Disorder-Specific Mitochondrial, Telomere, and Plasticity Signatures Revealed by Cross-Disorder TWAS in OCD, MDD, and Bipolar Disorder

BackgroundRecent large-scale genome-wide association studies have identified hundreds of risk loci for obsessive-compulsive disorder (OCD), major depressive disorder (MDD), and bipolar disorder (BIP). Even so, the biological pathways that separate these clinically distinct conditions are still not well resolved. Much of the post-GWAS work to date has relied on broad enrichment analyses, which can blur signals that are specific to one disorder. To address that gap, we conducted a targeted transcriptome-wide association study (TWAS) centered on linked biological pathways involved in NAD metabolism, mitochondrial function, cellular senescence, telomere maintenance, and synaptic plasticity.MethodsWe applied S-PrediXcan to summary statistics from the 2025 flagship GWAS of MDD, BIP, and OCD and meta-analyzed predicted gene expression across six GTEx v8 brain regions. We then tested a hypothesis-driven panel of 40 gene sets covering NAD metabolism, mitochondrial energetics and biogenesis, senescence and telomere pathways, GLP-1 signaling, apoptosis, and synaptic plasticity using competitive permutation-based enrichment.ResultsMDD showed the clearest negative enrichment for mitochondrial and oxidative phosphorylation gene sets, with especially strong downregulation of AMT and SIRT4. BIP was marked by pronounced telomere-maintenance collapse together with upregulation of the PGC-1alpha biogenesis program and strong downregulation of NT5C. OCD showed the most negative senescence-program signature and a relative positive shift in long-term potentiation genes compared with MDD. Pairwise analyses confirmed meaningful differences between disorders, despite only modest overall concordance across them, with sign concordance ranging from 0.46 to 0.61. Gene-level discordance was common, including direction flips in COX8A and PARP1, whereas PARP10, CD40, and NDUFS2 were consistently downregulated across all three disorders.ConclusionsThese findings suggest that shared genetic liability can be routed through different transcriptomic pathways in OCD, MDD, and BIP. The results point to disorder-specific pathway signatures that may be useful for biomarker-enriched trials, biologically informed stratification, and precision subtyping.