Difference in performance of fecal immunochemical tests with the same hemoglobin cutoff concentration in a nationwide colorectal cancer screening program
Tsung-Hsien Chiang, Shu-Lin Chuang, Sam Li-Sheng Chen, Han-Mo Chiu, Amy Ming-Fang Yen, Sherry Yueh-Hsia Chiu, Jean Ching-Yuan Fann, Chu-Kuang Chou, Yi-Chia Lee, Ming-Shiang Wu, Hsiu-Hsi Chen, Tsung-Hsien Chiang, Shu-Lin Chuang, Sam Li-Sheng Chen, Han-Mo Chiu, Amy Ming-Fang Yen, Sherry Yueh-Hsia Chiu, Jean Ching-Yuan Fann, Chu-Kuang Chou, Yi-Chia Lee, Ming-Shiang Wu, Hsiu-Hsi Chen
Abstract
Background & aims: We investigated whether 2 quantitative fecal immunochemical tests (FITs) with the same cutoff concentration of fecal hemoglobin perform equivalently in identifying patients with colorectal cancer (CRC).
Methods: A total of 956,005 Taiwanese subjects, 50 to 69 years old, participated in a nationwide CRC screening program to compare results from 2 FITs; 78% were tested using the OC-Sensor (n = 747,076; Eiken Chemical Co, Tokyo, Japan) and 22% were tested using the HM-Jack (n = 208,929; Kyowa Medex Co Ltd, Tokyo, Japan), from 2004 through 2009. The cutoff concentration for a positive finding was 20 μg hemoglobin/g feces, based on a standardized reporting unit system. The tests were compared using short-term and long-term indicators of performance.
Results: The OC-Sensor test detected CRC in 0.21% of patients, with a positive predictive value of 6.8%. The HM-Jack test detected CRC in 0.17% of patients, with a positive predictive value of 5.2%. The rate of interval cancer rate was 30.7/100,000 person-years among subjects receiving the OC-Sensor test and 40.6/100,000 person-years among those receiving the HM-Jack test; there was significant difference in test sensitivity (80% vs 68%, P = .005) that was related to the detectability of proximal CRC. After adjusting for differences in city/county, age, sex, ambient temperature, and colonoscopy quality, significant differences were observed between the tests in the positive predictive value for cancer detection (adjusted relative risk = 1.29; 95% confidence interval, 1.14-1.46) and the rates of interval cancer (0.75; 95% confidence interval, 0.62-0.92). Although each test was estimated to reduce CRC mortality by approximately 10%, no significant difference in mortality was observed when the 2 groups were compared.
Conclusions: Different brands of quantitative FITs, even with the same cutoff hemoglobin concentration, perform differently in mass screening. Population-level data should be gathered to verify the credibility of quantitative laboratory findings.
Keywords: Colorectal Cancer; Interval Cancer; Population Screening; Screening Test Sensitivity.
Copyright © 2014 AGA Institute. Published by Elsevier Inc. All rights reserved.
Source: PubMed