Table 4.

Genome-wide meta-analysis of overall test and/or interaction test with the stratified analysis for the association with hyperinsulinemia and multiple Cox regression for the genotypes of NR5A2 rs10919774 for predicting breast cancer risk, stratified by BMI

BMI < 30.0 kg/m2BMI ≥ 30.0 kg/m2
Interaction test for BMI(n = 7,862)(n = 3,247)
SNPGenetic modelAllelea (Ref/Alt)PQORbPbQbHRc (95% CI)PORbPbQbHRc (95% CI)P
NR5A2 rs10919774GenotypicGG1.45E−060.7070.930.4210.319Referent1.982.53E−080.726Referent
GA0.31 (0.09–0.99)0.0490.48 (0.06–3.65)0.478
AA0.29 (0.09–0.90)0.0330.69 (0.10–4.99)0.716
DominantGG/GA + AA0.29 (0.09–0.91)0.0330.67 (0.09–4.84)0.695
  • NOTE: Only SNPs that are significantly genome-wide associated with hyperinsulinemia in overall/interaction (G*E or subgroup) analysis and breast cancer were included. Numbers in bold face are statistically significant.

  • Abbreviations: Alt, alternative allele; Q, Cochran's Q; Ref, reference allele.

  • aAdditive genetic model regressed in genome-wide meta-analysis.

  • bResults from genome-wide meta-analysis of interaction test for the association with hyperinsulinemia.

  • cHR adjusted by age, education, annual family income, family history of breast cancer, depressive symptom, smoking, physical activity, dietary alcohol in g/day, % calories from SFAs/day, waist-to-hip ratio, hysterectomy ever, ages at menarche and menopause, oral contraceptive use, exogenous estrogen–only use, and exogenous estrogen plus progestin use.