Modifiable factors affects cancer-specific survival: findings from a large population-based prospective cohort study

Modifiable factors affect cancer's survival but literature did not differentiate prior to versus after cancer diagnosis. It is essential to provide references for the intervention prioritized at different stages.

In this prospective cohort study, we analyzed national data from the UK Biobank, including 121,399 participants, to assess the association of modifiable factors with cancer-specific survival (CSS) in two independent cohorts: a pre-cancer cohort (n = 78,027) and a post-cancer cohort (n = 43,372). Additionally, a weighted standardized score was derived to evaluate the joint effects across different domains. Interactions between the six domains and age at diagnosis, sex, and cancer site were evaluated using likelihood ratio tests. Subgroup analyses were then performed for factors showing significant effect modification. Population-attributable fractions (PAF) of different domains on 5-year cancer-specific death were calculated.

Our study comprehensively presented the differential patterns of modifiable factors' impact on CSS among pre-cancer and post-cancer cohorts, sexes and different cancer sites. In the pre-cancer cohort, CSS were predominantly attributable to smoking/alcohol consumption (PAF 9·2%) and daily activity (PAF 10·6%). Men exhibited a higher risk than women for dietary habits (HR:1·25 versus 1·18), daily activity (HR:1·50 versus 1·29) and living environment (HR:1·13 versus 1·03). The impact of modifiable factors, including daily activity, smoking/alcohol consumption, and physical measures, on CSS varied across different cancer sites. In the post-cancer cohort, 18·6% of 5-year cancer-specific deaths were attributable to unfavourable mental health. In subgroup analysis, the risk of CSS in the domain of smoking/alcohol consumption was higher in men than that in women (HR: 1·58 versus 1·34). The impact of modifiable factors, including smoking/alcohol consumption, mental health and physical measures, on CSS varied across different cancer sites.

Our findings suggested that targeted prevention and early intervention strategies should be implemented to reduce the risk of cancer-related deaths.