Scientific rationale
Cancers arise in diverse organs but share biological hallmarks such as genomic instability, immune evasion and metabolic reprogramming. Yet, actionable biomarkers for early detection, prognosis and treatment remain limited. The UK Biobank, with its large, well-phenotyped cohort, long follow-up, and rich omics, imaging and environmental data, offers an unparalleled resource for pan-cancer discovery and validation.
Research questions
1.Which molecular, imaging and clinical signatures are shared or cancer-specific?
2.Which germline variants and polygenic scores stratify susceptibility and prognosis?
3.Which modifiable exposures causally affect risk and interact with genetics?
4.Can longitudinal biomarkers and imaging identify pre-diagnostic signals?
5.Which genes or pathways are suitable therapeutic targets?
Objectives
* Perform GWAS and exome analyses; develop and validate polygenic and multi-omic risk models.
* Construct transparent prediction models using machine learning with ancestry-aware validation.
* Apply Mendelian randomisation to identify causal, modifiable factors.
* Integrate QTL, pathway and drug-target data to nominate therapeutic opportunities.
* Evaluate clinical and economic utility of risk-stratified screening and follow-up.
* Share phenotype definitions and aggregate outputs while preserving privacy.
Impact
This project will uncover biomarkers and risk factors across cancers, enabling early detection, better prognostication and personalised prevention, advancing precision medicine and optimising healthcare resources.
