Introduction

Phosphorus (³¹P) MR spectroscopy is a promising technique to reflect energy metabolism and membrane synthesis non-invasively in vivo. Localized ³¹P MR spectroscopy could detect important phosphate metabolites in pancreatic cancer to explore its pathology. Although there are few pancreas ³¹P MRS studies at 7 Tesla [1], it is extremely challenging to perform ³¹P MRS in human pancreas at 3 Tesla due to limited coil sensitivity and lower SNR. To our knowledge, there are no reports of ³¹P spectroscopy in human pancreas at 3 Tesla yet. This study aimed to preliminarily investigate the feasibility of a localized ³¹P spectroscopy in normal human pancreas at 3 Tesla.

Methods

A 2D ³¹P MRS sequence with Image-Selected In vivo Spectroscopy(ISIS) localization and Chemical Shift Imaging(CSI) phase encoding was applied in 11min. 20sec. With local IRB approval and written informed consents, four healthy volunteers were scanned on a clinical 3T MR scanner (Elition X, Philips Healthcare, Best, the Netherlands) using a vendor-supplied transmit-receive ³¹P flex coil with a 14cm diameter. Additional parameters of the 2D ³¹P CSI sequence included: TR/TE = 4000/0.22 ms, acquired voxel size = 35x35 mm², acquired matrix = 4x4, reconstructed voxel size = 28x28 mm², FOV = 140 mm/140 mm, slice thickness = 50 mm, adiabatic excitation pulse, NSA = 14, number of samples = 2048, spectral bandwidth = 3000 Hz, broadband NOE, and broadband proton decoupling. In addition, a T2 weighted multi-slice multivane sequence was scanned in 3min. 45sec. to provide anatomical reference for the ³¹P spectra, and the sequence parameters were: FOV = 350x350x89 mm³, TR/TE = 945/72 ms, acquired voxel size = 1.5x1.5x4 mm³, reconstructed voxel size = 0.55x0.55x4 mm³. All ³¹P spectra were processed using the scanner software (SpectroView, Philips Healthcare), with gaussian (24Hz) and exponential (-3Hz) apodization. Peak fitting was performed in a range of -25 ppm to 15 ppm for a variety of metabolites including PCr, α-ATP, β-ATP, γ-ATP, Pi, PC, PE, GPC, and GPE, and the peak areas of these metabolites were obtained for each of the voxels. PE and PC was conventionally combined to represent phosphomonoester(PME), while the addition of GPE and GPC means phosphomonoester(PDE). Various ratios of these phosphate metabolites were calculated and compared to previous findings at 7 Tesla[1].

Results

Three voxels were included in each pancreas and analyzed individually. ³¹P spectra of three normal pancreases were excluded as the PCr peaks were higher than corresponding α-ATP peaks. ³¹P spectrum of the other pancreas at the central voxel was considered as reasonable with lower-than-αATP PCr peak, as shown in Figure 1. Corresponding fitted peak areas under the curve of β-ATP, α-ATP, γ-ATP, PCr, PDE(GPE+GPC), Pi, PME (PE+PC) were 414, 1179, 1331, 713, 597, 1657 and 987, respectively, as demonstrated in Figure 2. Furthermore, Figure 2 also listed relevant phosphate ratios PME/PDE=1.65, PME/Pi=0.6, PDE/Pi=0.36, PME/α-ATP=0.84, PDE/α-ATP=0.84, PME/β-ATP=2.39 and γ-ATP/α-ATP=1.13, respectively.

Discussion

Despite small cohort size, we successfully obtained a similar localized ³¹P spectrum in the central pancreas body of a healthy volunteer at 3 Tesla, with reference to a previous 7 Tesla finding [1]. Our PME/PDE ratio 1.65 was comparable to the reported ratio 1.9±0.8 at 7 Telsa. However, the PCr peak of pancreas body was relatively higher than the example spectrum at 7 Tesla, and only one spectrum of the total 12 analyzed pancreas voxels was valid for comparison in our study. This indicates that localized ³¹P spectroscopy at 3 Tesla are more vulnerable to PCr contamination from neighboring muscle or gallbladder than ³¹P MRS at 7 Telsa. Thanks to the high SNR in ultrahigh magnetic field, ³¹P spectroscopy at 7 Tesla could utilize smaller voxel size (20x20x36mm³ vs. 28x28x50mm³ , 4 voxels vs. 3 voxels) than ³¹P MRS acquisition at 3 Tesla. Hence, it suffered less from PCr contamination. Nevertheless, considering our acquisition time was only half of the 7 Tesla acquisition(11min.20sec. vs. 22min.37sec.), we could also reduce the acquisition voxel size and double number of signal average (NSA) to increase success rate and SNR at 3 Tesla in the future investigation.

Conclusion

This preliminary study demonstrated that in vivo localized ³¹P spectroscopy of human pancreas on clinical 3T scanners is possible and promising although challenging. It could potentially detect various phosphate metabolites including PME and PDE as early biomarkers of abnormal energy metabolism and membrane synthesis in pancreatic cancer.

Acknowledgements

No acknowledgement found.

References

1. Comparison of 31P MRS in human pancreas and liver at 7 Tesla, Leonard W.F. Seelen et. al., ID0513, ISMRM 2023 Proceedings.