Magnetic resonance spectroscopy (MRS)1-4 is a method for non-invasively probing metabolism. In the heart, phosphorus 31P MRS has proved particularly valuable because it probes the creatine-kinase system which buffers and actively transports ATP from its site of generation in the mitochondria to its site of utilisation in the myofibrils, where it drives contractile work, pumps blood around the body, and sustains life. In most heart diseases, the concentrations and/or the rates of interconversion of 31P metabolites are altered. Spectroscopy benefits from ultra-high fields in two main ways: (i) the signal-to-noise ratio for a fixed scan time increases approximately proportional to B0; and (ii) the frequency separation between the peaks from different metabolites increases proportional to B0 which makes robust spectral fitting easier. This session reviews the first human cardiac spectroscopy studies at 7T.

The human heart consumes 6kg of adenosine triphosphate (ATP) per day as the primary fuel driving contraction of the cardiac myocytes (muscle cells). As the heart beats, it turns chemical energy (ATP) into mechanical (pV) work pumping blood around the body, which is essential to life. Cardiac myocytes take up glucose, free fatty acids and ketone bodies, using them to synthesise ATP in the mitochondria. The CK energy shuttle buffers and transports ATP to the myofibrils, in the form of phosphocreatine (PCr), to power contraction of the sarcomeres. The CK reaction at the myofibrils is:5

MgADP¯+ PCr2¯ + H+ ⇌ MgATP2¯ + Cr,

and the reverse reaction dominates in the mitochondria. In animal models, magnetic resonance spectroscopy (MRS) methods have measured: [PCr2¯] and [MgATP2¯] (or their ratio); intracellular [Mg2+]; the effective first order rate constant kf = k1 [MgADP¯][H+]; and the flux of MgATP2¯. [MgADP¯] can be computed by also measuring pH and [Cr], and knowing K=1.66 nM-1. ∆G~ATP for ATP hydrolysis in the myofibrils (ATP + H2O → ADP + Pi) can be computed from [Pi]. ∆G~ATP measures the power available to a myocyte. Together these processes are termed “high-energy phosphate” (HEP) metabolism.7 HEP metabolism is deranged in: heart failure, ischaemia, cardiomyopathies, myocardial infarction (MI), and in systemic diseases such as diabetes and obesity.6

