Diet, Fecal Bile Acids, and Neutral Sterols in Carcinoma of the Colon MORRY MOSKOVITZ, MD, COLIN WHITE, MD, ROY N. BARNETT, MD, SYDNEY STEVENS, RN, EDITH R U S S E L L , AB, DAVID VARGO, MS, and MARTIN H. FLOCH, MD
Increased concentrations of fecal bile acids and neutral sterols or their degradation products have been linked to certain diets and are implicated in colonic carcinogenesis. We measured fecal bile acid and neutral sterol concentrations by thin-layer and gas-liquid chromatography in 15 patients with colonic adenocarcinoma, 23 controls, and 16 patients with nongastrointestinal cancer. We compared these results with dietary intake. Detailed dietary histories showed no differences among the groups in the ingestion of calories, protein, fiber, fat, or carbohydrate. A wide variation in fecal concentration of individual bile acids and neutral sterols was f o u n d within each group, but no significant differences in the total bile acid or total neutral sterol per gram dry weight feces were found. Decreased coprostanol, coprostanone, and lithocholic acid excretion Was found in the colon cancer group compared with controls. The fecal excretion of all bile acids and neutral sterols was lower significantly in the nongastrointestinal cancer patients with liver metastases as compared with those without, We conclude that total bile acid and total neutral sterol excretion is similar in the three groups, all ingesting similar diets. We cannot confirm reported increased excretion of total bile acids nor excessive bacterial conversion to degradation products in colonic cancer patients. Hepatic metastases correlate with decreased fecal excretion of both bile acids and neutral sterols, which may be due to diminished hepatobiliary excretion.
Epidemiological investigations have shown a higher incidence of colonic cancer in Europe and the United States when compared with populations in Africa, Asia, and South America (1-3). It has been postulated that colon tumors may be etiologically related to environmental carcinogens, hereditary factors, or dietary differences (3, 5). Diet has been most closely linked with the disparity in the geographic incidence of colonic cancer. The Western high-risk societies consume foods characterized by high animal fat, high protein, and low fiber content. It is suggested that certain comManuscript received November 24, t978; revised manuscript received April 5, 1979; accepted May 14, 1979. From the Department of Medicine, Section of Gastroenterology, Norwalk Hospital and the Yale University School of Medicine, Norwalk and New Haven, Connecticut. This study was supported in part by National Cancer Institute grant CA 16090. Address for reprint requests: Dr. Martin H. Floch, Norwalk Hospital, Norwalk, Connecticut 06856.
746
ponents of the high-risk diet may alter the intestinal flora in such a way as to increase its ability to degrade various lumenal constituents, resulting in the production of carcinogens (4, 5). Hill concluded that no dietary carcinogen had thus far been isolated which could explain this correlation between diet and colonic cancer (6). He therefore postulated that certain bacteria metabolize endogenous primary bile acids to produce substances which then act as cocarcinogens. The secondary bile acids, lithocolic and deoxycholic acid, have been demonstrated to be tumor promotors in experimental animals. (10, 11). Diet may be a determinant of the fecal bile acid concentration (7, 8, 14) and the intestinal flora (9) and therefore may determine the amount of carcinogen or cocarcinogen produced. Hill and his associates (12) correlated high-risk populations with higher fecal concentrations of bile acids and neutral sterols as compared with African and Eastern countries at lower risk. In another Digestive Diseases and Sciences, Vol. 24, No. 10 (October 1979)
0163-2116/79/1000-0746503.00/1 9 1979DigestiveDisease Systems, Inc.
FECAL STEROIDS IN COLON CANCER study, Hill found that patients with colonic cancer had higher fecal total bile acids, neutral sterols, and d e o x y c h o l a t e excretion than controls (13). This was felt to be evidence supportive of the hypothesis that cancer of the colon is induced by high concentrations of bile acid derivatives produced by certain anaerobic bacteria. H o w e v e r , despite data that diet m a y affect fecal sterol output, the relationship of diet to fecal sterol concentration was not fully explored. In this study we investigate nutrient ingestion as compared with fecal bile acid and neutral sterol output in patients with colon cancer (CC) and two control groups: one group without gastrointest i n a l d i s e a s e (N); a n d the o t h e r w i t h nongastrointestinal cancer (NGI). In addition, we studied the effect of the presence of hepatic metastases upon fecal sterol excretion. M A T E R I A L S A N D METHODS Patients. Fifty-four subjects, all residents of Fairfield County, Connecticut, form the basis of this study. All were informed of the goals of the study and consent was obtained. The study was approved by our hospital Committee of Human Investigation. Most subjects were inpatients in Norwalk (Conn.) Hospital, but some healthy controls were utilized. Detailed medical histories were obtained. There were 15 newly diagnosed patients with colonic carcinoma (CG), 16 patients with nongastrointestinal cancer (NGI), and 23 controls (N). Histologically proven adenocarcinoma originated from various colonic segments: right colon, 4; transverse, 3; descending, 1; and rectosigmoid, 7. Only 2 CC patients had liver metastases at surgical exploration. The mean age of the CC patients was 71 and comprised 3 men and 12 women. All studies were done in the CC patients preoperatively. In the NGI patients the primary site of tumor was lung, 8; breast, 4; ovarian, 1; myeloma and prostatic, 1; lymphoma, I; and carcinoma of unknown site, 1. Their mean age was 62 and the group consisted of 7 men and 10 women. In the NGI group, liver metastases, as documented histologically at operation, or suggested by perfusion defects on liver scan and later confirmed by biopsy, or at postmortem examination, were present in 8 of 16 patients. All patients with liver metastases had abnormal liver tests and three had clinically overt obstructive jaundice. Serum liver tests were obtained in all patients by subsequently described methods. The N group of healthy volunteers or patients hospitalized with nonmalignant, nongastrointestinal illnesses, including hypertension, coronary heart disease, psoriasis, and phlebitis, had an average age of 63 and was comprised of 15 men and 9 women. Excluded from the study were patients with coincidental, active gastrointestinal disease or those given antibiotics within two weeks of evaluation. Detailed dietary histories were obtained from all subjects by a trained nurse-techniciart. This method was used to determine dietary intake since it was impossible Digestive Diseases and Sciences, Vol. 24, No. I0 (October 1979)
