A meta-analysis of sensitive periods for the effects of childhood adversity on dna methylation

Childhood adversity is a potent determinant of mental health across development. Recent evidence suggests these stressful events may be particularly detrimental during sensitive periods, or high-risk stages in the lifespan when life experiences can induce more potent and long-term impacts. Although the mechanisms underlying these effects remain unclear, DNA methylation (DNAm) may capture some time-dependent effects of early life adversity. However, most studies have only investigated the effects of time-varying adversity in a single cohort, limiting their statistical power and generalizability across socio-demographic contexts. Here, we present results from the first meta-analysis of time-varying exposures to childhood adversity and DNAm. We pooled data from up to seven population-based studies (n=2,347-3,279) with repeated measures of childhood adversity and epigenome-wide blood or saliva DNAm measured between age 7-17 (on either 450K or EPIC arrays). Within and across cohorts, we examined up to five types of childhood adversity (caregiver physical/emotional abuse; family instability; financial hardship; maternal psychopathology; and growing-up in a one adult household). To test the impact of time-varying adversity on DNAm, we extended a statistical technique that simultaneously tests different life course models – called the Structured Life Course Modeling Approach (SLCMA) – for use in high-dimensional and meta-analytic contexts. Using the SLCMA, we investigated six life course models relating exposure to each childhood adversity and DNAm: 1-4) exposure during one of four sensitive periods (ages 0-1, 2-3, 4-5, or 6-7); 5) accumulation of risk (total exposures, regardless of timing); and 6) any exposure across childhood (ever-exposed). Across all adversities, our meta-analysis results showed more associations with sensitive period models than other exposure paradigms (accumulation or ever-exposed). For instance, of the 27 DNAm loci meeting a discovery p < 1 × 10-5 threshold (6 family instability; 8 financial hardship; 8 maternal psychopathology; 3 one adult households; 2 parental cruelty), all but one CpG were linked to sensitive period exposures (56% linked to age 6-7 exposures). Three loci passed an FDR < 0.05 threshold, two of which were associated with maternal psychopathology (ages 0-1 or 6-7) and one with financial hardship (age 6-7). Our findings suggest that sensitive period effects may be present across different socio-demographic contexts. These results point to the advantages of moving beyond traditional analyses of “exposed” versus “unexposed” in epigenome-wide studies. These findings suggest potential periods when interventions aimed at mitigating the deleterious effects of childhood adversity on mental health might be more effective.