Polygenic and composite risk scores consistently stratify relative cancer risk but show modest discrimination gains, supporting their use for risk stratification rather than stand-alone clinical tools.
Does the integration of genomic and exposome data into combined risk scores improve cancer risk prediction and stratification in human populations?
Current evidence supports the use of combined risk scores primarily for risk stratification and prioritization rather than as stand-alone clinical decision systems, due to modest discrimination gains and limited multi-ancestry validation.
Polygenic risk scores (PRS) have emerged as important tools for quantifying inherited susceptibility to cancer, and are increasingly combined with environmental and lifestyle factors into composite risk scores (CRS). In this context, environmental inputs should be understood not as isolated covariates, but as components of the human exposome, encompassing cumulative, time-varying, and interacting exposures across the life course, which fundamentally shape cancer risk alongside inherited susceptibility. These approaches are often discussed as candidates for precision prevention and screening, yet their evidentiary basis spans heterogeneous study designs, outcomes, and methodological assumptions. Here, we provide an integrated review of genetic, environmental, and composite cancer risk models, explicitly distinguishing etiologic association from predictive performance and clinical translation. We synthesize evidence from large genome-wide association studies, cohort and case-control analyses, and recent CRS evaluations using both narrative assessment and structured quantitative summaries. Across cancer sites, PRS and CRS consistently stratify relative risk, with monotonic increases in odds ratios across score percentiles. However, gains in discrimination metrics such as the area under the curve or C-index are generally modest and heterogeneous, and calibration performance varies substantially across populations and settings. External validation and multi-ancestry evaluations remain limited, and methodological challenges, including overfitting, population stratification, and model transportability, are frequently under-reported. We argue that current evidence supports the use of PRS and CRS primarily as tools for risk stratification, prioritization, and risk-enriched research designs, rather than as stand-alone clinical decision systems. The most near-term translational value lies in targeted screening strategies, prevention trials, and population-level risk assessment, provided that calibration, governance, and equity considerations are explicitly addressed. We conclude by outlining key methodological and data requirements needed to advance CRS from exploratory models toward robust, population-appropriate tools in cancer prevention.
Sarigiannis et al. (Mon,) conducted a review in Cancer. Combined Risk Score (CRS) and Polygenic Risk Score (PRS) was evaluated. Polygenic and composite risk scores consistently stratify relative cancer risk but show modest discrimination gains, supporting their use for risk stratification rather than stand-alone clinical tools.