Abstract
Abstract Colorectal cancer (CRC) mortality is highly dependent on stage at diagnosis, and measures to increase the likelihood of early detection are urgently needed. Many countries have ongoing screening programs, but the addition of blood-based pre-screening biomarkers for risk stratification could reduce costs and increase compliance. In this study, we used prospectively collected plasma samples from the population-based Northern Sweden Health and Disease Study (NSHDS) to evaluate panels of protein biomarkers in relation to CRC risk. Using a two-tiered approach, we first included a unique set of 58 CRC cases and 58 matched controls with time-matched, repeated, prediagnostic samples (10 years apart, with the second sample collected three months to five years before diagnosis). Relative plasma concentrations of 161 proteins were analyzed by multiplex immunoassay (Olink proteomics, inflammation panel and oncology II panel). In the second tier, 15 of these proteins were selected for further analysis, together with six promising early detection protein markers from the literature. The second sample set included prediagnostic samples from 450 CRC cases and 450 matched controls, of which 33 case/control pairs had repeated samples. Conditional logistic regression and linear mixed models were used to estimate odds ratios for CRC risk. In the second-tier analyses, Lasso regression was used to develop multi-marker models, and the resulting protein combinations were tested using ROC analysis. Of the 15 proteins showing promise as predictive biomarkers in the first phase, none retained statistical significance in the second phase. The most promising marker candidate was FGF-21 (multivariable OR: 1.14 95% CI: 0.99-1.30), a metabolic hormone involved in stimulation of glucose uptake that has been associated with body mass index. FGF-21 was also the only protein consistently selected by the Lasso algorithm. Sensitivity analysis for FGF-21 in subgroups based on tumor site, stage and molecular subtypes including BRAF and KRAS mutations did not reach statistical significance. Our findings highlight the challenge of identifying cancer biomarkers that can be used in the prediagnostic window of opportunity for early detection. For early risk stratification, with the vision of achieving effective precision screening, single biomarkers or small marker panels may not be sufficient. Large studies using prospectively collected samples and machine learning to identify biomarker risk patterns may be a more promising approach. Citation Format: Sophia Harlid, Justin Harbs, Robin Myte, Carl Brunius, Marc Gunter, Xijia Liu, Richard Palmqvist, Bethany Van Guelpen. A two-tiered targeted proteomics approach to identify biomarkers of colorectal cancer risk [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 2353.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have