to record the regular home dietary pattern using food intake diaries while the patients were hospitalized. Interviews included an estimation of the daily ingestion of total calories, protein, carbohydrates, fat, fiber, and cholesterol according to the method of Burke (15). Although there is an inherent degree of error in determining dietary intake through dietary histories, the error should be random. This method has been confirmed as a reliable approximation of the diet (16). Collection of Feces. The stool collection was similar to that used by Hill et al (8, 13). A specimen was obtained from each patient after admission to the hospital. The specimens from CC patients were collected following diagnostic studies and prior to administering antibiotics in preparation for bowel surgery. Most specimens were collected two days after the initial bowel examination. Outpatients passed their stool specimens in the hospital. All samples were taken to an anaerobic glove box within 20 min. Once in the glove box they were weighed and homogenized without any dilution. A small portion of the sample was processed for bacteriological studies reported elsewhere (20). The remainder was frozen and freezedried on a Virtis freeze-dryer (model 10-010). Duplicate one gram subsamples were weighed out after the dry sample was ground in a mortar and pestle. Neutral Sterols and Bile Acids: The thin-layer chromatography and gas-liquid chromatography techniques of Miettinen et al for the neutral sterols and of Grundy et al for the bile acids were used with slight modifications from Reddy et al (17-19). Recovery was monitored by adding a known amount of [26-14C]cholesterol and [24-14]cholic acid. The radioactivity remaining at the end was counted on a Packard Tri-Carb liquid scintillation spectrophotometer (model 3330). Quenching was determined by adding and counting a known amount of [14C]toluene in an aliquot of the final solution. Based upon split.-sample determinations, the reproducibility of the fecal sterol procedure had a coefficient of variation of 16% of the mean. In four cases of nonintestinal cancer the determination of bile acids was rendered impossible by the presence of some compound that entirely quenched the radioactivity of both [24-14C]ch01ic acid and [14C]toluene. The Student's t test and median test were used to analyze the data. Serum Liver Tests. The serum bilirubin determination was performed by the method of Jendrassik-Grof utilizing the Auto-Analyzer. Serum alkaline phosphatase determinations were likewise performed with the Auto-Analyzer and used a Gilford 2000 spectrophotometer employing pnitrophenyl phosphate as the substrate. Serum glutamic oxaloacetic transaminase (SGOT) was also determined on the Auto-Analyzer.
RESULTS
Table 1 summarizes the data calculated from the diet histories of the three groups. There were no significant differences between the three groups with relation to any of the diet components: total calories, protein, carbohydrate, fiber, fat or cholesterol.
747
MOSKOVITZ ET AL TABLE 1. COMPARISON OF DAILY DIETARY VARIABLES IN THREE GROUPS: NORMALS (N), NONGASTROINTESTINALCANCERS (NGI), ANn COLON CANCERS (CC) (MEAN • SE)*
Variable
N
NGI
Calories 1699.0 + 107.0 Protein(g) 82.9 • 7.6 Carbohydrate 189.1 • 13.0 Fiber(g) 4.4 • 0.4 Fat(g) 73.0 • 10.4 Cholesterol(mg) 360.9 ___ 36.8
1670.0 69.4 189.8 3.6 71.8 378.7
CC
• 93.0 1583.0 • 82.0 • 4.0 68.2 + 2.9 • 13.3 176.9 • 15.4 • 0.3 3.9 • 0.4 • 5.9 65.2 • 4.7 • 36.1 429.9 • 47.3
*P was greater than 0.05 for all comparisons: N vs CC, N v s NGI, CC vs NGI.
TABLE 2. MEAN FECAL CONCENTRATIONSOF STEROLS AND BILE ACIDS IN THREE GROUPS: NORMALS (N), COLON CANCER (CC), AND NONGASTROINTESTINALCANCER (NGI)*
Dried stool (mg/g +- SE) N Coprostanone Coprostanol Cholesterol Total sterols Lithocholic Deoxycholic Chenic Cholic Totalbile acids
NGI
CC
3.6 16.6 8.9 29.2
• 1.5 - 2.9 --_ 2.0 _+ 4.1
2.2 15 9.4 26.6
--+ 0.8 - 3.4 • 2.5 -+ 3.5
2.8 8.0 7.9 18.6
_ • _
1.9 2.2 2.2 2.8
3.5 3.8 0.6 1.1 8.9
-+ 0.5 + 0.8 -4- 0.2 - 0'.5 -4- 1.3
2.6 5.5 1.0 0.8 10.5
_ 0.6 - 2.1 - 0.6 --- 0.3 --- 3.1
1.9 2.5 1.0 0.4 5.8
• • • •
0.6 0.5 0.3 0.1 1.0
*P > 0.05 for all comparisons.
Thus, all patients ingested a similar "Western" diet. Table 2 summarizes the fecal concentrations of bile acids and neutral sterols. There was a wide variation in the concentration of these compounds among subjects within each group. There were no statistically significant differences in the mean values of any of the bile acid or neutral sterol fecal c o n c e n t r a t i o n s or the totals among the three groups; colon cancers (CC), controls (N), and nongastrointestinal cancers (NGI). The data seemed to exhibit a non-Gaussian distribution. When the median test was applied, the total
bile acid and neutral steroi concentrations were again not significantly different among the three groups; however, the CC group contained more low values. There was significantly less lithocholate, coprostanol, and coprostanone concentrations in the CC group than in the N control group (P < 0.05). This is shown in Table 3. It was evident from the fecal samples that the NGI and CC stools were usually less formed. The increased moisture content of the stools indicated b y the wet/dry ratio is statistically significant in both groups when compared to the N group (NGI vs N, P < 0.05; CC vs N, P < 0.01), but not when the CC is compared to the NGI. Because of the large number of patients with liver metastases in the NGI group, those data were analyzed separately (Table 4). Here, the NGI group with liver metastases (NGIm) was compared with the NGI group without liver metastases (NGIo) using the mean and standard error. In every case, the NGI group with liver metastases showed a lower c o n c e n t r a t i o n of acid and neutral sterols. This reached statistical significance in total acid and total neutral sterols, lithocholic acid, and in deoxycholic acid fecal concentration. Also shown are the corresponding mean serum liver tests for these two groups and their ranges. DISCUSSION As reported by other investigators, fecal acid and neutral sterol values show a skewed distribution (22). It was therefore necessary to determine the significance of the data using a nonparametric statistical test. Our results suggest that the colon cancer and control populations have a similar pattern of total bile acid and total neutral sterol fecal excretion. Concerning specific bile acids, the CC group had a decrease in lithocholate, the bacterial conversion p r o d u c t of c h e n o d e o x y c h o l i c acid; and coprostanol and coprostanone, conversion products of cholesterol.
