Multi-cancer risk prediction in asymptomatic adults using urinary glycosaminoglycan profiling.

Abstract

10522 Background: Current screening guidelines are based on established risk factors such as age and tobacco use. Currently, no biomarker is routinely used to predict cancer risk. We investigated urinary glycosaminoglycan profiles - aggregated in a "GAGome score" - for multi-cancer risk prediction. Methods: In this population-based case-cohort study, we included adults from the Lifelines Cohort Study, Netherlands presumed healthy at baseline. All cases who self-reported any-type cancer or died by the 5-year study visit were confirmed in the Dutch Cancer Registry and matched 1:5 to randomly selected controls. We developed a multivariable Cox proportional hazard regression to estimate the hazard ratio (HR) for incident cancer given the GAGome score and adjusted for established risk factors. Model improvement was assessed using the likelihood ratio test. Then, we used 5-year risk predictions to stratify subjects in four groups: “Low” (<0.15%), “High” (>2%), and “Very high” (>8.5%) risk, or “Intermediate” if otherwise (reference group). Using sampling weights, we estimated the 5-year observed risk in each group, as well as the potentially screen-detectable rate and false positive rate across population and cancer subsets. Results: We included 5436 adults (median age = 49 years, 58% females) of whom 827 were diagnosed with incident cancer within 5 years. A standard deviation increase in the GAGome score had an HR = 1.66 (95% CI: 1.62–1.70) for incident cancer – ranging 1.27 for endometrial cancer to 3.4 for cancer of unknown primary - explaining 36% of the variance (p < 0.0001). The observed 5-year risk for “Low”, “Intermediate”, “High”, and “Very high GAGome risk” were 0.18%, 2.3%, 7.6%, and 32%, respectively. A "High GAGome risk" or higher prediction would detect 54% of incident cancers (including 54% of in situ carcinomas) with a 17% false positive rate. Conclusions: Implementing urine GAGomes as a strategy for risk-stratified targeted screening could pave the way for personalized surveillance approaches and potentially identify a broader range of adults with increased risk of cancer that are not being captured by current screening programs.