis andcontractile function of the heart. The ATP concentrationin ventricular myocardium is remarkably constant andquite independent of work load. Under normal con-ditions, the main part (40-70%) of ATP is formedintramitochondrially, fuelled by ^-oxidation of fattyacids. At the inner mitochondrial membrane, the high-energy phosphate bond of ATP is transferred to creatineto yield phosphocreatine and adenosine diphosphate(ADP), a reaction catalysed by mitochondrial creatinekinase. Through the phosphocreatine shuttle, phospho-creatine is then transported to the cytosol and thecontractile elements where it is split by cytoplasmiccreatine kinase (isoenzymes CK-MM and CK-MB) inorder to form ATP and creatine. At the sarcomerelevel, ATP is split by myosine-ATPase, resulting in theformation of ADP, inorganic phosphate (PJ and freeenergy — elements essential for contraction.

During myocardial ischaemia, glucose becomesthe predominant fuel whereby lactate is formed byanaerobic glycolysis. Accumulation of NADH, H+ andlactate inhibits several regulatory sites of metabolism,

0195-668X/96/081158 + 09 $18.00/0 © 1996 The European Society of Cardiology

Review 1159

including glycolysis. During the early stages of ischae-mia, the myocardial ATP concentration is preserved atthe expense of phosphocreatine, which decreases rapidlyduring ischaemia.

Heart failure can lead to metabolic changessimilar to those observed in ischaemia. The concen-trations of creatine and phosphocreatine are diminished,and this is accompanied by reduced total activity ofcreatine kinase and altered isoenzyme distributionof creatine kinase. The mechanisms behind theseadaptations are discussed later in the section entitled'magnetization transfer'.

31 P magnetic resonance spectroscopy

Magnetic resonance spectroscopy is performed with ahomogeneous magnetic field at a field strength of at least1.5 Tesla. Only nuclei which possess a 'magnetic spinmoment' due to an odd number of protons and/orneutrons are suitable for the technique. When nucleiwith a 'magnetic spin moment' are situated in a strongexternal magnetic field, and are subjected to a radiofre-quency pulse at the 'resonance frequency' of the nucleus,their spin state is altered. Following the radiofrequencypulse, the spins will precess around the direction of theexternal magnetic field and produce a radiofrequencysignal which is detected by an antenna ('surface coil'),positioned as close as possible to the tissue of interest.The signal received by the surface coil is a time-dependent voltage. After mathematical processing byapplication of Fourier transformation, the frequencyinformation is obtained as a spectrum. The separation ofthe peaks in a spectrum is caused by a phenomenoncalled 'chemical shift'. Identical nuclei within onemolecule or in different molecules experience a slightlydifferent local magnetic field caused by differences inchemical environment. This local magnetic field of anucleus gives rise to a specific resonance frequency.Differences in resonance frequency or 'chemical shifts'are expressed in Hertz, or as dimensionless parts permillion (p.p.m.), making it independent of the appliedmagnetic field strength. The area under each peak isproportional to the number of nuclei resonating atthat frequency and reflects the relative concentrationof specific metabolites within the volume of interest('sample volume').

Compared to hydrogen (*H), phosphorus (3lP) isless magnetic resonance sensitive, but has a wider chemi-cal shift range. The in vivo human 31P spectrum allowsthe identification of signals from several important high-energy phosphate metabolites (Fig. I), such as phospho-creatine, the three separate signals (a, fi and y) fromATP, the overlapping signals from phosphomonoesters,2,3-diphosphoglycerate and inorganic phosphate (P,),and finally a combined signal from phosphodiestersand phosphatidylcholine (mainly serum phospholipids).Because technical difficulties preclude the measurementof accurate absolute metabolite concentrations relativeconcentrations are most frequently used'3"51. In a recent

study by Yabe el al.[6i, 'absolute' phosphocreatine andATP concentrations were reported. However, thesemeasurements were based on a number of estimationsand assumptions resulting in high standard deviations'61.Nevertheless, the use of 'absolute' instead of 'relative'concentrations has advantages especially for diseasestates which are associated with reduced or increasedconcentrations of both phosphocreatine and ATP. Theratios of phosphocreatine/ATP and phosphocreatine/Piare considered to represent the phosphate potential(energy charge) of the myocardium within the volumeof interest'7"101. In studies of the human heart, thephosphocreatine/ATP ratio is most often used as anindication of energy charge since signal from Pj cannotusually be observed as a separate peak in the spectrum.This is due to the overlapping 2,3-diphosphoglyceratesignal which arises from blood within the volume ofinterest. The chemical shift of P, and of 2,3-diphosphoglycerate are almost identical but in somecases the resonances can be resolved using a techniquecalled proton-decoupling1"1. When the P, peak is re-solved, it can be used to estimate the pH within thevolume of interest because the chemical shift of P; ispH-dependent at the physiological pH range whereas thechemical shift of phosphocreatine is practically pH-independent. The difference in chemical shift betweenthe Pj and phosphocreatine peaks in a spectrum there-fore provides a non-invasive measure of intracellularpH[l ' '. The phosphocreatine/ATP or phosphocreatine/P,ratios are usually determined after 'curve fitting' of thespectrum to estimate peak areas and chemical shifts.In most cases the area under the curve rather thanthe height of a peak is used to calculate ratios'121. The/J-ATP signal is preferably used to determine thephosphocreatine/ATP ratio since it is not contaminatedwith signal from ADP, NAD+ or NADH"21, as is thecase for the a and y ATP peaks.

