683. Genomic foundations of precision treatment: polygenic risk and pleiotropy in predicting therapeutic response
Abstract Background Major depressive disorder (MDD) is the most common mental disorder and is associated with substantial functional impairment, socioeconomic costs and reduced quality of life. One third of patients presents treatment-resistant depression (TRD), a major clinical challenge that highly contributes to the socioeconomic burden. Molecular underpinnings of TRD are only partially known. While TRD has distinct heritability, only few underlying genetic determinants have been identified by genome-wide association studies (GWAS). Pleiotropy, that is the association of genetic variants with more than one phenotype, and in-silico post-GWAS functional analyses, can be leveraged to increase our knowledge on the molecular underpinnings of TRD. Aims & Objectives We used a suite of cutting-edge methods on large genome-wide association datasets to identify novel genetic determinants associated with TRD. Method Preliminary analyses included genome-wide summary statistics from the Vanderbilt University Medical Center (VUMC) Synthetic Derivative and the Mass General Brigham (MGB) Research Patient Data Registry GWAS meta-analysis (PMID: 38745458), in which TRD was defined based on treatment with electroconvulsive therapy (VUMC: 225 TRD cases and 106,564 MDD controls; MGB: 242 cases and 78,378 MDD controls, all of European ancestry). Cross-trait analyses with the conditional false discovery rate (condFDR) method to identify novel genetic variants associated with TRD conditioning on inflammatory marker levels (serum C-reactive protein, CRP), metabolic traits (body mass index and type 2 diabetes) as well as aging biomarkers (leukocyte telomere length [LTL] and a multivariate GWAS of aging [mvAge]) are ongoing. Summary-data-based Mendelian randomization (SMR) was used to identify genes associated with TRD via changes in brain gene expression based on data from the PsychENCODE project (RNA-seq data from 1.387 prefrontal cortex samples). Results Using condFDR, we identified eight, four and ten independent genetic variants significantly associated with TRD conditioning on CRP levels, LTL and mvAge, respectively. Two loci located in genes previously associated with TRD, i.e. FTO and MCHR1, and an intergenic locus with AC093326.3 as the nearest gene, were associated with TRD conditioning on all three phenotypes. Two novel loci (located in the JAZF1 and HSD17B12 genes) were identified conditioning either on CRP or mvAge. In addition, novel loci were identified conditioning on CRP levels (CRTAM, CDKL1, WWP2), LTL (RP11-774G5.1) or mvAge (KCNK3, TCF7L2, RNU2-23P, WWP2, ZBTB46). Using SMR, increased expression of XPNPEP3 on chromosome 22 was associated with TRD (top single nucleotide polymorphism: rs2143610, beta SMR = 0.04, p = 4.5E-06, adjusted p = 0.04). Analyses on metabolic phenotypes as well as additional analyses including summary statistics from a larger GWAS that evaluated TRD risk and treatment resistance in MDD across three Nordic countries (PMID: 40571737) are currently ongoing, to further explore and extend the present findings. Discussion & Conclusions Analytical approaches based on pleiotropy and quantitative trait loci can be leveraged to identify novel loci potentially associated with TRD.