- Email: [email protected]
Optimal Dietary Conditions for Hemoccult Testing
GASTROENTEROLOGY 1982 ;82 :899-903
Optimal Dietary Conditions for Hemoccult Testing FINLAY A. MACRAE, D. JAMES B. ST. JOHN, PAUL CALIGIORE, LEANNE S. TAYLOR, and JOHN W. LEGGE Departments of Gastroenterology and Nutrition, The Royal Melbourne Hospital and Department of Biochemistry, University of Melbourne, Melbourne, Victoria, Australia
Preliminary rehydration of Hemoccult II slides increases slide sensitivity for blood and lowers the false-negative rate for colorectal cancer. To test the effect of this modification on the false-positive rate due to peroxidase-containing foods, 156 healthy young subjects crossed between diets of differing peroxidase content. Hemoccult II slides were prepared in duplicate for testing with and without rehydration. When developed without rehydration, only seven (0.4%) of 1856 slides were positive, the positive results occurring with challenge diets containing 250 g rare red meat. With preliminary rehydration, 53 (5.7%) of 926 slides were positive, in 26 (17%) of the 156 subjects on a challenge diet that included rare red meat and uncooked fruit and vegetables. Well-cooked red meat also gave positive tests. Diets excluding red meat but with large quantities of uncooked fruit and vegetables produced only five positive results in 314 rehydrated tests, in 3 of 53 subjects. On a strict low-peroxidase diet, two of 310 tests, in 2 of 52 subjects, were positive. Rehydration of Hemoccult slides and exclusion of red meat and certain other high-peroxidase foods should give optimal sensitivity and specificity for colorectal cancer detection. The effect of these conditions on Received April 14, 1981. Accepted December 28, 1981. Address requests for ~eprints to: Dr. D.J.B. St. John, Department of Gastroenterology, The Royal Melbourne Hospital c/o The Royal Melbourne Hospital Post Office, Melbourne, Victoria 3050, Australia. Dr. F.A. Macrae was a Medical Postgraduate Research Scholar of the National Health and Medical Research Council of Australia. A portion of this study was presented to the meeting of the Gastroenterological Society of Australia in Noumea in October 1980.
This project was supported by the Anti-Cancer Council of Victoria. The authors thank the members of the staff of the University of Melbourne Department of Biochemistry and Dr. J.W. Wales for their support, Mr. J. Lee and Mrs. M. Clarke for their technical and secretarial assistance, and the 172 participants in the study for their enthusiasm and cooperation. © 1982 by the American Gastro~nterological Association. 0016/5085/82/050899-05$02.50
yield of lesions and false-positive rates should be tested in screening programs.
Although agreement has been reached about many technical aspects of Hemoccult testing (Smith Kline Diagnostics, Sunnyvale, Calif.), there is still no uniformity in the method of use of Hemoccult II slides in colorectal cancer screening programs (1). Although not currently recommended by the manufacturer, rehydration of slides before addition of peroxide developer solution is employed in some programs (2,3), although to a lesser extent in one of those programs (3). Dietary advice also varies, ranging from no restriction (4) through to exclusion of all types of meat (5) or red meat plus uncooked fruit and vegetables (6,7). Our previous study examined the relationship between bleeding from colorectal cancers and Hemoccult II slide sensitivity (8). Rehydration increased sensitivity of the slides for blood independently of storage delay, almost doubling it at the low levels of bleeding commonly encountered with cancers in the descending colon, sigmoid colon, and rectum. Testing over any 3-day period, the falsenegative rate for cancer fell from 31% to 9% when slides were rehydrated. Before rehydration can be accepted as an appropriate modification, it is necessary to exclude any adverse effect on specificity of the test for pathological bleeding, particularly with respect to the falsepositive rate due to peroxidase-containing foods. We therefore studied the influence of diets of differing peroxidase content on nonrehydrated and rehydrated Hemoccult II tests in healthy young subjects, to resolve the present uncertainty.
Methods A total of 172 healthy subjects under 40 yr of age enrolled in the study. Their mean age was 22.1 ± 4.4 yr (±SD). Subjects were only considered for inclusion if they
GASTROENTEROLOGY Vol. 82, No.5, Part 1
900 MACRAE ET AL.
Table 1. Diets Used in the Study Meat
Fruit
Vegetables
Full-challenge diet
250 g rare rump steak/day
3 pieces of selected fresh fruiUday plus cantaloupe or grapefruit
Rare red meat, but no fresh fruit or vegetables Fresh fruit and vegetables, but no red meat Strict low-peroxidase dieta
250 g rare rump steak/day
No uncooked fruit except oranges or strawberries As with full-challenge diet No uncooked fruit except oranges or strawberries
3 of the following uncooked vegetables/day: 1 carrot, 1 small cucumber, 2 servings green salad, cauliflower Cooked vegetables only
Diet
a
Chicken, ham, pork, bacon, white fish Chicken, ham, pork, bacon, white fish
As with full-challenge diet Cooked vegetables only
Subjects were instructed to take bran, whole-grain bread, and peanuts daily while on the strict low-peroxidase diet.
