195 Table 10
Spearman correlations between neurocognitive and sMRI data in SAD with EDT and SAD without EDT groups
Correlations in SAD without EDT
Neurocognitive test Caudate(L) ACC(L) ACC(R) Amygdala(L) Amygdala(R) Thalamus(L) WMS immediate -0.206(.384) .503(.020)* .235(.292) -.170(.449) -.133(.554) -.007(.977) HVLT immediate -.137(.565) .195(.398) .285(.199) -.325(.140) -.305(.167) -.150(.516) HVLT delayed -.308(.186) .418(.059) .456(.033)* -.283(.202) -.397(.068) -.083(.720)
Correlations in SAD with EDT
Neurocognitive test Caudate(L) ACC(L) ACC(R) Amygdala(L) Amygdala(R) Thalamus(L) WMS immediate .008(.970) -.313(.155) .197(.392) .155(.481) -.300(164) .013(.954) WMS delayed .018(.934) -.187(.403) .352(.118) .147(.502) -.330(.124) .024(.914) HVLT immediate -.199(.362) -.155(.610) -.064(.783) .021(.922) -.059(.791) -.392(.064) HVLT delayed -.177(.594) -.215(.337) .080(.732) .012(.956) -.050(.821) -.237(.277) GPB dominant -.434(.039)* .057(.800) -.206(.370) -.171(.434) .258(.235) -.073(.741) GPB non-dominant -.210(.336) .162(.472) -.324(.152) -.302(.161) .168(.444) -.068(.756) Stroop Colour .006(.978) -.016(.943) -.026(.910) .163(.456) -.065(769) -.366(.086) Stroop Word/Colour .119(.588) .087(.700) -.034(.885) .251(.249) .136(.535) -.037(.865) WCST NPE .102(.645) .187(.403) .064(.783) .220(.313) .427(.042)* .104(.637) WCST CLR -.301(.163) -.196(.383) -.024(.918) -.216(.321) -.436(.037)* .018(.935)
Note. These correlations were only run the findings from the pairwise comparisons that were
statistically different between the groups. This was an intentional focus as to further examine the relationship between grey matter differences and neurocognition. Abbreviations: WMS: Wechsler Memory Scales, HVLT = Hopkins Verbal Learning Test; GPB = Grooved
Pegboard; WCST = Wisconsin Card Sorting Test; ACC = Anterior Cingulate Cortex All brain volumes (Caudate; ACC; Amygdala; and Thalamus are measured in mm3)
* = P < 0.05, are all statistically significant.
196
7.4.1 Verbal memory correlations
In the SAD without EDT group there was a moderate positive correlation with the Wechsler memory scales logical memory immediate recall test and the left ACC (r = .503, p = 0.020). We found no correlations in the SAD with EDT group in the WMS logical memory
immediate recall test and brain volumes. In the SAD without EDT group there was a
moderate positive correlation with the HVLT logical memory delayed recall test and the right ACC (r = .456, p = 0.033). We found no correlations in the SAD with EDT group in the WMS logical memory delayed recall test and brain volumes.
7.4.2 Motor function correlations
In the SAD with EDT group there was a moderate negative correlation with the GPB dominant hand test and the left caudate nucleus (r = -.434, p = 0.039). No significant
correlations were found in the GPB non-dominant hand and grey matter volumes in the SAD with EDT group.
7.4.3 Executive functioning correlations
No significant correlations were found in the Stroop tests (colors and colors/words) and grey matter volumes in the SAD groups. In the SAD with EDT group we found correlations in the WCST non-perseverative errors and the right amygdala (r = -.427, p = 0.042) and the WCST conceptual level responses and the right amygdala (r = .436, p = 0.037).
197
7.5 Graph theoretical analysis (GAT) of sMRI brain volume data
Univariate analysis of group differences found specific grey matter differences in the SAD with EDT group compared to the control and SAD without EDT groups. We conducted a further analysis of grey matter differences, across diffuse neural networks using graph theoretical analysis (GAT) tools (see Figure 13 and Figure 14). This was conducted in order to examine the specificity of differences between the SAD groups. Only partial GAT analysis was conducted, up until just before full network hub analysis, to establish specificity of grey matter differences. The GAT analysis found marginal differences between grey matter
structures and a number of network nodes. We used the node network analysis to differentiate grey matter network structure differences between the SAD with EDT and SAD without EDT groups. We also conducted an analysis of regional network measures of between-ess and found differences suggesting smaller ACC, right cingulate cortex, right occipital lobe, right occipito-temporal lateral fusiform thickness and right orbital-H-shaped thickness in the SAD with EDT group compared with the SAD without EDT group. In particular, these findings support the ACC as an important component structure in the pathophysiology of SAD within the context of EDT. There is suggestion of specificity of neural network functioning
involving ACC interconnectivity in the SAD with EDT group compared with both the SAD without EDT and control groups.
Using the GAT approach is based on the assumption that positive correlations between morphometric/volumetric parameters of different brain regions are an indication of
connectivity (Bernhardt et al., 2008; Hosseini, et al, 2012). Previous studies have revealed that some of the tractography maps obtained from diffusion tensor imaging are strikingly similar to the pattern of correlations in cortical thickness produced by GAT analysis (Hosseini, 2012). The present study investigated between-group differences in regional network measures, specifically nodal between-ness, on networks threshold-ed. It is however important to note that none of these regions survived after correction for multiple
comparisons (P < 0.05), therefore we caution against the interpretation of these findings, as the analysis was exploratory in its nature.
198
Figure 13 represents Normalised between Node differences between all grey matter volumes from Free Surfer data between the SAD with EDT and SAD without EDT groups. Significant differences set at P < 0.05 for target structures, were found between ACC volumes between the two SAD groups. The ACC was an important area of focus in statistcial analysis and Node analyses was used to further examine these differences between the SAD groups.
199
Figure 14 represents the clustering coefficients of all grey matter structures of the SAD with EDT and SAD without EDT groups. Significant level set at P < 0.05.
200
7.6 Genotyping
7.6.1 Primary analysis of RGS2 and TPH2 genes
The genotype counts and associated frequencies (total n = 56) for the polymorphisms in the tryptophan hydroxylase 2 (TPH2) and regulator of G-protein signaling (RGS2) genes were as follows: TPH2 rs4570625 G/G: 42 (57.6%), T/G: 27 (37.0%), and T/T: 4 (5.5%); RGS2 rs4606 C/C: 40 (54.8%), G/C: 22 (30.1%) and G/G: 11 (15.1%) (SeeTable 11 for more details). Genotype distributions for both polymorphisms were in accordance with those expected under Hardy-Weinberg equilibrium (p = 0.333 and p = 0.091 for TPH rs4570625 and RGS2 rs4606, respectively).
Across-group genotype comparisons were undertaken. NHST regression analysis (logistic regression for categorical data and linear regression for continuous data) was selected to analyze differences in the sMRI left amygdala volume and 1H-MRS neurometabolite data
between TPH2 and RGS2 variants. The model was adjusted for age and gender.
Neurometabolites and regional brain volumes that were found to be statistically different in previous sMRI and 1H-MRS analysis of the groups with SAD only were included in the
analysis. Total intracranial volume (ICV) was used as a covariate in the analysis of sMRI data.