To locate the volume of interest, several localiz-ation techniques have been developed, such as DepthREsolved Surface coil Spectroscopy (DRESS), ImageSelected In vivo Spectroscopy (ISIS), and ChemicalShift Imaging (CSI), each with its own advantagesand disadvantages. Depending on the technique used,localization is possible in one, two or three dimensions.Figure 2 shows an example of volume selectionwith two-dimensional ISIS combined with one-dimensional CSI. It is important to realize that, due tothe sensitivity profile of the surface coil, spectra can beobtained from the anterior part of the left ventricularwall only. In obese patients, and in females with amore than average breast size, the distance betweenthe heart and the surface coil can even be too largeto obtain signal from the anterior myocardial wall.For a detailed description of the magnetic resonancespectroscopy technique we refer to more specificliterature17"91.

Localization techniques are necessary to mini-mize contamination of the spectra with signal fromoutside the volume of interest. Signal contaminationis of serious concern in 3IP magnetic resonance

Eur Heart J, Vol. 17, August 1996

1160 H. P. Beyerbacht et al.

15 10 -5 -10 -15 -20 -25ppm - - •

Figure 1 Myocardial 3 IP magnetic resonance spectrumobtained from a healthy volunteer, (a) Spectrumbefore curve-fitting, (b) Spectrum after curve fitting.p.p.m. = parts per million; PCr=phosphocreatine; ATP =adenosine triphosphate; PME=phosphomonoesters;a = 2,3-diphosphoglycerate (2,3-DPG, with 3 IP at position3); b=inorganic phosphate (P,)+2,3-DPG (with 3 IP atposition 2); PDE=phosphodiesters, mainJy serumphospholipids (SPL).

Figure 2 Sample volume selection with a combination oftwo localization techniques called 'Image Selected In vivoSpectroscopy' (ISIS) and 'Chemical Shift Imaging' (CSI).A transverse spin-echo image through the chest is used todetermine the appropriate size and orientation of thesample volume in two dimensions. The 3 IP spectra arecollected with a 10-cm diameter surface coil positioneddirectly above the left ventricular wall. The column pro-jected over the heart consists of sections, 1 cm thick, fromwhich separate spectra are collected.

spectroscopy1""131. Currently, most scanners, used forclinical magnetic resonance spectroscopy operate at afield strength of 1.5 Tesla and require volumes of interestwhich exceed 25 cm3 to acquire a spectrum withsufficient signal-to-noise ratio in a reasonable totalexamination time1141. The use of stronger magnetic fieldstrengths improves spatial resolution but is expensive.

To date, the smallest volume of interest reported is8 cm3, which requires an external magnetic field of 4Testa'15'161. Cardiac 31P magnetic resonance spectra mayoften contain signal from blood, skeletal muscle fromthe chest wall as well as from the diaphragm and signalfrom the liver*5'""13'. Such contamination will consider-ably influence the phosphocreatine/ATP ratio. For ex-ample, the phosphocreatine/ATP ratio of skeletal muscleis much higher than the myocardial phosphocreatine/ATP ratio. Correction for this type of contaminationafterwards is not always possible'12-'3'. The contributionof signal arising from ATP in blood to the total signal ofATP in a cardiac 31P spectrum can be corrected for bydetermining the amount of 2,3-diphosphoglycerate inthe 3IP spectrum. The ratio ATP/2,3-diphosphoglyceratein blood is relatively constant, and blood contains2,3-diphosphoglycerate in the absence of phospho-creatine151 U2 '17].

The severity of signal contamination by extra-cardiac tissues may differ between measurements evenwithin the same subject'18'. Many factors are importantin this respect, including positioning of the surface coil,the site of the sample volume, motion of the subject, andthe used localization technique1101181. Consequently, theresults from different research centres vary widely. Thereported limits of 'normal' for the phosphocreatine/ATPratio in human myocardium range from 09 to 2-l[5'12',but are still a matter of debate. Small changes inacquisition technique and data processing may consid-erably alter the results, which makes it difficult tocompare data from different centres'12'. Interpretationand comparison of 3IP magnetic resonance spectroscopydata would benefit from a consensus about the appro-priate acquisition technique and data processing1512>18'.

Magnetization transfer

Magnetic resonance spectroscopy can also be appliedto determine the rate of enzymatic reactions using atechnique called 'magnetization transfer''5'19"24'. Withthis technique the kinetics of creatine kinase havebeen investigated in the myocardium of animals andmen'22"24'. The creatine kinase activity in patients withdilated cardiomyopathy or left ventricular hypertrophy(LVH) was 40% lower than in healthy controls'19'. Therelative contribution of the creatine kinase isoenzymesCK-MB and CK-BB was higher, and that of the iso-enzyme CK-MM lower in the myocardium of the dilatedcardiomyopathy patients. The altered expression of thecreatine kinase isoenzyme genes appears to be stimu-lated by increased wall stress'19'. Similar changes wereobserved in patients with ischaemic heart disease1'9'.Some investigators have suggested that, under stressfulconditions, the relative increase in the creatine kinaseisoenzymes CK-MB and CK-BB partly compensates fordecreased energy reserve caused by depletion of phos-phocreatine and the decreased total creatine kinaseactivity1'922"24'. Because of a higher affinity to ADP, theB-containing isoenzymes are considered to be more

Eur Heart J, Vol. 17, August 1996

Review 1161

efficient than the MM isoenzyme in the formation ofATP from phosphocreatine15'19'22-241.

