Does Physical Activity Lower Breast Cancer Risk? What Causality Says

Regular movement has a credible link to lower breast cancer risk. The best prospective device-based study found a 26% lower incidence rate in the most active women than in the least active women. Genetic analyses now support the possibility that the relationship is partly causal. They do not yet justify a precise cancer-prevention prescription.

The distinction matters. People who move more often differ from less-active people in many ways: body size, diet, smoking, income, screening behavior, and underlying health all muddy observational research. A breast cancer that has begun developing may also reduce activity before diagnosis. The question is not whether activity correlates with lower risk. It does. The harder question is whether lifelong inactivity itself contributes to risk.

The observational evidence is consistent

In a prospective UK Biobank study of 48,286 women, 836 developed breast cancer over a median 5.4 years. Women in the highest quartile of accelerometer-measured total physical activity had a hazard ratio of 0.74 compared with those in the lowest quartile (95% confidence interval [CI], 0.61–0.91). That is an association with 26% lower incident breast cancer, and the CI excluded no difference.

The result did not depend solely on formal hard exercise. In the 2023 UK Biobank analysis by Parra-Soto et al., light-intensity activity was independently associated with lower breast cancer incidence (HR 0.79, 95% CI 0.64–0.96), as was combined moderate-to-vigorous physical activity, or MVPA (HR 0.78, 95% CI 0.64–0.96).

Those estimates do not prove that a prescribed amount of walking prevents a prescribed number of cancers. They compare groups whose activity was recorded during one week, not randomly assigned exercise programs. Still, device measurement avoids one familiar problem: people are famously optimistic about how much they exercise. Their calendars may agree; accelerometers tend to be less diplomatic.

Steps provide a useful, but indirect, dose-response clue

For a practical frame of reference, a 2023 Lancet UK Biobank study followed 86,556 participants and recorded 5,577 incident cancers across 13 cancer types previously linked to physical activity. Compared with 5,000 steps per day, 10,000 steps per day was associated with a hazard ratio of 0.81 (95% CI 0.73–0.90) for that composite outcome.

The study also modeled what happened when sedentary time was replaced. Reallocating 30 minutes per day to MVPA was associated with an HR of 0.96 (95% CI 0.94–0.98), while reallocating one hour per day to light-intensity activity was associated with an HR of 0.94 (95% CI 0.92–0.96). The composite curve appeared to level off around 9,000 to 10,000 daily steps.

That plateau is often repeated as if it were a breast-cancer threshold. It is not. The endpoint combined 13 cancer types, and the analysis did not provide breast-cancer-specific step-count splines. A later 2025 UK Biobank analysis extended the work with individual cancer-site models, step counts, intensity, and substitution analyses, but no supplied estimate establishes a breast-cancer-specific daily-step target.

Genetics can test causality, within limits

Dixon-Suen et al.'s 2022 Mendelian randomization study took a different approach. Mendelian randomization uses genetic variants associated with an exposure as proxies for a lifelong tendency toward that exposure. Because inherited variants are assigned before adult behavior and before cancer, this design can reduce reverse causation and some conventional confounding.

The researchers used five variants associated with accelerometer-measured overall activity in 91,105 UK Biobank participants. Those variants explained only 0.10% of the variation in activity. They then tested the variants against invasive breast cancer in the Breast Cancer Association Consortium: 130,957 women of European ancestry, including 69,838 invasive breast cancer cases and 54,452 controls.

Each standard-deviation increase in genetically predicted activity, roughly 8 milligravities of acceleration and approximately 50 additional minutes of brisk walking per week, was associated with an odds ratio of 0.48 for invasive breast cancer (95% CI 0.30–0.78). The direction is striking and statistically significant, but the estimate is imprecise: the interval allows anything from a 22% to a 70% lower odds. It also does not tell a person that adding 50 minutes of brisk walking next week will cut her risk in half.

