The study, led by Jie Ding, Ben Schöttker, Hermann Brenner and Michael Hoffmeister, exploited the prospective design of the UK Biobank, a cohort of more than half a million middle-aged and older adults recruited across England, Wales and Scotland between 2006 and 2010. After excluding people with prevalent cancer at recruitment, incomplete lifestyle data or less than two years of follow-up, the researchers followed 287,745 participants through national cancer and death registries for a median of 11.5 years. During that window, 12,947 participants received a first cancer diagnosis. Because the cohort was healthy at baseline and the cancers emerged afterwards, the analysis captures genuine temporal sequence rather than retrospective recall.

The technical heart of the study is the healthy lifestyle score, or HLS, a previously benchmarked instrument that assigns one point to each of five healthy behaviours: non-smoking, moderate alcohol intake, a healthy diet, adequate physical activity and a healthy body mass index. Non-smoking was defined as never having smoked regularly or having been a former smoker with fewer than 30 pack-years. Moderate drinking meant no more than 12 grams of alcohol per day for women and 24 grams for men. Diet quality was assessed by placing participants in the highest 40 percent of a cohort-specific healthy diet score distribution. Adequate physical activity required at least 500 MET-hours per week of walking, moderate or vigorous activity, and a healthy BMI meant more than 18.5 but less than 25 kilograms per square metre. Scores were grouped into low (0–1), medium (2–3) and high (4–5) categories.

Cox proportional hazards models, adjusted for age, sex, ethnicity, region, deprivation, household income, education, cancer screening history, family history of cancer, hypertension, diabetes and height, revealed a 29 percent lower risk of overall cancer for the high versus the low HLS group. But the site-specific results are where the story becomes dramatic. After correction for multiple comparisons using the Benjamini-Hochberg procedure, strong adherence was associated with reduced risks of oral cancer (hazard ratio 0.57), pharyngeal cancer (0.41), laryngeal cancer (0.12), oesophageal cancer (0.41), stomach cancer (0.49), colorectal cancer (0.65), liver cancer (0.44), pancreatic cancer (0.52), lung cancer (0.19), soft tissue sarcoma (0.61), kidney cancer (0.73), bladder cancer (0.52) and postmenopausal breast cancer (0.79). The laryngeal figure is extraordinary: an 88 percent risk reduction, meaning that people with four or five healthy behaviours almost never developed this tumour in the cohort.

The researchers went one step further and calculated population attributable fractions, or PAFs, a statistical measure of the proportion of cancer cases that would be attributable to incomplete adherence to a healthy lifestyle. Using the graphPAF R package with 1000 bootstrap replications, they found significant PAFs for 14 of the 27 cancer types. The highest fractions clustered in the upper aerodigestive and respiratory tracts: 65.5 percent for laryngeal cancer, 51.9 percent for lung cancer and 30.2 percent for pharyngeal cancer. Digestive cancers also carried heavy burdens, with 29.9 percent for oesophageal, 25.9 percent for stomach, 24.0 percent for pancreatic, 22.3 percent for liver and 16.6 percent for colorectal cancer. In plain terms, if everyone in the cohort had achieved four or five healthy behaviours, roughly two-thirds of laryngeal cancers and half of lung cancers might never have occurred.

When the team dissected the score into its individual components, smoking and overweight emerged as the dominant contributors. Non-smoking was significantly associated with 12 of the 27 cancers, with hazard ratios ranging from 0.13 to 0.82, including an 87 percent lower risk of lung cancer, an 82 percent lower risk of laryngeal cancer and a 67 percent lower risk of pharyngeal cancer. Normal BMI was linked to reduced risks of seven cancers: oesophageal, colorectal, pancreatic, kidney, central nervous system, postmenopausal breast and uterine. The other components, including diet, physical activity and moderate alcohol intake, showed associations with fewer cancer types. Smoking showed significant PAFs for 12 cancer types and overweight for 10, far more than any other factor, leading the authors to conclude that smoking cessation and weight control should be prioritised in cancer prevention strategies.

The analysis also uncovered some counterintuitive patterns that demand careful interpretation. A healthy lifestyle was associated with an increased risk of prostate cancer (hazard ratio 1.11), and non-smoking was linked to higher risks of melanoma, uterine and prostate cancer. Normal BMI showed positive associations with pharyngeal and lung cancer, a paradox previously reported in the literature and often attributed to confounding by smoking, although the evidence remains inconclusive. Moderate alcohol intake was associated with reduced risks of kidney cancer and multiple myeloma, consistent with some prior studies, and one proposed mechanism is that alcohol may increase insulin sensitivity. The authors caution that these paradoxical results require further study before any biological conclusions can be drawn.

Subgroup analyses revealed that the protective associations varied by age and sex. The link between a healthy lifestyle and reduced pancreatic cancer risk was stronger in participants under 60, while the associations with kidney and postmenopausal breast cancer were stronger in those aged 60 and over, suggesting that healthy behaviours may accumulate over time and have a more pronounced effect at older ages. Oesophageal and colorectal cancer showed stronger protective associations in men than in women, in line with prior European cohort evidence. For laryngeal cancer, no cases at all were observed among women with a high HLS, an outcome so rare that the hazard ratio for that group could not even be defined.

The study has notable strengths, including its simultaneous investigation of a broad spectrum of cancers within the same population, which enhances the comparability of cancer-specific results, its large sample size, its prospective design and its detailed lifestyle assessment. But the authors are candid about limitations. The UK Biobank population is predominantly White and healthier than the general population, which may limit generalisability. The dietary component was based on a cohort-specific diet quality score rather than a validated dietary index. Most lifestyle factors were self-reported, leaving room for measurement error and residual confounding. Lifestyle changes during follow-up may have caused regression dilution bias, which would actually mean the true effect sizes are stronger than estimated. And for rare cancers, statistical power may have been insufficient to detect relevant associations.

What makes this analysis resonate is its scope and its clarity. Global cancer incidence is projected to reach 28.4 million cases by 2040, and while age-standardised rates in the UK have been relatively stable since 1993, less common cancers such as liver, oral and melanoma skin cancer have shown notable increases. By demonstrating that a simple five-point checklist is associated with lower risks of both common and rare cancers, and by quantifying how much of the cancer burden is attributable to incomplete adherence, the study provides a technical foundation for targeted prevention. The message for public health is unambiguous: if the goal is to prevent the greatest number and the widest variety of cancers, the first priorities are keeping cigarettes out of reach and keeping weight in check, with the remaining healthy behaviours reinforcing the effect across a remarkable range of tumour types.

Subject of Research: Cancer epidemiology

Article Title: Healthy lifestyle and cancer risk: a pan-cancer analysis in the UK Biobank prospective cohort

Article References: Ding, J., Schöttker, B., Brenner, H., & Hoffmeister, M. (2026). Healthy lifestyle and cancer risk: a pan-cancer analysis in the UK Biobank prospective cohort. British Journal of Cancer. https://doi.org/10.1038/s41416-026-03613-9

Image Credits: AI Generated

DOI: 10.1038/s41416-026-03613-9

Keywords: cancer prevention, healthy lifestyle, UK Biobank, smoking, obesity, population attributable fraction, pan-cancer analysis, cancer epidemiology, Healthy, lifestyle, cancer, risk