Synopsis

Inorganic Phosphate is a resonance that holds important information on the metabolic state of tissues. From its resonance frequency, intracellular pH can be derived. The ratio of P_i to PCr or ATP are also important markers. Unlike in other tissues, myocardial P_i is frequently hidden underneath blood 2,3-DPG signals. Using 7 T STEAM's T_M delay to be one cardiac cycle, blood-pool originating signals are gone and the Pi resonance is clearly visible. In 3 subjects, P_i signal was detected and quantified. The signal was around 4.89±0.02ppm, corresponding to a pH of 7.08±0.02. This is a breakthrough for the investigation of cardiac metabolism.

Introduction

Investigating cardiac metabolism and its pathological alterations are essential in understanding various diseases. Cardiac ³¹P MRS is one of very few techniques capable of measuring cardiac metabolism in-situ and non-invasively. So far, the inorganic phosphate signal (P_i), commonly found in ³¹P MRS data has escaped unequivocal detection. The aim of this project is to be able to reliably quantify P_i and hence pH in human subjects at 7 T using ³¹P MRS.

In protocols with short acquisition delay or T_E contributions from blood-pool 2-3 diphospho-glycerate (DPG) signals around 5.5 ppm give rise to overlapping resonances due to imperfect localisation and limited spectral resolution [1]. The STEAM sequence offers two interesting properties: First, 90 deg RF pulses are comparatively easy to achieve, second, the magnetisation during the T_M time is stored along the longitudinal axis where T₁ decay occurs. P_i has a long T₁ of about 5.1 s [2]. Long T_M short T_E STEAM should give rise to signal from myocardium while suppressing fast moving blood.

Methods

A 7 T (Magnetom, Siemens, Germany) STEAM sequence (Figure 1) was adapted to ECG triggering twice, before the first and the third pulse, resulting in an effective T_M of one heartbeat. Thereby, a significant fraction of the blood pool has left the voxel before the echo is acquired. Furthermore, the spoiler gradients during the T_E intervals act as motion-sensitising gradients, probably contributing significantly to blood suppression [3]. Unfortunately, the weak B₁⁺ field (16 channel array, RAPID, Germany) requires comparatively long pulses (4.5 ms truncated sinc shapes) that produce significant chemical shift displacement artefacts. This was circumvented using interleaved acquisitions with shifted excitation: One at 0 Hz (PCr) and one at 570 Hz (P_i) every T_R (Figure 1). Effective repetition times for each metabolite were six heart-beats.

Healthy subjects were scanned lying supine. Localiser images in the main cardiac orientations were acquired before spectroscopy acquisition. Voxels were placed to cover the septum; dimensions varied between subjects (148±86 ml) (Figure 2). WSVD [4] coil combination was used based on the PCr signal in both scans. Matlab AMARES [5] was used to fit peaks, first the PCr-only spectrum and then Pi, whose line widths was constrained to that of PCr. The phase of PCr was used as an initial value for the P_i phase. This helped the fit converge to reasonable values judged by visual inspection.

Results and Conclusions

Cardiac ³¹P single voxel spectroscopy has not been attempted at 7 T before, we think. Unlike in UTE-CSI data (Figure 3), the 2,3 DPG signal is gone and P_i nicely observable (Figure 4). Using this approach, PCr and in particular P_i signals were measured in 3 subjects (SNR P_i > 4, SNR PCr > 50), so far. Cardiac intracellular pH values were 7.078 ± 0.017. The P_i signal is 0.09 ± 0.02 times weaker than PCr. These values are very similar to those reported in patients with hypertrophic cardiomyopathy (HCM) by high resolution CSI [6].

Unlike previous reports, this technique resulted in spectra with P_i clearly visible and without a predominant 2,3-DPG signal in the 5 ppm spectral area in all subjects after a few minutes' scan time even in healthy subjects with low P_i concentrations and normal left myocardial wall thickness (unlike the HCM patients mentioned above). The chemical shift displacement is really quite strong so even ATP in the PCr scan is significantly affected. The small PCr peak in Figure 4 is a residual originating from some completely different location.

The ability to non-invasively quantify intracellular cardiac P_i in a matter of minutes will be a significant boost to the utility of cardiac ³¹P MRS and a new tool for cardiology research.

Acknowledgements

References

[1] CT Rodgers et al. Magn Reson Med; 2014; doi: 10.1002/mrm.24922

[2] L Valkovic et al. “Quantification of human cardiac Pi”; submitted to ISMRM 2018

[3] TE Reese et al. Magn Reson Med 1995; doi: 10.1002/mrm.1910340603

[4] CT Rodgers et al. Magn Reson Med 2010; doi: 10.1002/mrm.22230

[5] LAB Purvis et al. PloS One 2017; doi: 10.1371/journal.pone.0185356

[6] WT Clarke “Human cardiac magnetic resonance spectroscopy“; 2016; http://solo.bodleian.ox.ac.uk/OXVU1:LSCOP_OX:oxfaleph020840395