Breast Cancer Risk and Insulin Resistance: Post Genome-Wide Gene-Environment Interaction Study Using a Random Survival Forest

Abstract

Obesity-insulin connections have been considered potential risk factors for postmenopausal breast cancer, and the association between insulin resistance (IR) genotypes and phenotypes can be modified by obesity-lifestyle factors, affecting breast cancer risk. In this study, we explored the role of IR in those pathways at the genome-wide level. We identified IR-genetic factors and selected lifestyles to generate risk profiles for postmenopausal breast cancer. Using large-scale cohort data from postmenopausal women in the Women's Health Initiative Database for Genotypes and Phenotypes Study, our previous genome-wide association gene-behavior interaction study identified 58 loci for associations with IR phenotypes (homeostatic model assessment-IR, hyperglycemia, and hyperinsulinemia). We evaluated those single-nucleotide polymorphisms (SNP) and additional 31 lifestyles in relation to breast cancer risk by conducting a two-stage multimodal random survival forest analysis. We identified the most predictive genetic and lifestyle variables in overall and subgroup analyses [stratified by body mass index (BMI), exercise, and dietary fat intake]. Two SNPs (LINC00460 rs17254590 and MKLN1 rs117911989), exogenous factors related to lifetime cumulative exposure to estrogen, BMI, and dietary alcohol consumption were the most common influential factors across the analyses. Individual SNPs did not have significant associations with breast cancer, but SNPs and lifestyles combined synergistically increased the risk of breast cancer in a gene-behavior, dose-dependent manner. These findings may contribute to more accurate predictions of breast cancer and suggest potential intervention strategies for women with specific genetic and lifestyle factors to reduce their breast cancer risk. SIGNIFICANCE: These findings identify insulin resistance SNPs in combination with lifestyle as synergistic factors for breast cancer risk, suggesting lifestyle changes can prevent breast cancer in women who carry the risk genotypes.

Figure 1.

Overall analysis: the second stage of random survival forest (RSF) with 10 single-nucleotide polymorphisms and 12 behavioral factors selected from the first stage of RSF analysis A. Comparing minimal depth and VIMP rankings. (BMI, body mass index; E+P, exogenous estrogen + progestin; VIMP, variable of importance) B. Out-of-bag concordance index (c-index). (Improvement in out-of-bag c-index was observed when the top 6 variables [●] were added to the model, whereas other variables [○] did not further improve the accuracy of prediction.)

Figure 2.

Cumulative breast cancer incidence rate for the 8 most influential variables (3 SNPs and 5 behavioral factors) based on a random survival forest analysis. (BMI, body mass index; E+P, exogenous estrogen + progestin; SNPs, single-nucleotide polymorphisms. Dashed red lines indicate 95% confidence intervals.)

Figure 3.

Contour plot of cumulative breast cancer incidence rate for the combination of SNP (MKLN1 rs117911989) and oral contraceptive use or E+P use, stratified by physical activity (E+P, exogenous estrogen + progestin; MET, metabolic equivalent; SNPs, single-nucleotide polymorphisms. Cumulative incidence rate estimated from the random survival forest model was adjusted by age, body mass index, waist to hip ratio, depressive symptom, age at menarche, age at menopause, dietary alcohol, daily vegetables, % calories from protein, and % calories from carbohydrates.) A1. Active Group (MET ≥ 10) A2. Active Group (MET ≥ 10) B1. Inactive Group (MET < 10) B2. Inactive Group (MET < 10)