Developing a Model to Predict the ACTH Response to a Social Stressor Using Clinical Variables and Genotype

Turner CA, Hagenauer M, Ramdas S, Mayer S, Li J, Young E, Watson SJ, Abelson J, Akil H
Annual Meeting of the American College of Neuropsychopharmacology. 2015.


Background: The ability to predict an individual’s stress response is essential to understanding disorders that are often precipitated by stress, including mood disorders. This understanding will also aid in the identification of individuals as stress resilient or stress vulnerable. This work is especially important for military populations where the incidence of stress-related disorders is quite high.

Methods: We performed a social stressor on controls, as well as individuals with major depressive disorder some of whom also had an anxiety disorder or post-traumatic stress disorder. Specifically, we measured the ACTH response to the Trier Social Stress Test (TSST) as an endophenotype. We also collected demographic data and a range of clinical rating scales. We then performed genotyping using Illumina’s HumanOmniExpress BeadChips. We pre-specified 62 candidate genes based on the neuroendocrine literature and on previous work in human postmortem brains. The panel consisted of circadian, immune, stress and fibroblast growth factor genes. After filtering the SNPs in PLINK, we combined our entire genotyped dataset of 209 subjects with HapMap (release 23a) and determined that the first two principal components of variation in our dataset were likely to represent the effects of population stratification. After removing SNPs with linkage disequilibrium greater than 0.5, we had 448 SNPs to interrogate. We then performed linear regression on the 63 subjects that had neuroendocrine data controlling for demographic and clinical variables to identify informative SNPs.

Results: Models integrating clinical variables and genetic variants were able to explain approximately half of the variance in the ACTH response to the TSST. Both clinical information and SNPs made significant contributions. A full model, explaining a significant proportion of neuroendocrine variability in response to a standardized psychosocial stressor, will be presented.

Conclusions: Extension and replication in larger groups is needed, but the data suggest that the endocrine response to a laboratory social stressor may represent a useful endophenotype for studying both genetic and environmental determinants of stress reactivity, which may well have relevance to vulnerability to mood disorders and PTSD. These results could provide useful biomarkers and help identify individuals who may be more vulnerable to stress-related disorders.