Introduction

Phosphorus MR spectroscopy (³¹P-MRS) is a valuable technique for non-invasive measurement of high-energy metabolites, e.g. adenosine triphosphate (ATP) and phosphocreatine (PCr), that allows assessment of tissue energy metabolism in vivo. Interest in ³¹P-MRS is great in cardiovascular medicine, as the PCr/ATP concentration ratio in the heart changes in most major disease states^1,2. Nevertheless, the full potential of cardiac ³¹P-MRS is not robustly accessible as the quantification of cardiac inorganic phosphate (Pi), important for ATP synthesis, is hindered by the dominant signal of 2,3-diphosphoglycerate (2,3-DPG), originating from the blood. We have recently demonstrated the ability to detect cardiac Pi in vivo using long TR and adiabatic excitation at 7T and use it to calculate cardiac pH, which is important as acidosis can limit heart’s ability to pump blood in disease3. However, even with acquisitions using relatively long TR, partial saturation of Pi signal cannot be fully excluded due to generally long T₁ times of phosphorus metabolites^4,5, making straightforward quantification without T₁ correction unfeasible. Therefore, our aim was to determine the T₁ of Pi in human heart at 7T using adiabatic excitation with a dual TR technique⁶. We also report Pi concentration, quantified using our measured T₁, relative to ATP and PCr in 8 healthy volunteers.

Materials and Methods

All measurements were performed on a 7T MR system (Siemens) equipped with a custom dual-tuned RF-coil, comprising a quadrature ³¹P coil (two 15cm loops, with overlap decoupling)⁷ and a single ¹H loop (10cm in diameter) for localization. First, four healthy subjects (3M/1F) were recruited for determination of the T₁ of cardiac Pi in compliance with local regulations. Two 3D UTE-CSI⁸ acquisitions (matrix size 8x16x8 and FOV of 240x240x200mm³) were made, one with the shortest possible and one matching our previous study³. The SAR limits for the narrow-banded AHP pulse9 we used did not allow TR< 3s. Our chosen TRs were therefore: TR₁ = 3s and TR₂ = 6s. As the AHP pulse has a relatively narrow bandwidth at the achievable B₁⁺, the excitation was centred at +600Hz from the PCr resonance frequency. Simulations of the Bloch Equations were used to estimate the expected errors in calculated T₁ for resonance offsets range from -180Hz to 180Hz from centre frequency, for the possible T₁s ranging from 1s to 10s. Second, data from eight healthy volunteers (7M/1F) were used to quantify the Pi/ATP and Pi/PCr ratios. We used our previously reported protocol for cardiac ³¹P-MRS using interleaved AHP excitation at 7T⁹. In short, the AHP excitation is interleaved to excite the PCr and ATP region in odd acquisitions and the Pi and DPG region in even acquisitions. The effective TR was 6s and other parameters match the T₁ measurements. Six septal voxels were fitted in all in vivo data using a Matlab implementation of AMARES¹⁰ and corrected for T₁ relaxation using the measured values for Pi and literature values for ATP and PCr⁵. The T₁ of Pi in chest muscles (again taking six voxels) was evaluated for comparison.

Results and Discussion

Figure 1 depicts the simulated off-resonance error in T₁ estimation using the AHP pulse for the lowest B₁⁺ expected in the heart, i.e. 16μT⁷. This shows that even in the worst case, the potential error is below 20% for a range of frequencies ±30Hz, i.e. 0.25ppm. The asymmetry of the AHP pulse starts to play a role for larger frequency offsets; the smoother side of the pulse, which faces PCr here, gives smaller errors. Figure 2 shows ³¹P-MRS spectra from the chest muscles and septum of one volunteer acquired at TR₁ and TR₂. The inter-subject mean T₁ of Pi measured in skeletal muscles was 6.3±0.7s and in human heart it was 5.1±0.2s. The chest muscle T₁ value is in good agreement with the measurements in calf muscle⁴ at 7T and the measured T₁ of cardiac Pi follows the trend of slightly shorter ³¹P T₁s in the human heart compared to skeletal muscle⁵. Values for each individual subject are given in Table 1. Table 2 shows the Pi ratios to ATP and PCr, together with the PCr/ATP ratio, for each individual subject. The inter-subject mean Pi/PCr ratio was 0.15±0.02, which is in good agreement with literature values of 0.12±0.03 determined at 1.5T using proton decoupling to separate Pi from 2,3-DPG signals (success rate in healthy volunteers was ~50% in these studies)^11,12.

Conclusion

This study provides a longitudinal relaxation time of cardiac Pi at 7T measured by the dual-TR technique with adiabatic excitation. The measured T₁ enables us to report the saturation-corrected Pi/PCr ratio for the first time in the human heart at 7T.

Acknowledgements

We acknowledge financial support from the Sir Henry Dale Fellowship awarded jointly by the Wellcome Trust and the Royal Society (098436/Z/12/Z), and Slovak Grant Agencies VEGA (2/0001/17) and APVV (15-0029).