Guaiac-based fecal occult blood (FOB) tests commonly are used in screening programs for colorectal carcinoma. Guaiac is a chemical test for peroxidase activity that is present in both hemoglobin (Hb) and certain foods.1, 2 To improve readability, the guaiac FOB test HemoccultSENSA (HOS) (Beckman-Coulter, Inc., Primary Care Diagnostics, Palo Alto, CA) was introduced.3, 4 This test led to improved sensitivity for significant neoplasia (colorectal carcinoma or adenomas ≥ 1 cm in size), but, conversely, to reduced specificity due to increased false positivity from dietary peroxidases.4–8
The overall impression was that despite its improved sensitivity for significant neoplasia, the high positivity rate of HOS and cumbersome dietary restrictions recommended limited its usefulness as an annual screening test.9, 10 The alternatives considered were to change to or add an immunochemical test specific for human Hb, which also has an additional advantage in that no dietary restrictions are required.6, 8, 9–12 The immunochemical test we evaluated is FlexSure OBT (FS) (Beckman-Coulter, Inc., Primary Care Diagnostics, Palo Alto, CA).
Meanwhile, both experimental studies and clinical experience showed that delaying development of HOS until 3 days after preparing the last FOB card significantly reduced its false positivity rate and improved its specificity.5, 9 This apparently allowed time for the breakdown of dietary vegetable peroxidases.13 Further experience also showed that by this delay in development of HOS of at least 3 but not more than 14 days, dietary restrictions before and during HOS preparation were not required in our population and could be eliminated.14
We examined the comparative usefulness of each test within the framework of an ongoing cumulative evaluation of different FOB tests in an endoscopic screening/follow-up study.5, 9, 15 The objective being to draw conclusions and recommend which FOB test to use for a population-screening program.
MATERIALS AND METHODS
The FOB tests were given to consecutive persons attending a colorectal cancer screening/follow-up service or to symptomatic patients coming for evaluation of abdominal complaints. Inclusion criteria were agreement to comply with medication limitations, ability to prepare the slides (see below), and history of a recent normal endoscopic examination of their large bowel. Those who had not received an endoscopic examination underwent one after the FOBT study. Exclusions included noncooperation, active rectal bleeding, and known ulcerative colitis.
Initially, all examinees received verbal and written instructions to eat a low peroxidase-containing diet for 3 days before and during the days of preparing the stool tests.9 Medications such as aspirin and other nonsteroidal antiinflammatory drugs were withdrawn 1 week before preparing the stool tests. Further experience showed that dietary limitations could be eliminated when using HOS without affecting FOBT positivity in the neoplasia free screenees. So, after the first 403 examinees, no diet was used.14 However, the examinees were requested not to take vitamin C supplements for 3 days before and during the period of FOB test preparation.
In our program, in addition to an annual FOB test, screenees at above average risk have a colonoscopic examination every 3–5 years, and the others have a flexible sigmoidoscopy examination every 3–5 years.15 Persons with a neoplasm detected during flexible sigmoidoscopy or have a positive FOB test or who are being followed up after having had an adenomatous polyp or cancer or are symptomatic have a colonoscopic examination. All lesions were measured, biopsied, and/or removed at endoscopy and sent for examination by the one pathologist.
Fecal Occult Blood Tests
HOS is a standard guaiac test with two windows on each of the three sample collection cards. The FS immunochemical test collection cards are similar in appearance and preparation to the guaiac test. In this study, the HOS and FS collection cards were attached together, so that after each bowel movement the subjects prepared both FOB cards from the same stool specimen. To keep the stool sample dry during collection, examinees were provided with a disposable paper stool-collection device that was placed on the toilet seat. Samples were taken with a wooden spatula from two different parts of the stools and smeared on the top and lower sections of both windows of both FOB sample collection cards. The cards were prepared on three consecutive occasions, kept in a cool place and hand-returned. The examinees recorded on each card the date of preparing that test card. These dates and the date of developing the FOB test were registered.
