Gender Difference in Reproducibility of Quantitative Fecal Immunochemical Test in Asymptomatic Individuals

Abstract

Introduction: Quantitative fecal immunochemical test (QTFIT) has become an important as a screening tool for colorectal neoplasms (CRN). However, as the optimal cutoff value and number of tests are arbitrarily determined, the number samples required for most effective CRN screening remains unknown. While it is thus critical to know the reproducibility of QTFIT, there have been few previous reports. The aim of this study was to determine the reproducibility of QTFIT. Methods: Between January 1 2004 and August 31 2011, 3,827 asymptomatic individuals underwent a general health check-up including 2-day QTFIT and colonoscopy. First and second samples for QTFIT were collected within 3 days prior to colonoscopy. QTFIT was processed via the OC-MICRO instrument (Eiken Chemical Co., Tokyo, Japan). We excluded 371 individuals in whom at least 1 QTFIT result was missing. We conducted cross-sectional study using 3,456 consecutive persons who completed both full colonoscopy and 2-day QTFIT. To determine the reproducibility of QTFIT, we calculated intraclass correlation coefficient (ICC) and performed the Bland-Altman method. Results: Mean age (SD) of 3456 participants was 51 (11) years; 2603 (75%) were male. Median (range) ng/mL of first and second QTFIT was 7 ng/mL (0-912 ng/mL) and 7 ng/mL (0-1000 ng/mL), respectively. Six hundred sixty-one cases of adenoma, 230 of hyperplastic polyp, and 3 of colorectal cancer were diagnosed. Median of second QTFIT was significantly lower than first QTFIT (P=0.0064, Wilcoxon Signed-Rank test). Intraclass correlation coefficient (ICC) of QTFIT was 0.305 (95% confidence interval [CI] 0.27, 0.34). Stratified by gender, QTFIT ICC in women and men was 0.49 (95% CI 0.43, 0.54) and 0.22 (95% CI 0.19, 0.26), respectively. According to the Bland-Altman method, 93 cases (2.7%) of 3,456 fell outside the limits of agreement. Twenty cases (2.34%) of 853 women, and 73 cases (2.8%) of 2,603 men fell outside the limits of agreement (Figure 1). We observed poor agreement in both genders and either subset of women and men. Difference between each day’s QTFIT tended to increase as the mean of 2 values increased in both male and female.Figure 1Conclusion: QTFIT demonstrated poor reproducibility with variability tending to increase as the mean of QTFITs increase. As reproducibility of QTFIT is significantly poorer in men compared to women, the most appropriate number of QTFIT may differ by gender.