Current issues of ACP Journal Club are published in Annals of Internal Medicine


Diagnosis

Fecal occult blood screening for colorectal cancer was insensitive

ACP J Club. 1993 July-Aug;119:25. doi:10.7326/ACPJC-1993-119-1-025


Source Citation

Ahlquist DA, Wieand HS, Moertel CG, et al. Accuracy of fecal occult blood screening for colorectal neoplasia. A prospective study using Hemoccult and HemoQuant tests. JAMA. 1993 Mar 10;269:1262-7.


Abstract

Objective

To determine the accuracy of Hemoccult and HemoQuant fecal occult blood screening tests against a diagnostic standard for detecting colorectal cancer in patients at risk.

Design

Independent, blinded comparison of fecal blood test results with structural colorectal evaluations and follow-up.

Setting

Communities, primary care centers, and referral centers in the United States and Canada.

Patients

Two patient groups were studied: 1217 patients having routine structural colorectal surveillance evaluations within 4 months of curative resection of a colorectal tumor that preserved at least half of the colorectum, and 12 312 first-degree relatives (age ≥ 50 y) of colorectal cancer patients. Exclusion criteria were inflammatory bowel disease, familial adenomatous polyposis, or defecation through a stoma.

Description of tests and diagnostic standard

Patients refrained from red meat, nonsteroidal anti-inflammatory drugs, iron supplements, and vitamin C for 5 days before and during stool collections. Sealed plastic tubes were used for mailing 3 fecal samples. The Hemoccult II test was done by experienced technicians as recommended in kit instructions. The HemoQuant test was done using a semiautomated modification of the originally described assay. The diagnostic standard was complete colorectal evaluations by colonoscopy or proctosigmoidoscopy, or by colon roentgenogram plus proctosigmoidoscopy, in all postresection patients. Missed cancers in the group of relatives were estimated from the proportion of 900 patients who presented with cancer within 36 months of screening.

Main outcome measures

Sensitivity and specificity of fecal blood tests for detecting colorectal neoplasia.

Main results

In the postresection group, structural surveillance evaluations detected 46 malignant colorectal cancer and 402 polyps. At a matched specificity of 95%, both tests had a sensitivity of 26% (95% CI 13% to 39%). Hemoccult sensitivity for detecting intraluminal recurrence, all new primary tumors, and Dukes A or B cancers was 21%, 33%, and 29%, respectively. HemoQuant sensitivity for the same cancers was 24%, 28%, and 29%, respectively. The sensitivity of detecting polyps ≥ 2.0 cm was 20% for both tests. In the group of relatives, the estimated sensitivity for colorectal cancer at 1 to 3 years follow-up was 25% to 33% by Hemoccult and 29% to 43% by HemoQuant.

Conclusions

Fecal occult blood screening using Hemoccult and HemoQuant tests for detecting colorectal neoplasia was insensitive. Most cancers and a majority of polyps were missed.

Source of funding: National Institutes of Health.

For article reprint: Dr. D.A. Ahlquist, Division of Gastroenterology, Mayo Clinic, Rochester, MN 55905, USA. FAX 507-284-0538.


Commentary

The clever study by Ahlquist and colleagues adds to growing doubt about the value of fecal occult blood tests (FOBTs) for colorectal cancer screening. The authors recognized that patients with resected colorectal cancer would have regular colonoscopy (or barium enema) and obtained stools from them before their annual structural examinations. In contrast to previous studies, they did complete evaluations in persons with both positive and negative FOBTs, providing an unbiased assessment of sensitivity and specificity.

The study showed that fecal occult blood tests failed to detect 70% of cancers and 80% of large (≥ 2.0 cm) polyps. HemoQuant, a more expensive and cumbersome test, offered no advantage over Hemoccult at matched sensitivity. Study strengths included large sample size, prospective design, careful follow-up, and meticulous laboratory methods. The study population and test conditions differed somewhat from previous studies. Stool samples were mailed in sealed containers, although this may actually be superior to fecal smears. All patients had had previous colonic resection. Perhaps stool characteristics or transit rate after partial colectomy reduces FOBT sensitivity. The fact that sensitivity was similarly low in family members and in other recent reports suggests that the results are not an aberration.

Clinicians should not necessarily abandon Hemoccult tests based on this study. A final decision should await the large, randomized trials currently under way. These trials will tell us whether FOBT saves lives, which is our ultimate goal. If physicians choose to use Hemoccult, however, both they and their patients need to recognize its serious shortcomings. The test will miss more than half the cancers and most polyps. Many colonoscopies will be done for false-positive tests. Hemoccult may detect some cancers in asymptomatic persons, but it is an insensitive and inefficient method. There is no place for HemoQuant.

Robert S. Sandler, MD, MPH
University of North CarolinaChapel Hill, North Carolina, USA