Initial results with magnetic resonancespectroscopy

Using magnetic resonance spectroscopy, the high-energyphosphate metabolism has been evaluated most exten-sively in animal studies, allowing repeat experimentsunder controlled conditions. Examination of a heart invitro using a Langendorff preparation prevents contami-nation of the spectrum by signal arising from blood andextracardiac tissue, and high spatial resolution can beobtained by the application of a very strong externalmagnetic field. Contamination of the spectrum by signalarising from extracardiac tissue cart be markedlyreduced in vivo with the surface coil directly at themyocardium during an open chest experiment inanimals'25"271. The absence of physiological circum-stances, and the observed differences between differentspecies in myocardial high-energy phosphate metab-olism when subjected to stress'281, raise the question ofhow results from animal studies can be extrapolated tofindings in man.

Despite all differences of opinion about the limitsof 'normal' for the in vivo human myocardialphosphocreatine/ATP ratio, and about the optimal pro-tocol for measuring the phosphocreatine/ATP ratio,much useful information about the cardiac high-energyphosphate metabolism of both animals and men hasbeen gathered under a variety of physiological andpathophysiological states, such as myocardial ischaemia,transplanted hearts, cardiomyopathies and left ventricu-lar hypertrophy. The following sections will deal withthese clinical disease states.

Myocardial ischaemia

The effect of an ischaemic period on the myocardialhigh-energy phosphate metabolism depends upon thelength and the severity of the ischaemic insult'25"341.Both factors also influence the capability of recovery ofthe heart following reperfusion'27'29'30'321. In a study ofZhang et a/.'251 the left anterior descending coronaryartery of a canine heart was partially occluded forseveral hours. After onset of occlusion, the myocardialphosphocreatine concentration and phosphocreatine/ATP ratio initially decreased considerably, accompaniedby a loss of contractile function. Several hours ofhypoperfusion led to a normalization of the myocardialphosphocreatine concentration and phosphocreatine/ATP ratio, although ischaemic myocardial segmentsremained hypokinetic. Apparently, an adaptation pro-cess takes place which resembles the adaptation de-scribed for 'hibernating' myocardium. Consequently, anormal phosphocreatine/ATP ratio does not exclude thepresence of ischaemia. Abnormal contractile function,

as can be evaluated using magnetic resonance imaging,provides additional diagnostic information.

Weiss et a/.'351 examined 11 healthy volunteersand 16 patients with left main or left anterior descendingcoronary artery disease. Both groups were studied atrest and during isometric handgrip testing. At rest themyocardial phosphocreatine/ATP ratio was 1-72 ± 0 1 5for controls and 1-45 ±0-31 for patients. During exer-cise, the myocardial phosphocreatine/ATP ratio forhealthy volunteers remained unchanged, but in patientswith left main or left anterior descending coronaryartery disease, the phosphocreatine/ATP ratio decreasedsignificantly to 0-91 ± 0-24. In these patients the myocar-dial phosphocreatine/ATP ratio recovered to 1-27 ± 0-38within 2 min of discontinuing exercise. Following re-vascularization, the myocardial phosphocreatine/ATPratio normalized to 1 -60 ± 0-70 at rest and to 1-62 ±018during exercise.

In 27 patients with severe left anterior descend-ing coronary artery disease, Yabe et alP6] compared theresults obtained with magnetic resonance spectroscopyto the findings with thallium-201 scintigraphy. Patientswith a persistent scintigraphic defect did not show asignificant decrease in myocardial phosphocreatine/ATPratio during exercise (1-24 ±0-30 at rest versus1 • 19 ± 0-28 during exercise), whereas patients with areversible scintigraphic defect had a significant decreasein phosphocreatine/ATP ratio from 1-60 ±0-19 at rest to096 ± 0-28 during exercise. The healthy controls studiedby Yabe et a/.'361 had a myocardial phosphocreatine/ATP ratio of 1-85 ±0-28 at rest versus 1-90 ± 0 2 3during exercise. These studies indicate that 3IP magneticresonance spectroscopy is a technique that can clinicallydetect ischaemia of the anterior left ventricular wall.

Stress testing by means of physical exercise oradministration of pharmacological agents, such asdobutamine, will play an important role in future 3IPmagnetic resonance spectroscopy studies'5'18'28'35'361.Hearts that are hardly capable of maintaining theirenergy reserves at rest will show a decrease in myocar-dial phosphocreatine/ATP ratio when subjected tophysically or pharmacologically induced stress'5'28'35'361.The effects of a successful therapeutic intervention canbe deduced from a smaller decrease of the myocardialphosphocreatine/ATP ratio during stress compared tothe baseline situation.

Yabe et aL[6] studied 41 subjects with significantleft anterior descending coronary artery disease, anddivided them into a group with fixed perfusion defectsand a group with reversible defects as assessed bythallium-201 scintigraphy. The myocardial phospho-creatine concentration (expressed in umol. g ~' wetheart tissue) was significantly lower in the group withfixed perfusion defects compared with the group withreversible defects (3-94 ± 2-21 versus 7-64 ±300, re-spectively). Also, the myocardial ATP concentration(expressed in umol. g ~ ' wet heart tissue) was signifi-cantly lower in the patients with fixed defects comparedwith the patients with reversible defects (6-35 ±3-17versus 4-35 ± 1-52, respectively). The authors concluded

Eur Heart J, Vol. 17, August 1996

1162 H. P. Beyerbacht et al.

that measurement of the ATP concentration in thehuman heart with 31P magnetic resonance spectroscopycan be used clinically for the evaluation of myocardialischaemia and viability. Clinical use, however, ishampered by the fairly large standard deviations of thereported values'61.

