Date of Award

Summer 8-15-2019

Author's School

Graduate School of Arts and Sciences

Author's Department


Degree Name

Doctor of Philosophy (PhD)

Degree Type



Background: Stress is among the most robust predictors of psychopathology. However, the factors that influence vulnerability to its pathogenic effects and the neural pathways through which such risk is expressed remain poorly understood. Preclinical and human studies show that stress exposure and stress-related disorders are characterized by alterations in the function and structure of corticolimbic brain regions including the amygdala and hippocampus. This raises the intriguing possibility that these regions may play a prominent role in risk for stress-related psychopathology. Polygenic indices of psychiatric risk generated by applying single nucleotide summary statistics derived from genomewide association studies (GWASs) of psychiatric phenotypes represent a promising approach to quantify individual risk that can be used to assess whether individual differences in these putative biological mechanisms represent genomically-conferred sources of vulnerability to stress-related psychopathology. Methods: Using data from college students of non-Hispanic European ancestry who completed the Duke Neurogenetics Study (n=526; PI: Hariri), I examined whether polygenic risk for posttraumatic stress disorder (PTSD) is associated with anxiety symptoms as well as amygdala and hippocampal gray matter volume and threat-related amygdala activity (study 1). Here, PTSD polygenic risk scores (PRS) were generated using summary statistics for the European Ancestry subsample provided by the Psychiatric Genomics Consortium Posttraumatic Stress Disorder group (PCG-PTSD) PTSD GWAS meta-analysis (Duncan et al., 2017). Voxel-based morphometry was applied to magnetic resonance imaging (MRI) data to quantify gray matter volume (GMV) of the hippocampus and amygdala. Threat-related amygdala reactivity was assessed using a widely used emotion face matching functional MRI (fMRI) challenge paradigm (Carre, Fisher, Manuck, & Hariri, 2012). In study 2, I tested whether these neural phenotypes (i.e., amygdala and hippocampal GMV and threat-related amygdala reactivity) are associated with genetic risk for broad anxiety disorder expression and mediate associations with the expression of anxiety symptoms. More specifically, we investigated whether two genome-wide significant loci identified from the Anxiety NeuroGenetics (ANGST) GWAS (Otowa, Hek, Lee, Byrne, Mirza, Nivard, Bigdeli, Aggen, Adkins, Wolen, Fanous, Keller, Castelao, Kutalik, der Auwera, et al., 2016), were predictive of concurrent or future anxiety through variability in amygdala and hippocampal structure and function. Furthermore, given recent research which has supported the role of amygdala reactivity in predicting vulnerability to future life stress, we also investigated whether amygdala reactivity interacted with recent life stress to predict future anxiety. Results: In study 1, I found that PTSD PRS was positively associated with increased right hippocampal volume; however no other significant associations were observed with left hippocampus or bilateral amygdala structure or function. Findings from study 2 showed that one of the genome-wide significant loci predictive of anxiety factor score, CAMKMT rs1067327, was significantly associated with increased right amygdala reactivity (Faces>Shapes contrast). However, contrary to prior research, such increased reactivity of the amygdala was not associated with concurrent or future anxiety. Moreover, no significant interactions were found between amygdala reactivity and recent life stress in predicting the development of future anxiety. Discussion: As studies often investigate neural abnormalities among individuals who already have the disorder, it is often impossible to disentangle whether such variability is indicative of premorbid or acquired risk. Although we find some support for the predispositional nature of genetic risk on amygdala reactivity and hippocampal volume, the results are inconsistent with our hypotheses as PTSD PRS was correlated with hippocampal volume in the opposite direction as might be expected and only one of the genomewide significant loci was correlated with amygdala reactivity and. Further, we failed to observe any association between these phenotypes with concurrent or future anxiety. A host of factors may have influenced our findings including our relatively small sample, use of a resilient sample (i.e., college students), questionable task-related fMRI reliability, and poor genomic estimates of risk. It is possible that prior brain-based associations with psychopathology are attributable to the expression of psychopathology or related environmental factors (e.g., childhood stress) or reflect false positives attributable to initial small samples in clinical neuroscience research. As more precise estimates of genomic risk arise from larger GWASs it will be important to conduct further research into biological mechanisms through which such genomic risk may be expressed.


English (en)

Chair and Committee

Ryan Bogdan

Committee Members

Deanna Barch, Agrawal Arpana, Barch Deanna, Nelson Elliot,


Permanent URL: https://doi.org/10.7936/j9z8-k732

Available for download on Tuesday, August 15, 2119