Evaluation of Glutathione T₂ in the Human Brain Using J-Difference MRS at 3 T: Multicenter Multivendor Study

The need to quantify brain glutathione (GSH) accurately by J-difference spectroscopy has stimulated assessment of the TE effects on GSH edited signals at the popular field strength 3 T. We performed multiple-TE J-difference MRS at two sites to evaluate the GSH T₂ relaxation and TE dependence of the GSH signal resolution. Two 10-ms spectrally selective Gaussian editing RF pulses were implemented in 3 T MEGA-PRESS sequences at two sites having different vendors. The sequences were optimized, with numerical and phantom analyses, for editing of the GSH 2.95 ppm resonance. The timings of the editing pulses within the sequences were tailored for high-amplitude GSH signal production for a TE range of 58-160 ms. In vivo human brain data were collected at five TEs (58, 70, 88, 116, and 150 ms) from five subjects at each site. Following LCModel analysis of difference and edit-off spectra independently between the sites, metabolite T₂ values were estimated with mono-exponential regression of the signal estimates. Simulations and phantom data indicated that the MEGA-edited GSH peak amplitude was progressively larger with increasing TE up to 125-140 ms and the maximum amplitude was 2- to 2.5-fold greater than the amplitude at TE of 58 ms. For in vivo data, the edited GSH peak was the largest at TE of 88 ms among the five TEs. Brain GSH T₂ was measured as 88 ± 11 ms from 10 subjects, with no significant difference between the sites. The LCModel-returned correlation coefficient between GSH and co-edited N-acetylaspartate (NAA) multiplet was significantly smaller at short TEs than at long TEs. Our data suggest that MEGA-edited GSH signal undergoes extensive attenuation with increasing TE due to the fast T₂ relaxation, and the edited GSH signal can be well resolved at short TEs with small interferences from adjacent co-edited NAA multiplet.