When the target tooth was missing, the second molar in the same side or the first incisor in the opposite side was examined. The deepest PPD was recorded for each tooth. Periodontal disease was defined as positive if a woman had at least one tooth with a PPD of 3.5 mm or deeper. Among the 1157 women, 131 cases of periodontal disease were identified using this definition. The 1026 remaining participants were eligible to serve as control subjects, but seven women were excluded because of missing

data on the factors under study; thus, 1019 women were classified as control subjects. In the baseline survey, each participant filled out a questionnaire and mailed it to the data management AZD3965 centre. Research technicians completed missing or illogical data by telephone interview. The questionnaire in the baseline survey included questions about smoking habits, household income, education, toothbrushing frequency and use of an interdental brush.

A history of smoking was defined as having smoked at least once per day for at least 1 year. Research technicians or subjects themselves collected buccal specimens with BuccalAmp swabs (Epicenter BioTechnologies, Madison, WI, USA). Genomic Inhibitor Library purchase DNA was extracted using a QIAmp DNA mini kit (Qiagen, Inc., Valencia, CA, USA). Genotyping of VDR SNPs was performed using TaqMan SNP Genotyping Assays on a StepOnePlus machine (Applied Biosystems, Foster City, CA, USA), according to the manufacturer’s instructions. Departures from Hardy–Weinberg equilibrium were tested among the control subjects using the Chi-square test. Linkage disequilibrium was examined using Haploview software version 4.2 (Broad Institute, Cambridge, MA, USA) [23]. Estimations of crude odds ratios (ORs) and 95% confidence intervals (CIs) for periodontal disease associated with the SNPs under

study Silibinin were made by means of logistic regression analysis, with the reference category being the homozygote of the major allele. Multiple logistic regression analysis was used to control for age at oral examination, region of residence, education, smoking, toothbrushing frequency and use of interdental brush. The statistical power calculation was performed using QUANTO version 1.2 [24]. Haplotypes and their frequencies were inferred according to the expectation maximization algorithm. For differences in haplotype frequency between the cases and control groups, crude ORs and 95% CIs were estimated based on the frequency of each haplotype relative to all other haplotypes combined. We examined multiplicative and additive interactions between the SNPs under study and smoking with regard to the risk of periodontal disease. The multiplicative interaction was estimated by introducing a multiplicative term into a multiple logistic regression model.