The effects of a myocardial infarction on thehigh-energy phosphate metabolism of the human hearthave been studied as well. In 20 patients with a previousmyocardial infarction, Mitsunami et a/.'3' measured adecrease in myocardial concentrations of phosphocre-atine and ATP. The phosphocreatine concentration inpost-myocardial infarction patients was 7-4 ± 26 versus11 -3 ± 3-7 umol . g ~ ' myocardium in healthy volunteersand the ATP concentration was 4-9 ± 20 versus7-4 ± 2-9 umol. g~ ' myocardium, respectively. Todetermine whether myocardium is infarcted requiresthe measurement of absolute rather than relativeconcentrations of phosphocreatine and ATP13"51.

In summary, 31P magnetic resonance spectros-copy in the assessment of myocardial ischaemia andviability can be applied clinically, but technical restraintslimit its volume of interest to the anterior part of the leftventricular wall and the interventricular septum. Thisrestriction is due to the loss of signal strength at greaterdistance from the surface coil. It is difficult to improvespatial resolution and to increase the depth at whichspectra can be obtained, but if these limitations could beovercome the clinical value of 3IP magnetic resonancespectroscopy in the assessment myocardial ischaemiaand viability would increase considerably.

Transplanted hearts

Animal studies in rats have shown a significant corre-lation between the myocardial phosphocreatine/ATPratio and histological evidence of rejection wherebya decrease of the phosphocreatine/ATP ratio evenpreceded the histological evidence of rejection'37'38'.

Inspired by these experimental studies'37'38' onewould expect that 31P magnetic resonance spectroscopycould provide a non-invasive means of early detection ofrejection in patients who received a heart transplant.Several human studies, however, showed no reliablerelationship between the myocardial phosphocreatine/APT ratio and the degree of rejection'5'17'39'. Thepatients with severe rejection in particular, showed adecreased myocardial phosphocreatine/ATP ratio, butthe specificity of this finding is low as there were alsopatients with clear histological evidence of rejection anda normal ratio'5'17'39]. Presumably, patients with severerejection have — like patients with a previous myocar-dial infarction — decreased myocardial phosphocreatineand ATP concentrations in proportional quantities,leaving their ratio unchanged.

Evanochko et a/.'40' found a normal myocardialphosphocreatine/ATP ratio (181 ± 006) in five patientswith mild rejection, and a diminished phosphocreatine/ATP ratio in four transplant patients with moderate

rejection as compared to healthy controls (1-13 ±0-17versus 1 76 ±011). In a study by Wolfe et a/.'41', 13patients who received a heart transplant had a similarmyocardial phosphocreatine/ATP ratio if the heartsshowed mild (n = 8) or moderate (n = 5) rejection(1-29 ±0-13 versus 1-16 ± 110). In 13 patients who didnot show rejection, the myocardial phosphocreatine/ATP ratio was significantly higher (1-45 ±009).Bottomley et a/P9] examined 14 patients on 19 occasionsfrom 1 month to 6 years after transplanation. Theauthors did not find a good relationship between themyocardial phosphocreatine/ATP ratio and histologicalbiopsy scores for rejection. One explanation for thedifferences between human and animal studies might bethe absence of immunosuppressive medication in theanimal studies, leading to more severe acute rejection inrats(5).

In summary, both experimental and clinicalresults make it unlikely that, in the near future, 3IPmagnetic resonance spectroscopy will become availableas a diagnostic modality for the management of patientsreceiving a heart transplant. Future developments how-ever, such as the use of stronger magnetic fields, yieldinghigher sensitivity and resolution, and the measurementof absolute instead of relative concentrations, may helpto overcome the present limitations.

Hypertrophic cardiomyopathy

Weiss et a/.'351 were the first to study patients withhypertrophic cardiomyopathy, and they observed nosignificant decrease in the myocardial phosphocreatine/ATP ratio. However, only five patients were studied.Conversely, subsequent 31P magnetic resonance spec-troscopy studies by De Roos et a/.'"', Masuda et al.l42]

and Sakuma et a/.'43' showed a significantly lowerphosphocreatine/ATP ratio in hearts of patients withhypertrophic cardiomyopathy compared with healthycontrols. De Roos et al.lu] found a lower myocardialphosphocreatine/ATP ratio in eight hypertrophic car-diomyopathy patients compared with nine healthy con-trols (1-32 ± 0-29 vs 1 -65 ± 0-26). In the study of Masudaet a/.'42', 12 patients with hypertrophic cardiomyopathyhad a myocardial phosphocreatine/ATP ratio of1 43 ± 0-36 vs 209 ± 0-44 in 15 controls. Lastly, Sakumaet a/.'43' found in 19 hypertrophic cardiomyopathypatients a significant decrease in the phosphocreatine/ATP ratio when compared to controls, with values of1-07 ±0-10 and 1-71 ±013 , respectively. They observedno relationship between perfusion abnormalitiesassessed by thallium-201 scintigraphy and the decreasedmyocardial phosphocreatine/ATP ratio, indicating anuncoupling of perfusion and metabolism. As moststudies show a decreased myocardial phosphocreatine/ATP ratio in hypertrophic cardiomyopathy patients, 3IPmagnetic resonance spectroscopy may find a clinicalapplication in the diagnostic process when hypertrophiccardiomyopathy is suspected. The pathophysiologicalmechanism responsible for the lowered myocardial

Eur Heart J, Vol. 17. August 1996

Review 1163

phosphocreatine/ATP ratio in hypertrophic cardio-myopathy patients has not been elucidated yet, but alower vasodilator reserve, as demonstrated by Camiciet a/.'441, could be a contributing factor.