had no history of bleeding hemorrhoids or other significant gastrointestinal disease. None was taking antiinflammatory or analgesic drugs. Each received both verbal and written instruction on the purpose of the study. The protocol was approved by the hospital Board of Medical Research and the university Human Experimentation Committee. All of the subjets had medical or paramedical training, and all gave free and informed consent. Sixteen of the 172 later withdrew because of failure to strictly adhere to the diets. Each subject followed a defined diet for 6 days and then followed a second diet for a further 6 days; a record was kept of items of food consumed. Meat, fruit, and vegetables were provided free of charge. Four different diets were prescribed (Table 1). Foods were selected for their high or low peroxidase content as determined by measurement of peroxidase activity (9). The peroxidase-rich foods included cantaloupe and cauliflower (5-10 g having a peroxidase activity equivalent to 1.0 ml blood), cucumber (10...,.20 g equivalent to 1.0 ml blood), and grapefruit and carrot (2050 g equivalent to 1.0 rp.l blood). Orange and strawberries have no detectable peroxidase (9) and were allowed in the strict low-peroxidase diet. The quantities chosen were considered to be the maximum that could reasonably be consumed each day by the participants, 74 of whom were women. All participants had access to cooking facilities and personally supervised preparation of their own food. Ascorbic-acid fortified products (10) and salicylates or other analgesics weie forbidden during the study. In the first section of the study, 52 hospital staff members (group 1) followed a strict low-peroxidase diet for 6 days and then crossed to the full-challenge diet of rare red meat and fresh fruit and vegetables. In the second section, 104 medical students were randomly assigned to one of two groups. Of 51 students in group 2, 26 followed the fullchallenge diet and then crossed to the diet containing rare red meat but no fresh fruit or vegetables. The other 25 had these diets in reverse order. Of the 53 students in group 3, 28 started with the full-challenge diet and crossed to the diet of fresh fruit and vegetables without red meat, and 25 followed these diets in reverse order. Women arranged their participation to avoid the time of menstruation. Eleven subjects who returned positive Hemoccult results on a rare red meat diet were requested to extend their
study for 6 days, this time taking well-cooked red meat. Duplicate sets of Hemoccult II slides were prepared on the last 3 days of each 6-day dietary period, for subsequent testing with and without rehydration. Stool specimens were collected without coming in contact with water in the toilet bowl. Two boxes on each of two Hemoccult cards were smeared each day. Instructions were given to ensure that the same stool sample was used to smear corresponding boxes of the duplicate cards. The subjects were asked to test the first, if any, bowel action each' day. If constipated, the diet was extended to allow the cards to be completed. Cards were returned to the laboratory for testing within 5 days of preparation. Rehydration was performed by adding one drop of water to the under-surface of each box. Sufficient time was allowed for the water to fully penetrate the guaiac-impregnated paper before addition of the peroxide developer solution. Masked slides were read independently by two observers who were also reading masked slides from patients with colorectal disease. Results were recorded as po~itive only if both observers agreed. It was not possible to mask rehydrated from nonrehydrated slides, but a colorchange chart was used to standardize readings semiquantitatively. Hemoccult II batches were sensitivity-tested in vitro for between-batch quality control (11). The minimal limits of sensitivity were determined by testing with aqueous preparations of bovine hemoglobin and human whole blood lysates. No differences were found. All subjects in group 1 retrieved additional stool samples on the last 3 days of each dietary period for quantitative measurement of fecal peroxidase activity. Samples were stored in airtight containers. Peroxidase activity was measured in aqueous extracts of the samples within 5 days, by the Worthington method (12). Five grams of stool in a small volume of distilled water was ground to a uniform consistency and water was added to a total volume of 25 ml. followed by 1.0 ml of anesthetic grade diethylether to disrupt cell membranes and release intracellular proteins. There was no evidence from the rate measurements that any traces of the antioxidants added in the ether affected the assay. After standing for 60 min, the extracts were centrifuged at 20,000 g for 10 min. Volumes of 0.1 ml of supernatant were added to a cuvette containing 1.5 ml of 1.7 mM H2 0 2 in 0.2 M phosphate buffer (pH
DIET FOR OCCULT BLOOD TESTING 901
May 1982
7.0), and 1.4 ml of 2.5 mM 4-amino antipyrine and 0.17M phenol in water. The rate of increase in absorbance at 510 nm was then followed in a recording spectrophotometer. A similar aliquot (0.1 ml) of 100-fold dilution of blood (Hb 140-160 gIL) was run as a reference with each batch of samples. It was assumed that the initial rates were proportional to the amount of hemoglobin displaying pseudoperoxidase activity, 1000 times the measured activity being proportional to 1.0 ml blood. Results were expressed as the concentration of hemoglobin that would give the same peroxidase activity as measured in the stool samples. The significance of differences in the proportions of positive Hemoccult results was assessed by testing for differences in proportions using a one-tailed test (13).
Results The results are shown in Table 2. Within each of the three groups, Hemoccult results have been combined for subjects following the same diet, irrespective of the order of cross-over. Comparison of Diets With nonrehydrated slides, positive tests were infrequent (0.4%) and occurred only with challenge diets containing rare red meat. With preliminary rehydration of slides, significantly more tests were positive (p < 0.001) on the full-challenge diet, compared with the strict lowperoxidase diet (group 1). When fresh fruit and vegetables were omitted from the challenge diet (group 2) the proportion of positive tests remained unchanged. Exclusion of red meat from the challenge diet (group 3) led to a significant (p < 0.001) fall from 6.6% to 1.6% in Hemoccult positivity. The numbers of positive slides on the strict low-peroxidase diet and the diet of fresh fruit and vegetables but no red meat were two and five, respectively, too small for valid statistical comparison. Analyzing results for individual subjects rather than tests, statistical differences noted between diets for each group were: group 1, p < 0.01; group 2, not significant; group 3, p < 0.05. Comparison of Methods of Development Comparing tests with and without rehydration, significant increases (p < 0.001) occurred with rehydration for diets that included rare red meat. There were only small numbers of positive tests with each method of development for diets excluding red meat. Of 11 subjects who had positive Hemoccult results when eating rare red meat, 4 had positive rehydrated tests and 1 a positive nonrehydrated test when on well-cooked red meat.
