Background: Alcohol is a modifiable dietary factor and a known carcinogen, yet there is limited prospective evidence regarding the relationship between alcohol consumption and breast cancer risk in low- and middle-income countries (LMICs). This is especially true in areas where women's alcohol intake tends to be low, which creates uncertainty about how to translate etiological findings into actionable prevention strategies effectively. The newly launched Latin American and Caribbean Code Against Cancer (IARC/PAHO) explicitly recommends avoiding alcoholic drinks for cancer prevention, highlighting the importance of assessing risk in the low-intake range that is common among women in Mexico.
?Objective: To assess the impact of alcohol consumption on the incidence of breast cancer among Mexican women
Methods: We conducted a prospective cohort analysis using data from the Mexican Teachers’ Cohort (MTC). The baseline was defined as the date of the first valid food frequency questionnaire (FFQ), administered in either 2006 or 2008. Of the initial 115,273 women, we excluded 20,762 who provided no alcohol information in a valid FFQ, 1,322 who were pregnant when reporting alcohol use, and 623 prevalent breast cancer cases. Adjusted Cox models included 95,535 women. We modeled alcohol intake using a natural cubic spline with knots at the 25th, 50th, and 75th percentiles to characterize the dose–response relationship flexibly. We also reported spline-predicted hazard ratios (HRs) at policy-relevant intake levels (0, 5, 10, 15, and 30 g/day), with 0 g/day as the reference. Women were followed from the FFQ date until they were diagnosed with incident breast cancer, died, or until December 31, 2023. Cox proportional hazards models estimated hazard ratios (HR) and 95% confidence intervals (CI), adjusting for age, speaking an indigenous language, socioeconomic status, menopausal status, physical activity, oral contraceptive use, parity, age at first birth, breastfeeding, smoking, family history of breast cancer, age at menarche, and dietary covariates (total cholesterol, protein, saturated fat, polyunsaturated fat, and fiber category). Sensitivity analyses included adjustments for BMI.
?Results: Among the 95,535 women, there were 1,891 incident breast cancer cases during the follow-up period (mean follow-up of 15.0 years). Overall, alcohol intake was low; the mean consumption among drinkers was 1.75 g/d. In multivariable spline models anchored at 0 g/day, the predicted breast cancer hazard increased even at policy-relevant low doses: HR=1.13 (95% CI=1.004–1.27) at 5 g/day, HR=1.17 (95% CI=0.97–1.42) at 10 g/day, HR=1.22 (95% CI=0.92–1.63) at 15 g/day, and HR=1.37 (95% CI=0.75–2.45) at 30 g/day. The uncertainty of the estimates increased at higher intake levels, consistent with sparse data in that range. Categorical models were directionally consistent: compared with 0 g/day, the HR was 1.07 (95% CI=0.97–1.18) for >0–<5 g/day, HR=1.46 (95% CI=1.12–1.89) for 5–<10 g/day, and HR=1.002 (95% CI=0.59-1.71) for ≥10 g/day. Sensitivity analyses, including BMI as a confounder, did not change the estimators.

?Conclusions/Implications: Even at low-to-moderate intake of alcohol, the spline suggests a progressively higher breast cancer hazard relative to abstention, underscoring prevention relevance in LMIC settings where the exposure distribution is concentrated at low levels. 

Splines for acohol consumption and breast cancer hazard