TABLE 3. MEDIAN EXCRETION OF FECAL STEROLS 1N THREE GROUPS: N, NGI, AND CC (MG/G DRIED STOOL)
Neutral sterols Coprostanone Copro s tanol Cholesterol
Total Sterols
N
NGI
CC
0.96 13.49 4.62
0.47 12.75 4.47
0.31" 4.35 * 3.50
30.20
26.30
18.10
A c i d sterols Lithocholic Deoxycholic Chenic Cholic Total Acids
N
NGI
CC
2.94 2.97 0.27 0.08
1.84 2.02 0.01 0.05
1.07* I. 93 0.62 0.14
6.86
3.65
6.38
*CC is significantly less than N in these categories, P < 0.05.
748
Digestive Diseases and Sciences, Vol. 24, No. 10 (October 1979)
FECAL STEROIDS IN COLON CANCER TABLE 4. FECAL STEROL EXCRETION AND SERUM LIVER STUDIES IN 8 NGI PATIENTSWITH LIVER METASTASES(NGIM) v s 9 NGI PATIENTS WITHOUT LIVER METASTASES ( N G I o ) (MG/G DRIED STOOL) NGIm
Neutral sterols* Coprostanone Coprostanol Cholesterol Total sterols Bile acids* Lithocholic acid Deoxycholic Chenic Cholic acid Totalbile acids Serum liver tests* Total Bilirubin (mg/100 ml) Sample range Normal range Alkaline phosphatase (IU/liter) Sample range Normal range SGOT (IU/liter) Sample range Normal range
0.83 10.66 7.22 18.72
0.54 0.72 0.05 0,33 1.63
4• • •
NGIo
.31 4.10 2.66 4.26
3.21 18.45 11.06 32.72
• 1.20 +- 5.00 • 3.91 -+ 4.31t
4- 1.70 4- 0.35 • 0.04 4- 0.23 • 0.53
3.81 8.53 1.64 2.02 16.02
4- 0.70t -- 2.92t _+ 0.96 -+ 1.09 -+ 3.95t
5.6 • 2.9 0.3-22.8 up to 1.2
0.4 +- 0.04 0.2-0.6 up to 1.2
471 • 216 54-1908 20-120 65.0 4- 17.9 8.0-153 5-35
91 • 10.9 55-130 20-120 23.8 +- 2.6~ 13.0-34 5-35
*Results reported as sample mean + SE. t P < 0.02. ~/P < 0,05.
This is in contrast with the results of Hill et al, who reported an increased excretion of total bile acids and neutral sterols along with an increase in the amount of bile acid degradation (13). The reason for the dissimilar findings is unclear. It is possible for dietary factors to modify biliary secretions and cholesterol excretions. Fecal bile acid and neutral sterol concentrations have been reported to be directly related to dietary fat intake (8). Reddy et al showed that total fecal bile acids did not change significantly during a period of consumption of a high meat diet compared with a period with a nonmeat diet (14). However, the fecal concentration of microbially modified bile acids and neutral sterols decreased and alterations in flora were observed when subjects eating a high meat diet transferred to a nonmeat diet. It is possible that daily variations in diet may have been an influence on the fecal sterol levels. It is also possible the differences in the two studies were due to the fecal collection technique. We deliberately reduplicated the methods of Hill et al, which collect only one specimen (8, 13). To eliminate this possible source of variation, it would be necessary to employ a sample collection method similar to ones used in fecal sterol balance studies. Digestive Diseases and Sciences, Vot. 24, No. 10 (October 1979)
This would mean collecting a 48- to 72-h stool specimen and then analyzing a homogenized sample from the pooled collections. This may also allow for a more Gaussian distribution of the bile acid and neutral sterol data within the sample populations. Another possible source of variations is the inability to obtain an unbiased sampling of the population of colon cancer patients. Many colon cancer patients who have advanced stages of the disease experience alterations in colonic function. Constipation is a common problem, and many prospective CC subjects were not used in this study since they passed no stool during the days prior to surgery. This eliminates a segment of the population which may differ in sterol excretion from that of the patients who were analyzed. The stool from the CC group also differed from the control group in respect to its moisture content. The CC group had a higher wet/dry weight ratio, indicating a greater water content than the controls. This was also apparent during the collection of the stools. It was determined through the medical histories that loose bowel movements were not common during the time corresponding to the early stages of the disease. This makes it impossible to draw conclusions as to the concentration of the sterols from specimens obtained during the symptomatic period. The comparison of the NGI group to the N group did not show any differences in any of the parameters studied. Within the NGI group we did find a significantly decreased bile acid and neutral sterol fecal concentration in the patients with liver metastases as compared to the patients without liver metastases. Several studies have shown that serum bile acids are elevated in the presence of various hepatobiliary disorders (23, 24). It would therefore be presumed that there is diminished biliary excretion of bile acids and neutral sterol in our patients with liver metastases. It is conceivable that with their diffuse parenchymal hepatic disease (with tumor replacement), the liver was unable to excrete bile acids and neutral sterols in a normal fashion. This resulted in a decreased fecal concentration of these components in the patients with hepatic metastases. Thus the decreased excretion of acid and neutral sterols was probably a result of their disease. It is possible that some colon cancer patients used in the studies had undetected liver metastases which may have had an influence in the bile acid and neutral sterol levels. The dietary histories indicated that our cancer
749
MOSKOVITZ ET AL and control groups all had been eating similar " W e s t e r n " diets during their adult lives and that during hospitalization there were no qualitative changes from their previous dietary intake, although there was a general decrease in consumption of all foods. Dietary histories may not be the most accurate method for obtaining dietary intake, but it is the best available method for obtaining data during the asymptomatic period of the disease. The diets of the groups appear to be hypocaloric and are not true values of the dietary intake. We have previously determined that the differences between actual dietary intake and the calculated intake from dietary histories is a decrease in calories of approximately 20%. A n o t h e r factor may have been the decrease in appetite experienced by hospitalized patients which could have affected their judgment of the size of their normal food portions. Since all were hypocaloric, it was assumed that this would not invalidate the comparison of the diets. N o differences in the dietary factors which we measured was observed between the groups. The variations in fecal sterols were probably not caused by any major differences in diet. Recently it was reported that there may be slight variations in diets between individuals with cancer of the colon or rectum as compared to healthy controls. Graham et al, in a study of over 900 cancer patients, detected a correlation between the decreased frequency of ingestion of vegetables with an increasing risk of colon and rectal cancer (25). The intake of animal fat or protein, which has been suggested as a factor influencing risk, did not show any correlation. In conclusion, our results suggest that patients with colon cancer and control subjects ingesting similar diets have a similar excretion of total acid and neutral sterols. Colon cancer patients, however, show less excretion of the bacterial-transformed sterols, mainly lithocholate, coprostanol, and coprostanone. This is in contrast to the findings previously reported using similar collection methods and laboratory techniques. We also observed a wide variation in bile and neutral sterol excretion among individuals within the groups which may have been associated with the dissimilar results. Difficulties in obtaining feces from colon cancer patients by an unbiased sampling technique make the value of using such patients questionable to establish whether fecal sterol determinations m a y be useful in predicting an individual's risk for colon cancer. Finally, the presence of liver metastasis correlates with a significantly lower fecal excretion of
750
both bile acids and neutral sterols, a factor which must be taken into account in future studies. ACKNOWLEDGMENTS The authors gratefully acknowledge sistance of Mr. David Vargo and Mrs. Hans-Martin Fuchs, MD, Sean May, Ramos, MD, for specimen collection clinical data; and Mrs. Gail Krebs for ance.