To allow the breakdown of unstable dietary peroxidases, development of the FOB tests was not performed sooner than 3 days after the last test card was prepared, but not later than 14 days after the first sample collection so that the stool specimens would not dry out completely. According to the manufacturer's instructions, this will still allow detection of low, but clinically significant levels of blood in stools.16
HOS was developed by applying two drops of the specific developing reagent to each window and one drop to the control (Performance Monitor) area on each card. If positive, a blue coloration appeared within 1 minute.9, 16 The FS immunochemical detection of human fecal Hb was performed by removing a small paper tab containing the stool sample from the collection card and transferring it onto a test card device. To extract the sample, three drops of buffer then were applied to the test card. This extract migrated through a pad containing anti-human Hb conjugate, which binds to Hb if it is present. The complex continued to migrate chromatographically along the test strip. When the antigen–antibody complex reached the reaction area, where anti-human Hb is bound, a red line appeared within 5 minutes if the test was positive (test line). A second control line always appeared indicating normal performance of the test. Each day, before developing the FS, a standard positive and negative control was used to ensure that the buffer-developing reagent was active.9
Our performance proficiency and reading threshold sensitivity for HOS and FS had been evaluated by our 3 examiners developing and blindly reading, in parallel, 60 samples smeared with stools having 0, 0.2, 0.5, 1.0, or 2.0 mL blood added to 100-g stools. Our reading threshold for HOS lay between 0.5 and 2 mL of blood, equivalent to 70–280 mg Hb/100 g feces. The FS reading threshold sensitivity was for 0.2 mL added blood equivalent to 28 mg Hb/100 g stool.9
Clinical and Statistical Evaluation
All FOB tests were distributed by the same medical interviewers, and development was performed by the one technician who was blinded to the clinical diagnosis. If any one FOB test was positive, then the test was regarded as positive and total colonoscopy was recommended. The presence of an adenomatous polyp of any size or cancer was considered a positive finding, and patients were classified by their largest adenoma detected or presence of cancer. The findings were analyzed as all cases having any colorectal neoplasia detected, all cases having only adenomas detected, and those having clinically significant neoplasia, adenomas ≥ 1.0 cm in diameter or cancers. Nonneoplastic endoscopic findings were regarded as a negative examination for neoplasia. Asymptomatic persons having a positive FOB test but normal colonoscopic examinations were not investigated any further but remained under observation.
To evaluate the duration of time elapsing between patient preparing and our developing each type of FOB test, each smeared HOS window and FS window was considered an individual test. So, there were 6 HOS and 3 FS tests per examinee available for evaluation. Not all the examinees had a total colonoscopic examination, so specificity was estimated from the results of total colonoscopy or negative flexible sigmoidoscopic examinations in the absence of any positive FOB test. The estimated specificity, relative sensitivity, and predictive values of each type of FOB test were evaluated as were those resulting when both the immunochemical test and the guaiac test were positive in the same patient. The 95% confidence intervals for proportions were determined, and chi-square analysis was used to compare differences in proportions between the tests.
The screening program was authorized by the hospital's ethics committee in 1978 and this FOB study in 1996. All endoscopic procedures were performed after obtaining informed consent.
The protocol of performing all FOB tests and an endoscopic examination was completed by 1410 persons. They were 1367 (97%) consecutive asymptomatic persons (average risk, 21%; family history of colorectal neoplasia, 47%; cured breast carcinoma, 6%; follow-up of adenoma, 19%; follow-up of cured colorectal carcinoma, 7%) coming to the Colorectal Cancer Screening/Follow-up Service for the first time (11%) or for repeated annual visits (89%),15 and 43 (3%) symptomatic patients coming for evaluation of abdominal complaints. None had active rectal bleeding. Their mean age (± standard deviation [SD]) was 60.9 ± 11 years, and 53% were women.
The screening/follow-up protocol or symptomatology required 730 (51.8%) of the examinees to undergo a total colonoscopic examination and the remaining 680 (48.2%) to undergo a flexible sigmoidoscopic examination, either at this time or within 4 years of this annual FOB test. The details are as follows.