Dilated cardiomyopathy

In patients with dilated cardiomyopathy a relationshiphas been established between the severity of heart failureand the decrease of the myocardial phosphocreatine/ATP ratio'51. Neubauer et al.[4S] studied 19 dilatedcardiomyopathy patients and found a myocardialphosphocreatine/ATP ratio of 1-94 ±0-43 in patientswith only mild symptoms of heart failure whereaspatients with severe signs of heart failure had a ratio of1-44 ±0-52. The correlation between New York HeartAssociation (NYHA) class and phosphocreatine/ATPratio was highly significant (r= -0-60, P<0005). Treat-ment of six patients with severe heart failure during 3months, led to an average improvement in NYHA classby 08 ± 0-3 points, which was accompanied by a rise inmyocardial phosphocreatine/ATP ratio from 1-51 ± 0-32to 215 ± 027. The authors'451 also reported that such arelation could not be established for the left ventricularejection fraction and the myocardial phosphocreatine/ATP ratio.

De Roos et a/.'1'' compared nine healthy controls(phosphocreatine/ATP 1-65 ±0-26) with nine patientswith dilated cardiomyopathy. The dilated cardiomyo-pathy patients showed, on average, a normal myocardialphosphocreatine/ATP ratio (1 -52 ± 0-58), although someindividual dilated cardiomyopathy patients did showa lowered phosphocreatine/ATP ratio compared tohealthy controls (Fig. 3). Differences in myocardialphosphocreatine/ATP ratio between dilated cardio-myopathy patients may be due to differences in severityof the disease between the patients. Measurement ofaccurate myocardial phosphocreatine/ATP ratios indilated cardiomyopathy patients is hampered by a rela-tively thin anterior left ventricular wall. The samplevalue may contain insufficient myocardium to generate aspectrum with an acceptable signal-to-noise ratio1"1.Follow-up studies in dilated cardiomyopathy patientswith progressive heart failure and assessment of efficacyof therapy may become a clinical application of 3IPmagnetic resonance spectroscopy.

Left ventricular hypertrophy

Left ventricular hypertrophy is known to be associatedwith a decreased density of capillaries and a reducedvasodilator reserve'46'471. In rats left ventricular hyper-trophy has been shown to be associated with a decreasedmyocardial ATP concentration and increased sensitivityto ischaemia, when analysed in a Langendorff perfusionmodel148"501.

Patients with left ventricular hypertrophy repre-sent a heterogenous group which may be divided into

15.0 10.0 5.0 0.0 -5.0 -10.0 -15.0 -20.0 -25.0p p m - *•

Figure 3 Myocardial 3 IP magnetic resonance spectrumfrom a patient with dilated cardiomyopathy and from ahealthy control. Clearly visible is the larger 2,3-diphosphoglycerate signal in tbe spectrum of the dilatedcardiomyopathy patient due to the larger amount of bloodin the sample volume; in this particular dilated cardio-myopathy patient the phosphocreatine/ATP ratio wassignificantly lower than in die average healthy control.This was not the case for the dilated cardiomyopathypatients as a group. (With permission from de RoosetaLiUi.)

those with pressure overload (hypertension, aortic valvestenosis), volume overload (aortic valve insufficiency ormitral valve regurgitation), or physical training (athlete'sheart) as causal factor15'1.

Conway et al.[52] studied 14 patients with leftventricular hypertrophy, six of whom were treated forheart failure (NYHA classes II-III). Only these sixpatients could be distinguished from healthy controls onthe basis of the myocardial phosphocreatine/ATP ratio(1-11 ±0-3 vs l-5±0-2), whereas the patients withoutclinical signs of heart failure had an average myocardialphosphocreatine/ATP ratio in the normal range(1-6 ±0-2).

In 15 patients with left ventricular hypertrophydue to aortic valve disease, Neubauer et a/.'531 showedthat the myocardial phosphocreatine/ATP ratio wascorrelated to NYHA class, fractional shortening, pul-monary capillary wedge pressure and right atrialpressure. No correlation was found between the myocar-dial phosphocreatine/ATP ratio and the left ventricularejection fraction or cardiac output'531. Dell' Italiaet a/.'54' studied 12 patients with left ventricular hyper-trophy due to aortic stenosis both before and 3 monthsafter aortic valve replacement. Before surgery theymeasured a myocardial phosphocreatine/ATP ratiowhich was significantly lower than in controls(105 ± 0-22 vs 1-30 ±015). Aortic valve replacementsurgery led to a significant functional improvement ofNYHA class, but not to an increase in myocardial

Eur Heart J, Vol. 17, August 1996

1164 H. P. Beyerbacht et al.

phosphocreatine/ATP ratio. In concordance with thesefindings are the results of a study by Ito et a/.'551, whoobserved that, after regression of left ventricular hyper-trophy in rats, the heart remained more sensitive toischaemia. They showed that the vascular resistance,which was higher in hypertrophic hearts, did not nor-malize parallel with regression of left ventricular hyper-trophy. They considered an increased vascular resistanceas an independent risk factor for cardiac morbidity.Longer follow-up periods than those used in the pre-viously mentioned studies'54'551 may be warranted toobserve whether or not a normalization of the myocar-dial phosphocreatine/ATP ratio follows left ventricularhypertrophy regression.

31P magnetic resonance spectroscopy of the heartmight become of clinical value in patients with leftventricular hypertrophy. Possible applications are im-proved detection of the optimal moment for surgery inpatients with aortic or mitral valve disease, an assess-ment of efficacy of therapy in patients with valve diseaseor hypertension.