the technical asVidya Teredesai; MD, and Rafael and gathering of secretarial assist-
REFERENCES 1. Doll R: The geographical distribution of cancer. Br J Cancer 23:1-8, 1969 2. Burkitt DP: Epidemiologyof cancer of the colon and rectum. Cancer 28:3-13, 1971 3. Kassira E, Parent L, Vahovny G: Colon Cancer: An epidemiologic survey. Am J Dig Dis 21:205-214, 1976 4. Nigro ND, Bhadracnari N, Chomchai, C: A rat model for studying colon cancer: Effect of cholestyramine on induced tumors. Dis Colon Rectum 16:438-443, 1973 5. Weisburger JH: Colon carcinogens: Their metabolism and mode of action. Cancer 28:60-70, 1971 6. Hill MJ: Metabolic epidemiology of dietary factors in large bowel cancer. Cancer Res 35:3398-3402, 1975 7. Antonis A. Bersohn I: The influence of diet on faecal lipids in South African white and bantu prisoners. Am J Nutr 11:142-155, 1962 8. Hill MJ: The effect of some factors on the fecal concentration of acid sterols, neutral sterols and urobilins. J Pathol 104:239-244, 1971 9. HoffmanK: Untersuchvngen uber die zusammensetzungder stuhl flora wahrend eines langdovernden ernahrungsversuchs mit kolenhydratrier, mit fettreichen und mit eiwrissneicher. Zentr Bacterio-Praisitenk, Abt I Orig 192:500-508, 1964 10. Barisawa I, Magadia NE, Weisburger JH, Wynder EL: Promoting effect of bile acids on co/on carcinogenesis after intra rectal installation of N-methyl-N-nitrosoguanidinein rats. J Natl Cancer Inst 55:1093-1097, 1974 11. Reddy BS, Narisawa T, Weisburger JH: Promoting effect of sodium deoxycholate on adenocarcinomas in germ free rats. J Natl Cancer Inst 56:441-442, 1976 12. Hill MJ, Draser BS, Aries VC, Crowther JS, Hawksworth GM, Williams, REO: Bacteria and etiology of cancer of the large bowel. Lancet 1:95-100, 1971 13. Hill MJ, Draser BS, Williams REP, Meade TW, Cox AG, Simpson JEP, Morson BC: Fecal bile acids and clostridia in patients with cancer of the large bowel. Lancet 2:535-539, 1975 14. Reddy BS, Weisburger JH, Wynder EL: Effects of high risk and low risk diets for colon carcinogenesis on fecal microflora and steroids in man. J Nutr 105:878-884, 1975 15. Burke BS: The dietary history as a tool in research. J Am Dietet Assoc 23:1041, 1947 16. Reshef H, Epstein LM: Reliability of a dietary questionnaire. Am J Clin Nutr 25:91-95, 1972 17. Miettinen TA, Ahrens EH, Jr., Grundy SM: Quantitative isolation and gas-liquid chromatographic analysis of total dietary and fecal neutral steroids. J Lipid Res 6:411-424, 1965 Digestive Diseases and Sciences, Vol. 24, No. 10 (October 1979)
FECAL STEROIDS IN COLON CANCER 18. Grundy SM, Ahrens EH, Jr., Miettinen TA: Quantitative isolation and gas-liquid chromatographic analysis of total bile acids. J Lipid Res 6:397-410, 1965 19. Reddy BS, Weisburger JH, Wynder EL: Effects of high risk and low risk diets for colon carcinogenesis on fecal microflora and steroids in man. J Nutr 105:878-884, 1975 20. Floch MH, Moskovitz M, Vargo D, Ramos R, May S, Fuchs HM: The spectrum of fecal bacterial flora in colon cancer and nongastrointestinal cancer subjects. Clin Res 26:318A, 1978 21. Avigan J, Steinberg D, Vroman H: Sterol and bile acid excretion in man and the effects of dietary fat. J Clin Invest 44:1845-1856, 1965
Digestive Diseases and Sciences, Vol. 24, No. 10 (October 1979)
22. Mower HF, Ray RM, Stemmerman GN, Nomura A, Glober GA: Analysis of fecal bile acids and diet among the Japanese in Hawaii. J Nutr 108:1289-1296, 1978 23. Makino I, Nagagawa S, Mashimo K: Conjugated and unconjugated serum bile acid levels in patients with hepatobiliary diseases. Gastroenterology 56:1033-1039, 1969 24. Makino I, Hashimoto H, Shinozaki K, Yoshino K, Nagagawa S: Sulfated and nonsulfated bile acids in urine, serum and bile of patients with hepatobiliary diseases. Gastroenterology 68:545-553, 1975 25. Graham S, Dayal H, Swanson J, Mittelman A, Wilkinson G: Diet in the epidemiology of cancer of the colon and rectum. J Natl Cancer Inst 61:709-714, 1978
751
Increased concentrations of fecal bile acids and neutral sterols or their degradation products have been linked to certain diets and are implicated in colonic carcinogenesis. We measured fecal bile acid and neutral sterol concentrations by thin-layer and gas-liquid chromatography in 15 patients with colonic adenocarcinoma, 23 controls, and 16 patients with nongastrointestinal cancer. We compared these results with dietary intake. Detailed dietary histories showed no differences among the groups in the ingestion of calories, protein, fiber, fat, or carbohydrate. A wide variation in fecal concentration of individual bile acids and neutral sterols was f o u n d within each group, but no significant differences in the total bile acid or total neutral sterol per gram dry weight feces were found. Decreased coprostanol, coprostanone, and lithocholic acid excretion Was found in the colon cancer group compared with controls. The fecal excretion of all bile acids and neutral sterols was lower significantly in the nongastrointestinal cancer patients with liver metastases as compared with those without, We conclude that total bile acid and total neutral sterol excretion is similar in the three groups, all ingesting similar diets. We cannot confirm reported increased excretion of total bile acids nor excessive bacterial conversion to degradation products in colonic cancer patients. Hepatic metastases correlate with decreased fecal excretion of both bile acids and neutral sterols, which may be due to diminished hepatobiliary excretion.