At the time of this round of FOB testing, 681 of the 1410 (48%) underwent an endoscopic examination. They were examined by flexible sigmoidoscopy only (256 persons, 18%) or colonoscopy (425 persons, 30%). The latter group included asymptomatic subjects belonging to a high risk group, those having colonic symptoms, all persons having a neoplastic finding at flexible sigmoidoscopy, and examinees having a positive FOB test.
The remaining 729 subjects (52%) who were asymptomatic did not have FOB detected at this annual test. They previously had repeatedly negative annual FOB tests and a normal flexible sigmoidoscopic examination (424 persons), mean (± SD) 2.8 ± 1.2 years before, or a colonoscopic examination (305 persons), mean (± SD) 2.5± 1.0 years, before performing this annual FOB test.
FOB Tests: Compliance and Evaluation
The 3 cards for each test were correctly prepared by 1391 (98.6%) of the examinees; the remainder prepared only 2 cards.
All tests were negative in 1322 examinees (93.7%). Overall, HOS was positive in 74 persons (5.2%), a mean of 2.4 tests of 6 per positive person. FS was positive in 24 (1.7%), a mean of 1.6 tests of 3 per positive person. Both FOB tests were in agreement and positive in 11 persons (0.8%) HOS only was positive in 64 (4.5%), and FS only was positive in 14 (1.0%). Of the 74 persons having a positive HOS, 70 (94.6%) were evaluated by colonoscopy, and 4 persons refused colonoscopy and only had a normal flexible sigmoidoscopy examination. Clinical follow-up of > 2 years has not revealed overt evidence of neoplasia in these 4 persons. All persons having a positive FS test had a colonoscopic examination.
For 1355 persons having a negative endoscopic examination for neoplasia at present or in the recent past, HOS was positive in 57 persons (4.2%). In the same persons, FS was positive in 14 (1.0%) (P < 0.05). Both HOS and FS were positive in 11 of such examinees (0.8%).
The mean (± SD) duration of time elapsing after HOS and FS patient preparation and development of each window was 6.9 ± 2.8 days, which is approximately 3.9 days after returning the kits. There was no significant difference in duration before development between those with or without having any neoplasia detected and having or not having a positive FOB test (6.7 ± 3.0 days vs. 7.0 ± 2.0 days for HOS, and 6.9 ± 3.4 days vs. 6.7 ± 1.9 days for FS).
The 7 cancers detected were Dukes A (3 cases), B (3 persons), and C (1 case); both HOS and FS were positive in 3 of the 7 cancers. Adenomas ≥ 1.0 cm were found in 13 examinees; 7 had a positive HOS tests, and 4 had positive FS tests. Adenomas > 1.0 cm were found in 35 cases; 9 had positive HOS tests, and 3 had positive FS tests.
These are summarized in Tables 1 and 2. When including all cases of colorectal neoplasia (adenoma or carcinoma) detected (Table 1), the sensitivity was highest with HOS (35% vs. 18% for FS, P < 0.05). The specificity and positive predictive values were significantly higher with FS (99% and 42% vs. 96 and 25% for HOS, P < 0.05).
Table 1. Analysis for Neoplasia (Adenomas of All Sizes or Carcinomas) by Each FOB Test and When Both HOS and FS Results Were Positive
|CI (95%)||CI (95%)||CI (95%)|
|Sensitivity (%)||35a||22, 47||18a||8, 28||13a||4, 22|
|Specificity (%)||96a||95, 97||99a||98, 100||100a||99, 100|
|False-positive (%)||4.2a||3.1, 5.3||1.0a||0.5, 1.6||0.3a||0.01, 0.6|
|False-negative (%)||66a||53, 78||82a||72, 92||87a||79, 96|
|Predicted positive (%)||25a||23, 27||42a||39, 44||64a||61, 66|
|Predicted negative (%)||97||97, 98||97||96, 98||97||96, 98|
Table 2. Analysis for Significant Neoplasia, Adenomas ≥ 1 cm or Carcinoma by Each FOB Test and When Both HOS and FS Were Positive
|CI (95%)||CI (95%)||CI (95%)|
|Sensitivity (%)||50a||28, 72||35||14, 56||25a||6, 44|
|Specificity (%)||95a||94, 96||99a||98, 99||100a||99, 100|
|False-positive (%)||4.9a||3.6, 5.9||1.2a||0.7, 1.8||0.4a||0.1, 0.8|
|False-negative (%)||50a||28, 72||65||44, 86||75a||56, 94|
|Predicted positive (%)||13a||11, 15||29a||27, 32||46a||43, 48|
|Predicted negative (%)||99||99, 100||99||99, 100||99||98, 100|
Analyzing the ability to detect the cases of clinically significant neoplasia, cancer, or adenomas ≥ 1.0 cm (Table 2), the sensitivities of both FOB tests were improved and not significantly different (50% for HOS and 35% for FS). HOS had the lower specificity (95%) and predictive positive values (13%) as compared with FS (99% and 29%, respectively, P < 0.05).