Table 2.
Summary of Hemoccult Results on 156 Normal Subjects on Specified Diets
Group 1: 52 subjects Full-challenge Strict low-peroxidase Group 2: 51 subjects Full-challenge Rare red meat, but no fresh fruit or vegetables Group 3: 53 subjects Full-challenge Fresh fruit and vegetables , but no red meat Group 1, 2, and 3 combined: 156 subjects Full-challenge
Hemoccult n°
Rehydrated Hemoccult lIb
0/306 c (O)d 0% 0/310 (0) 0%
13/306 (8) 4.2% 2/310 (2) 0.6%
2/304 (1) 0.7% 2/304 (1) 0.7%
19/304 (9) 6.3% 20/304 (11) 6.6%
3/316 (2) 0.9% 0/3 16 (0) 0%
21/316 (9) 6.6% 5/314 (3) 1.6%
5/926 (3) 0.5%
53/926 (26) 5.7%
°Hemoccult n developed without rehydration.
b Hemoccult II developed with preliminary rehydration. C Proportion of individual tests that were positive. d Number of individual subjects, shown in parentheses, with one or more positive tests.
Assay of Stools
Direct assay of stools showed no detectable peroxidase activity in 154 of 155 stool samples from the 52 subjects following the strict low-peroxidase diet. In subjects receiving the full-challenge diet, mean peroxidase activity was equivalent to 0.32 mg Hb/g stool (range 0-2.92). In the subgroup on wellcooked red meat, fecal peroxidase was detected in five of 24 samples, from 3 of 8 subjects (mean 0.30 mg Hb/g stool). Excluding a subject with a single sample containing 4.85 mg Hb/g stool, the mean fecal peroxidase was 0.11 mg Hb/g stool (range 0-0.86).
Discussion The confusion about dietary recommendations during guaiac-based fecal occult blood testing stems from the paucity of clinical studies of the effects of diet on test results, and the lack of information about peroxidase content of foods and resultant effects on fecal peroxidase activity. The practical consequences of this uncertainty have become much greater with the introduction of fecal occult blood testing as a method for mass screening for colorectal cancer. Small increments in the rate of positivity due to peroxidase-containing
902
MACRAE ET AL.
foods may lead to unnecessary inconvenience, expense, and even morbidity as a result of the diagnostic procedures required for follow-up of positive tests. However, insistence on too rigorous a diet may adversely affect participation rates in screening programs. The strategy of first testing without diet and then repeating any positive tests on a restricted diet carries the risk of missing lesions that are bleeding intermittently, as well as providing another opportunity for default in compliance. In the present study, well-motivated healthy young subjects were used to assess the effects of diet on Hemoccult results. It was considered that young people without gastrointestinal symptoms would be unlikely to have occult gastrointestinal bleeding. Also, the cross-over design should have offset any bias introduced by inclusion of subjects with unsuspected occult bleeding. Whether younger subjects digest peroxidase any differently than older subjects is unknown. In the absence of any information to the contrary, we have assumed that observations on the rate of positive Hemoccult tests in our study groups are relevant to people more than 40 yr of age. The concept of challenge diets was introduced to maximize any differences in results between diets and to simulate conditions that could be encountered with subjects who eat large quantities of peroxidase-containing foods. Indeed, the level of fecal peroxidase activity on the full-challenge diet frequently was in the range expected for the level of bleeding encountered with cancers in the descending colon, sigmoid colon, and rectum (8). The absence of peroxidase activity in all but one specimen from subjects following the strict low-peroxidase diet confirmed compliance to the diet as well as the absence of bleeding in that group of subjects. It also suggested that peroxidase activity, as measured by Hemoccult testing and spectrophotometry, is more dependent on diet than on intestinal micro flora in the individual subject. Dietary fiber content is difficult to quantitate biochemically, and measurement using promising radiolabeling techniques has only been applied to animal models (14). Nevertheless, the inclusion of large quantities of fresh fruit and vegetables in two of our challenge diets almost certainly produced a higher fiber content in those diets. The possibility that the positive tests in subjects taking fresh fruit and vegetables were due to bleeding from mucosal abrasions produced by roughage cannot be refuted. However, it should be noted that there was no significant difference in the rates of positivity for rehydrated slides in the group crossing between the rare red meat diet (6.6% tests positive) and the fullchallenge diet (6.3% tests positive). In addition, the subjects following the strict low-peroxidase diet
GASTROENTEROLOGY Vol. 82, No.5, Part 1
were asked to include bran, whole-meal bread, and peanuts in the diet. No peroxidase activity was measurable quantitatively in their stools, suggesting that at least this type of roughage is nonabrasive in healthy young subjects. In a recent study using Hemoccult I slides developed without rehydration, significantly more positive results were attributed to an unrestricted diet compared with a strict low-peroxidase diet (7). The investigation was not cross-over in design and involved patients with colarectal lesions. Radiochromium measurements of overall gastrointestinal blood loss were used to separate normal from pathological bleeding. Hemoccult may detect pathological bleeding from colorectallesions even though overall 1 5 Cr-Iabeled blood loss falls within the so-called "normal" range (8). In clinical studies involving patients with colorectal lesions, the