Epidemiological investigations have shown a higher incidence of colonic cancer in Europe and the United States when compared with populations in Africa, Asia, and South America (1-3). It has been postulated that colon tumors may be etiologically related to environmental carcinogens, hereditary factors, or dietary differences (3, 5). Diet has been most closely linked with the disparity in the geographic incidence of colonic cancer. The Western high-risk societies consume foods characterized by high animal fat, high protein, and low fiber content. It is suggested that certain comManuscript received November 24, t978; revised manuscript received April 5, 1979; accepted May 14, 1979. From the Department of Medicine, Section of Gastroenterology, Norwalk Hospital and the Yale University School of Medicine, Norwalk and New Haven, Connecticut. This study was supported in part by National Cancer Institute grant CA 16090. Address for reprint requests: Dr. Martin H. Floch, Norwalk Hospital, Norwalk, Connecticut 06856.
746
ponents of the high-risk diet may alter the intestinal flora in such a way as to increase its ability to degrade various lumenal constituents, resulting in the production of carcinogens (4, 5). Hill concluded that no dietary carcinogen had thus far been isolated which could explain this correlation between diet and colonic cancer (6). He therefore postulated that certain bacteria metabolize endogenous primary bile acids to produce substances which then act as cocarcinogens. The secondary bile acids, lithocolic and deoxycholic acid, have been demonstrated to be tumor promotors in experimental animals. (10, 11). Diet may be a determinant of the fecal bile acid concentration (7, 8, 14) and the intestinal flora (9) and therefore may determine the amount of carcinogen or cocarcinogen produced. Hill and his associates (12) correlated high-risk populations with higher fecal concentrations of bile acids and neutral sterols as compared with African and Eastern countries at lower risk. In another Digestive Diseases and Sciences, Vol. 24, No. 10 (October 1979)
0163-2116/79/1000-0746503.00/1 9 1979DigestiveDisease Systems, Inc.
FECAL STEROIDS IN COLON CANCER study, Hill found that patients with colonic cancer had higher fecal total bile acids, neutral sterols, and d e o x y c h o l a t e excretion than controls (13). This was felt to be evidence supportive of the hypothesis that cancer of the colon is induced by high concentrations of bile acid derivatives produced by certain anaerobic bacteria. H o w e v e r , despite data that diet m a y affect fecal sterol output, the relationship of diet to fecal sterol concentration was not fully explored. In this study we investigate nutrient ingestion as compared with fecal bile acid and neutral sterol output in patients with colon cancer (CC) and two control groups: one group without gastrointest i n a l d i s e a s e (N); a n d the o t h e r w i t h nongastrointestinal cancer (NGI). In addition, we studied the effect of the presence of hepatic metastases upon fecal sterol excretion. M A T E R I A L S A N D METHODS Patients. Fifty-four subjects, all residents of Fairfield County, Connecticut, form the basis of this study. All were informed of the goals of the study and consent was obtained. The study was approved by our hospital Committee of Human Investigation. Most subjects were inpatients in Norwalk (Conn.) Hospital, but some healthy controls were utilized. Detailed medical histories were obtained. There were 15 newly diagnosed patients with colonic carcinoma (CG), 16 patients with nongastrointestinal cancer (NGI), and 23 controls (N). Histologically proven adenocarcinoma originated from various colonic segments: right colon, 4; transverse, 3; descending, 1; and rectosigmoid, 7. Only 2 CC patients had liver metastases at surgical exploration. The mean age of the CC patients was 71 and comprised 3 men and 12 women. All studies were done in the CC patients preoperatively. In the NGI patients the primary site of tumor was lung, 8; breast, 4; ovarian, 1; myeloma and prostatic, 1; lymphoma, I; and carcinoma of unknown site, 1. Their mean age was 62 and the group consisted of 7 men and 10 women. In the NGI group, liver metastases, as documented histologically at operation, or suggested by perfusion defects on liver scan and later confirmed by biopsy, or at postmortem examination, were present in 8 of 16 patients. All patients with liver metastases had abnormal liver tests and three had clinically overt obstructive jaundice. Serum liver tests were obtained in all patients by subsequently described methods. The N group of healthy volunteers or patients hospitalized with nonmalignant, nongastrointestinal illnesses, including hypertension, coronary heart disease, psoriasis, and phlebitis, had an average age of 63 and was comprised of 15 men and 9 women. Excluded from the study were patients with coincidental, active gastrointestinal disease or those given antibiotics within two weeks of evaluation. Detailed dietary histories were obtained from all subjects by a trained nurse-techniciart. This method was used to determine dietary intake since it was impossible Digestive Diseases and Sciences, Vol. 24, No. I0 (October 1979)
