Accuracy of fecal occult blood screening for colorectal neoplasia. A prospective study using Hemoccult and HemoQuant tests

Abstract

OBJECTIVES: To define the validity of fecal blood as a marker for colorectal neoplasia in the screening setting and to compare yields by Hemoccult and HemoQuant fecal occult blood screening tests. DESIGN: A multicenter masked comparison of fecal blood test results against structural colorectal evaluations and longitudinal follow-up, serving as criterion standards, in nonreferred subjects at risk for colorectal neoplasia. SETTING: Communities, primary care centers, referral centers. PARTICIPANTS: Two groups: (1) 1217 patients aged at least 18 years undergoing routine structural surveillance evaluations following curative resection of a colorectal tumor and (2) 12312 relatives of colorectal cancer patients aged at least 50 years. INTERVENTIONS: Blinded Hemoccult II and HemoQuant testing on three mailed-in stool samples per subject. MAIN OUTCOME MEASURE: Sensitivity of fecal blood tests for colorectal neoplasia. RESULTS: In the postresection group, surveillance evaluations revealed 46 malignant colorectal neoplasms and 402 polyps. At matched specificity, sensitivity of either test for cancer was 26% (95% confidence interval, 13% to 39%). Hemoccult was positive in 21% of intraluminal recurrences, 33% of all new primary tumors, and 29% of Dukes A or B cancers; HemoQuant was elevated in 24%, 28%, and 29%, respectively. Sensitivity for polyps 1.0 cm or larger was 13% by Hemoccult and 11% by HemoQuant. In the group of relatives, estimated sensitivity for cancer at 1 to 3 years of follow-up was 25% to 33% by Hemoccult, not significantly different from the 29% to 43% by HemoQuant. CONCLUSIONS: Based on our observations in the screening setting, fecal blood appears to be a poor marker for colorectal neoplasia. Most cancers and the vast majority of polyps will be missed. Hemoccult and HemoQuant are similarly insensitive.