Conclusions3IP magnetic resonance spectroscopy is a valuable diag-nostic tool but is currently restricted mainly to researchapplications. Most likely in the future 31P magneticresonance spectroscopy of the heart will become avail-able as a clinically useful diagnostic modality. Thecombined information of structure, function and metab-olism which can be obtained in a single study offers aunique method for investigating these modalities in a'one-stop-shop' fashion.

The use of slightly differing data-acquisition pro-tocols and localization techniques by several researchgroups warrants the development of a consensus about anormal value for the 'true' myocardial phosphocreatine/ATP ratio. Standardization of hardware, software andstudy protocols would improve the reproducibility of31P magnetic resonance spectroscopy studies performedby different research groups'5'121. To date, reproducibil-ity may even present a problem within one researchgroup'12'181. This may be due, at least partly, to adifferent extent of signal contamination with differentmeasurements, as the quantity of extracardiac tissuewithin the sample volume varies among differentmeasurements. Small variations in sample volume selec-tion may strongly affect the measured myocardialphosphocreatine/ATP ratio"'- | 3 l8l.

Another limitation to the clinical use of 3IPmagnetic resonance spectroscopy is the ability of theheart to maintain its high-energy phosphate metabolismwithin normal ranges. Changes in the high-engergyphosphate metabolism may therefore manifest relativelylate151. Studies in patients with left ventricular hypertro-phy or dilated cardiomyopathy*513-45521 showed myocar-dial phosphocreatine/ATP ratios which were decreasedonly if symptoms of heart failure were already present.Stress testing during magnetic resonance spectroscopy

offers a means of showing latent shortages in high-energy phosphates and may improve sensitivity fordetecting the early stages of heart disease1'5'18'28'35'361.Evaluation of therapy in patients with dilated cardio-myopathy, hypertrophic cardiomyopathy, left ventricu-lar hypertrophy or myocardial ischaemia located in theanterior wall of the left ventricular, may become arealistic clinical application for 3IP magnetic resonance.The range of clinical applications may increase whenabsolute instead of relative metabolite concentration aremeasured. Then 31P magnetic resonance spectroscopywill allow evaluation of disease states in which myo-cardial phosphocreatine and ATP concentrations de-crease or increase by equal amounts. Until now, the lowspatial resolution of >8 cm3, due to a low intrinsicsensitivity and low abundance of phosphorus containingmolecules, the relatively long exposure times, and therelatively high costs of magnetic resonance machinesand dedicated software packages still limit major clinicalapplications. Ongoing technical progress will facilitatethe implementation of 31P magnetic resonancespectroscopy in clinical cardiology.

References

[1] Gadian DG, Hoult DI, Radda GK, Seely PJ, Chance B,Barlow C. Phosphorous nuclear magnetic resonance studies innormoxic and ischemic cardiac tissue. Proc Natl Acad SciUSA 1976 1976; 73: 4446-8.

[2] Bottomley PA. Noninvasive study of high-energy phosphatemetabolism in human heart by depth-resolved P NMRspectroscopy. Science 1985; 229: 769-72.

[3] Mitsunami K, Okada M, Inoue T, Hachisuka M, KinoshitaM, Inubushi T. In vivo 3 IP nuclear magnetic resonancespectroscopy in patients with old myocardial infarction. JpnCirc J 1992; 56: 614-9.

[4] Tofts PS, Wray S. A critical assessment of methods ofmeasuring metabolite concentrations by NMR spectroscopy.NMR Biomed 1988; 1: 1-10.

[5] Bottomley PA. MR spectroscopy of the human heart: thestatus and the challenges. Radiology 1994; 191: 593-612.

[6] Yabe T, Mitsunami K, Inubushi T, Kinoshita M. Quantitativemeasurements of cardiac phosphorus metabolites in coronaryartery disease by 31P magnetic resonance spectroscopyCirculation 1995; 92: 15-23.

[7] Osbakken MD, Mitchell MD. Magnetic resonance spectros-copy to study myocardial metabolism and cellular function.In: Marcus ML, Skorton DJ, Schelbert HR, Wolf GL, eds.Cardiac imaging: a companion to Braunwald's Heart disease.Philadelphia: W. B. Saunders; 1991: 841-63.

[8] Gadian DG. Nuclear magnetic resonance and its applicationto living systems. Oxford: Clarendon, 1982.

[9] Brown JJ, Mirowitz SA, Sandstrom JC, Perman WH.MR spectroscopy of the heart. AJR Am J Roentgenol 1990;155: 1-11.

[10] Stamper JJ, Pohost GM. Assessing myocardial ischemic insultusing nuclear magnetic resonance spectroscopy. Am J CardImaging 1992, 6: 244-58.

[11] de Roos A, Doombos J, Luyten PR, Oosterwaal LJMP, vander Wall EE, den Hollander JA. Cardiac metabolism inpatients with dilated and hypertrophic cardiomyopathyassessment with proton-decoupled P-31 MR spectroscopyJ Magn Reson Imaging 1992; 2: 711-9.

[12] Bottomley PA. The trouble with spectroscopy papers.Radiology 1991; 181: 344-50.

Eur Heart J. Vol. 17, August 1996

Review 1165

[13] Higgins CB, Saeed M, Wendland M, Chew WM. Magneticresonance spectroscopy of the heart. Overview of studies inanimals and man. Invest Radiol 1989; 24: 962-8.