to record the regular home dietary pattern using food intake diaries while the patients were hospitalized. Interviews included an estimation of the daily ingestion of total calories, protein, carbohydrates, fat, fiber, and cholesterol according to the method of Burke (15). Although there is an inherent degree of error in determining dietary intake through dietary histories, the error should be random. This method has been confirmed as a reliable approximation of the diet (16). Collection of Feces. The stool collection was similar to that used by Hill et al (8, 13). A specimen was obtained from each patient after admission to the hospital. The specimens from CC patients were collected following diagnostic studies and prior to administering antibiotics in preparation for bowel surgery. Most specimens were collected two days after the initial bowel examination. Outpatients passed their stool specimens in the hospital. All samples were taken to an anaerobic glove box within 20 min. Once in the glove box they were weighed and homogenized without any dilution. A small portion of the sample was processed for bacteriological studies reported elsewhere (20). The remainder was frozen and freezedried on a Virtis freeze-dryer (model 10-010). Duplicate one gram subsamples were weighed out after the dry sample was ground in a mortar and pestle. Neutral Sterols and Bile Acids: The thin-layer chromatography and gas-liquid chromatography techniques of Miettinen et al for the neutral sterols and of Grundy et al for the bile acids were used with slight modifications from Reddy et al (17-19). Recovery was monitored by adding a known amount of [26-14C]cholesterol and [24-14]cholic acid. The radioactivity remaining at the end was counted on a Packard Tri-Carb liquid scintillation spectrophotometer (model 3330). Quenching was determined by adding and counting a known amount of [14C]toluene in an aliquot of the final solution. Based upon split.-sample determinations, the reproducibility of the fecal sterol procedure had a coefficient of variation of 16% of the mean. In four cases of nonintestinal cancer the determination of bile acids was rendered impossible by the presence of some compound that entirely quenched the radioactivity of both [24-14C]ch01ic acid and [14C]toluene. The Student's t test and median test were used to analyze the data. Serum Liver Tests. The serum bilirubin determination was performed by the method of Jendrassik-Grof utilizing the Auto-Analyzer. Serum alkaline phosphatase determinations were likewise performed with the Auto-Analyzer and used a Gilford 2000 spectrophotometer employing pnitrophenyl phosphate as the substrate. Serum glutamic oxaloacetic transaminase (SGOT) was also determined on the Auto-Analyzer.
RESULTS
Table 1 summarizes the data calculated from the diet histories of the three groups. There were no significant differences between the three groups with relation to any of the diet components: total calories, protein, carbohydrate, fiber, fat or cholesterol.
747
MOSKOVITZ ET AL TABLE 1. COMPARISON OF DAILY DIETARY VARIABLES IN THREE GROUPS: NORMALS (N), NONGASTROINTESTINALCANCERS (NGI), ANn COLON CANCERS (CC) (MEAN • SE)*
Variable
N
NGI
Calories 1699.0 + 107.0 Protein(g) 82.9 • 7.6 Carbohydrate 189.1 • 13.0 Fiber(g) 4.4 • 0.4 Fat(g) 73.0 • 10.4 Cholesterol(mg) 360.9 ___ 36.8
1670.0 69.4 189.8 3.6 71.8 378.7
CC
• 93.0 1583.0 • 82.0 • 4.0 68.2 + 2.9 • 13.3 176.9 • 15.4 • 0.3 3.9 • 0.4 • 5.9 65.2 • 4.7 • 36.1 429.9 • 47.3
*P was greater than 0.05 for all comparisons: N vs CC, N v s NGI, CC vs NGI.
TABLE 2. MEAN FECAL CONCENTRATIONSOF STEROLS AND BILE ACIDS IN THREE GROUPS: NORMALS (N), COLON CANCER (CC), AND NONGASTROINTESTINALCANCER (NGI)*
Dried stool (mg/g +- SE) N Coprostanone Coprostanol Cholesterol Total sterols Lithocholic Deoxycholic Chenic Cholic Totalbile acids
NGI
CC
3.6 16.6 8.9 29.2
• 1.5 - 2.9 --_ 2.0 _+ 4.1
2.2 15 9.4 26.6
--+ 0.8 - 3.4 • 2.5 -+ 3.5
2.8 8.0 7.9 18.6
_ • _
1.9 2.2 2.2 2.8
3.5 3.8 0.6 1.1 8.9
-+ 0.5 + 0.8 -4- 0.2 - 0'.5 -4- 1.3
2.6 5.5 1.0 0.8 10.5
_ 0.6 - 2.1 - 0.6 --- 0.3 --- 3.1
1.9 2.5 1.0 0.4 5.8
• • • •
0.6 0.5 0.3 0.1 1.0
*P > 0.05 for all comparisons.
Thus, all patients ingested a similar "Western" diet. Table 2 summarizes the fecal concentrations of bile acids and neutral sterols. There was a wide variation in the concentration of these compounds among subjects within each group. There were no statistically significant differences in the mean values of any of the bile acid or neutral sterol fecal c o n c e n t r a t i o n s or the totals among the three groups; colon cancers (CC), controls (N), and nongastrointestinal cancers (NGI). The data seemed to exhibit a non-Gaussian distribution. When the median test was applied, the total
bile acid and neutral steroi concentrations were again not significantly different among the three groups; however, the CC group contained more low values. There was significantly less lithocholate, coprostanol, and coprostanone concentrations in the CC group than in the N control group (P < 0.05). This is shown in Table 3. It was evident from the fecal samples that the NGI and CC stools were usually less formed. The increased moisture content of the stools indicated b y the wet/dry ratio is statistically significant in both groups when compared to the N group (NGI vs N, P < 0.05; CC vs N, P < 0.01), but not when the CC is compared to the NGI. Because of the large number of patients with liver metastases in the NGI group, those data were analyzed separately (Table 4). Here, the NGI group with liver metastases (NGIm) was compared with the NGI group without liver metastases (NGIo) using the mean and standard error. In every case, the NGI group with liver metastases showed a lower c o n c e n t r a t i o n of acid and neutral sterols. This reached statistical significance in total acid and total neutral sterols, lithocholic acid, and in deoxycholic acid fecal concentration. Also shown are the corresponding mean serum liver tests for these two groups and their ranges. DISCUSSION As reported by other investigators, fecal acid and neutral sterol values show a skewed distribution (22). It was therefore necessary to determine the significance of the data using a nonparametric statistical test. Our results suggest that the colon cancer and control populations have a similar pattern of total bile acid and total neutral sterol fecal excretion. Concerning specific bile acids, the CC group had a decrease in lithocholate, the bacterial conversion p r o d u c t of c h e n o d e o x y c h o l i c acid; and coprostanol and coprostanone, conversion products of cholesterol.