false-positive rate should therefore not be equated with the proportion of tests positive in the range 0-2 mg Hb/g stool. Several studies have found a minimal effect of diet on nonrehydrated guaiac slides. Hemoccult tests were negative in a single volunteer who ate 1 kg of rare beef (15). Positive results were obtained with the guaiac Fecatest (Finnpipette Ky; Helsinki, Finland) in 7 subjects who consumed 480 g red meat daily, but none were recorded when those subjects had 240 g daily (16). Hemoccult results were rarely positive in 20 healthy volunteers having an unspecified quantity of meat (17). In the present study, 156 subjects on the full-challenge diet, which included 250 g rare red meat daily for 6 days, had only 0.5% positive nonrehydrated Hemoccult tests. This suggests that individuals with positive results in programs conducted without dietary restriction, and without rehydration of Hemoccult slides, should all be investigated for a source of bleeding. With rehydrated slides, our findings show that inclusion of red meat in the diet is unacceptable. Chicken, pork, bacon, ham, and white fish are acceptable as fecal peroxidase was undetectable in 51 of the 52 subjects on the strict low-peroxidase diet, and only 0.6% of slides were positive. Rehydration of Hemoccult slides had little effect on the rate of positivity in the group on large quantities of uncooked fruit and vegetables, as well as in the group on the strict low-peroxidase diet of white meat and cooked fruit and vegetables. Statistical comparison of the results of rehydrated tests for the two groups was inappropriate because of the small number of positive tests and the absence of cross-over between these diets. Another study would have to be performed to determine whether the slight difference was of significance. The subjects in the challenge group were having much more uncooked fruit and vegetables than would be expected in usual circum-
May 1982
stances. When Hemoccult tests are developed with rehydration it would therefore be appropriate to allow most uncooked fruit and vegetables in moderate amounts. However, it would be prudent to exclude individual items, such as uncooked radish, horseradish, cantaloupe, and cauliflower, which have particularly high peroxidase contents. A mass of 5-10 g of each of these items has a peroxidase activity equivalent to 1.0 ml blood (9). Data about the effect of dietary exclusions on compliance in screening programs are conflicting and relate only to exclusion of meat (18,19). Until more information becomes available, the optimal strategy might be to recommend the least restrictive diet compatible with an acceptable test specificity, particularly where subjects are not self-selected.
References 1. Winawer SJ. The complexities of a "simple test." Gastroenter-
ology 1979;77:193-4. 2. Gilbertsen VA. McHugh R. Schuman L. et al. The earlier detection of colorectal cancers: a preliminary report of the results of the occult blood study. Cancer 1980;45:2899-901. 3. Winawer SJ, Andrews M. Flehinger B. et al. Progress report on controlled trial of fecal occult blood testing for the detection of colorectal neoplasia. Cancer 1980;45:2959-64. 4. Kurnick JE. Walley LB . Jacob HH. et al. Colorectal cancer detection in a community hospital screening program. JAMA 1980;243:2056-7. 5. Greegor DH. Diagnosis of large-bowel cancer in the asymptomatic patient. JAMA 1967;201:943- 5. 6. Illingworth DG. Influence of diet on occult blood tests. Gut 1965;6:595-8. 7. Bassett ML. Goulston KJ. False positive and negative Hemoccult reactions on a normal diet and effect of diet restriction. Aust NZ J Med 1980;10:1-4.
DIET FOR OCCULT BLOOD TESTING 903
8. Macrae FA. St. John DJB . Relationship between patterns of bleeding and Hemoccult sensitivity in patients with colorectal cancers or adenomas. Gastroenterology; 1982;82:891-8. 9. Caligiore p. Macrae FA. St. John DJB. et al. Peroxidase levels in food: relevance to colorectal cancer screening. Am J Clin Nutr (in press). 10. Jaffe RM. Kasten B. Young DS. et al. False-negative stool occult blood tests caused by ingestion of ascorbic acid (vitamin C). Ann Intern Med 1975;83:824-6. 11 . Fleisher M. Schwartz MK. Winawer SJ. Laboratory studies on the Hemoccult slide for fecal occult blood testing. In: Winawer S. Schottenfeld D. Sherlock P. eds. Colorectal cancer: prevention. epidemiology and screening. New York: Raven. 1980:181-7. 12. Worthington Biochemical Corporation. Peroxidase. In: Enzymes and related biochemicals. Freehold. New Jersey: Worthington. 1978:145-6. 13. Walpole RE. Myers RH. Probability and statistics for engineers and scientists. 2nd ed. New York: Collier Macmillan International Editions. 1978. 14. Malagelada J-R. Carter SE. Brown ML. et al. Radiolabeled fiber: aphysiologic marker for gastric emptying and intestinal transit of solids. Dig Dis Sci 1980;25:81-7 15. Rosenfield RE. Kochwa S. Kaczera Z. et al. Nonuniform distribution of occult blood in feces. Am J Clin Pathol 1979;71 :204-9. 16. Brault J. Favre H. Evaluation en pratique hospitaliere du Fecatest. un nouvel examen de laboratoire pour la recherche de sang occulte dans les selles. Schweiz Med Wochenschr 1979;109:73-6. 17. Ostrow JD. Mulvaney CA. Hansell JR. et al. Sensitivity and reproducibility of chemical tests for fecal occult blood with an emphasis on false-positive reactions. Am J Dig Dis 1973;18:930-40. 18. Elwood TW. Erickson A. Leiberman S. Comparative educational approaches to screening for colorectal cancer. Am J Public Health 1978;68:135-8. 19. Feifel G. Manner C. v. Liebe S. The Haemoccult test without dietary restriction. In: Goerttler K. ed. Early detection of colorectal cancer. Nurnberg: Verlog DE. Wachholz KG. 1980:111-6.
