Ulrike Peters

PhD, MPH

Fred Hutchinson Cancer Research Center

Associate Director

PLCO (Learn more about this study)

PLCO-872

Dec 8, 2021

Using PRS for CRC screening

Screening is an established method for decreasing colorectal cancer (CRC) incidence and mortality. However, despite guidance supporting CRC screening initiation (e.g., 45-50 years), relatively little is known about what to do for the subsequent 30+ years after screening. This is particularly concerning given that over 40% of individuals who undergo screening are found to have a pre-cancerous polyp and are instructed to undergo frequent colonoscopies (termed surveillance), ranging from 3-10 years, because of their elevated risk for CRC. To help guide patients and providers on who needs surveillance, guidelines have categorized patients as high- or low-risk based solely on the presence or absence of polyps from the baseline (first) colonoscopy. Although this polyp-based risk stratification scheme has been widely adopted in clinical practice, its ability to predict the risk of advanced neoplasia is low. Further exacerbating the low predictive value of the current guideline-recommended risk stratification scheme is the well-documented under- and overutilization of surveillance colonoscopy, which may lead to increased healthcare costs as well as undue harm to patients from unnecessary procedures or missed cancers.

A risk prediction tool that includes clinical, genetic, and endoscopic risk factors has been proposed as a potential strategy to tailor surveillance; identify patients at highest risk of advanced neoplasia; and reduce under- and overutilization of surveillance. However, few studies have developed a prediction tool utilizing clinical risk factors outside of traditional polyp findings and none have included genetic predictors. With recent advances in genomic technology and plummeting genotyping costs, patients have broader accessibility to genetic data, particularly polygenic risk scores (PRS) that aggregate individual common genetic risk variants into a single score, which has been strongly linked with multiple clinical conditions. Despite growing clinical acceptance of PRS for cancer screening (e.g., WISDOM trial for breast cancer screening), there is currently limited data on its ability to predict advanced neoplasia following initial CRC screening. Therefore, determining the effect of PRS on advanced neoplasia risk after screening, and understanding its impact in a comprehensive prediction model is urgently needed as more individuals seek for a personalized approach to tailor surveillance and healthcare systems strive to optimize surveillance services for CRC prevention.

The overall goals of this proposal are to examine the role of PRS on determining advanced neoplasia risk among patients undergoing surveillance; and develop and validate a comprehensive model for advanced neoplasia among patients who had previously undergone colonoscopic screening. To achieve this goal, we will draw on recent advances in genomic research and our work demonstrating the significant added value of using PRS for CRC screening. Our overarching hypothesis is that integrating PRS with clinicopathological assessment will improve risk stratification of patients with a prior history of screening and better inform optimal surveillance follow-up. To test this hypothesis, we will leverage data from a contemporary, community-based cohort of >20,000 patients, including nearly 4,000 ethnic minority patients, nested within a fully integrated health services delivery organization, with a stable membership broadly representative of its local communalities. This cohort’s detailed data includes genome-wide genotype arrays coupled with prior screening and clinicopathologic data, and surveillance outcomes (i.e., advanced neoplasia). In addition, we will externally validate our findings in an independent dataset with complete genetic and clinicopathologic data. Utilizing these rich resources and environment, we propose the following specific aims:

AIM 1: Develop and internally validate a clinical (e.g., age, sex, body mass index, family history of CRC) and endoscopic (e.g., polyp findings) risk score to predict advanced neoplasia after screening; and determine the added value of incorporating a polygenic risk score to this model.
AIM 2: Validate the performance of the comprehensive risk model using an independent dataset with complete genetic data, and prior screening and clinicopathologic data.
AIM 3: Identify the optimal strategy for surveillance using information from Aims 1-2 by modeling within the established microsimulation model (MISCAN).
AIM 4: Engage key stakeholders (i.e., patients, providers, and healthcare administrators) of the model and elicit their willingness on implementing this tool in clinical practice