Objective: To identify the interactions of susceptive genes with related to the genetic polymorphism of metabolism enzymes (CYP1A1, GSTT1 and GSTM1) and their impacts on the risk of breast cancer; and to test the feasibility of using Multifactor Dimensionality Reduction (MDR) model in analyzing gene-gene interactions.
Methods: A paired case-control study, matched by age and menstruate state, was conducted. From December 2003 to September 2004, 78 pairs of people with and without breast cancers were investigated. The variant genotypes of CYP1A1 Msp I and GSTT1/M1 were identified by PCR-RFLP and multiplex PCR assays. The gene-gene interactions were analyzed with the MDR model. Based on the result of the MDR model, a conditional logistic regression model was constructed as the final cause-effect interpretative model.
Results: The interaction between CYP1A1 Msp I variant genotype (vv) and GSTT1 null genotype gave the best MDR model with statistical significance (Sign Test, P = 0.05). The model Testing Balance Accuracy was 0. 5920. The Cross-Validation consistency was 10/10. The final conditional logistic regression based on the MDR model showed that passive smoking, abortion and gene-gene interaction were risks of breast cancers, with an OR (95% confidence interval) of 12.234 (1.7459-85.7279), 4.554 (1.3250-15.6507) and 9.597 (1.5783-58.3599), respectively.
Conclusion: The MDR model may be an effective method for estimating risks of breast cancers due to gene-gene and gene-environment interactions. The gene-gene interaction with related to the genetic polymorphism of metabolism enzymes (CYP1A1 and GSTT1) may increase the risk of breast cancer by disturbing the metabolism of estrogen.