Biomarkers in Early Detection of Colorectal Cancer: Does Setting Matter?
Posted: Monday, August 27, 2018
Although an inflammatory biomarker combination performed well in a clinical setting for early detection of colorectal cancer, diagnostic performance was substantially lower under true screening conditions, reported a recent study in the Journal of Clinical Epidemiology. Most studies of cancer detection identify inflammatory markers in clinically manifest cases, but they may not be evaluated under true screening conditions.
“Our results underscore the necessity to validate biomarkers for colorectal cancer early detection discovered in colorectal cancer cases from clinics in a true screening setting, with strict quality control as an indispensable step prior to translation of the results to clinical practice,” Hermann Brenner, MD, MPH, of the University of Heidelberg, Germany, and colleagues concluded.
The authors quantified 92 inflammatory proteins in baseline plasma samples from individuals with colorectal cancer and neoplasm-free controls. A biomarker panel was chosen and evaluated in samples from a clinical validation setting (n = 318) as well as a true screening validation setting (n = 126). The former included patients who were recruited and blood samples were taken after diagnosis, whereas in the true screening setting, cases were evaluated through screening colonoscopy and blood samples prior to diagnosis.
A 5-biomarker signature chosen from the training set performed well and provided an AUC of 0.85 and a 60.9% sensitivity to detect colorectal cancer at 90% specificity. The panel was applied in a clinical setting and AUC decreased slightly to 0.85 and sensitivity dropped to 49.5%. Diagnostic performance dropped further in a true screening validation setting, providing an AUC of 0.59 and a 28.6% sensitivity.