Optimal Dietary Conditions for Hemoccult Testing FINLAY A. MACRAE, D. JAMES B. ST. JOHN, PAUL CALIGIORE, LEANNE S. TAYLOR, and JOHN W. LEGGE Departments of Gastroenterology and Nutrition, The Royal Melbourne Hospital and Department of Biochemistry, University of Melbourne, Melbourne, Victoria, Australia
Preliminary rehydration of Hemoccult II slides increases slide sensitivity for blood and lowers the false-negative rate for colorectal cancer. To test the effect of this modification on the false-positive rate due to peroxidase-containing foods, 156 healthy young subjects crossed between diets of differing peroxidase content. Hemoccult II slides were prepared in duplicate for testing with and without rehydration. When developed without rehydration, only seven (0.4%) of 1856 slides were positive, the positive results occurring with challenge diets containing 250 g rare red meat. With preliminary rehydration, 53 (5.7%) of 926 slides were positive, in 26 (17%) of the 156 subjects on a challenge diet that included rare red meat and uncooked fruit and vegetables. Well-cooked red meat also gave positive tests. Diets excluding red meat but with large quantities of uncooked fruit and vegetables produced only five positive results in 314 rehydrated tests, in 3 of 53 subjects. On a strict low-peroxidase diet, two of 310 tests, in 2 of 52 subjects, were positive. Rehydration of Hemoccult slides and exclusion of red meat and certain other high-peroxidase foods should give optimal sensitivity and specificity for colorectal cancer detection. The effect of these conditions on Received April 14, 1981. Accepted December 28, 1981. Address requests for ~eprints to: Dr. D.J.B. St. John, Department of Gastroenterology, The Royal Melbourne Hospital c/o The Royal Melbourne Hospital Post Office, Melbourne, Victoria 3050, Australia. Dr. F.A. Macrae was a Medical Postgraduate Research Scholar of the National Health and Medical Research Council of Australia. A portion of this study was presented to the meeting of the Gastroenterological Society of Australia in Noumea in October 1980.
This project was supported by the Anti-Cancer Council of Victoria. The authors thank the members of the staff of the University of Melbourne Department of Biochemistry and Dr. J.W. Wales for their support, Mr. J. Lee and Mrs. M. Clarke for their technical and secretarial assistance, and the 172 participants in the study for their enthusiasm and cooperation. © 1982 by the American Gastro~nterological Association. 0016/5085/82/050899-05$02.50
yield of lesions and false-positive rates should be tested in screening programs.
Although agreement has been reached about many technical aspects of Hemoccult testing (Smith Kline Diagnostics, Sunnyvale, Calif.), there is still no uniformity in the method of use of Hemoccult II slides in colorectal cancer screening programs (1). Although not currently recommended by the manufacturer, rehydration of slides before addition of peroxide developer solution is employed in some programs (2,3), although to a lesser extent in one of those programs (3). Dietary advice also varies, ranging from no restriction (4) through to exclusion of all types of meat (5) or red meat plus uncooked fruit and vegetables (6,7). Our previous study examined the relationship between bleeding from colorectal cancers and Hemoccult II slide sensitivity (8). Rehydration increased sensitivity of the slides for blood independently of storage delay, almost doubling it at the low levels of bleeding commonly encountered with cancers in the descending colon, sigmoid colon, and rectum. Testing over any 3-day period, the falsenegative rate for cancer fell from 31% to 9% when slides were rehydrated. Before rehydration can be accepted as an appropriate modification, it is necessary to exclude any adverse effect on specificity of the test for pathological bleeding, particularly with respect to the falsepositive rate due to peroxidase-containing foods. We therefore studied the influence of diets of differing peroxidase content on nonrehydrated and rehydrated Hemoccult II tests in healthy young subjects, to resolve the present uncertainty.
Methods A total of 172 healthy subjects under 40 yr of age enrolled in the study. Their mean age was 22.1 ± 4.4 yr (±SD). Subjects were only considered for inclusion if they
GASTROENTEROLOGY Vol. 82, No.5, Part 1
900 MACRAE ET AL.
Table 1. Diets Used in the Study Meat
Fruit
Vegetables
Full-challenge diet
250 g rare rump steak/day
3 pieces of selected fresh fruiUday plus cantaloupe or grapefruit
Rare red meat, but no fresh fruit or vegetables Fresh fruit and vegetables, but no red meat Strict low-peroxidase dieta
250 g rare rump steak/day
No uncooked fruit except oranges or strawberries As with full-challenge diet No uncooked fruit except oranges or strawberries
3 of the following uncooked vegetables/day: 1 carrot, 1 small cucumber, 2 servings green salad, cauliflower Cooked vegetables only
Diet
a
Chicken, ham, pork, bacon, white fish Chicken, ham, pork, bacon, white fish
As with full-challenge diet Cooked vegetables only
Subjects were instructed to take bran, whole-grain bread, and peanuts daily while on the strict low-peroxidase diet.
