F6. DISSECTING THE DIFFERENT EPIGENETIC ARCHITECTURES OF EPILEPSY AND EPILEPSY WITH COMORBID PSYCHOSIS

Background Epilepsy is a brain disorder that causes recurring, unprovoked seizures; in 5–6% of the patients, psychosis is a common comorbidity. The etiology and molecular mechanisms linking both conditions remain incompletely understood. DNA methylation (DNAm) may be seen as the mechanism linking genetic susceptibility, disease-related biological processes, treatment exposure, immune activity, and other molecular influences with downstream effector pathways. We aimed to characterize the DNAm architecture among individuals affected by epilepsy (EP), psychosis (PSY), and comorbid epilepsy-psychosis (EPSY). Additionally, we explored correlations associated with DNAm sites and multiple variables, in a deeply phenotyped sample. Methods Whole-blood DNAm was profiled in 30 individuals (9 with EP, 9 with PSY, and 12 with EPSY) using the Illumina MethylationEPIC array. Differential DNAm was assessed using linear models adjusted by sex, age, smoking status, BMI, estimated blood cell-type proportions and the first five principal components. Statistical significance was considered after inflation and Bonferroni correction (p < 5.6 × 10-8) and a nominal significance threshold of false discovery rate was set at FDR < 0.05. DMRs were identified using a comb-p-based regional approach and prioritized by regional Šidák significance, number of CpGs, and directional consistency. A correlation analysis between prioritized DMRs and clinical, sociodemographic, and neuroimaging variables was performed. Enrichment analysis of annotated genes was performed using Metascape, considering q < 0.05 as statistically significant. Results Eleven statistically significant CpGs and additional 98 nominally significant were identified. Differences were found between EP_vs_PSY (74 CpGs); PSY_vs_EPSY (17 CpGs) and EP_vs_EPSY (7 CpGs). After quality control and FDR correction, 184 DMRs were identified, including 180 DMRs with CpGs in a consistent direction of effects. These prioritized DMRs mapped to multiple loci, such as ZFP57, PCDHGA/PCDHGB, LRRC28/TTC23, PM20D1/PM20D1-AS1, RUNX1, NR4A2, RUFY1, and HOXA. The most significant DMR regional signal was ZFP57 between PSY_vs_EPSY, including 20 CpGs (β=8.73, p=5.7X × 10-50; FDR=1.4X × 10-47). DMRs included 272 unique genes and, in the global non-directional analysis, were significantly enriched in homophilic cell-cell adhesion (p=1.4 × 10-24; q=3.1 × 10-20) and cell-cell adhesion (p=4.19 × 110-15; q=4.55 × 10-11), mainly driven by PCDHGA/PCDHGB genes. 15 non-duplicate associations between DMRs and the well-characterized phenotypes were detected. Relevant associations included disorganized speech, hallucinations, delusions, and total clinical scores. The most significant correlation involved a hypermethylated DMR within the PCDHGA/PCDHGB cluster between PSY_vs_EPSY, which in turn negatively correlated with disorganized speech of EPSY (rho=-0.87; p=5.5 × 10-4). Discussion Significant methylation differences among patients with one disorder compared to those with the comorbidity were identified. It is worth noting that these changes were not observed in isolation at individual CpG sites but rather corresponded to regional epigenetic changes. The significant loci mapped to multiple regulatory and neurodevelopmentally associated relevant genes. These DMRs may also be correlated to symptom dimensions. This is the first study exploring the epigenetic architecture of the effects of psychosis in epilepsy patients.