MP09-17: Predicting incidence versus mortality for cancers associated with overdiagnosis: an overlooked issue for polygenic risk scores and cancer screening

Andrew Vickers*, NEW YORK, NY, Amit Sud, London, United Kingdom, Jonine Bernstein, NEW YORK, NY, Richard Houlston, London, United Kingdom

Introduction: Population-based cancer screening programs, including prostate-specific antigen (PSA) screening for prostate cancer, are generally directed at all healthy individuals in a given age stratum. It has recently been proposed that PSA screening could be risk-stratified using polygenic risk scores (PRSs) based on panels of single-nucleotide polymorphisms (SNPs): those with high PRS would be screened whereas those with low PRS would be exempted. However, prostate cancer PRSs were generated using the endpoint of cancer incidence rather than cancer-specific mortality, and it has been demonstrated that the PRSs do not differentially predict lethal vs. indolent disease. As such, use of PRSs to risk-stratify screening will not address overdiagnosis, the major problem associated with PSA screening.

Methods: We developed a simple net benefit formulation to comparing different screening strategies, given as: deaths prevented – number screened / NWS – number diagnosed / (NWD -1), where NWS and NWD are the number of individuals we would be willing to, respectively, screen and diagnose, to prevent one death. We applied the formula to empirical data on PSA screening and varied NWS and NWD to assess the effects on our findings.

Results: Risk-stratifying PSA screening using a PRS that does not differentially discriminate between incident and lethal cancer did not improve screening outcomes in our model. Net benefit was 4 per 1000 for screening all eligible men. For a PRS with a C-index of 0.75 (higher than is typically found) but which did not discriminate between incidence and mortality, net benefit was 3.80, 3.36 and 1.78 for restricting screening to those with the highest 50%, 25% and 10% PRS. Net benefit was far lower for a concordance index of 0.65, closer to what is often reported. Results were robust for different choices of NWS and NWD. In contrast, risk stratifying screening based on PSA, which does predict mortality better than incidence, was associated with a net benefit up to 5.64, higher than that for screening all men.

Conclusions: Using PSA levels to determine which men should continue screening will improve screening outcomes; using PRS to risk-stratify screening will do more harm than good. If PRS-based cancer screening is to be effective, research needs to focus on identifying PRSs associated with cancer mortality.

Source of Funding: This work was supported in part by the National Institutes of Health/National Cancer Institute (NIH/NCI) with a Cancer Center Support Grant to Memorial Sloan Kettering Cancer Center [P30 CA008748], a SPORE grant in Prostate Cancer to Dr. H. Scher [P50-CA92629], the Sidney Kimmel Center for Prostate and Urologic Cancers. R.H. acknowledges grant support from Cancer Research UK (C1298/A8362) and the Wellcome Trust (214388). A.S. is in receipt of a National Institute for Health Research (NIHR) Academic Clinical Lectureship and funding from the Royal Marsden Biomedical Research Centre. This is a summary of independent research supported by the NIHR Biomedical Research Centre at the Royal Marsden NHS Foundation Trust and the Institute of Cancer Research. The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health.