We first investigated the relationship between opiate abuse and striatal CREM by studying two independent postmortem opiate abuse populations. In the first population, composed of an ethnically heterogeneous US population23, 24, rs12765063 was associated with elevated risk of opiate abuse (Χ2=24.30, p<0.0001, Fig. 3a inset). To eliminate the possibility that this pronounced effect was due to ancestral differences between the two groups, we separately analyzed on the basis of race and still detected significant associations (Mantel–Haenszel Test, Χ2=5.03, p=0.025, Fig. 3a). In an independent, ethnically homogenous European population25, rs12765063 associated with measures of abuse severity, namely years of use preceding lethal intoxication (t43=2.16, p=0.036, Fig. 3b) and history of previous overdose (Χ2=4.32, p=0.038, Fig. 3c), but not heroin abuse (Χ2=0.31, p=0.58). These data suggest that CREM associates with hazardous patterns of opioid use and a more rapid progression to lethal intoxication. In a large COGA dataset26 for which use of other substances was measured, rs12765063 significantly associated with DSM-IV alcohol dependence symptom count (β=0.063, p=0.002), and total symptom count for alcohol, marijuana, cocaine and opiate dependence (β=0.062, p=0.04, Fig. 3d).
