use or other addictive behavior in their lifetime, psychiatric interviews were conducted in the Manhattan population. As such, we were able to ascertain these patterns, and only alcohol use was evident in some control subjects. Although undocumented addictive phenotype in controls is an important consideration, significant psychopathology in the control group would have limited the possibility to detect associations and gene-gene interactions with opiate abuse, but and as evidenced, this was not the case. Finally, there may be population stratification, since we examined multiple populations, and each was of relatively small sample size and of different ethnic and racial composition. Despite these limitations, our a priori study design allowed us significant power to detect the gene-gene association with heroin abuse.
