Steatorrhea results from impaired digestion or absorption of dietary fats and it can also be caused by multiple diseases including cystic fibrosis (CF), chronic pancreatitis, cholestatic liver disease, celiac disease, and inflammatory bowel disease (1). If untreated, fat malabsorption may result in malnutrition, growth failure, and deficiencies of fat-soluble vitamins A, E, D, and K, with resultant skin and visual changes, neurological deficits, osteoporosis/rickets, and coagulopathy.
Presently the gold standard test to diagnose steatorrhea remains the 72-hour fat balance method that is based on the premise that fat intake minus fat output equals fat absorbed (2). A coefficient of fat absorption (CFA), calculated from the intake/output data, is the standard value used to indicate malabsorption. This test is time-consuming and logistically difficult because it requires a 3- to 5-day stool collection and a complete dietary intake record. Additionally, most patients with fat malabsorption have diarrhea and therefore accurate and complete collection is difficult, particularly in infants and children. These limitations make the 72-hour fat balance method impractical in the clinical setting and stress the need for a facile, accurate test of fat malabsorption. Despite previous attempts to develop simpler methods than the 72-hour fat balance method, none have proven to be easier than and as reliable as the gold standard.
Preliminary investigations in mice and rats have provided evidence that the technique of administering a small amount of sucrose polybehenate (SPB) is a potentially viable method for assessing stool fat without collecting 72 hours of stool (3). Early work with humans suggested that either linoleic or linolenic acid would be an excellent fatty acid for comparison; however, it was found that dietary contamination and bacterial degradation made them unsatisfactory. Subsequently, it was determined that lauric acid was a fatty acid better suited as a marker of absorption because it is present in minimal amounts (0.8 g/day) in the normal human diet (http://www.barc.usda.gov/bhnrc/foodsurvey/home.htm) and it does not undergo appreciable bacterial alteration.
Before our present study, we studied 4 adult control participants. In 2 participants, the 72-hour stool collections were obtained and both the fecal behenate method and fat balance method concurrently performed. The fecal behenate method indicated absorption of 95.1% and 91.9% of dietary fat, whereas the 72-hour fat balance method indicated absorption of 67.5% and 95.1%, respectively. In the other 2 participants, only the fecal behenate method was performed and fat absorption was calculated to be 99.9% and 100% (unpublished results), respectively. These results suggested that the fecal behenate method was robust in control participants and indicated that the method may be useful for differentiating normal versus excessive fecal fat loss. Those studies encouraged us to pursue the present investigations in patients with CF.
The present study was designed to test the hypothesis that the ratio of stool lauric:behenic acid would be predictive of the CFA using the 72-hour fat balance method in children and adults with CF when they were not taking their customary pancreatic enzyme supplements and when taking enzyme supplements before or throughout the meal.
PATIENTS AND METHODS
Participants were recruited from the Cystic Fibrosis Clinic at Cincinnati Children's Hospital Medical Center (CCHMC), The Children's Hospital (Aurora, CO), and the pulmonary clinic at the University of Cincinnati. Eligible participants were patients ages 7 to 50 years with CF confirmed by sweat electrolytes and/or genotype with pancreatic insufficiency treated with pancreatic enzyme replacement therapy (PERT) and stable or increasing weight for at least 3 months before recruitment. Participants were required to be receiving stable PERT for the preceding 3 to 6 months. Exclusion criteria included pregnancy, the presence of other diseases causing fat malabsorption, need for total parenteral nutrition, and any change in their usual antibiotic therapy in the month preceding the periods of stool collection. Participants were approached by the lead investigators (J.D., M.R.N.) at each site, and assessed regarding the inclusion/exclusion criteria. If they met the entry criteria, a verbal and then written explanation of the study was provided to each participant. The study was approved by The Children's Hospital and the University of Colorado Denver, the CCHMC and University of Cincinnati institutional review boards, and the scientific advisory boards of the General Clinical Research Center (GCRC) at CCHMC and The Children's Hospital.
