Projecting individualized probabilities of lifetime total cancer risk across a population

Introduction : Technological advances and direct-to-consumer marketing have unearthed significant organic demand from patients for cancer screening and prevention. However, in the absence of strong data or guidelines, physicians have minimal support on how to approach patients in clinical practice. Methods : We projected individualized probabilities of 10-year and lifetime cancer risk across a population as well as potential improvement with healthy behaviors in the UK Biobank (UKB). We developed cancer-specific models to simultaneously select and quantify the impact of risk factors identified in the literature. We then used iCARE package to quantify and project absolute cancer risks for individual participants. We evaluated (1) the distribution of total cancer risk by age and sex and (2) the potential impact of achieving healthy risk factor profiles at the population level. Results : The final study sample included 446,795 patients. A total of 118 distinct variables were included across 38 cancer-specific models. The distribution of lifetime cancer risk had a rightward skew and wide variation for both men and women. The median lifetime cancer risk was 29.5% for men (interquartile range (IQR) 8.4%) and 21.0% for women (IQR 8.8%). The lifetime risk of cancer at the 90 th percentile (42.61%) was 1.4x higher compared to the median and 1.8x higher than the 10 th percentile (23.78%) for men. Similarly, the lifetime risk of cancer at the 90 th percentile (35.46%) was 1.7x higher compared to the median and 2.3x higher than the 10 th percentile (15.28%) for women. If all modifiable risk factors were set to the ideal state, this decreased to 20.5% for men (IQR 3.9%) and 16.5% for women (IQR 4.9%). There was considerable overlap between age groups, with men aged 50-59 at the 90 th percentile having greater risk (11.9%) than men aged 60-70 at the 25 th percentile (11.8%), and women aged 40-49 at the 90 th percentile having greater risk (7.4%) than women aged 50-59 at the 60 th percentile (6.8%) and women aged 60-70 at the 20 th percentile (7.3%). Conclusions : Lifetime cancer risk varies widely across the UK Biobank cohort, but this risk decreases substantially with healthy behaviors. There was considerable overlap in 10-year cancer risk between age groups, suggesting that future multicancer screening guidelines should account for more than age and sex as more evidence becomes available in the future. Our results underscore the potential for precision prevention approaches that allocate preventive resources and screening intensity according to comprehensive, individualized risk rather than age-based criteria alone.