A two-tiered targeted proteomics approach to identify pre-diagnostic biomarkers of colorectal cancer risk

• Published in: Scientific Reports
• Main Authors: Sophia Harlid, Justin Harbs, Robin Myte, Carl Brunius, Marc J. Gunter, Richard Palmqvist, Xijia Liu, Bethany Van Guelpen
• Format: Article in Journal/Newspaper
• Language: English
• Published: Nature Portfolio 2021
• Subjects: Medicine; R; Science; Q; Northern Sweden
• Online Access: https://doi.org/10.1038/s41598-021-83968-6; https://doaj.org/article/ecce1d3ceeb4472caa4b3a0b21c3609a

Abstract Colorectal cancer prognosis is dependent on stage, and measures to improve early detection are urgently needed. Using prospectively collected plasma samples from the population-based Northern Sweden Health and Disease Study, we evaluated protein biomarkers in relation to colorectal cancer risk. Applying a two-tiered approach, we analyzed 160 proteins in matched sequential samples from 58 incident colorectal cancer case–control pairs. Twenty-one proteins selected from both this discovery phase and the literature were then analyzed in a validation set of 450 case–control pairs. Odds ratios were estimated by conditional logistic regression. LASSO regression and ROC analysis were used for multi-marker analyses. In the main validation analysis, no proteins retained statistical significance. However, exploratory subgroup analyses showed associations between FGF-21 and colon cancer risk (multivariable OR per 1 SD: 1.23 95% CI 1.03–1.47) as well as between PPY and rectal cancer risk (multivariable OR per 1 SD: 1.47 95% CI 1.12–1.92). Adding protein markers to basic risk predictive models increased performance modestly. Our results highlight the challenge of developing biomarkers that are effective in the asymptomatic, prediagnostic window of opportunity for early detection of colorectal cancer. Distinguishing between cancer subtypes may improve prediction accuracy. However, single biomarkers or small panels may not be sufficient for effective precision screening.