We did observe several principal components significantly associated with abstinence. Gizer sellckchem et al. noted significant heterogeneity in association of the VNTR with ADHD diagnosis, including differences in the association direction and strength due to genetic ancestry differences (Gizer et al., 2009). Inclusion of covariates of genetic ancestry will correct for continental and some subcontinental population genetic ancestry differences, but will not address genetic differences due to rare variation that have arisen in the recent expansion of the human population (Keinan & Clark, 2012). Rare-linked variants have been associated with ADHD diagnosis (Grady et al., 2003, 2005; Tovo-Rodrigues et al., 2012). CYP2B6 exhibits extensive coding and noncoding variation within and between continental population groups that influence CYP2B6 expression and function (Lang et al.

, 2001). Sampling effects within analysis categories may result in differences in relevant pharmacogenetic variation (Lee et al., 2007). Differences in treatment between the trials encompass the different behavioral treatments, with the Brown et al. (2007) trial focused on evaluating differences between effects of cognitive�Cbehavioral treatment for depression added to standard cognitive�Cbehavioral smoking cessation treatment (CBTD) versus standard cognitive�Cbehavioral smoking cessation treatment (CBT), while the behavioral treatment in the Lerman et al. (2003) trial consisted of cognitive�Cbehavioral therapy in a group counseling setting. While no differences were observed between CBTD and CBT on point prevalence abstinence in the Brown et al.

(2007) trial (Brown et al., 2007), both behavioral treatments in the Brown et al. (2007) trial were more intensive (in the number and length of counseling sessions) than the treatments in the Lerman et al. (2003) trial. The Lerman et al. (2003) sample, as reported in this analysis, reports lower abstinence (point prevalence and continuous abstinence) prevalence than the Brown et al. (2007) sample reported in Leventhal et al. (2012) for approximately 75% of the 24 comparisons (four genotype by pharmacotherapy treatment categories, three timepoints, two abstinence outcomes). Differences in behavioral treatment intensity between the trials might account for some of these differences.

Differences in the efficacy of behavioral treatment might also influence relative efficacy by pharmacotherapy and genotype, and account Entinostat for some of the treatment by genotype interaction association differences observed between the trials. Differences in the values of some behavioral or genetic characteristics, genetically based covariates, or relative treatment efficacy by genotype, might explain a portion of the differences in association results observed between the results of Leventhal et al. (2012) and our analyses, although the role of chance, in either previously reported findings or in differences between analysis results, cannot be ignored.