Brain-Wide Versus Genome-Wide Vulnerability Biomarkers for Severe Mental Illnesses

Authors

Peter Kochunov
Yizhou Ma
Kathryn S Hatch
Si Gao
Neda Jahanshad
Paul M Thompson
Bhim M Adhikari
Heather Bruce
Andrew Van der Vaart
Eric L Goldwaser
Aris Sotiras
Mark D Kvarta
Tianzhou Ma
Shuo Chen
Thomas E Nichols
L Elliot Hong

Publication Date

11-1-2022

Journal

Human Brain Mapping

Abstract

Severe mental illnesses (SMI), including major depressive (MDD), bipolar (BD), and schizophrenia spectrum (SSD) disorders have multifactorial risk factors and capturing their complex etiopathophysiology in an individual remains challenging. Regional vulnerability index (RVI) was used to measure individual's brain‐wide similarity to the expected SMI patterns derived from meta‐analytical studies. It is analogous to polygenic risk scores (PRS) that measure individual's similarity to genome‐wide patterns in SMI. We hypothesized that RVI is an intermediary phenotype between genome and symptoms and is sensitive to both genetic and environmental risks for SMI. UK Biobank sample of N = 17,053/19,265 M/F (age = 64.8 ± 7.4 years) and an independent sample of SSD patients and controls (N = 115/111 M/F, age = 35.2 ± 13.4) were used to test this hypothesis. UKBB participants with MDD had significantly higher RVI‐MDD (Cohen's d = 0.20, p = 1 × 10−23) and PRS‐MDD (d = 0.17, p = 1 × 10−15) than nonpsychiatric controls. UKBB participants with BD and SSD showed significant elevation in the respective RVIs (d = 0.65 and 0.60; p = 3 × 10−5 and .009, respectively) and PRS (d = 0.57 and 1.34; p = .002 and .002, respectively). Elevated RVI‐SSD were replicated in an independent sample (d = 0.53, p = 5 × 10−5). RVI‐MDD and RVI‐SSD but not RVI‐BD were associated with childhood adversity (p < .01). In nonpsychiatric controls, elevation in RVI and PRS were associated with lower cognitive performance (p < 10−5) in six out of seven domains and showed specificity with disorder‐associated deficits. In summary, the RVI is a novel brain index for SMI and shows similar or better specificity for SMI than PRS, and together they may complement each other in the efforts to characterize the genomic to brain level risks for SMI.

Keywords

big data, DTI, ENIGMA, meta‐analysis, RVI, structural deficit patterns

Comments

Supplementary Materials

PMID: 36040723