[14] Barrett EJ, Alger JR, Zaret BL. Nuclear magnetic resonancespectroscopy: its evolving role in the study of myoardialmetabolism. J Am Coll Cardiol 1985; 6: 497-501.

[15] Menon RS, Hendrich K, Hu X, Ugurgil K. 31P NMRspectroscopy of the human heart at 4 T: detection of substan-tially uncontaminated cardiac spectra and differentiation ofsubepicardium and subendocardium. Magn Reson Med 1992;26: 368-76.

[16] Hetherington HP, Luney DJ, Vaughan T et al 3D 31Pspectroscopic imaging of the human heart at 4.1 T. MagnReson Med 1995; 33: 427-31.

[17] van Dobbenburgh JO, de Jonge N, flopping C, Lahpor JR,Woolley SR, van Echteld CJ. Postischemic alterations inmyocardial energy metabolism in heart transplant patients[abstract]. Circulation 1993; 88: 1378.

[18] Lamb HJ, Doornbos J, den Hollander JA, Beyerbacht HP,de Roos A. Strategies for cardiac "P-MR spectroscopy atrest and during dobutamine stress: a study of reproduc-ibility [abstract]. Society of Magnetic Resonance: Thirdscientific meeting and exhibition; August 19-25, 1995; Nice,France. Berkely Calif Society of Magnetic Resonance, 1995:96.

[19] Ingwall JS. Is cardiac failure a consequence of decreasedenergy reserve? Circulation 1993; 87: VII58-62.

[20] Ugurbil K. Magnetization-transfer measurements of individ-ual rate constants in the presence of multiple reactions.J Magn Reson 1985; 64- 207-19.

[21] Bittl JA, Ingwall JS. Intracellular high-energy phosphatetransfer in normal and hypertrophied myocardium. Circu-lation 1987; 75. 196—1101.

[22] Ingwall JS, Kramer MF, Fifer MA et al. The creatine kinasesystem in normal and diseased human myocardium N Engl JMed 1985; 313: 1050-4.

[23] Ingwall JS, Atkinson DE, Clarke K, Fetters JK. Energeticcorrelates of cardiac failure: changes in the creatine kinasesystem in the failing myocardium. Eur Heart J 1990; 11:108-15

[24] Smith SH, Kramer MF, Reis I, Bishop SP, Ingwall JS.Regional changes in creatine kinase and myocyte size inhypertensive and nonhypertensive cardiac hypertrophy. CircRes 1990; 67: 1334-44

[25] Zhang J, Path G, Chepuri V et al. Responses of myocardialhigh energy phosphates and wall thickening to prolongedregional hypoperfusion induced by subtotal coronary stenosis.Magn Reson Med 1993; 30: 28-37.

[26] Schaefer S, Camacho SA, Gober J et al. Response of myocar-dial metabolites to graded regional ischemia: 31P NMR spec-troscopy of porcine myocardium in vivo. Circ Res 1989; 64:968-76.

[27] Wroblewski LC, Aisen AM, Swanson SD, Buda AJ. Evalu-ation of myocardial viability following ischemic and reper-fusion injury using phosphorus 31 nuclear magnetic resonancespectroscopy in vivo. Am Heart J 1990; 120: 31-9.

[28] Schaefer S, Schwartz GG, Steinman SIC, Meyerhoff DJ,Massie BM, Weiner MW. Metabolic response of the humanheart to inotropic stimulation: in vivo phosphoms-31 studiesof normal and cardiomyopathic myocardium. Magn ResonMed 1992; 25: 260-72.

[29] Heusch G, Schulz R. Chronische Myokardischamie —hibernating myocardium: Merkmale und Grenzen. Z Kardiol1993; 82(Suppl 5): 133-41.

[30] Wolfe CL, Moseley ME, Wikstrom M G d u l Assessment ofmyocardial salvage after ischemia and reperfusion using mag-netic resonance imaging and spectroscopy. Circulation 1989;80: 969-82.

[31] Baer FM, Theissen P, Schneider CA, Voth E, Schicha H,Sechtem U. Magnetresonanztomografische Untersuchungs-techniken fur die myokardiale Vitalitatsdiagnostik. Herz 1994;19: 51-64.

[32] Rehr RB, Tatum JL, Hirsch JI, Wetstein L, Clarke G.Effective separation of normal, actutely ischemic, and reper-fused myocardium with P-31 MR spectroscopy Radiology1988; 168: 81-9.

[33] Bottomley PA, Smith LS, Brazzamano S, Hedlund LW,Redington RW, Herfkens RJ. The fate of inorganic phosphateand pH in regional myocardial ischemia and infarction: anoninvasive 3 IP NMR study. Magn Reson Med 1987; 5:129-42.

[34] Rehr RB, Tatum FL, Hirsch JI, Quint R, Clarke G.Reperfused-viable and reperfused-infarcted myocardium: dif-ferentiation with in vivo P-31 MR spectroscopy. Radiology1989; 172: 53-8.

[35] Weiss RG, Bottomley PA, Hardy CJ, Gerstenblith G. Re-gional myocardial metabolism of high-energy phosphates dur-ing isometric exercise in patients with coronary artery disease.N Engl J Med 1990; 323: 1593-600.

[36] Yabe T, Mitsunami K, Okada M, Mirikawa S, Inubushi T,Kinoshita M. Detection of myocardial ischemia by 31Pmagnetic resonance spectroscopy during handgrip exercise.Circulation 1994; 89: 1709-16.

