Data from Smoking and Selected DNA Repair Gene Polymorphisms in Controls: Systematic Review and Meta-Analysis

Background: When the case-only study design is used to estimate statistical interaction between genetic (G) and environmental (E) exposures, G and E must be independent in the underlying population, or the case-only estimate of interaction (COR) will be biased. Few studies have examined the occurrence of G–E association in published control group data.

Methods: To examine the assumption of G–E independence in empirical data, we conducted a systematic review and meta-analysis of G–E associations in controls for frequently investigated DNA repair genes (XRCC1 Arg399Gln, Arg194Trp, or Arg280His, XPD Lys751Gln, and Asp312Asn, and XRCC3 Thr241Met), and smoking (ever/never smoking, current/not current smoker, smoking duration, smoking intensity, and pack-years).

Results: Across the 55 included studies, single nucleotide polymorphisms SNP-smoking associations in controls (OR_z) were not reliably at the null value of 1.0 for any SNP-smoking combinations. Two G–E combinations were too heterogeneous for summary estimates: XRCC1 399 and ever-never smoking (N = 21), and XPD 751 and pack-years (N = 12). OR_z ranges for these combinations were: [OR_z (95% confidence interval (CI)] 0.7 (0.4, 1.2)–1.9 (1.2, 2.8) and 0.8 (0.5, 1.3)–2.3 (0.8, 6.1), respectively). Estimates for studies considered homogeneous (Cochran's Q P-value <0.10) varied 2- to 5-fold. No study characteristics were identified that could explain heterogeneity.

Conclusions: We recommend the independence assumption be evaluated in the population underlying any potential case-only study, rather than in a proxy control group(s) or pooled controls.

Impact: These results suggest that G–E association in controls may be population-specific. Increased access to control data would improve evaluation of the independence assumption. Cancer Epidemiol Biomarkers Prev; 19(12); 3055–86. ©2010 AACR.