Development of an algorithm combining blood-based biomarkers, fecal immunochemical test, and age for population-based...

Implementation of screening modalities has reduced the burden of colorectal cancer (CRC), but high false positive rates pose a major problem for colonoscopy capacity. We aimed to create a tailored screening algorithm that expands the fecal immunochemical test (FIT) with a blood specimen and current age to improve selection of individuals for diagnostic colonoscopy.

In this prospective multicenter study, 8 blood-based biomarkers (carcinoembryonic antigen, ferritin, high-sensitivity C-reactive protein, human epididymis protein 4, Cyfra21-1, hepsin, interleukin 8, and osteoprotegerin) were investigated in 1977 FIT-positive individuals from the Danish national CRC screening program undergoing follow-up colonoscopy. Specimens were analyzed on Architect i2000, Architect c8000 (both from Abbott, Chicago, Ill, USA), or Luminex xMAP machines (MilliporeSigma, St. Louis, Mo, USA). FIT analyses and blood-based biomarker data were combined with clinical data (ie, age and colonoscopy findings) in a cross-validated logistic regression model (algorithm) benchmarked against a model solely using the FIT result (FIT model) applying different cutoffs for FIT positivity.

The cohort included individuals with CRC (n = 240), adenomas (n = 938), or no neoplastic lesions (n = 799). The cross-validated algorithm combining the 8 biomarkers, quantitative FIT result, and age performed superior to the FIT model in discriminating CRC versus non-CRC individuals (area under the receiver operating characteristic curve, 0.77 vs 0.67, respectively; P < .001). When discriminating individuals with either CRC or high- or medium-risk adenomas versus low-risk adenomas or clean colorectum, the areas under the receiver operating characteristic curve were 0.68 versus 0.64 for the algorithm and FIT model, respectively.

The algorithm presented here can improve patient allocation to colonoscopy, reducing colonoscopy burden without compromising cancer and adenoma detection rates or vice versa.