A validation of the diathesis-stress model for depression in Generation Scotland

Abstract

Depression has well-established influences from genetic and environmental risk factors. This has led to the diathesis-stress theory, which assumes a multiplicative gene-by-environment interaction (GxE) effect on risk. Recently, Colodro-Conde et al. empirically tested this theory, using the polygenic risk score for major depressive disorder (PRS, genes) and stressful life events (SLE, environment) effects on depressive symptoms, identifying significant GxE effects with an additive contribution to liability. We have tested the diathesis-stress theory on an independent sample of 4919 individuals. We identified nominally significant positive GxE effects in the full cohort (R² = 0.08%, p = 0.049) and in women (R² = 0.19%, p = 0.017), but not in men (R² = 0.15%, p = 0.07). GxE effects were nominally significant, but only in women, when SLE were split into those in which the respondent plays an active or passive role (R² = 0.15%, p = 0.038; R² = 0.16%, p = 0.033, respectively). High PRS increased the risk of depression in participants reporting high numbers of SLE (p = 2.86 × 10^-4). However, in those participants who reported no recent SLE, a higher PRS appeared to increase the risk of depressive symptoms in men (β = 0.082, p = 0.016) but had a protective effect in women (β = -0.061, p = 0.037). This difference was nominally significant (p = 0.017). Our study reinforces the evidence of additional risk in the aetiology of depression due to GxE effects. However, larger sample sizes are required to robustly validate these findings.

Fig. 1. Prediction of depression symptoms and SLE using the PRS for MDD.

a Association between polygenic risk scores (PRS) and depression score (main effects, one-sided tests). PRS were generated at 8 p-threshold levels using summary statistics from the Psychiatric Genetic Consortium MDD GWAS (released July 2016) with the exclusion of Generation Scotland participants. The depression score was derived from The General Health Questionnaire. The Y-axis represents the % of variance of depression score explained by PRS main effects. The full cohort (yellow) was split into men (blue) and women (red). In Colodro-Conde et al. PRS for MDD significantly explained up to 0.46% of depression score in their sample (~0.39% in women and ~0.70% in men). b Association between reported number of SLE and depression score (main effect, one-sided tests, results expressed in % of variance in depression score explained). SLE were self-reported through a brief life-events questionnaire based on the List of Threatening Experiences and categorized into: total number of SLE reported (TSLE), “dependent” SLE (DSLE) or “independent” SLE (ISLE). The full cohort (yellow) was split into men (blue) and women (red). In Colodro-Conde et al. “personal” SLE significantly explained up to 12.9% of depression score variance in their sample (~11.5% in women and ~16% in men)

Fig. 2. Association between GxE effect and depression score.

The plots show the percentage of depression score explained by the interaction term (two-sided tests) fitted in linear mixed models to empirically test the diathesis-stress model. Red numbers show significant interactions p-values. *Shows the significance of the difference in variance explained between sexes. The full cohort (yellow) was split into men (blue) and women (red). PRS were generated at 8 p-threshold levels using summary statistics from the Psychiatric Genetic Consortium MDD GWAS (released July 2016) with the exclusion of Generation Scotland participants. The interaction effect was tested with the number of: (a) SLE (TSLE), (b) “dependent” SLE (DSLE) and c) “independent” SLE (ISLE). In Colodro-Conde et al., the variance of depression score explained in their sample by GxE was 0.12% (p = 7 × 10⁻³). GxE were also significant in women (p = 2 × 10^-3) explaining up to 0.25% of depression score variation, but not in men (p = 0.059; R² = 0.17%; negative/protective effect on depression score)

Fig. 3. Scatterplot of diathesis-stress interactions on depression score.

Interactions with PRS at which nominally significant GxE effects were detected in (a) full cohort (p-threshold = 0.01) and (b) in women (p-threshold = 1 × 10^–5) are shown. At bottom, the remaining samples (i.e., full cohort, women or men) at same p-threshold are shown for comparison. The X-axis represents the direct effect of PRS (standard deviation from the mean) based on (a) p-threshold = 0.01 and (b) p-threshold = 1 × 10^–5_, using the total number of SLE reported by each participant (dot) as environmental exposures at three SLE levels represented by colors. Blue: 0 SLE, “no stress”, n = 1 833/1 041/792; green: 1 or 2 SLE, “low stress”, n = 2 311/1 459/852; red: 3 or more SLE, “high stress”, n = 775/490/285; in the full cohort, women and men, respectively. Y-axis reflects the depression score standardized to mean of 0 and standard deviation of 1. Lines represent the increment in risk of depression at a certain degree of “stress” dependent on a genetic predisposition ( = diathesis)