APT imaging has become a useful method to assess the tumor and stroke in vivo based on endogenous mobile proton concentration and pH changes. However, other tissue and experimental parameters such as water proton concentration and water longitudinal relaxation time (T_1w) may also affect the APT signal.^1,2 One interesting issue raised recently is how the relationship between T_1w and water concentration affects the observed APT signal, which has become a controversial topic in evaluating the signal origins for APT imaging.³ In this study, we analyzed datasets acquired from multiple animal disease models with two quantification methods (MTR_asym(3.5ppm) and recently proposed APT^# and NOE^# methods⁴) to investigate the correlations between the quantified APT-MRI signals and other parameters (water proton concentration and T_1w).

MRI experiment: Six human glioblastoma (hGB)-bearing rats, six ischemic rats with MCAO at two time points (2 hours and 1 day), and five U87-bearing rats treated using XRT (fractional XRT with 3 Gy/day during 8 days) at three time points (pre XRT, post XRT, and post XRT after 6 days) were scanned at 4.7T. CEST datasets with 61 frequency offsets (S₀ and -15~15ppm@0.5ppm intervals) were acquired with a long continuous-wave RF saturation pulse (power/time = 1.3 μT/4 s). WASSR dataset with 26 frequency offsets (-0.6~0.6ppm@0.05ppm intervals) were acquired with 0.5 μT RF saturation power. High SNR APT images were acquired using two frequency offsets (±3.5ppm) with sixteen signal averages. The T_1w map with seven IRs (0.05~3.5 s), the T_2w map with seven TEs (30~90 ms), the ADC map with seven b-values (0~1000 s/mm²), and the CBF map with ASL technique were also acquired.

Data processing: Quantification of the APT^# and the NOE^# analysis procedures were performed with following approaches³: (i) B₀-corrected Z-spectrum was fitted to Henkelman's two-pool MTC model with the super-Lorentzian lineshape. (ii) Limited data points (+15~+7ppm) were fitted to avoid possible CEST and NOE contributions. (iii) The EMR signals (Z_EMR) in the whole offset frequency ranges were obtained using fitted parameters, and the differences between Z_EMR and experimental data at 3.5ppm (APT^#) and -3.5ppm (NOE^#) were calculated. The T_1w and T_2w maps were fitted by I=I₀+A∙exp(-TI/T_1w) and I=I₀∙exp(-TE/T_2w), and we assumed that [water proton]≈I₀. To remove the T₂ effect in the [water proton] map, it was corrected with an exp(-TE/T_2w) factor. For the correlation analysis between the CEST signals and T_1w/[water proton], the Pearson's correlation coefficients (r) and p values were calculated.

In the hGB model, large APT^# signal and NOE^# signal were clearly observed (Fig. 1). The tumor signal was also clearly showed hyperintensities on APT^# and MTR_asym(3.5ppm), compared to the normal tissue. In the results between GM and WM, the GM signal intensities were slightly higher than the WM in the APT^# and MTR_asym(3.5ppm). Notably, the APT^#, NOE^#, and MTR_asym(3.5ppm) values showed no significant correlations with T_1w/[water proton] (all |r|<0.19; p>0.05).

Based on the ischemic group, the APT^# and MTR_asym(3.5ppm) signal intensities in ischemic lesions were lower than in normal tissue lesions (Fig. 2). The NOE^# signals seemed relatively larger than the APT^# signals, and this made the MTR_asym(3.5ppm) to be negative. In the cases of correlations between CEST signals and T_1w/[water proton], there were no significant correlations (all |r|<0.20; p>0.05).

The results of the XRT treatment group also showed large APT^# and NOE^# signals (Fig. 3). Notably, the APT^# and MTR_asym(3.5ppm) signal intensities were decreased according to the XRT progresses. For the correlations results between CEST signals and T_1w/[water proton], no significant correlations were found (all |r|<0.19; p>0.05). In addition, unlike previous theoretical indications¹, the results from the treated tumor group clearly showed that the APT signal is not a simple proportional relationship with the T_1w signal. Instead, although the T_1w values were unchanged or even increased, the APT values were decreased during the XRT treatment progress.^5,6

Tumor tissues showed hyperintensities, compared with normal tissues, on the ADC, T_1W, [Water proton], APT^#, and MTR_asym(3.5ppm) maps (Fig. 4). On the T_2w map, hyperintensities were observed at the ischemic and XRT treated tumor lesions. These conditions may reflect the increased edema signals in both lesions. In addition, T_1w/[water proton] map showed negligible signal differences between the tumor and normal tissues, and the ischemia and normal tissues.

Our results clearly showed the effect of combined T_1w and water concentration on the APT signals, which was mostly canceled out in multiple disease models. Therefore, the signal changes on APT imaging significantly reflect the mobile amide proton concentration and/or tissue pH variation (related with amide proton exchange rate).