Head-to-head comparison of the test performance of self-administered qualitative vs. laboratory-based quantitative fecal immunochemical tests in detecting colorectal neoplasm

Abstract

Background:

Fecal immunochemical tests (FITs) are the most widely used non-invasive tests in colorectal cancer (CRC) screening. However, evidence about the direct comparison of the test performance of the self-administered qualitative a laboratory-based quantitative FITs in a CRC screening setting is sparse.

Methods:

Based on a CRC screening trial (TARGET-C), we included 3144 pre-colonoscopy fecal samples, including 24 CRCs, 230 advanced adenomas, 622 non-advanced adenomas, and 2268 participants without significant findings at colonoscopy. Three self-administered qualitative FITs (Pupu tube) with positivity thresholds of 8.0, 14.4, or 20.8 μg hemoglobin (Hb)/g preset by the manufacturer and one laboratory-based quantitative FIT (OC-Sensor) with a positivity threshold of 20 μg Hb/g recommended by the manufacturer were tested by trained staff in the central laboratory. The diagnostic performance of the FITs for detecting colorectal neoplasms was compared in the different scenarios using the preset and adjusted thresholds (for the quantitative FIT).

Results:

At the thresholds preset by the manufacturers, apart from the qualitative FIT-3, significantly higher sensitivities for detecting advanced adenoma were observed for the qualitative FIT-1 (33.9% [95% CI: 28.7–39.4%]) and qualitative FIT-2 (22.2% [95% CI: 17.7–27.2%]) compared to the quantitative FIT (11.7% [95% CI: 8.4–15.8%]), while at a cost of significantly lower specificities. However, such difference was not observed for detecting CRC. For scenarios of adjusting the positivity thresholds of the quantitative FIT to yield comparable specificity or comparable positivity rate to the three qualitative FITs accordingly, there were no significant differences in terms of sensitivity, specificity, positive/negative predictive values and positive/negative likelihood ratios for detecting CRC or advanced adenoma between the two types of FITs, which was further evidenced in ROC analysis.

Conclusions:

Although the self-administered qualitative and the laboratory-based quantitative FITs had varied test performance at the positivity thresholds preset by the manufacturer, such heterogeneity could be overcome by adjusting thresholds to yield comparable specificities or positivity rates. Future CRC screening programs should select appropriate types of FITs and define the thresholds based on the targeted specificities and manageable positivity rates.