had no history of bleeding hemorrhoids or other significant gastrointestinal disease. None was taking antiinflammatory or analgesic drugs. Each received both verbal and written instruction on the purpose of the study. The protocol was approved by the hospital Board of Medical Research and the university Human Experimentation Committee. All of the subjets had medical or paramedical training, and all gave free and informed consent. Sixteen of the 172 later withdrew because of failure to strictly adhere to the diets. Each subject followed a defined diet for 6 days and then followed a second diet for a further 6 days; a record was kept of items of food consumed. Meat, fruit, and vegetables were provided free of charge. Four different diets were prescribed (Table 1). Foods were selected for their high or low peroxidase content as determined by measurement of peroxidase activity (9). The peroxidase-rich foods included cantaloupe and cauliflower (5-10 g having a peroxidase activity equivalent to 1.0 ml blood), cucumber (10...,.20 g equivalent to 1.0 ml blood), and grapefruit and carrot (2050 g equivalent to 1.0 rp.l blood). Orange and strawberries have no detectable peroxidase (9) and were allowed in the strict low-peroxidase diet. The quantities chosen were considered to be the maximum that could reasonably be consumed each day by the participants, 74 of whom were women. All participants had access to cooking facilities and personally supervised preparation of their own food. Ascorbic-acid fortified products (10) and salicylates or other analgesics weie forbidden during the study. In the first section of the study, 52 hospital staff members (group 1) followed a strict low-peroxidase diet for 6 days and then crossed to the full-challenge diet of rare red meat and fresh fruit and vegetables. In the second section, 104 medical students were randomly assigned to one of two groups. Of 51 students in group 2, 26 followed the fullchallenge diet and then crossed to the diet containing rare red meat but no fresh fruit or vegetables. The other 25 had these diets in reverse order. Of the 53 students in group 3, 28 started with the full-challenge diet and crossed to the diet of fresh fruit and vegetables without red meat, and 25 followed these diets in reverse order. Women arranged their participation to avoid the time of menstruation. Eleven subjects who returned positive Hemoccult results on a rare red meat diet were requested to extend their
study for 6 days, this time taking well-cooked red meat. Duplicate sets of Hemoccult II slides were prepared on the last 3 days of each 6-day dietary period, for subsequent testing with and without rehydration. Stool specimens were collected without coming in contact with water in the toilet bowl. Two boxes on each of two Hemoccult cards were smeared each day. Instructions were given to ensure that the same stool sample was used to smear corresponding boxes of the duplicate cards. The subjects were asked to test the first, if any, bowel action each' day. If constipated, the diet was extended to allow the cards to be completed. Cards were returned to the laboratory for testing within 5 days of preparation. Rehydration was performed by adding one drop of water to the under-surface of each box. Sufficient time was allowed for the water to fully penetrate the guaiac-impregnated paper before addition of the peroxide developer solution. Masked slides were read independently by two observers who were also reading masked slides from patients with colorectal disease. Results were recorded as po~itive only if both observers agreed. It was not possible to mask rehydrated from nonrehydrated slides, but a colorchange chart was used to standardize readings semiquantitatively. Hemoccult II batches were sensitivity-tested in vitro for between-batch quality control (11). The minimal limits of sensitivity were determined by testing with aqueous preparations of bovine hemoglobin and human whole blood lysates. No differences were found. All subjects in group 1 retrieved additional stool samples on the last 3 days of each dietary period for quantitative measurement of fecal peroxidase activity. Samples were stored in airtight containers. Peroxidase activity was measured in aqueous extracts of the samples within 5 days, by the Worthington method (12). Five grams of stool in a small volume of distilled water was ground to a uniform consistency and water was added to a total volume of 25 ml. followed by 1.0 ml of anesthetic grade diethylether to disrupt cell membranes and release intracellular proteins. There was no evidence from the rate measurements that any traces of the antioxidants added in the ether affected the assay. After standing for 60 min, the extracts were centrifuged at 20,000 g for 10 min. Volumes of 0.1 ml of supernatant were added to a cuvette containing 1.5 ml of 1.7 mM H2 0 2 in 0.2 M phosphate buffer (pH
DIET FOR OCCULT BLOOD TESTING 901
May 1982
7.0), and 1.4 ml of 2.5 mM 4-amino antipyrine and 0.17M phenol in water. The rate of increase in absorbance at 510 nm was then followed in a recording spectrophotometer. A similar aliquot (0.1 ml) of 100-fold dilution of blood (Hb 140-160 gIL) was run as a reference with each batch of samples. It was assumed that the initial rates were proportional to the amount of hemoglobin displaying pseudoperoxidase activity, 1000 times the measured activity being proportional to 1.0 ml blood. Results were expressed as the concentration of hemoglobin that would give the same peroxidase activity as measured in the stool samples. The significance of differences in the proportions of positive Hemoccult results was assessed by testing for differences in proportions using a one-tailed test (13).
Results The results are shown in Table 2. Within each of the three groups, Hemoccult results have been combined for subjects following the same diet, irrespective of the order of cross-over. Comparison of Diets With nonrehydrated slides, positive tests were infrequent (0.4%) and occurred only with challenge diets containing rare red meat. With preliminary rehydration of slides, significantly more tests were positive (p < 0.001) on the full-challenge diet, compared with the strict lowperoxidase diet (group 1). When fresh fruit and vegetables were omitted from the challenge diet (group 2) the proportion of positive tests remained unchanged. Exclusion of red meat from the challenge diet (group 3) led to a significant (p < 0.001) fall from 6.6% to 1.6% in Hemoccult positivity. The numbers of positive slides on the strict low-peroxidase diet and the diet of fresh fruit and vegetables but no red meat were two and five, respectively, too small for valid statistical comparison. Analyzing results for individual subjects rather than tests, statistical differences noted between diets for each group were: group 1, p < 0.01; group 2, not significant; group 3, p < 0.05. Comparison of Methods of Development Comparing tests with and without rehydration, significant increases (p < 0.001) occurred with rehydration for diets that included rare red meat. There were only small numbers of positive tests with each method of development for diets excluding red meat. Of 11 subjects who had positive Hemoccult results when eating rare red meat, 4 had positive rehydrated tests and 1 a positive nonrehydrated test when on well-cooked red meat.
