A comparison of fecal occult-blood tests for colorectal-cancer screening

Summary

Traditional guaiac-based fecal occult blood tests are relatively insensitive for the detection of adenomas and cancer. Rehydrated stool guaiac cards are more sensitive and have been shown to decrease mortality from colon cancer, but they are less specific and lead to many false positives. Newer approaches to FOB testing include more sensitive guaiac cards (such as Hemoccult II Sensa), and an immunochemical test for human hemoglobin (HemeSelect). This study was designed to compare three of these methods and a combination test.

Methods

• Subjects: Patients of Kaiser Permanente Medical Center in Oakland, CA, over 50 years of age, scheduled for a "personal health appraisal" examination between October 1990 and October 1991.
   
• Stool cards: Each patient received three stool cards, each containing three stool tests -- Hemoccult II (conventional guaiac), Hemoccult II Sensa (more sensitive guaiac test) and HemeSelect (immunochemical test).
   
• Specimen collection: Collection during the three days prior to the clinic visit, using a paper saddle device (to avoid contact of the stool with water). Multiple dietary and medication exclusions during the week prior to the sampling.
   
• Stool card analysis: Each of the three tests was developped and recorded separately. A combination test was also defined, using the Hemoccult II Sensa and the HemeSelect -- it was positive if both tests were positive, but negative for any other combination. Not all submitted samples were interpretable; a subject was considered screened by a given method if there was at least one interpretable result by that method.
   
• Follow-up: Patients with a positive test result were evaluated, usually by colonoscopy, except for patients who had a positive Hemoccult II Sensa but negative other tests. When most of these patients were found to have negative colonoscopies, the protocol was changed to recommending sigmoidoscopy followed by repeat stool testing in 6 months.
   
• Test evaluation: Tests were evaluated for sensitivity and specificity using the incidence of polyps > 1cm. and colon cancer detected during the two years following the stool collection.

• 8104 subjects were screened with at least one test. Age groups 50-59, 60-69 and over 70 were equally represented.
   
• At least one test was positive in 1312 patients (16.2 %). Of these, 57% had only a positive Hemoccult II Sensa test.
   
• Colonoscopy was performed in over 80% of patients with positive results on a test other than Hemoccult II Sensa. 142 neoplasms were detected, 107 polyps and 35 carcinomas.
   
• The sensitivity and specificity of the four tests for polyps over 1 cm and carcinoma were:
   
• Hemoccult II: sensitivity 32%, specificity 98%
• Hemoccult II Sensa: sensitivity 71%, specificity 88%
• HemeSelect: sensitivity 67%, specificity 95%
• Combination test: sensitivity 54%, specificity 98%

The authors feel that the combination test (Hemoccult II Sensa and, it this test is positive, confirmatory HemeSelect testing) represents the best approach. They assume that, as in their study, a positive Hemoccult II Sensa with a negative HemeSelect result will be followed by flexible sigmoidoscopy and Hemoccult II testing at 6 and 12 months. Although HemeSelect alone has a higher sensitivity than the combination test, it has a lower specificity and a much higher cost ($3.31 for Hemoccult II; $3.82 for Hemoccult II Sensa and $19.23 for HemeSelect). Since about 2/3 of the combination tests would not require HemeSelect testing, the cost of the combination test would be significantly less than HemeSelect testing alone.

Comment

In an accompanying editorial, D. Ransohoff and C. Lang state "Because neoplasms bleed intermittently, however, the next step in the evaluation of the combination test is to show that the sensitivity remains high when the test is applied to sequential specimens obtained at two different times, as Allison et al. suggest, instead of to the same specimens, as in their study." This is not my understanding of the combination test, however, which is not clearly defined in the article. My interpretation of the combination test is that all patients would provide simultaneous samples for both tests, and that the immunochemical test would only be performed if the guaiac test turned out to be positive. This is a very important distinction. Having a patient obtain three stool samples, followed by three more samples for the 1/3 of patients with a positive guaiac test would not be acceptable or practical.

The authors recommend performing sigmoidoscopy and repeat stool testing at 6 and 12 months for those with positive guaiac but negative immunochemistry tests. Unfortunately, they do not give data on the incidence of polyps and tumors in these patients, compared to those with negative guaiac tests. If these incidences turned out to be the same, even the flex sig and repeat stool testing might not be necessary, although it is hard to argue against them.

Assuming that the combination test is made available as a single, multi-specimen card, it certainly would seem to represent an improved fecal occult blood test approach.

2/27/96

Reader comments

 Remember that flexsig only studies about half of the colon. Aircontrast barium enema or a good single contrast barium enema is necessary, because about 30% of the significant polyps or cancers would otherwise be missed.
 

• In this study, flexible sig was only performed in patients with an isolated positive sensitive guaiac test (and negative immunochemical test) -- which the authors felt represented an almost certainly falsely positive test.

For screening in asymptomatic patients, FOB and flexible sigmoidoscopy have a role. Some have argued for a single colonoscopy at age 50, particularly for high risk patients. The problem with using barium enema as part of a screening strategy (in guaiac positive patients) is that any findings such as polyps would then require a colonoscopy for polypectomy. BE vs. colonoscopy in symptomatic patients is another question, not totally resolved as far as I know. -- mj

Letters to the Editor about this article from the NEJM

The question of the proper interpretation of the combination test, which I raised in my comments about this article, were addressed and answered here.

Journal Club on the Web home page

Submit a comment about this article