EARLY DETECTION OF COLORECTAL NEOPLASIA: APPLICATION OF A BLOOD-BASED SEROLOGICAL PROTEIN TEST ON SUBJECTS UNDERGOING POPULATION-BASED SCREENING.
Jakob Kleif*1, Lars N. Jørgensen2,3, Jakob Hendel4, Mogens R. Madsen5, Jesper Vilandt6, Søren Brandsborg7, Ali Khalid8, Linnea Ferm1, Gerard J. Davis9, Susan H. Gawel9, Frans Martens10, Ib J. Christensen1, Hans J. Nielsen1,3
1Hvidovre Hospital - Copenhagen - Denmark, Hvidovre, Denmark; 2Bispebjerg and Frederiksberg Hospital, Copenhagen, Denmark; 3Copenhagen University, Copenhagen, Denmark; 4Herlev and Gentofte University Hospital, Herlev, Denmark; 5Herning Hospital, Herning, Denmark; 6Nordsjaellands Hospital, Hillerød, Denmark; 7Horsens Hospital, Horsens, Denmark; 8Viborg Hospital, Viborg, Denmark; 9Abbott Laboratories, Abbott park, IL, IL; 10Amsterdam Medical Center, Amsterdam, Netherlands
Introduction: Direct colonoscopy is the "gold" standard for colorectal cancer screening. Limitation in colonoscopy capacities makes direct colonoscopy as a nationwide screening test for colorectal cancer unrealistic. Nationwide colorectal cancer screening programs use FIT or gFOBT with compliance rates below the desirable rate of 90%. A blood-based screening test may increase compliance, but if it is to be used as frontline screening test for colorectal cancer it must be discovered, tested and validated in a true screening population. Several blood-based signature tests for colorectal cancer have shown promising results, but most have been performed in symptomatic populations, preformed in limited studies, or without proper internal or external validation. We aimed to develop a blood-based signature test using multiple blood-based proteins for the detection of colorectal cancer in a large screening population. The blood-based proteins were chosen based on previous studies in symptomatic patients.
Methods: The study population is based on a subset, training set, from the Endoscopy III Project including 8,415 FIT positive (hemoglobin ≥ 100 ng/mL; OC-Sensor platform) participants from the Danish Colorectal Cancer Screening Program. Participants were included prior to colonoscopy and blood samples were collected just before colonoscopy. The cohort was divided into a training set and a test set in a consecutive order so that the test set included at least 240 colorectal cancer cases. EDTA plasma samples from the training set were analyzed at the Abbott Center of Excellence in Amsterdam, where CEA, hs-CRP, TIMP-1, Pepsinogen-2, HE4, CyFra21-1, Galectin-3, Ferritin and B2M were measured using the Architect Automated immunoassay platform (Abbott, Chicago, IL, USA). The primary aim of the study was to develop a blood-based signature test using the multiple blood-based proteins, age, and gender for discriminating between CRC versus all others. Secondary aims were to develop a test for discriminating between CRC, high risk-, and medium risk adenomas versus all others, and CRC versus clean colon. Logistic regression was used for model identification. Blood markers were log transformed.
Results: The distribution of the training set (n=4,048) is 242 CRC, 2,064 adenomas, and 1,742 clean colon. The model for discriminating between CRC and all others had an AUC of 0.70 (95% CI: 0.66-0.74). When focusing on CRC, high-, and medium risk adenomas versus all others the AUC was 0.61 (95% CI: 0.59-0.63). For CRC versus clean colon the AUC was 0.73 (95% CI: 0.69-0.77).
Conclusion: A blood-based signature test based on CEA, hs-CRP, TIMP-1, Pepsinogen-2, HE4, CyFra21-1, Galectin-3, Ferritin, B2M, age, and gender needs future performance enhancement with additional orthogonal covariates as well as validation as a frontline screening test for CRC.
Back to 2020 Abstracts