Table 2.
Summary of Hemoccult Results on 156 Normal Subjects on Specified Diets
Group 1: 52 subjects Full-challenge Strict low-peroxidase Group 2: 51 subjects Full-challenge Rare red meat, but no fresh fruit or vegetables Group 3: 53 subjects Full-challenge Fresh fruit and vegetables , but no red meat Group 1, 2, and 3 combined: 156 subjects Full-challenge
Hemoccult n°
Rehydrated Hemoccult lIb
0/306 c (O)d 0% 0/310 (0) 0%
13/306 (8) 4.2% 2/310 (2) 0.6%
2/304 (1) 0.7% 2/304 (1) 0.7%
19/304 (9) 6.3% 20/304 (11) 6.6%
3/316 (2) 0.9% 0/3 16 (0) 0%
21/316 (9) 6.6% 5/314 (3) 1.6%
5/926 (3) 0.5%
53/926 (26) 5.7%
°Hemoccult n developed without rehydration.
b Hemoccult II developed with preliminary rehydration. C Proportion of individual tests that were positive. d Number of individual subjects, shown in parentheses, with one or more positive tests.
Assay of Stools
Direct assay of stools showed no detectable peroxidase activity in 154 of 155 stool samples from the 52 subjects following the strict low-peroxidase diet. In subjects receiving the full-challenge diet, mean peroxidase activity was equivalent to 0.32 mg Hb/g stool (range 0-2.92). In the subgroup on wellcooked red meat, fecal peroxidase was detected in five of 24 samples, from 3 of 8 subjects (mean 0.30 mg Hb/g stool). Excluding a subject with a single sample containing 4.85 mg Hb/g stool, the mean fecal peroxidase was 0.11 mg Hb/g stool (range 0-0.86).
Discussion The confusion about dietary recommendations during guaiac-based fecal occult blood testing stems from the paucity of clinical studies of the effects of diet on test results, and the lack of information about peroxidase content of foods and resultant effects on fecal peroxidase activity. The practical consequences of this uncertainty have become much greater with the introduction of fecal occult blood testing as a method for mass screening for colorectal cancer. Small increments in the rate of positivity due to peroxidase-containing
902
MACRAE ET AL.
foods may lead to unnecessary inconvenience, expense, and even morbidity as a result of the diagnostic procedures required for follow-up of positive tests. However, insistence on too rigorous a diet may adversely affect participation rates in screening programs. The strategy of first testing without diet and then repeating any positive tests on a restricted diet carries the risk of missing lesions that are bleeding intermittently, as well as providing another opportunity for default in compliance. In the present study, well-motivated healthy young subjects were used to assess the effects of diet on Hemoccult results. It was considered that young people without gastrointestinal symptoms would be unlikely to have occult gastrointestinal bleeding. Also, the cross-over design should have offset any bias introduced by inclusion of subjects with unsuspected occult bleeding. Whether younger subjects digest peroxidase any differently than older subjects is unknown. In the absence of any information to the contrary, we have assumed that observations on the rate of positive Hemoccult tests in our study groups are relevant to people more than 40 yr of age. The concept of challenge diets was introduced to maximize any differences in results between diets and to simulate conditions that could be encountered with subjects who eat large quantities of peroxidase-containing foods. Indeed, the level of fecal peroxidase activity on the full-challenge diet frequently was in the range expected for the level of bleeding encountered with cancers in the descending colon, sigmoid colon, and rectum (8). The absence of peroxidase activity in all but one specimen from subjects following the strict low-peroxidase diet confirmed compliance to the diet as well as the absence of bleeding in that group of subjects. It also suggested that peroxidase activity, as measured by Hemoccult testing and spectrophotometry, is more dependent on diet than on intestinal micro flora in the individual subject. Dietary fiber content is difficult to quantitate biochemically, and measurement using promising radiolabeling techniques has only been applied to animal models (14). Nevertheless, the inclusion of large quantities of fresh fruit and vegetables in two of our challenge diets almost certainly produced a higher fiber content in those diets. The possibility that the positive tests in subjects taking fresh fruit and vegetables were due to bleeding from mucosal abrasions produced by roughage cannot be refuted. However, it should be noted that there was no significant difference in the rates of positivity for rehydrated slides in the group crossing between the rare red meat diet (6.6% tests positive) and the fullchallenge diet (6.3% tests positive). In addition, the subjects following the strict low-peroxidase diet
GASTROENTEROLOGY Vol. 82, No.5, Part 1
were asked to include bran, whole-meal bread, and peanuts in the diet. No peroxidase activity was measurable quantitatively in their stools, suggesting that at least this type of roughage is nonabrasive in healthy young subjects. In a recent study using Hemoccult I slides developed without rehydration, significantly more positive results were attributed to an unrestricted diet compared with a strict low-peroxidase diet (7). The investigation was not cross-over in design and involved patients with colarectal lesions. Radiochromium measurements of overall gastrointestinal blood loss were used to separate normal from pathological bleeding. Hemoccult may detect pathological bleeding from colorectallesions even though overall 1 5 Cr-Iabeled blood loss falls within the so-called "normal" range (8). In clinical studies involving patients with colorectal lesions, the