Fecal Immunochemical Testing in Patients With Low-Risk Symptoms of Colorectal Cancer: A Diagnostic Accuracy Study

Abstract

Background: The fecal immunochemical test (FIT) is recommended for triaging primary care patients in England with low-risk symptoms of colorectal cancer (CRC). The evidence underpinning recommendations by the National Institute for Health and Care Excellence had limitations, with a paucity of primary care evidence. This study examines the diagnostic accuracy of FIT in a defined low-risk symptom primary care population.

Patients and methods: Consecutive symptomatic adult patients referred for a FIT between October and December 2019 were included. Patients were derived from 225 primary care practices in England. Serious colorectal diseases (CRC, high-risk polyps, and inflammatory bowel disease [IBD]) were identified through patient follow-up over 18 months, using both primary and secondary healthcare records. Performance characteristics of FIT are reported according to differing thresholds, including the currently recommended threshold of ≥10 μg hemoglobin per gram of feces (μg/g).

Results: A total of 3,506 patients were included in the final analysis. Of these, 708 had a positive FIT result (≥10 μg/g). The prevalence of CRC was 1.3%. FIT positivity declined from 20.2% to 5.8% and 4.5% at cutoffs of 10, 80, and 120 μg/g, respectively. The sensitivity of FIT at ≥10 μg/g to detect CRC was 91.1% (95% CI, 77.9%-97.1%); its specificity was 80.7% (95% CI, 79.3%-82.0%); the positive predictive value (PPV) was 5.8% (95% CI, 4.2%-7.8%); and the negative predictive value (NPV) was 99.9% (95% CI, 99.6%-99.95%). The area under the receiver operating characteristic curve was 0.93 (0.91-0.96). PPV and specificity increased, whereas sensitivity and NPV decreased when serious colorectal diseases (CRC, high-risk polyps, and IBD) were combined. Age, sex, socioeconomic deprivation, and anemia did not significantly influence FIT sensitivity on subgroup analysis.

Conclusions: Utilization of FIT at a threshold ≥10 μg/g can safely triage patients with low-risk symptoms in primary care, with negative results effectively ruling out CRC.