When excluding the cases having cancer, the sensitivity for adenomas of all sizes (not shown) was highest with HOS (33%) and lowest with FS (15%) (P < 0.05), but HOS had the lower specificity (96% vs. 99%, P < 0.05). Predictive positive value was highest with FS (29% vs. 21%, P < 0.05).
When both HOS and FS tests were positive in the same cases, an even higher specificity of 100% was obtained for all cases of neoplasia, but with a reduced sensitivity of 13–25% (Tables 1 and 2). This combination of tests decreased the number of adenomas detected (not shown). So, when limited to identifying only cancers or adenomas ≥ 1.0 cm (Table 2), the sensitivity of this combination was only 25% but the specificity enhanced to 100%; the predictive positive value positive became 46% and better than HOS alone (13%, P < 0.05).
To our knowledge, this large comparative study demonstrated for the first time that, even without dietary restrictions, the analytic sensitivity for clinically significant neoplasia of the sensitive guaiac HOS was not significantly different from that of the more specific immunochemical FS test. In addition, the HOS test identified more patients with adenomas, so its sensitivity for any colorectal neoplasia was significantly better than FS, but with significantly less specificity. It also demonstrated the limited clinical value of using a combination of both types of FOBTs.
The main strength of this study is that all the examinees underwent endoscopic examinations. Thus, the estimations of specificity are highly accurate. However, even though the population examined was asymptomatic, they are not at average risk for colorectal neoplasia but mainly at increased risk. Even so, because they were already in a long term screening/follow-up program, they were unlikely now to be harboring clinically significant neoplasia. For this reason, we included a small number of symptomatic persons (3%) to better mimic population screening and to obtain data on relative sensitivity.
The analytic sensitivity of HOS for large adenomas and cancer or even for any neoplasia was as good as or better than with FS. This is surprising because FS is a specific immunochemical test for human Hb and able to detect lower levels of fecal blood than HOS.9 However, this conclusion is based on in vitro studies, in which human blood is mixed with stool before smearing the test card. In vivo, the situation is not the same because the rate and degree of human blood degradation will clearly depend on factors such as transit time, differences between body and room temperature, and the different moisture content of intact stool and air-dried test cards, all of which could all affect the rate of Hb degradation by bacteria. The relatively poor sensitivity of FS for neoplasia in our ambulatory population setting was lower than that reported in symptomatic hospital-based colorectal carcinoma or adenoma cases, by using other immunochemical FOB tests.4, 12, 17 The differences might be related to the type of immunochemical test used and its ability to detect Hb or its products.9
This higher sensitivity of HOS for neoplasia is at the expense of significantly lower specificity than FS. Because FS is specific for human Hb, it has significantly better predictive negative and positive values than HOS. The lower specificity of HOS is unexplained and probably related to unidentified non-Hb sources of fecal peroxidases.
This dilemma, of choosing either FOB test sensitivity or specificity, is most acute when screening large size populations for colorectal carcinoma. In the recently completed randomized population screening studies, the positivity of the guaiac FOB test used (standard Hemoccult) varied from a high 9.8% (due to rehydration) to markedly low values of 0.6–2.1%.18–20 Even with the lowest positivity rates, within a multiyear screening program, there was a significant reduction in large bowel carcinoma mortality. The reduction was highest (33%) when using the most sensitive FOB test annually and so leading to a large number of colonscopic examinations, and lowest (15–18%) when using the less sensitive test biennially. Our one-time test with HOS identified only 50% of clinically significant neoplasia. However, a FOB screening program is not a one-time examination, and only by consistent retesting can the maximum program benefit be obtained.