TABLE 3. MEDIAN EXCRETION OF FECAL STEROLS 1N THREE GROUPS: N, NGI, AND CC (MG/G DRIED STOOL)
Neutral sterols Coprostanone Copro s tanol Cholesterol
Total Sterols
N
NGI
CC
0.96 13.49 4.62
0.47 12.75 4.47
0.31" 4.35 * 3.50
30.20
26.30
18.10
A c i d sterols Lithocholic Deoxycholic Chenic Cholic Total Acids
N
NGI
CC
2.94 2.97 0.27 0.08
1.84 2.02 0.01 0.05
1.07* I. 93 0.62 0.14
6.86
3.65
6.38
*CC is significantly less than N in these categories, P < 0.05.
748
Digestive Diseases and Sciences, Vol. 24, No. 10 (October 1979)
FECAL STEROIDS IN COLON CANCER TABLE 4. FECAL STEROL EXCRETION AND SERUM LIVER STUDIES IN 8 NGI PATIENTSWITH LIVER METASTASES(NGIM) v s 9 NGI PATIENTS WITHOUT LIVER METASTASES ( N G I o ) (MG/G DRIED STOOL) NGIm
Neutral sterols* Coprostanone Coprostanol Cholesterol Total sterols Bile acids* Lithocholic acid Deoxycholic Chenic Cholic acid Totalbile acids Serum liver tests* Total Bilirubin (mg/100 ml) Sample range Normal range Alkaline phosphatase (IU/liter) Sample range Normal range SGOT (IU/liter) Sample range Normal range
0.83 10.66 7.22 18.72
0.54 0.72 0.05 0,33 1.63
4• • •
NGIo
.31 4.10 2.66 4.26
3.21 18.45 11.06 32.72
• 1.20 +- 5.00 • 3.91 -+ 4.31t
4- 1.70 4- 0.35 • 0.04 4- 0.23 • 0.53
3.81 8.53 1.64 2.02 16.02
4- 0.70t -- 2.92t _+ 0.96 -+ 1.09 -+ 3.95t
5.6 • 2.9 0.3-22.8 up to 1.2
0.4 +- 0.04 0.2-0.6 up to 1.2
471 • 216 54-1908 20-120 65.0 4- 17.9 8.0-153 5-35
91 • 10.9 55-130 20-120 23.8 +- 2.6~ 13.0-34 5-35
*Results reported as sample mean + SE. t P < 0.02. ~/P < 0,05.
This is in contrast with the results of Hill et al, who reported an increased excretion of total bile acids and neutral sterols along with an increase in the amount of bile acid degradation (13). The reason for the dissimilar findings is unclear. It is possible for dietary factors to modify biliary secretions and cholesterol excretions. Fecal bile acid and neutral sterol concentrations have been reported to be directly related to dietary fat intake (8). Reddy et al showed that total fecal bile acids did not change significantly during a period of consumption of a high meat diet compared with a period with a nonmeat diet (14). However, the fecal concentration of microbially modified bile acids and neutral sterols decreased and alterations in flora were observed when subjects eating a high meat diet transferred to a nonmeat diet. It is possible that daily variations in diet may have been an influence on the fecal sterol levels. It is also possible the differences in the two studies were due to the fecal collection technique. We deliberately reduplicated the methods of Hill et al, which collect only one specimen (8, 13). To eliminate this possible source of variation, it would be necessary to employ a sample collection method similar to ones used in fecal sterol balance studies. Digestive Diseases and Sciences, Vot. 24, No. 10 (October 1979)
This would mean collecting a 48- to 72-h stool specimen and then analyzing a homogenized sample from the pooled collections. This may also allow for a more Gaussian distribution of the bile acid and neutral sterol data within the sample populations. Another possible source of variations is the inability to obtain an unbiased sampling of the population of colon cancer patients. Many colon cancer patients who have advanced stages of the disease experience alterations in colonic function. Constipation is a common problem, and many prospective CC subjects were not used in this study since they passed no stool during the days prior to surgery. This eliminates a segment of the population which may differ in sterol excretion from that of the patients who were analyzed. The stool from the CC group also differed from the control group in respect to its moisture content. The CC group had a higher wet/dry weight ratio, indicating a greater water content than the controls. This was also apparent during the collection of the stools. It was determined through the medical histories that loose bowel movements were not common during the time corresponding to the early stages of the disease. This makes it impossible to draw conclusions as to the concentration of the sterols from specimens obtained during the symptomatic period. The comparison of the NGI group to the N group did not show any differences in any of the parameters studied. Within the NGI group we did find a significantly decreased bile acid and neutral sterol fecal concentration in the patients with liver metastases as compared to the patients without liver metastases. Several studies have shown that serum bile acids are elevated in the presence of various hepatobiliary disorders (23, 24). It would therefore be presumed that there is diminished biliary excretion of bile acids and neutral sterol in our patients with liver metastases. It is conceivable that with their diffuse parenchymal hepatic disease (with tumor replacement), the liver was unable to excrete bile acids and neutral sterols in a normal fashion. This resulted in a decreased fecal concentration of these components in the patients with hepatic metastases. Thus the decreased excretion of acid and neutral sterols was probably a result of their disease. It is possible that some colon cancer patients used in the studies had undetected liver metastases which may have had an influence in the bile acid and neutral sterol levels. The dietary histories indicated that our cancer
749
MOSKOVITZ ET AL and control groups all had been eating similar " W e s t e r n " diets during their adult lives and that during hospitalization there were no qualitative changes from their previous dietary intake, although there was a general decrease in consumption of all foods. Dietary histories may not be the most accurate method for obtaining dietary intake, but it is the best available method for obtaining data during the asymptomatic period of the disease. The diets of the groups appear to be hypocaloric and are not true values of the dietary intake. We have previously determined that the differences between actual dietary intake and the calculated intake from dietary histories is a decrease in calories of approximately 20%. A n o t h e r factor may have been the decrease in appetite experienced by hospitalized patients which could have affected their judgment of the size of their normal food portions. Since all were hypocaloric, it was assumed that this would not invalidate the comparison of the diets. N o differences in the dietary factors which we measured was observed between the groups. The variations in fecal sterols were probably not caused by any major differences in diet. Recently it was reported that there may be slight variations in diets between individuals with cancer of the colon or rectum as compared to healthy controls. Graham et al, in a study of over 900 cancer patients, detected a correlation between the decreased frequency of ingestion of vegetables with an increasing risk of colon and rectal cancer (25). The intake of animal fat or protein, which has been suggested as a factor influencing risk, did not show any correlation. In conclusion, our results suggest that patients with colon cancer and control subjects ingesting similar diets have a similar excretion of total acid and neutral sterols. Colon cancer patients, however, show less excretion of the bacterial-transformed sterols, mainly lithocholate, coprostanol, and coprostanone. This is in contrast to the findings previously reported using similar collection methods and laboratory techniques. We also observed a wide variation in bile and neutral sterol excretion among individuals within the groups which may have been associated with the dissimilar results. Difficulties in obtaining feces from colon cancer patients by an unbiased sampling technique make the value of using such patients questionable to establish whether fecal sterol determinations m a y be useful in predicting an individual's risk for colon cancer. Finally, the presence of liver metastasis correlates with a significantly lower fecal excretion of
750
both bile acids and neutral sterols, a factor which must be taken into account in future studies. ACKNOWLEDGMENTS The authors gratefully acknowledge sistance of Mr. David Vargo and Mrs. Hans-Martin Fuchs, MD, Sean May, Ramos, MD, for specimen collection clinical data; and Mrs. Gail Krebs for ance.
