Thresholds adjustments and impact on clinical performance of three FIT assays in a colorectal cancer screening program

OBJECTIVES: Fecal immunochemical tests (FITs) from various manufacturers yield different results, leading to variable positivity rates and clinical performance. These differences can influence the healthcare costs of a colorectal cancer screening program (CRC-SP), specifically when switching among FIT manufacturers. The analytical and clinical performance of three FITs was investigated to determine adjusted thresholds using positivity rate harmonization. METHODS: A cohort of 6,600 participants from the Quebec CRC-SP received collection kits from three different manufacturers to sample the same fresh stool. Participants with positive results were referred for a colonoscopy which was considered positive if advanced neoplasia (AN) was detected. Positivity rates were determined for each FIT at an unadjusted threshold. Adjusted thresholds were determined using a positivity rate harmonization strategy. Clinical performance was then evaluated for each supplier at unadjusted and adjusted thresholds. RESULTS: Among 5,513 participants who fulfilled the inclusion criteria, 327 underwent colonoscopy. The FIT positivity rates at unadjusted thresholds differed between manufacturers. The adjusted thresholds determined to yield the same positivity rate were different for each FIT. A total of 69 participants were diagnosed with AN. Significant differences in concordance, discordance, sensitivity, and specificity were observed when using the unadjusted threshold. After applying adjusted thresholds, differences in clinical performance between manufacturers were no longer statistically significant. CONCLUSIONS: Threshold adjustment using a positivity rate harmonization strategy render differences in clinical performance statistically nonsignificant and leads to a stable colonoscopy rate between manufacturers. CRC screening programs should determine adjusted thresholds when changing FIT suppliers or when using multiple assays.