Ransohoff and Lang noted that a FOB test specificity of less than 95% would lead to an excessive number of colonoscopic examinations.21 Within a population-screening program, this could be an expensive drain on the medical resources and increased iatrogenic morbidity due to the invasive diagnostic procedures. HOS specificity is at the border of this level. The alternative FS has a better specificity but a lower sensitivity for clinically significant neoplasia. Within an annual FOB screening program, a lower one-time test sensitivity might be acceptable by looking at overall program sensitivity, namely, the diagnostic-therapeutic results obtained over a period of some years. It remains to be proven that FS could significantly reduce colorectal carcinoma mortality and at what cost, because the immunochemical test is more expensive than the guaiac FOB tests. There is a favorable experience with a national screening program in Japan, using only an immunochemical test for human Hb, and because of their high intake of dietary peroxidases, it is believed to be better than if a guaiac test were to be used.22
Another possibility that is being considered, to improve the specificity of using HOS and so reduce screening costs, is to combine it with FS, which would only be developed if the HOS was positive, the “two-tier test.”8–11 The assumption is that most significant neoplasia would be identified by a positive HOS test and the false positivity would be reduced by confirmation with FS, a specific test for human Hb. This combination test is being examined prospectively in a population screening trial (Allison JE, personal communication). The analysis of our own results showed that the combination gave a significantly lower sensitivity for any colorectal neoplasm than with HOS alone but markedly improved its specificity. When considering sensitivity only for clinically significant neoplasia (large adenomas and cancer), it was significantly lower than that with HOS alone and FS alone (NS), but with maximum specificity. This was because of the lower sensitivity of FS for large or small adenomas So, we do not see any clinical advantage in this two-tier approach.
One of the disadvantages of HOS is its high sensitivity to dietary peroxidases.9, 13 We and others found that this high positivity could be reduced by delaying its development > 3 days after the FOB test card preparation.13, 14 In our population, this eliminated the need for severe dietary restrictions of almost 1-week duration each year. This is important because it could lead to reduced compliance for annual FOB retesting, which is required to obtain a maximum reduction in colorectal cancer mortality.18–20 This change removes one of the main advantages in using a more expensive human Hb-specific FOB test, which obviously is not sensitive to diet. However, note that our population has a low intake of red meat, which usually is koshered, i.e., the blood has been drained out. So, the recommendation to ignore dietary restriction by changing the timing of FOB test development requires confirmation from other countries having a higher red meat intake.
HOS is the most common FOB test used in Israel and is being considered for use in our National Screening Program, but without routine flexible sigmoidoscopy.23 We had to decide which approach is more practical, either to use the sensitive HOS test, which detects more neoplasia initially at a high cost financially and strain on medical resources, or to use a more gradual approach with the less sensitive but more specific FS annual immunochemical test. Currently, there is not enough long term experience with the immunochemical test to conclude that the program sensitivity will adequately reduce colorectal carcinoma mortality, and at a cost equivalent to, or less than that obtained with a sensitive guaiac test. Because the incidence of colorectal adenomas is lower in Israel than in the U.S. or Europe, and we have well developed endoscopy facilities, the more sensitive HOS test has been chosen.15
In conclusion, within an endoscopic screening program, we have shown that the sensitive guaiac HOS test, without dietary restrictions, had a better sensitivity for colorectal neoplasia than the specific immunochemical FS test. This was done by identifying more adenomas of all sizes. The specificity of HOS although lower than FS was still acceptable for our population screening program.
The authors thank Dr. M. Baratz for the pathology results, Dr. Y. Villa for the statistical evaluation, Ms. S. Zimmerman for the secretarial help, and Mr. R. Schoengold and Ms. Josefina Baker (California) for their advice.