Cancers arise in different organs but often share the same biological wiring. A pan-cancer view asks what is common across tumours and what is unique to each type. Understanding this can enable earlier detection, better prevention, and more personalised care.
Using UK Biobank’s large, well-characterised cohort with linked health records and available omics, we will study multiple common cancers together to find molecular, imaging, metabolic and epigenetic signals that (a) are shared across cancers and (b) are specific to individual cancers. Our key questions are:
1. Which biological signatures are common across cancers and which are cancer-specific?
2. Can these signatures improve prediction of who is at higher risk and who is likely to have poorer outcomes?
3. Which lifestyle or biological factors are likely to play a causal role and are therefore potentially modifiable?
4. Do prediction tools perform fairly across age, sex and ancestry groups?
Objectives. Build transparent risk and prognosis models; prioritise potentially causal and actionable pathways; and produce resources (phenotype definitions, derived variables, code) that other researchers and clinicians can reuse.
Scientific rationale. Many hallmarks of cancer (e.g., DNA damage response, immune evasion, metabolic rewiring) cut across tumour sites. A pan-cancer, multi-omics approach in a single population with consistent data capture is well suited to separate shared from disease-specific biology. UK Biobank’s scale, longitudinal follow-up and linkage to cancer registries, hospital records, prescribing and mortality data provide the endpoints needed to evaluate clinical relevance. Findings could inform risk-stratified screening and more targeted prevention and follow-up pathways.
All analyses will be privacy-preserving, with only aggregate outputs exported and no re-identification attempts.
