1. Structural neuroimaging phenotypes of a novel multi-gene risk score in youth bipolar disorder
- Author
-
Ronak Patel, Mikaela Dimick, Benjamin I. Goldstein, Maria Tampakeras, Kody G. Kennedy, Alvi H. Islam, Jaime Cazes, Natalie Freeman, Clement C. Zai, Lisa Fiksenbaum, Bradley J. MacIntosh, and James L. Kennedy
- Subjects
Candidate gene ,Bipolar Disorder ,Ventrolateral prefrontal cortex ,Adolescent ,Neuroimaging ,Biology ,03 medical and health sciences ,0302 clinical medicine ,Risk Factors ,medicine ,Humans ,Bipolar disorder ,Allele ,Genetic association ,Genetics ,Framingham Risk Score ,medicine.disease ,Magnetic Resonance Imaging ,030227 psychiatry ,Psychiatry and Mental health ,Clinical Psychology ,Cross-Sectional Studies ,Phenotype ,medicine.anatomical_structure ,Sample size determination ,030217 neurology & neurosurgery ,Genome-Wide Association Study - Abstract
Background Bipolar disorder (BD) is among the most heritable psychiatric disorders, particularly in early-onset cases, owing to multiple genes of small effect. Here we examine a multi-gene risk score (MGRS), to address the gap in multi-gene research in early-onset BD. Methods MGRS was derived from 34 genetic variants relevant to neuropsychiatric diseases and related systemic processes. Multiple MGRS were calculated across a spectrum of inclusion p-value thresholds, based on allelic associations with BD. Youth participants (123 BD, 103 healthy control [HC]) of European descent were included, of which 101 participants (58 BD, 43 HC) underwent MRI T1-weighted structural neuroimaging. Hierarchical regressions examined for main effects and MGRS-by-diagnosis interaction effects on 6 regions-of-interest (ROIs). Vertex-wise analysis also examined MGRS-by-diagnosis interactions. Results MGRS based on allelic association p≤0.60 was most robust, explaining 6.8% of variance (t(226)=3.46, p=.001). There was an MGRS-by-diagnosis interaction effect on ventrolateral prefrontal cortex surface area (vlPFC; β=.21, p=.0007). Higher MGRS was associated with larger vlPFC surface area in BD vs. HC. There were 8 significant clusters in vertex-wise analyses, primarily in fronto-temporal regions, including vlPFC. Limitations Cross-sectional design, modest sample size. Conclusions There was a diagnosis-by-MGRS interaction effect on vlPFC surface area, a region involved in emotional processing, emotional regulation, and reward response. Vertex-wise analysis also identified several clusters overlapping this region. This preliminary study provides an example of an approach to imaging-genetics that is intermediate between candidate gene and genome-wide association studies, enriched for genetic variants with established relevance to neuropsychiatric diseases.
- Published
- 2021
- Full Text
- View/download PDF