A GENETIC APPROACH TO UNDERSTANDING PSYCHOTROPIC DRUG EFFECTS ON BRAIN STRUCTURE

While neuroimaging studies have reported many brain alterations linked to psychiatric disorders, there has been substantial heterogeneity in findings. Effects of psychotropic medications on brain structure likely contribute to this inconsistency, yet disentangling disorder from drug-induced effects on brain structure is challenging. Moreover, small sample sizes limit most studies to investigate the associations between broad drug categories with brain structure, preventing fine-grained analyses of specific medications. Here, we leverage genetics to mitigate these issues and expand our understanding of the putative effects of psychotropic drug treatment on brain structure. Genome-wide association studies of brain morphology, white matter microstructure, and several psychiatric disorders were used. Leveraging the novel bivariate GSA-MiXeR, heritability enrichment (delta AIC > 0 and fold enrichment (FE) > 1) for gene sets defined by targets of 23 commonly used psychotropic drugs was estimated. Additional enrichment testing using MAGMA was employed (p < 0.05). If both a brain structure and psychiatric disorder were enriched for a psychotropic gene set, a genetic correlation (rg) was estimated within that gene set. Therefore, a significant rg suggests the psychotropic drug may affect disorder relevant mechanisms that alter brain structure and provides an expected direction for brain alterations. Future analyses will include (i) additional approved psychotropic drugs, (ii) spatial analyses of the pattern of genetic associations across subregions of the brain, and (iii) individual-level validation of associated brain structures. When grouping psychotropics into broad categories, small to moderate FE was observed using bivariate GSA-MiXeR. For example, the volume of subcortical brain structures was enriched for antidepressant genes (FE range=1.29-2.19) and antiepileptic mood stabilizer genes (FE range=1.55-1.67), meanwhile; measures of white matter microstructure were enriched for both antidepressant and antipsychotic genes (FE range=1.08-2.66). However, gene sets for broad psychotropic categories were not enriched using MAGMA. When investigating specific psychotropics, large variability in the strength of enrichment was observed. For example, strong gene set enrichment for the antidepressant tranylcypromine was observed for the volume of the amygdala (FE=17.25, MAGMA_p=0.04) and pallidum (FE=12.29, MAGMA_p=0.03); meanwhile, volumes of several subcortical brain structures were variably enriched for genes linked to the antipsychotic aripiprazole (FE range=3.56-12.21, MAGMA_p range=0.02-0.04). Several psychiatric disorders had a significant rg with brain structure within psychotropic gene sets. For example, negative rgs were observed between schizophrenia and cortical surface area within the paliperidone gene set (rg=-0.59, p=0.01) and the volume of the pallidum in the risperidone gene set (rg=-0.47, p=0.03). Our genetic study suggests that grouping drugs into broad categories may mask nuanced psychotropic-specific associations with brain structure. Further work will determine if neuroimaging metrics with a genetic link to a given psychotropic, particularly an observed rg, may serve as indicators of treatment response. Crucially, genetics can be leveraged to further our understanding of psychotropic drug effects on brain structure and potentially inform objective markers of treatment efficacy.