The study design was a crossover, wherein 1 group of participants studied was initially receiving PERT and then repeated off PERT; similarly the other group studied was initially off PERT and then repeated while receiving PERT. The independent variables analyzed were percentage fat absorption assessed by the lauric acid/behenic acid and 72-hour fat balance method, both continuous variables. The primary dependent variable in this study was fat malabsorption status. Participants were randomized to 3 days without pancreatic enzyme supplements or their usual supplements administered before meals. It was anticipated that most participants would be receiving PERT before meals and not throughout; however, this was noted specifically because it was relevant to the studies. Participants were encouraged to select a high-fat (at least 70g fat/day) diet, and a trained research dietitian (S.S., Janine Higgins, or Melanie Kasten) analyzed their intake to calculate total energy and fat intake. The participants were admitted to the GCRC at Children's Hospital Medical Center or the Clinical Translational Research Center at The Children's Hospital. Each adolescent/adult participant was fed a diet of his or her own choice containing at least 70 g fat. This fat intake represents 30% of the energy in a 2000-calorie diet. Children were also provided diets of their own choice with proportional fat and energy intake appropriate for their age. Before admission, participants were given a color marker (brilliant blue) by mouth. Participants, who were not to be receiving PERT for the admission, stopped taking their pancreatic enzymes 3 days before the ingestion of this marker. When the first marked stool was identified by the participant, he or she was admitted to the GCRC. Thereafter, stools were collected individually and then pooled when the collection was complete. Seventy-two hours after the identification of the color marker in the stool, a second marker was administered and all of the stools were collected including the second marked stool. On the second day of hospitalization, the first meal consisted of fat-free pudding with SPB, a component of fat used in the commercial preparation of fried snack food (a gift from the Procter & Gamble Company, Cincinnati, OH) and coconut oil. Coconut oil was purchased from a Kroger supermarket in Cincinnati. These components were blended at a weight ratio of 95:5 coconut oil:SPB and mixed with fat-free chocolate or vanilla pudding. One serving contained 485 kcal with 6.94 g protein, 46.04 g carbohydrate, and 30.91 g fat consisting of 3.38 g saturated fat, 6.34 g monosaturated fat, and 19.83 g polyunsaturated fat. The test meal was served as breakfast and supplemented with other fat-free products such as fat-free beverages including skim milk, fruit juice, black coffee/tea, and/or piece of fresh fruit. The participants did not eat for the next 3 hours and thereafter resumed their usual diet.
Approximately 100 mg of feces was removed from the next 2 to 4 stools passed after the test meal for lauric acid:behenic acid analysis by gas chromatography techniques. This volume of stool represents <0.1% of the total stool collected for 72 hours and had minimal impact on the results of the 72-hour collection. Results were expressed as the average of the ratio of lauric acid:behenic acid contained in 2 to 3 stools collected after consumption of the meal.
Fatty acid compositions of the pudding with added fat and of the feces were analyzed by gas chromatography using the method of Metcalfe et al (3,4). The fraction of fat absorbed was calculated using the following equation:
Equation (Uncited)Image Tools
where Fd = mass of dietary lauric acid in the test meal, Bd = mass of dietary behenic acid, Ff = mass of fecal lauric acid, and Bf = mass of fecal behenic acid.
The fat content of all food consumed was calculated to obtain total fat intake for each participant. At the end of the 72-hour collection, all of the stools were pooled and analysis performed using nuclear magnetic resonance (5,6) by Mayo Clinic Laboratories (Rochester, MN). All of the results were expressed as the CFA:
Equation (Uncited)Image Tools
Subsequently, participants either used their standard pancreatic enzyme supplements or no supplements for 3 days as outlined above in preparation for the second study period. Participants then repeated the study using their pancreatic enzyme supplements throughout the duration of the meal for the third study period (enzymes would be administered equally at the beginning, middle, and end of the meal).
Sample Size Calculation
The sample size was based on comparison of the methods when participants were either receiving or not receiving PERT. The expectation is that when the participants are receiving PERT, the amount of fat absorption would be 85%, as compared with 60% when they are not receiving supplements. Thus, the reduction in one group is 25% and in the other group it is –25%. The standard deviation (SD) is not known, so to be conservative, a SD equal to 20% was used. Using the sample size procedure for a 2-sample t test, the value of n required was computed. The software used was nQuery version 5.0 (Statistical Solutions Ltd, Boston, MA) for a 2-sided t test with a level of significance of 0.05 and a power of 80%. For this aim (aim 1), if 9 individuals crossed over, it would be sufficient. The assessment of differences between PERT during meals versus PERT before meals was considered exploratory, and therefore a pilot study (aim 2) was done. It was anticipated that the sample size selected for aim 1 may not be sufficient to determine the impact of using PERT before versus using PERT during the meal. The results of this preliminary study may be used to assess the potential size of a population needed to answer this question if future studies are undertaken. Using a crossover design, we projected that means of absorption may be 10% or −10% with an SD of 20%. For this approach a sample size of 34, calculated using nQuery version 5.0 for a paired t test, would be required to be able to detect the difference of 10% between participants ingesting pancreatic enzymes before versus participants ingesting pancreatic enzymes during the course of the meal. Because it was not practical to perform the entire study of aim 2, we compromised and initiated studies with 15 participants as a pilot that should inform us regarding a more accurate sample size for a definitive study.
Before any analysis, SPSS version 16.0.1 (SPSS Inc, Chicago, IL) was used to generate means, SDs, skewness, range, and other descriptive statistics. Analysis of variance was used to analyze the data from the crossover study (off PERT vs receiving PERT and receiving PERT before meals vs PERT during meals) to determine difference in fat absorption calculated by the gold standard 72-hour quantitative fecal fat determination. Comparisons of fat absorption by the behenate method versus the 72-hour fat balance method across all conditions were made using the Pearson correlation coefficient.