false-positive rate should therefore not be equated with the proportion of tests positive in the range 0-2 mg Hb/g stool. Several studies have found a minimal effect of diet on nonrehydrated guaiac slides. Hemoccult tests were negative in a single volunteer who ate 1 kg of rare beef (15). Positive results were obtained with the guaiac Fecatest (Finnpipette Ky; Helsinki, Finland) in 7 subjects who consumed 480 g red meat daily, but none were recorded when those subjects had 240 g daily (16). Hemoccult results were rarely positive in 20 healthy volunteers having an unspecified quantity of meat (17). In the present study, 156 subjects on the full-challenge diet, which included 250 g rare red meat daily for 6 days, had only 0.5% positive nonrehydrated Hemoccult tests. This suggests that individuals with positive results in programs conducted without dietary restriction, and without rehydration of Hemoccult slides, should all be investigated for a source of bleeding. With rehydrated slides, our findings show that inclusion of red meat in the diet is unacceptable. Chicken, pork, bacon, ham, and white fish are acceptable as fecal peroxidase was undetectable in 51 of the 52 subjects on the strict low-peroxidase diet, and only 0.6% of slides were positive. Rehydration of Hemoccult slides had little effect on the rate of positivity in the group on large quantities of uncooked fruit and vegetables, as well as in the group on the strict low-peroxidase diet of white meat and cooked fruit and vegetables. Statistical comparison of the results of rehydrated tests for the two groups was inappropriate because of the small number of positive tests and the absence of cross-over between these diets. Another study would have to be performed to determine whether the slight difference was of significance. The subjects in the challenge group were having much more uncooked fruit and vegetables than would be expected in usual circum-
May 1982
stances. When Hemoccult tests are developed with rehydration it would therefore be appropriate to allow most uncooked fruit and vegetables in moderate amounts. However, it would be prudent to exclude individual items, such as uncooked radish, horseradish, cantaloupe, and cauliflower, which have particularly high peroxidase contents. A mass of 5-10 g of each of these items has a peroxidase activity equivalent to 1.0 ml blood (9). Data about the effect of dietary exclusions on compliance in screening programs are conflicting and relate only to exclusion of meat (18,19). Until more information becomes available, the optimal strategy might be to recommend the least restrictive diet compatible with an acceptable test specificity, particularly where subjects are not self-selected.
References 1. Winawer SJ. The complexities of a "simple test." Gastroenter-
ology 1979;77:193-4. 2. Gilbertsen VA. McHugh R. Schuman L. et al. The earlier detection of colorectal cancers: a preliminary report of the results of the occult blood study. Cancer 1980;45:2899-901. 3. Winawer SJ, Andrews M. Flehinger B. et al. Progress report on controlled trial of fecal occult blood testing for the detection of colorectal neoplasia. Cancer 1980;45:2959-64. 4. Kurnick JE. Walley LB . Jacob HH. et al. Colorectal cancer detection in a community hospital screening program. JAMA 1980;243:2056-7. 5. Greegor DH. Diagnosis of large-bowel cancer in the asymptomatic patient. JAMA 1967;201:943- 5. 6. Illingworth DG. Influence of diet on occult blood tests. Gut 1965;6:595-8. 7. Bassett ML. Goulston KJ. False positive and negative Hemoccult reactions on a normal diet and effect of diet restriction. Aust NZ J Med 1980;10:1-4.
DIET FOR OCCULT BLOOD TESTING 903
8. Macrae FA. St. John DJB . Relationship between patterns of bleeding and Hemoccult sensitivity in patients with colorectal cancers or adenomas. Gastroenterology; 1982;82:891-8. 9. Caligiore p. Macrae FA. St. John DJB. et al. Peroxidase levels in food: relevance to colorectal cancer screening. Am J Clin Nutr (in press). 10. Jaffe RM. Kasten B. Young DS. et al. False-negative stool occult blood tests caused by ingestion of ascorbic acid (vitamin C). Ann Intern Med 1975;83:824-6. 11 . Fleisher M. Schwartz MK. Winawer SJ. Laboratory studies on the Hemoccult slide for fecal occult blood testing. In: Winawer S. Schottenfeld D. Sherlock P. eds. Colorectal cancer: prevention. epidemiology and screening. New York: Raven. 1980:181-7. 12. Worthington Biochemical Corporation. Peroxidase. In: Enzymes and related biochemicals. Freehold. New Jersey: Worthington. 1978:145-6. 13. Walpole RE. Myers RH. Probability and statistics for engineers and scientists. 2nd ed. New York: Collier Macmillan International Editions. 1978. 14. Malagelada J-R. Carter SE. Brown ML. et al. Radiolabeled fiber: aphysiologic marker for gastric emptying and intestinal transit of solids. Dig Dis Sci 1980;25:81-7 15. Rosenfield RE. Kochwa S. Kaczera Z. et al. Nonuniform distribution of occult blood in feces. Am J Clin Pathol 1979;71 :204-9. 16. Brault J. Favre H. Evaluation en pratique hospitaliere du Fecatest. un nouvel examen de laboratoire pour la recherche de sang occulte dans les selles. Schweiz Med Wochenschr 1979;109:73-6. 17. Ostrow JD. Mulvaney CA. Hansell JR. et al. Sensitivity and reproducibility of chemical tests for fecal occult blood with an emphasis on false-positive reactions. Am J Dig Dis 1973;18:930-40. 18. Elwood TW. Erickson A. Leiberman S. Comparative educational approaches to screening for colorectal cancer. Am J Public Health 1978;68:135-8. 19. Feifel G. Manner C. v. Liebe S. The Haemoccult test without dietary restriction. In: Goerttler K. ed. Early detection of colorectal cancer. Nurnberg: Verlog DE. Wachholz KG. 1980:111-6.