Synopsis

Keywords: Non-Proton, Spectroscopy, Phosphorus, pH, cancer

In this study, we analyzed pH values obtained via high-resolution ³¹P MRSI at 7T in different subjects and different regions of the human brain, and identified a trend towards lower pH values in the frontal region compared to other brain regions. The potential regional variation should be taken into account in cases, where only small pH changes between different tissues are expected and partial volume effects might dominate.

Introduction

The pH value in living tissue is of importance in cellular metabolism, and is a valuable biomarker for cancer imaging. ³¹P MRSI enables to non-invasively obtain pH values, and had recently been shown to acquire whole-brain maps of intra- and extracellular pH with reasonable spatial resolution at 7T¹. Though the attainable spatial resolution is in principle high enough to separate various tissue types, partial volume effects might still hide regional pH differences across the human brain, particularly when healthy or tumorous pH changes are not well-pronounced. Therefore, the aim of this study was to analyze the variability of pH values obtained via high-resolution ³¹P MRSI at 7T in different subjects and different regions of the healthy human brain, and to identify potential implications for its application to research of glioma.

Methods

Six healthy volunteers (4 male, age: 22-35) and two patients with low-grade glioma (2 male, age: 23&34) were examined on a 7T system (MAGNETOM, Siemens) using a ³¹P/¹H coil with 32 ³¹P receiver channels (Rapid Biomedical). The ³¹P MRSI data (isotropic nominal resolution: 12.5mm; effective voxel size V_eff=5.7ml; T_meas=51min) were acquired and processed as described in detail in^1,2. All data processing was performed in MATLAB R2021a (MathWorks). The quantification of localized ³¹P spectra was performed with a home-built MATLAB implementation of AMARES³ using the model described in², which includes intra- and extracellular inorganic phosphate resonances (Pi, ePi). Intra- and extracellular pH values (pH_i, pH_e) were calculated according to the modified Henderson-Hasselbalch⁴ equation using the quantified chemical shifts of Pi and ePi, respectively, and Phosphocreatine (PCr). For regional analysis, a frontal, parieto-temporal and occipital region-of-interest (ROI) was automatically selected in each subject with a home-build segmentation tool. This tool uses an automated segmentation for grey (GM) and white matter (WM) (SPM⁵) based on T1w and T2w images, combined with a brain atlas provided by the SPM extension automated anatomical labelling atlas⁶. In the patients, selection was performed only on the contralateral healthy brain tissue, and additionally a tumor ROI was selected by a radiologist using clinical ¹H images.

Results

Figure 1 illustrates a representative set of ³¹P spectra localized in the three defined brain regions of one volunteer (#1). Subtle regional differences in the chemical shifts of Pi and ePi are observed, which are visible in all subjects. The corresponding intracellular pH maps (Figure 2) reveal regional differences in healthy brain tissues accordingly, which appear to relate to GM and WM differences. Note that in the shown patient (#2, Figure 2 bottom row) pH_i is increased only in some voxels of the tumor region, compared to the contralateral healthy tissue. The distribution of pH values obtained in individual brain regions, like in the representative ROI analysis of volunteer #1 (Figure 3), indicates differences with a trend towards lower median pH values in the frontal lobe compared to the parieto-temporal and occipital lobe. These observations substantiate further across the group of volunteers (Figure 4). Figure 5 illustrates the ROI analysis for the regional pH differences of both patients, where the observations from the healthy subjects are resembled well. Note that in patient #1, the tumor volume is spatially well separated from the contralateral healthy ROIs, while the frontal ROI in patient #2 is in direct vicinity to the tumor, thus suffering from voxel bleed. In the tumor ROI, pH values show in some voxels pronounced differences compared to healthy tissue (cf. Figure 2), as indicated by the large value range. The median pH values across the whole tumor ROI appear to follow the median pH in the analyzed brain location.

Discussion

The group analysis indicates a certain robustness in detecting regional pH differences across the human brain of different subjects using ³¹P MRSI. pH differences between GM and WM were already reported earlier⁷ , and were also visible in our pH maps (Figure 2). Note that in the ROI analyses GM-WM differences are absent in all regions (Figure 3&4), because corresponding GM and WM voxels are too close to each other. In this study, a regional pH difference was found across the human brain (Figure 3-5). However, partial volume effects are questioning its origin. The potential voxel bleed from nearby muscle tissue to the parieto-temporal and occipital ROIs might explain the difference in pH values compared to the frontal ROI. A varying fraction of GM and WM content in the analyzed regions was not found across the subjects (data not shown), thus seems not to explain the regional differences. However, larger cohort sizes are required to poof the observations. Nonetheless, our findings highlight the importance of (I) achieving highest possible spatial resolution and (II) consistent placing of healthy control tissue ROIs, in clinical cases where ROI-averaged pH differences are small (Figure 2&5).

Conclusion

The presented data indicates that pH maps of the human brain with comparable regional variation in different subjects can be obtained via ³¹P MRSI at 7T. A trend towards lower pH values was identified in the frontal region compared to other brain regions. This regional variation should be taken into account in clinical cases, where only small pH changes compared to healthy tissues are expected and partial volume effects might dominate.

Acknowledgements

No acknowledgement found.