As seen in , the two groups were well matched in age, parental education and degree of drug use. The most robust difference between groups, aside from their degree of exposure to PVA, was in their degree of perceived financial stress growing up (d′ = −0.96). PVA subjects indicated that their family’s financial resources were on average adequate, while controls indicated that they were more than adequate. There were slight differences between the groups in gender distribution, years of education, alcohol use, and measures of memory and IQ that could serve as potential confounders based on a d′ > 0.1 criteria.
As expected, subjects in the PVA group had substantially higher levels of anxiety, depression, somatization and anger-hostility. Altogether, 48% of the subjects in the PVA group had a history of mood disorders, and 24% had a history of anxiety disorders. Almost all were currently in remission.
The most prominent neural finding was a significant increase in GMV in the left superior temporal gyrus (STG) in individuals exposed to VA (BA 22; Talairach’s coordinates x= −61 – −50, y= −34 – −18, z= −1 – 13, cluster size = 676, P = 0.004, corrected cluster level) (). On average there was a 14.1% increase in GMV in this cluster in the VA subjects. No other areas of increase were found with a corrected cluster probability value that approached significance.
A significant correlation was found between the left STG GMV and PVAS scores at the corrected cluster level (P = 0.002), FWE corrected voxel level (P = 0.001), and FDR corrected voxel level (P = 0.002). GMV of this region correlated significantly with PVAS scores across all subjects (r = 0.521, P = 0.001). This relationship was particularly strong for the subjects with PVA (r = 0.55, P = 0.01), but was not apparent in healthy control subjects (r = 0.238, P > 0.3).
Multiple regression analysis, using the covariates included in the VBM analysis, indicated that GMV in left STG (BA 22) correlated significantly with maternal and paternal VAS scores and parental education. As indicated in , the overall correlation was high (r = 0.810, adjusted r2 = 0.573, P < 0.0001) with the major determinants being maternal VAS (β = 0.544, P < 0.0001), paternal VAS (β = 0.300, P = 0.018), and level of parental education (β = −0.577, P < 0.0001) (). Interestingly, in the PVAS group maternal VAS (β = 0.763, P = 0.006) and paternal VAS (β = 0.629, P = 0.013) were significant independent variables, but parental education was not. In contrast, parental education (β = −0.705, P = 0.013) was a strong determinant in controls, but maternal and paternal VAS were not ().
Multiple Regression Analysis Indicating Relationship between Gray Matter Volume in Superior Temporal Gyrus and Exposure to Verbal Aggression
indicates the association between potential confounders and left superior temporal gyrus GMV. Overall, none of these potential confounders, either alone or in combination, correlated significantly with the dependent variable (r2 = 0.26, adjusted r2 = −0.19; F13,21 = 0.58, p > 0.8). This suggests that none of these parameters qualified as confounders that could account for the association between exposure to PVA and superior temporal gyrus GMV.
Multiple regression analysis indicated that STG GMV accounted for a modest portion of the variance in the verbal comprehension index (β = 0.25, P = 0.05) in addition to FSIQ (β = 0.67, P < 0.0001) (r2 = 0.4988, adjusted r2 = 0.4546, P < 0.0001). There was also a significant correlation between left STG GMV and consumption of hard liquor (r = 0.497, P < 0.002). There were no other significant corrected correlations between psychiatric symptom ratings and STG GMV.
Using a lower criteria for statistical significance revealed a 10.5% increase in GMV in the left parahippocampal gyrus (BA 36; Talairach’s coordinates x= −38, y= −37, z= −10, cluster size = 129, P < 0.001, uncorrected voxel level) in PVA subjects. Examination of voxels with reduced GMV in PVA subjects identified no significant corrected voxel level-cluster regions. One tiny region of reduced GMV in PVA subjects was identified. There was a 9.4% reduction in GMV in the right middle frontal gyrus (BA 9, x= 30 y= 32 z= 28, cluster size = 4) that was significant at the uncorrected voxel level (Z = 3.15, P = 0.001).