T20 - New Insights Into The Pharmacogenomics Of Antidepressant Response From The Gendep And Star*D Studies: Results Of Rare Variant Analysis And High-Density Imputation

Abstract

Genome-wide association studies have generally failed to identify polymorphisms associated with antidepressant response. Possible reasons are limited coverage of genetic variants, phenotypic heterogeneity and small sample size. This study investigated the genetic predictors of antidepressant efficacy in Genome-Based Therapeutic Drugs for Depression (GENDEP) and Sequenced Treatment Alternatives to Relieve Depression (STAR*D) samples trying to address some of the reported limitations. In detail: 1) coverage of genetic variants was increased by adding exome array data to previously available genome-wide data and by genotype imputation using the largest available reference panel (Haplotype Reference Consortium (HRC)); 2) a meta-analysis was performed using SNP methods and multi-marker tests at gene and pathway level. Each dataset was imputed using Minimac3 and the HRC panel after standard quality control. Both samples included patients with diagnosis of major depressive disorder. In GENDEP, patients were partially randomized to escitalopram or nortriptyline; STAR*D patients were treated with citalopram. Escitalopram is the active isomer of citalopram, thus the whole GENDEP sample and the escitalopram-treated subgroup were meta-analyzed with STAR*D. The phenotypes were depressive symptom improvement and remission at week 12 according to standard scales. SNP-level analysis was performed using PLINK (pngu.mgh.harvard.edu/~purcell/plink/); gene and pathway analyses were performed using MAGMA (http://ctglab.nl/software/magma). Covariates were age, baseline severity, center of recruitment and ancestry-informative principal components. NEWMEDS (http://www.newmeds-europe.com) consortium samples, excluding GENDEP, served for replication. 7,062,950 SNPs were analysed in GENDEP (n=738) and STAR*D (n=1409). There was no evidence of genomic inflation (lambda values were ≤ 1.01). No SNP was associated with symptom improvement or remission in the full meta-analysis. In the citalopram/escitalopram analysis, rs116692768 (MAF=0.033, p=1.87e-08, ITGA9 or integrin alpha 9 gene) and rs76191705 (MAF=0.012, p=2.39e-08, NRXN3 or neurexin 3 gene) were associated with symptom improvement. At gene level (whole sample), OR4K2 was associated with improvement (corrected p=0.04), but its effect was inconsistent between the samples. At pathway level (whole sample), the Gene Ontology terms GO:0005694 (chromosome) and GO:0044427 (chromosomal part) were associated with improvement (corrected p=0.007 and 0.045, respectively). Genome-wide significant SNPs were not replicated in NEWMEDS SSRI-treated sample (n=751) (p>0.05). ITGA9 and NRXN3 show a meaningful biological rationale for being involved in antidepressant effect. ITGA9 codes for a membrane glycoprotein receptor for neurotrophins and NRXN3 is a transmembrane neuronal adhesion receptor involved in post-synaptic and pre-synaptic differentiation, with relevant implications for synaptic activity and synaptic plasticity. Interestingly, rs76191705 is a non-sense mediated decay transcript variant and a polymorphism in complete LD with it (rs79302561) acts as an enhancer of gene expression (Ensembl GRCh37 release 84). However, no convincing replication of these findings was achieved and further studies may be useful to clarify the role of these genes in antidepressant efficacy.