No nominally significant gene-level or SNP-level main effects on cannabis dependence emerged. Of the 6 eCB genes tested, only MGLL demonstrated a significant interaction with CSA predicting symptoms of cannabis dependence (Table 4; Figure S1; Gene-Stat = 0.002 using top 14 SNPs; 6 of 24 SNPs nominally significant; p = .0085. Inspection of individual SNPs across the entire eCB gene set revealed one SNP, rs604300 within MGLL, that survived SNP-level Bonferroni correction (Table 5; Table S5; Figure 1; bGxE = -4.37, 95% CI [-6.69, -2.05], ΔR2 = .007, ΔF(1,1527) = 13.69, p = .0002). Within the full rs604300 model, and consistent with prior literature (Duncan et al., 2008), there was an overall main effect of CSA on cannabis dependence symptoms, such that increasing exposure to CSA was associated with a greater number of dependence symptoms (bE = 2.73, 95% CI [1.72, 3.73], p < .001). However, this effect was only present in G allele homozygotes (bE = 3.68, 95% CI [2.56, 4.81], p < .001); there was no association between CSA and cannabis dependence symptoms in heterozygotes (bE = -0.69,
