From Adversity to Psychopathology: Long-Term Epigenetic Consequences in Adversity-Divergent Twins

Background: Mental disorders are one of the leading causes of global disability and represent significant disease burden worldwide. Psychosocial adversity (PSA) has been established as one of the risk factors in the development of psychopathology. DNA methylation (DNAm), particularly at CpG sites, has been proposed as a candidate mechanism through which environmental and lifestyle factors become biologically embedded, potentially acting as a mediating factor in the relationship between PSA and psychopathology. However, these mechanisms are still not entirely understood, and isolating post-natal environmental influences without the confounding of any genetic predispositions poses a major challenge. The extent to which adversity-induced DNAm patterns predict or mediate mental health outcomes – apart from merely assessing exposure history – remains unclear. We hypothesize that (1) PSA leads to differential patterns in DNAm between the exposed and control twins, (2) the link between PSA and mental health outcomes is partially mediated by altered DNAm, and (3) the extent of mental disorder symptoms is linked to DNAm (scale model).
Aims: The overarching aim of this project was to understand the epigenetic mechanisms involved in the relationship between PSA experienced from pre-school age through early adulthood, and mental health outcomes. In order to assess our hypotheses, we (1) developed and validated methodology for quantifying PSA in monozygotic (MZ) and dizygotic (DZ) twin to identify twin pairs that diverge meaningfully to warrant further analyses, (2) examined whether specific CpG methylation sites mediate associations between PSA and psychological symptoms, and (3) evaluated whether aggregate poly-epigenetic scores (PES) predict psychological symptom severity and mediate the PSA – psychopathology relationship.
Methods: Three studies were conducted using data from the German TwinLife cohort. Study I screened 739 twin pairs (349 MZ, 390 DZ; ages 17–32) for PSA, using six validated adversity instruments mapping experiences like social and peer rejection, familial and household conflict, lack of social support, and general negative life events. We applied standardized score rankings, Euclidean distances, and standard deviation sanity checks to identify adversity-discordant pairs that would be invited for subsequent studies within the project. Study II examined 9 MZ twin pairs (N = 18) selected in Study I. Using the Illumina Infinium MethylationEPIC v2.0 BeadChip, we extracted genome-wide DNAm β-values, utilized the PSA scores established in Study I, and collected psychological symptom scores via structured clinical interviews based on the Mini-DIPS. Differential methylation analysis, dimension reduction (PLS-DA, sPLS), and bootstrap mediation models tested whether specific CpG sites quantified as β-values mediated the associations between PSA and psychological symptoms. Study III evaluated whether PES – constructed from 200 CpG sites weighted by variable importance projection (VIP) scores – predicted nine psychological symptom dimensions and mediated adversity effects in 22 participants (9 complete twin pairs plus 4 unpaired individuals).
Results: Study I identified 144 MZ and 173 DZ adversity-discordant twin pairs (approximately 41–44% of the sample). Additive genetic factors, Common (shared) environment, and E (unique or non-shared) environment (ACE) modelling revealed that 43% of adversity variance was attributable to unique environment, 23% to shared environment, and 33% to genetic factors, validating the rationale for the case co-twin design. Study II identified 73 differentially methylated CpG sites between adversity-exposed and control twins. Mediation analyses of biologically relevant candidate sites revealed significant indirect effects for several loci, with a predominant suppression pattern: adversity-induced hypomethylation associated with GTF2I (cg24977276) and PRKAR1B (cg17650397) was linked to altered depression and anxiety symptom scores, suggesting compensatory epigenetic mechanisms as a response to PSA exposure. In contrast, SS18 promoter hypermethylation (cg15060929) exhibited positive mediation (higher PSA exposure was linked to increased methylation), transmitting 94.9% of the adversity effect on bipolar symptoms. Study III replicated prior findings that PES significantly differentiated adversity-exposed from control twins (d = 1.12, p = 0.017), with within-pair adversity-PES correlations confirming the association (r = −0.69, p = 0.039). However, PES showed no significant associations with any psychological symptom dimension (all r ≤ 0.26, all p > 0.05), and mediation analyses across 45 models revealed no significant indirect effects.
Conclusion: These results provide observational evidence that suggests that PSA is biologically embedded through DNAm in the present sample, with specific CpG sites mediating adversity effects on psychological symptoms. The findings provide preliminary evidence that DNAm may function not only as a mechanism transmitting risk but also as a compensatory mechanism conferring protection. This view challenges traditional "molecular scar" interpretations of adversity-related epigenetic modifications, but is in line with emerging body of research that suggests that adversity-induced DNAm is highly context-dependent and not necessarily straightforward. The non-significant results of PES as a predictive or mediating factor in the link between PSA and psychological outcomes may be partially due to sampling limitations, but could also suggest that aggregate methylation scores may function as stress history biomarkers rather than measures of psychological vulnerability/resilience. These findings have implications for the understanding of adversity-induced DNAm as a biomarker and predictor of psychological ill-health development. Accounting for limitations such as small sample sizes, the preliminary results warrant further investigation. Future work should prioritize replication in larger cohorts, longitudinal investigation of perceived adversity rather than memory recounts, and functional validation of candidate loci rather than the focus solely on numerical quantifiers of DNAm. The identification of both risk-transmitting and protective epigenetic pathways offers a more nuanced understanding of biological embedding and suggests novel targets for epidemiological and diagnostic approaches.