the technical asVidya Teredesai; MD, and Rafael and gathering of secretarial assist-
REFERENCES 1. Doll R: The geographical distribution of cancer. Br J Cancer 23:1-8, 1969 2. Burkitt DP: Epidemiologyof cancer of the colon and rectum. Cancer 28:3-13, 1971 3. Kassira E, Parent L, Vahovny G: Colon Cancer: An epidemiologic survey. Am J Dig Dis 21:205-214, 1976 4. Nigro ND, Bhadracnari N, Chomchai, C: A rat model for studying colon cancer: Effect of cholestyramine on induced tumors. Dis Colon Rectum 16:438-443, 1973 5. Weisburger JH: Colon carcinogens: Their metabolism and mode of action. Cancer 28:60-70, 1971 6. Hill MJ: Metabolic epidemiology of dietary factors in large bowel cancer. Cancer Res 35:3398-3402, 1975 7. Antonis A. Bersohn I: The influence of diet on faecal lipids in South African white and bantu prisoners. Am J Nutr 11:142-155, 1962 8. Hill MJ: The effect of some factors on the fecal concentration of acid sterols, neutral sterols and urobilins. J Pathol 104:239-244, 1971 9. HoffmanK: Untersuchvngen uber die zusammensetzungder stuhl flora wahrend eines langdovernden ernahrungsversuchs mit kolenhydratrier, mit fettreichen und mit eiwrissneicher. Zentr Bacterio-Praisitenk, Abt I Orig 192:500-508, 1964 10. Barisawa I, Magadia NE, Weisburger JH, Wynder EL: Promoting effect of bile acids on co/on carcinogenesis after intra rectal installation of N-methyl-N-nitrosoguanidinein rats. J Natl Cancer Inst 55:1093-1097, 1974 11. Reddy BS, Narisawa T, Weisburger JH: Promoting effect of sodium deoxycholate on adenocarcinomas in germ free rats. J Natl Cancer Inst 56:441-442, 1976 12. Hill MJ, Draser BS, Aries VC, Crowther JS, Hawksworth GM, Williams, REO: Bacteria and etiology of cancer of the large bowel. Lancet 1:95-100, 1971 13. Hill MJ, Draser BS, Williams REP, Meade TW, Cox AG, Simpson JEP, Morson BC: Fecal bile acids and clostridia in patients with cancer of the large bowel. Lancet 2:535-539, 1975 14. Reddy BS, Weisburger JH, Wynder EL: Effects of high risk and low risk diets for colon carcinogenesis on fecal microflora and steroids in man. J Nutr 105:878-884, 1975 15. Burke BS: The dietary history as a tool in research. J Am Dietet Assoc 23:1041, 1947 16. Reshef H, Epstein LM: Reliability of a dietary questionnaire. Am J Clin Nutr 25:91-95, 1972 17. Miettinen TA, Ahrens EH, Jr., Grundy SM: Quantitative isolation and gas-liquid chromatographic analysis of total dietary and fecal neutral steroids. J Lipid Res 6:411-424, 1965 Digestive Diseases and Sciences, Vol. 24, No. 10 (October 1979)
FECAL STEROIDS IN COLON CANCER 18. Grundy SM, Ahrens EH, Jr., Miettinen TA: Quantitative isolation and gas-liquid chromatographic analysis of total bile acids. J Lipid Res 6:397-410, 1965 19. Reddy BS, Weisburger JH, Wynder EL: Effects of high risk and low risk diets for colon carcinogenesis on fecal microflora and steroids in man. J Nutr 105:878-884, 1975 20. Floch MH, Moskovitz M, Vargo D, Ramos R, May S, Fuchs HM: The spectrum of fecal bacterial flora in colon cancer and nongastrointestinal cancer subjects. Clin Res 26:318A, 1978 21. Avigan J, Steinberg D, Vroman H: Sterol and bile acid excretion in man and the effects of dietary fat. J Clin Invest 44:1845-1856, 1965
Digestive Diseases and Sciences, Vol. 24, No. 10 (October 1979)
22. Mower HF, Ray RM, Stemmerman GN, Nomura A, Glober GA: Analysis of fecal bile acids and diet among the Japanese in Hawaii. J Nutr 108:1289-1296, 1978 23. Makino I, Nagagawa S, Mashimo K: Conjugated and unconjugated serum bile acid levels in patients with hepatobiliary diseases. Gastroenterology 56:1033-1039, 1969 24. Makino I, Hashimoto H, Shinozaki K, Yoshino K, Nagagawa S: Sulfated and nonsulfated bile acids in urine, serum and bile of patients with hepatobiliary diseases. Gastroenterology 68:545-553, 1975 25. Graham S, Dayal H, Swanson J, Mittelman A, Wilkinson G: Diet in the epidemiology of cancer of the colon and rectum. J Natl Cancer Inst 61:709-714, 1978
751
