The PDM subscales include Automaticity, Craving, Loss of Control, and Tolerance. The find FAQ SDM subscales include Affiliative Attachment, Behavioral Choice, Cognitive Enhancement, Cue Exposure, Negative Reinforcement, Positive Reinforcement, Social Goads, Taste Property, and Weight Control (Baker et al., 2009; Piper et al., 2004). Genetic Data Blood samples were collected for genetic analyses, and genotype data were cleaned extensively. Primary genetic associations of this sample have been reported in prior publications (Bierut et al., 2007; S. F. Saccone et al., 2007; N. L. Saccone et al., 2009). We focused on four variants previously shown as associated with nicotine dependence in the large-scale meta-analyses: rs16969968 (CHRNA5 on chromosome 15q25), rs6474412 (upstream of CHRNB3 on chromosome 8p11), rs3733829 (EGLN2 near CYP2A6 on chromosome 19q13), and rs1329650 on chromosome 10q23.
Analyses Association analyses for dichotomous phenotypes and subphenotypes used logistic regression models with age, gender, and the single nucleotide polymorphism (SNP) as covariates. Genotypes were coded additively as the number of nonreference alleles, defined as the minor allele in the European ancestry population. CPD and TTF in the morning were dichotomized with median splits when compared with the other FTND subphenotypes. In order to capture the subphenotypes of the FTND associated with a specified variant, we tested the associations between the variant (as response variable) and all six FTND subphenotypes (as covariates with age and gender) in stepwise regression models where at each step, an independent variable not in the equation that had the smallest probability of F statistics was entered if that probability was sufficiently small.
Variables already in the regression equation were removed if their probability of F statistics became sufficiently large. The method terminated when no additional variables were eligible for inclusion or removal. Second, we examined if three FTND dimensional phenotypes (FTND score, CPD score, and TTF score) differed in their level of association with the tested variant. CPD and TTF were tested as quasi-continuous phenotypes with four levels, and Z-scores were used to standardize dimensional Anacetrapib phenotypes across measures. To test the difference in genetic associations across phenotypes, we modeled the difference in genetic associations between different phenotypes and each genetic variant using mixed models (Andrade, Eaton, & Chilcoat, 1994). The mixed model approach accounted for nonindependence of multiple phenotype measures within individuals. The interaction term between each measure and the genetic variant was a test for a significant differential genetic association.