illicit drug or alcohol use). The nine SNPs that were most strongly associated with heroin dependence in their sample included three OPRD1 SNPs, all of which were genotyped in the current report (see Supplementary Table 4). Two (rs2236857 and rs3766951) were among the four SNPs in our sample for which corrected p values were significant; the third (rs2236861) reached only nominal significance. The agreement observed in the OPRD1 findings across these reports is extremely encouraging. Our results contrast with those of another group (Zhang et al. 2008) who observed a significant association with rs1042114 after correction for multiple testing and no significant association involving either of our two most significantly associated SNPs (rs2236857 or rs2236855). Differences in design may have contributed to the divergence in findings. Their sample’s 103 opioid dependent cases included individuals recruited from settings other than ORT clinics including some who were dependent on opioids other than heroin. They genotyped samples using other methods (either polymerase chain reaction restriction fragment length polymorphism or the TaqMan technique). Finally, they report p values for case-control comparisons involving rs2236855 and rs2236857 that, while non-significant, differ substantially in magnitude, surprising given that these two SNPs are in complete LD (i.e., r2=1)
