Obesity-dependent selection of driver mutations in cancer

Abstract

Obesity is a risk factor for cancer, but whether obesity is linked to specific genomic subtypes of cancer is unknown. We examined the relationship between obesity and tumor genotype in two clinicogenomic corpora. Obesity was associated with specific driver mutations in lung adenocarcinoma, endometrial carcinoma and cancers of unknown primaries, independent of clinical covariates, demographic factors and genetic ancestry. Obesity is therefore a driver of etiological heterogeneity in some cancers.

Fig. 1. Oncogenic mutations are associated with BMI.

a, Statistical association between continuous BMI and genotype across gene–cancer type pairs. The −log₁₀(P values) and estimated sizes from univariate logistic regression are on the y and x axes, respectively. Statistically significant pairs are in black. b, Multivariate regression demonstrating that BMI categories (underweight, BMI < 18.5 kg m⁻²; healthy, 18.5 ≤ BMI < 25 kg m⁻²; overweight, 25 ≤ BMI < 30 kg m⁻²; obese, BMI ≥ 30 kg m⁻²) are associated with KRAS mutations independent of other clinical factors. Results for multivariate regression with BMI as a continuous variable are shown in Supplementary Table 3. Error bars represent the 95% confidence interval (CI). c, KRAS mutation frequency in patients with lung adenocarcinoma categorized by BMI and genetic ancestry. ASJ, Ashkenazi Jewish; EAS, East Asian; EUR, European. Error bars represent the s.e. d, EGFR (top) and KRAS (bottom) mutation frequency in BMI categories in the MSKCC cohort. Error bars represent the s.e. e, EGFR (top) and KRAS (bottom) mutation frequency in BMI categories in the DFCI cohort. Error bars represent the s.e. f, EGFR (top) and KRAS (bottom) are not associated with weight loss before cancer diagnosis in lung adenocarcinoma using the χ² test.

Extended Data Fig. 1. BMI Distribution Across Cancer Types.

BMI values across cancer types with over 200 samples. Each dot corresponds to a patient, with the y-axis representing the patient’s BMI. Cancer types are ordered by median BMI, with the lowest median BMI on the left and highest BMI on the right.

Extended Data Fig. 2. Silent mutations are not associated with body mass index.

Scatterplot comparing -log₁₀ expected p-values vs. -log₁₀ observed p-values in univariate logistic regression models for silent mutations.