[37] Fraser CD Jr, Chacko VP, Jacobus WE et al. Early phos-phorus 31 nuclear magnetic resonance bioenergetic changespotentially predict rejection in heterotopic cardiac allografts.J Heart Transplant 1990; 9: 197-204.

[38] Fraser CK Jr, Chacko VP, Jacobus WE et al. Metabolicchanges preceding functional and morphologic indices ofrejection in heterotopic cardiac allgrafts. A 31P nuclearmagnetic resonance study. Transplantation 1988; 46: 346-51

[39] Bottomley PA, Weiss RG, Hardy CJ, Baumgartner WA.Myocardial high-energy phosphate metabolism and allograftrejection in patients with heart transplants. Radiology 1991;181: 67-75.

[40] Evanochko WT, Bouchard A, Kirklin JK, Bourge RC, LuneyD, Pohost GM. Detection of cardiac transplant rejection inpatients by 3 IP NMR spectroscopy [abstract]. Society ofMagnetic Resonance in Medicine: Ninth annual scientificmeeting and exhibition; August 18-24, New York, NY USA.Berkeley Calif. Society of Magnetic Resonance in Medicine,1990: 246.

[41] Wolfe CL, Caputo G, Chew W et al. Detection of cardiactransplant rejection by magnetic resonance imaging and spec-troscopy [abstract]. Society of Magnetic Resonance in Medi-cine: Tenth annual scientific meeting and exhibition; August10-16, 1991, San Francisco, Calif. USA. Berkely Calif. Societyof Magnetic Resonance in Medicine, 1991; 574.

[42] Masuda Y, Tateno Y, Ikehira H et al. High-enegy phosphatemetabolism of the myocardium in normal subjects andpatients with various cardiomyopatnies. The study using ECGgated MR spectroscopy with a localization technique. JpnCirc J 1992; 56: 620-6.

[43] Sakuma H, Takeda K, Tagami T et al. 31P MR spectroscopyin hypertrophic cardiomyopathy: comparison with Tl-201myocardial perfusion imaging. Am Heart J 1993; 125: 1323-8.

[44] Camici P, Chiriatti G, Lorenzoni R et al. Coronary vaso-dilatation is impaired in both hypertrophied and non-hypertrophied myocardium of patients with hypertrophiccardiomyopathy: a study with nitrogen-13 ammonia andpositron emission tomography. J Am Coll Cardiol 1991; 17:879-86.

[45] Neubauer S, Krahe T, Schindler R et al. 31P magneticresonance spectroscopy in dilated cardiomyopathy and cor-onary artery disease. Altered cardiac high-energy phosphatemetabolism in heart failure. Circulation 1992; 86: 1810-8.

[46] Wexler LF, Lorell BH, Momomura S, Weinberg EO, IngwallJS, Apstein CS. Enhanced sensitivity to hypoxia-induceddiastolic dysfunction in pressure-overload left ventricularhypertrophy in the rat: role of high-energy phosphatedepletion. Circ Res 1988; 62: 766-75.

[47] Vatner SF, Shannon R, Hittinger L. Reduced subendocardialcoronary reserve: a potential mechanism for impaired diastolicfunction in the hypertrophied and failing heart. Circulation1990; 81: 1118-14.

Eur Heart J, VoL 17, August 1996

1166 H. P. Beyerbacht et al.

[48] Bache RJ, Arentzcn CE, Simon AB, Vrobel TR Abnor-malities in myocardial perfusion during tachycardia in dogswith left ventricular hypertrophy metabolic evidence formyocardial ischemia. Circulation 1984, 69' 409-17.

[49] Harmsen E, Schoemaker R, Yu J, Ruacka M, Leenen FHH.Sensitivity to ischemic ATP breakdown in different models ofcardiac hypertrophy in rats. J Hypertens 1994; 12: 49-57.

[50] Lorell BH, Wexler LF, Momomura S, Weinberg E, ApsteinCS. The influence of pressure overload left ventricular hyper-trophy on diastolic properties during hypoxia in isovolumi-cally contracting rat hearts. Circ Res 1986; 58- 653-63.

[51] Schaefer S, Gober JR, Schwartz GG, Twieg DB, Weiner MW,Massie B. In vivo phosphorus-31 spectroscopic imaging inpatients with global myocardial disease Am J Cardiol 1990;65: 1154-61.

[52] Conway MA, Allis J, Ouwerkerk R, Niioka T, RajagopalanB, Radda GK. Detection of low phosphocreatine to ATP ratioin failing hypertrophied human myocardium by 31P magneticresonance spectroscopy. Lancet 1991; 338: 973-6.

[53] Neubauer S, Horn M, Mader H et al Clinical and hemo-dynamic correlates of impaired cardiac high-energy phosphatemetabolism in patients with aortic valve disease [abstract].Proceedings of the Society fo Magnetic Resonance in Medi-cine; 1993 Augus 14-20; New York, NY USA. Berkeley,Calif.: Society of Magnetic Resonance in Medicine, 1993: 55.

[54] Dell' Italia LJ, Evanochko WT, Blackwel) GG, den HollanderJ, Singleton HR, Pohost GM. Dissociation between mech-anics and PCr/ATP in patients with volume overload hyper-trophy [abstract]. Proceedings of the Society of MagneticResonance in Medicine, 1993; August 14-20; New York, NYUSA. Berkeley, Calif.: Society of Magnetic Resonance inMedicine, 1993: 356.

[55] Ito N, Isoyama S, Kuroha M, Takishima T. Duration ofpressure overload alters regression of coroanry circulationabnormalities. Am J Physiol 1990. 258- HI753-60.

Eur Heart J. Vol. 17, August 1996