Age-related differences in GABA: Impact of analysis technique

Previous research using magnetic resonance spectroscopy (MRS) has indicated that GABA levels decline in multiple brain regions over the course of healthy aging. However, brain atrophy also occurs during healthy aging, and as a result the tissue composition of MRS voxels (i.e., the percentage of grey matter, white matter, and cerebrospinal fluid in the voxel) may also differ between age groups. Many authors therefore argue for applying a correction to GABA estimates in order to control for differences in tissue composition. Here, we use data from a large healthy aging study to investigate the influence of three tissue correction strategies on age-group differences in GABA. We also evaluate the use of different analysis packages and reference metabolites on group differences. A 3T MEGA-PRESS sequence was used to obtain spectra from seven voxels placed in the visual, auditory, and sensorimotor cortex of 58 young adults (aged 18-29 years) and 85 older adults. We obtained several different estimates of GABA concentrations from the spectra using two analysis software packages (Gannet 3.1 and LCModel), three reference metabolites (water, creatine and N-acetylaspartate) and four tissue correction strategies. Young adults consistently demonstrated significantly higher GABA concentrations in the visual, auditory, and sensorimotor cortex when we used an uncorrected GABA estimate referenced either to water or creatine. When uncorrected GABA was referenced to N-acetylaspartate, age-related differences were observed only in the right ventral visual cortex. Similarly, when any of the four tissue corrections were applied to the data, only age-related differences in the left and right ventral visual cortex voxels remained. Correlations between GABA concentration estimates obtained from different software packages were moderate, as were correlations between uncorrected GABA estimates when different baseline metabolites were used. Correlations between all tissue corrections we explored were extremely high. These results confirm that reports of age-related differences in GABA concentrations are driven, at least in part, by changes in tissue composition.