Since the first-in-man 31P-MRS experiments in 1985,8 many labs have translated the successes of animal 31P-MRS to human studies. However, 31P-MRS applications in humans have always been limited by the low intrinsic signal-to-noise of 31P-MRS scans compared to 1H MRI. Spectroscopy benefits from using stronger field MRI scanners in two main ways: (i) the signal-to-noise ratio for a fixed scan time increases approximately proportional to B0; and (ii) the frequency separation between the peaks from different metabolites increases proportional to B0 which makes robust spectral fitting easier. This session reviews the first human cardiac spectroscopy studies at 7T.In 2013, we performed the first cardiac 7T 31P-MRS scans in a study on 9 volunteers. We observed a 2.8x increase in signal-to-noise ratio (SNR) at 7T compared to 3T.9 We also obtained the T1 relaxation times of the key 31P metabolites.Recently, we completed a study comparing the performance of 31P-MRS at 3T and 7T in 25 patients with dilated cardiomyopathy.10 This was the first cardiac patient study at 7T in Oxford. We observed a 2.6x increase in phosphocreatine signal-to-noise ratio at 7T compared to 3T. Intriguingly, while we saw the expected reduction in standard deviation of PCR/ATP in control subjects from 3T to 7T, in the DCM cohort we saw very little reduction in the PCr/ATP standard deviation. Our study was not large enough for this difference to be statistically significant, but it suggests that cardiac 31P-MRS at 7T may be approaching the level of SNR where inter-subject biological variation becomes visible.Cardiac 31P spectra are often contaminated by signals from overlying skeletal muscle, which has ~2x greater PCr:ATP ratio than myocardium and different creatine-kinase kinetics. Traditionally, skeletal muscle signal is suppressed with additional radiofrequency pulses. However, at 7T, this is undesirable because we have limited radiofrequency power. Instead, we constructed a local B0-spoiling gradient insert coil. We demonstrated the ability to de-phase 31P signals arising from overlying skeletal muscle with minimal effects on cardiac 31P spectra in normal volunteers in vivo.11It is often necessary to know the B1+ in the heart in vivo to set appropriate pulse voltages or during post-processing. Existing methods (e.g. the dual-angle, or dual-TR method) are not well suited to 31P-MRS in the heart at 7T. We adapted the Bloch-Siegert method for flip angle measurement for use in 31P spectroscopy and found that it is an attractive approach for 31P-MRS B1-mapping.12We implemented cardiac 31P saturation transfer at 7T, and validated our method by measuring the creatine-kinase (CK) rate constant kf in calf muscle and in the mid-interventricular septum of 10 normal volunteers. The 2.8x higher signal-to-noise of 31P-MRS at 7T allows us to make 3D-resolved kf measurements in the human heart in vivo.13Our current standard RF coil for 7T 31P-MRS studies is an array with 16 receive elements and a single 28x28cm2 transmit element (Rapid Biomedical, Germany). This coil has the best receive performance of our coils in Oxford, but it gives a comparatively low peak transmit field strength in the heart.14 To overcome the limitation of low transmit field strength, we optimised a cardiac quadrature surface coil design by EM simulations. We have now built this coil, performed RF safety tests, and used it to obtain spectra from the anterior and inferior interventricular septum in the human heart. In vivo, the coil gives much higher transmit field strength in the anterior and mid regions of the heart compared to the 16-element array, but it has a lower receive sensitivity because it uses the two 15cm diameter loops for both transmit and receive.15 We have now also implemented an adiabatic half-passage excitation pulse (made possible by the increased transmit field strength from the quadrature coil) and used this to scan the hearts of 12 subjects. This proof-of-concept study shows the feasibility of adiabatic excitation in the anterior and mid-septal regions of the heart at 7T, which is a key step towards making absolute concentration measurements of metabolites, i.e. in mmol per kg wet weight tissue.16

Whole-body RF transmit

Prof Klomp’s group at UMC Utrecht have recently demonstrated a 31P whole-body RF birdcage coil. The 5 minute CSI spectroscopy dataset presented in their paper shows that this coil has much more uniform B1+ and receive sensitivity than the surface coils we have used for our studies to date. The wider availability of whole-body transmit coils for 31P-MRS will significantly enhance the opportunities for studying cardiac disease by 31P-MRS at 7T.

B0-shimming using 1H GRE field maps

In a collaborative study with Prof Vaughan at CMRR, University of Minnesota, we tested the impact on cardiac 31P-MRS of per-subject B0 shimming using a 16-element 1H coil for shimming and then swapping to our 16-element 31P coil. B0 shimming reduced the linewidth of the 31P spectra and in some subjects allowed us to resolve spectra from the inferior segments of the heart.17

Four channel cardiac 31P-MRS coil

Lu et al. from Auburn University presented bench and phantom results from a four-channel 7T cardiac 31P-MRS coil at ISMRM2013.18

Cardiac 1H MRS19 has been used successfully e.g. to quantify the fat fraction in the myocardium at 3T20, or to measure the total creatine concentration at 1.5T21. However, it is not straightforward to adapt these single voxel spectroscopy PRESS and STEAM methods for use in the heart at 7T because of the challenge of achieving sufficiently uniform and strong RF transmit (B1+) fields in the heart, and because PRESS and STEAM cause significant RF heating (SAR) at 7T. Nevertheless, there is clear potential for cardiac 1H-MRS at 7T e.g. it may become possible to distinguish the 1H creatine (Cr) and phosphocreatine (PCr) resonances at 7T, instead of only measuring “total creatine” (tCr = Cr + PCr).