Debras · PLoS medicine 2022 · Prospective population-based cohort study · n=102865

Artificial sweeteners and cancer risk: Results from the NutriNet-Santé population-based cohort study.

Cited 326 times in the scientific literature.

Level 3 - non-randomized controlled study

Prospective observational cohort study

PubMed 35324894 · doi:10.1371/journal.pmed.1003950 · record verified 2026-08-29

What was done

Researchers tracked 102,865 adults in the French prospective NutriNet-Santé cohort (2009–2021) over a median follow-up of 7.8 years. Dietary exposure to total artificial sweeteners and specific compounds (aspartame, acesulfame-K, sucralose) was measured using repeated 24-hour dietary records including commercial brand names. Cox proportional hazards models adjusted for age, sex, education, physical activity, smoking, BMI, weight change, family history of cancer, diabetes, and multiple baseline dietary quality variables were used to estimate hazard ratios for cancer incidence.

What was found

Across 3,358 total incident cancer cases, higher consumers of total artificial sweeteners had an increased risk of overall cancer compared to non-consumers (HR = 1.13, 95% CI 1.03 to 1.25, P-trend = 0.002). Increased overall cancer risk was specifically observed for aspartame (HR = 1.15, 95% CI 1.03 to 1.28, P = 0.002) and acesulfame-K (HR = 1.13, 95% CI 1.01 to 1.26, P = 0.007). Higher aspartame intake was associated with breast cancer (979 cases, HR = 1.22, 95% CI 1.01 to 1.48, P = 0.036). Obesity-related cancers (2,023 cases) were increased with total sweetener intake (HR = 1.13, 95% CI 1.00 to 1.28, P = 0.036) and aspartame (HR = 1.15, 95% CI 1.01 to 1.32, P = 0.026).

Why it matters

This study provides large-scale, brand-specific epidemiological data suggesting that common artificial sweeteners may not be inert sugar substitutes regarding long-term cancer risk, directly informing safety evaluations by regulatory agencies.

Limits

The observational design cannot establish causality. Dietary intake was self-reported and subject to measurement error. The volunteer cohort may carry selection bias, and residual confounding or reverse causality remains possible despite multivariable adjustment.

Cited by