The authors used weighted-median MR, MR-Egger, MR-PRESSO, leave-one-out analyses, and other checks. These did not detect directional pleiotropy, meaning no obvious evidence that the variants affected breast cancer through unrelated pathways. But five variants give those tests little power. An absence of a pleiotropy signal is reassuring; it is not a clean bill of health.

Current MR evidence provides preliminary causal support for physical activity as a breast-cancer risk factor. It does not yet deliver a dose-calibrated prevention protocol.

The central complication is body size

The activity-associated variants also co-associated with BMI and adiposity in genetic-database searches reported by Dixon-Suen et al. That creates a serious interpretive problem. Higher BMI is an established breast cancer risk factor, particularly for postmenopausal estrogen-receptor-positive disease. The genetic result may reflect a direct effect of activity, an effect mediated through adiposity, or some combination. Multivariable MR that conditions on BMI would better separate those paths; this analysis has not been reported here.

Other MR research points in the same general direction. Chen et al.'s 2022 analysis examined 23 established and suspected risk factors across breast cancer molecular subtypes, and Verdiesen et al.'s 2025 study evaluated causal risk factors across hormone-receptor subtypes. These studies provide corroborating context, not a substitute for stronger activity instruments or diverse populations.

Intensity is less settled than total movement

The same UK Biobank breast-cancer cohort complicates the usual "harder is better" story. Vigorous activity alone had an HR of 0.77 (95% CI 0.56–1.05), and moderate activity alone had an HR of 0.73 (95% CI 0.44–1.19). Both point toward lower risk, but neither reached conventional statistical significance because both CIs included 1.0. Light activity and combined MVPA did reach significance in that analysis.

This does not show that vigorous exercise lacks benefit. It shows that this cohort could not establish an independent vigorous-only association with precision. The available evidence supports building daily movement volume, including light activity, rather than treating occasional punishing workouts as the only route to lower risk.

Mechanisms have human support, but they are not proof of the full chain

Physical activity can plausibly affect breast cancer biology through insulin and insulin-like growth factor 1 (IGF-1). A 2022 systematic review by Swain et al. found randomized-trial evidence that physical activity reduces circulating insulin and IGF-1. A companion review by Drummond et al., drawing on prospective cohorts and MR studies, found that insulin/IGF perturbation is causally associated with breast cancer risk. Together, these findings make insulin/IGF-1 the most complete human evidence chain available here.

Inflammation is another plausible route. Lou et al.'s 2023 systematic review found human prospective and MR evidence linking inflammatory markers to breast cancer risk. These biomarkers support biological plausibility. They do not prove that every activity-related change in insulin, IGF-1, or inflammation causes fewer breast cancers.

Who can apply these findings?

Care is especially warranted for premenopausal women. The Nurses' Health Studies reported inverse associations predominantly for postmenopausal breast cancer, while the World Cancer Research Fund's 2018 assessment classified evidence for premenopausal breast cancer as limited-suggestive. Fortner et al.'s 2024 analysis adds menopausal and hormone-receptor detail, but it does not resolve the subgroup evidence gap. Parra-Soto et al. found no statistically significant interaction by menopausal status; a null interaction does not establish identical protection in both groups.

Generalizability is another constraint. The central MR studies used European-ancestry populations, and the UK Biobank activity-instrument sample was 96.6% white British. The causal estimate cannot simply be carried over to populations with different activity patterns, BMI-related genetics, social conditions, or breast cancer subtype distributions. The accelerometer cohorts also measured activity during a single seven-day period, not across adolescence, reproductive life, and menopause.

The practical verdict

Physical activity belongs in a breast-cancer risk-reduction strategy because the observational association is consistent, objective measures support it, and Mendelian randomization offers preliminary causal backing. The most defensible action is sustained movement across the week: make room for walking and other light activity, and include MVPA when it suits your health and capacity. Around 9,000 to 10,000 steps per day is a reasonable population-level orientation for the broader activity-related cancer composite, not a breast-cancer guarantee.

The evidence does not support treating a step target as oncology software. It supports moving more, consistently, while recognizing that age, alcohol, body composition, reproductive history, genetics, screening, and access to care also shape breast cancer risk.