summarizes the demographic, clinical, and cognitive composite scores in three study groups. A χ2-test of independence revealed no difference in the male-to-female ratios among the three study groups. There were also no group difference in years of formal education; however, MCI patients were significantly older than control and preMCI subjects. As expected, MCI patients had higher CDR sum of boxes scores and lower mean MMSE, memory and executive composite scores and than controls and preMCI subjects (see ).
Description of the three study groups
The first objective of this study was to compare hippocampal volumes in preMCI, MCI, and control subjects. Analyses revealed no significant group effect for hippocampal volume.
The second objective of this study was to compare fusiform volume in preMCI, MCI, and control subjects. We were unable to obtain Freesurfer parcellation for five control subjects, six preMCI subjects, and two MCI subjects due to issues with scan quality. Analyses of the remaining data revealed a significant main group effect for fusiform (F2,34 = 7.60, p = 0.002) volume (see ). Planned contrasts revealed that controls had more fusiform volume than MCI and preMCI subjects.
Mean volumes and metabolite ratios for three study groups
The third objective of this study was to compare the volumes of the ERC and parahippocampal gyrus in preMCI, MCI, and control subjects. Analyses revealed significant main group effects for both ERC (F2,34 = 7.62, p = 0.002) and parahippocampal (F2,34 = 3.58, p = 0.04) volume (see ). Planned contrasts revealed that controls had more ERC volume than MCI and preMCI and more parahippocampal volume than MCI.
The fourth objective of this study was to compare the patterns of GM atrophy in the four lobes of the brain in preMCI, MCI, and controls. Analyses revealed a significant main group effect for frontal GM volume (F2,47 = 5.91, p = 0.005). Planned contrasts showed that controls had more frontal GM volume than MCI (p = 0.016) and preMCI (p = 0.003).
The final objective of the study was to examine metabolite ratios in the posterior cingulate in preMCI, MCI, and controls. Single-voxel 1H MRS data was available for 9 CN, 9 preMCI and 13 MCI subjects. Analysis revealed a signifi-cant main group effect for NAA/mI (F2,30 = 5.26, p = 0.012). Planned contrasts showed that controls had higher levels of NAA/mI (1.57 ± 0.14) in the posterior cingulate region than MCI (1.32 ± 0.13, p = 0.005) and marginally more NAA/mI than preMCI subjects (1.55 ± 0.18, p = 0.06).
Because there were significant group differences in posterior cingulate NAA/mI, we used the Freesurfer parcellation in a post hoc analysis to examine whether there was a similar group difference in posterior cingulate volume. As shown in , the region where we placed the voxel for proton spectroscopic imaging was part of a region classified by Freesurfer as the isthmus of the cingulate. Post hoc analysis revealed a significant main group effect for volume (F2,34 = 3.45, p = 0.45). Planned contrasts showed that while MCI had less volume (p = 0.013) in this region of the brain than controls, there were no significant difference between MCI and preMCI subjects.
Examples Freesurfer parcellation of the (1) entorhinal cortex, (2) parahippocampal and (3) fusiform gyri, and (4) the isthmus of the cingulate gyrus. The black square denotes the approximate location of the single MRS voxel.