Neurogenetic plasticity and sex influence the link between corticolimbic structural connectivity and trait anxiety

Corticolimbic pathways connecting the amygdala and ventral prefrontal cortex (vPFC) are linked with trait anxiety, but it remains unclear what potential genetic moderators contribute to this association. We sought to address this by examining the inter-individual variability in neuroplasticity as modeled by a functional polymorphism (rs6265) in the human gene for brain derived neurotrophic factor (BDNF). Amygdala-vPFC pathway fractional anisotropy (FA) from 669 diffusion magnetic resonance images was used to examine associations with trait anxiety as a function of rs6265 genotype. We first replicated the inverse correlation between trait anxiety and amygdala-vPFC pathway FA in women. Furthermore, we found a moderating influence of rs6265 genotype such that the association between trait anxiety and right amygdala-vPFC pathway FA was strongest in women carrying the Met allele, which is linked with decreased activity-dependent neuroplasticity. Results indicate that the microstructural integrity of pathways supporting communication between the amygdala and vPFC help shape the expression of trait anxiety in women, and that this association is further modulated by genetically driven variability in neuroplasticity.


Neurogenetic plasticity and sex influence the link between corticolimbic structural connectivity and trait anxiety
M. Justin Kim, Reut Avinun, Annchen R. Knodt, Spenser R. Radtke & Ahmad R. Hariri

Genotype Data for the Healthy Subgroup
Consistent with the whole group analysis, genotype distribution did deviate from Hardy-Weinberg equilibrium across our entire healthy subgroup (χ 2 = 7.42, p = 0.006), but not for non-Hispanic Caucasian (χ 2 = 0.061, p = 0.805) or Asian (χ 2 = 0.217, p = 0.642) subgroups (the remaining subgroups did not yield sufficient number of individuals in one or more of the genotypes). There were 326 Val allele homozygotes, 173 heterozygotes, and 42 Met allele homozygotes.

Results for the Non-Hispanic Caucasian Subgroup
To test whether the findings from the whole group analysis were also observed in a subgroup with relatively homogeneous ancestry, the same analysis procedure was applied to non-Hispanic Caucasian participants only (n = 276; healthy subgroup n = 212). While the significance levels for some tests dropped, likely due to reduced power from smaller sample size, the overall findings were consistent with and in the same direction as the whole group analysis. A full summary of the results is described below.

Genotype Data for the Non-Hispanic Caucasian Subgroup
Genotype distribution did not deviate from Hardy-Weinberg equilibrium for the non-Hispanic Caucasian subgroup (χ 2 = 0.3, p = 0.583), as well as for the healthy subgroup (χ 2 = 0.222, p = 0.637). Across the entire non-Hispanic Caucasian subgroup, there were 180 Val allele homozygotes, 84 heterozygotes, and 12 Met allele homozygotes. As for the healthy subgroup, there were 139 Val allele homozygotes, 64 heterozygotes, and 9 Met allele homozygotes. Consistent with the whole group analysis, participants with the Val/Met and Met/Met allele were grouped together for further statistical analyses.

Demographic and Behavioral Characteristics
A total of 276 participants identified themselves as being non-Hispanic Caucasian (142 women, ages 18-22 years, mean age = 19.7 years). Average STAI-T scores were 36.2 (± 8.7), and there were no significant difference between men and women. Among them, 212 did not have past or current diagnosis for mental disorders (113 females, ages 18-22 years, mean age = 19.7 years; see Table S1 for full list of diagnoses). Average STAI-T scores for the healthy subsample were 35.3 (± 8), and once again there were no significant difference between men and women.

Amygdala-vPFC Structural Connectivity and Trait Anxiety
Similar to the whole group analysis, when all non-Hispanic Caucasian participants (n = 276) were included, hierarchical regression analyses revealed that adding amygdala-vPFC pathway FA in the second step did not significantly improve the model from the first step that included age, sex, and head motion, in

Sex Difference in the Present Brain-Anxiety Association
A significant moderating effect of sex was observed for the right amygdala-vPFC pathway in predicting trait anxiety, after controlling for the effects of age and head motion (ΔR 2 = 0.019, F(1, 370) = 5.434, β = -0.148, p = 0.021). Upon closer inspection, we found that the sex difference was driven by significant

Anxiety Association
There were no overall significant effects of BDNF genotype on the present brainanxiety association. However, a significant moderating effect of BDNF rs6265 genotype × sex interaction was observed for the right amygdala-vPFC pathway in predicting trait anxiety, after controlling for the effects of age, ancestry (the first two MDS components instead of four was used, since the analysis was restricted to the non-Hispanic Caucasian subsample), and head motion (ΔR 2 = 0.033, F(3,  Table S3. Figure S1. Moderating effect of rs6265 genotype on sex-dependent brainanxiety association within the non-Hispanic Caucasian subsample (n = 276). (ad) Consistent with the analysis in all participants, women Met allele carriers (blue circle) displayed an inverse correlation between the structural integrity of the right amygdala-vPFC pathway and trait anxiety, whereas men Met allele carriers (blue triangle) displayed a positive correlation. All correlation coefficients were calculated controlling for the effects of age, ancestry, and head motion. Figure S2. Moderating effect of rs6265 genotype on sex-dependent brainanxiety association when the analyses were limited to healthy non-Hispanic Caucasian subsample (n = 212). (a-d) Consistent with the analysis in all participants, women Met allele carriers (blue circle) displayed an inverse correlation between the structural integrity of the right amygdala-vPFC pathway and trait anxiety at a trend level, whereas men Met allele carriers (blue triangle) displayed a positive correlation. All correlation coefficients were calculated controlling for the effects of age, ancestry, and head motion. Table S1. Diagnoses for past or current mental disorders in the total study sample (n = 128) and the non-Hispanic Caucasian subgroup (n = 64).

Diagnosis Total Non-Hispanic Caucasian
Major depressive disorder 26 12 Bipolar disorder I or II 5 3 Bipolar disorder -Not otherwise specified 11 8 Hypomanic episode 12 4 Panic disorder 8 5 Agoraphobia 10 8 Social anxiety disorder 7 2 Generalized anxiety disorder 13 6 Obsessive-compulsive disorder 6 3 Posttraumatic stress disorder 1 0 Alcohol abuse/dependence 75 42 Marijuana abuse/dependence 23 10 Eating disorder 5 2 Antisocial personality disorder 1 0 Borderline personality disorder 1 0 The sum of the individual diagnoses is higher than the number of participants with mental disorders because a subsample of them has multiple comorbid diagnoses   Table S4. Summary of correlation coefficients for each sex × BDNF genotype subgroups for all participants. Coefficients were calculated controlling for the effects of age, ancestry, and head motion.