Eighteen participants met inclusion criteria and were enrolled in the study. Fifteen participants completed at least 2 of the arms of the study. Fourteen participants completed all 3 arms; 1 did not complete the arm in which PERT was consumed throughout the meal. Participants were ages 8 to 50 years with a mean age of 21.1 years and a median age of 19.5 years. There were 7 males and all of them were white. Participants were well nourished with a body mass index in adults (older than 18 years of age) of 17.8 to 30.3, whereas for children the body mass index ranged from the 10th to the 75th percentile for age. Nine participants were receiving antacids (ranitidine or a proton pump inhibitor), 3 were receiving ursodeoxycholic acid (all with mild CF-associated liver disease), 11 were receiving supplemental vitamins, 4 were receiving stable antibiotic regimens, and 10 were taking medications for pulmonary disease. One patient had diabetes mellitus and was receiving insulin. The mean dose of PERT administered was estimated to be 1528 U/kg per meal.
There was no difference between fat absorption measurements assessed by the 72-hour fat balance method compared with the test method (fecal behenate method) while not receiving PERT (Fig. 1). Fat absorption was similar (P = ns) when assessed by the fecal behenate method (51.7% ± 21.6%, mean ± SD) compared with the 72-hour fat balance method (51.5% ± 22.7%). There was, however, a significant difference (P = 0.03) between the fat absorption results assessed using the fecal behenate method (64.5% ± 28.3%) versus the 72-hour fat balance method (80.4% ± 18.4%) with individual results while receiving PERT (Fig. 2).
There was a trend toward concurrence for the 2 methods (Fig. 3). For all of the studies performed for the 3 arms of the study, there was a significant correlation between the fat absorption assessed by the 72-hour fat balance method compared with the fecal behenate method (r2 = 0.219, P = 0.001). We assessed fat absorption during periods when participants were not receiving PERT compared with when they were receiving PERT either before or during meals. For the fecal behenate method, fat absorption was 51.7% ± 21.6% for participants without PERT compared with 64.5% ± 28.3% for participants receiving PERT before meals and 67.6% ± 21.1% for participants receiving PERT during meals. When we compared fat absorption by the 72-hour fat balance method for all of the participants completing all 3 arms of the study, we found fat absorption of 51.5% ± 22.7% for participants without PERT compared with 80.4% ± 18.4% for participants with PERT before meals and 78.3% ± 15.7% for participants with PERT during meals. There was no significant difference (P = ns) in fat absorption when PERT was administered before or during consumption of the meal (Fig. 4) for either method.
The results of the present study showed that the fecal behenate method that measures the fecal lauric acid/behenic acid ratio after a test meal containing lauric acid mixed with other fatty acids correlates with the gold standard, the 72-hour quantitative fecal fat measurement. Unfortunately, the correlation is not sufficiently robust for us to confidently suggest that the behenate method could replace the 72-hour fat balance method for the management of pancreatic enzyme treatment of patients with CF. In our preliminary studies, fat absorption does not appear to be affected by whether enzymes are taken preceding or during the meal by patients with CF. To definitively answer this question, an extremely large sample (n = 329 with 80% power and α of 0.05) would have to be studied to indicate that there are differences in absorption resulting from the timing of enzyme administration.
The goal of this study was to determine whether we could find a facile way to estimate fat absorption in patients with CF that could be used to monitor responses to interventions directed at correcting fat absorption. Given the complications that can result from fat malabsorption, accurate and timely identification is critical in the diagnosis and management of these patients. Although other tests of fat malabsorption have been proposed, these tests are limited in application, accuracy, and feasibility. Several other diagnostic tests to identify fat malabsorption have been proposed. In 1961, Drummey et al (7) reintroduced the technique of microscopic examination of stool for fat globules and outlined a scale for grading this steatorrhea. Although the sensitivity of this method has been reported (8) to be as high as 96%, these results have not been reproducible (9,10). Additionally, this test is only a semiquantitative measure with relatively poor specificity (11), thus patients with fat malabsorption identified by this method still need a 72-hour fat balance study to quantify and confirm this steatorrhea. Another method, the triolein breath test, measures exhaled 14CO2 or 13CO2 after ingestion of a known amount of triglyceride labeled with 14C or 13C. Reports have been mixed as to the sensitivity and specificity of this test, with sensitivity varying from 64% to 100% (12,13). Unfortunately, several factors other than triolein absorption affect the rate of conversion to carbon dioxide, and this test may be inaccurate in many diseases, such as cholestatic liver disease and CF, that lead to fat malabsorption (12,13). Because of these limitations, the triolein breath test is not commercially available for clinical use. Finally, in 1981, Phuapradit et al (14) introduced the steatocrit, a screening method to estimate fecal fat that required only a small volume of stool, making it ideal for use in the pediatric population. This procedure was modified by Tran et al (15,16), who found that acidification of the stool before performing the steatocrit improved the fat separation and increased the sensitivity of this method. Unfortunately, the initial studies by Van den Neucker et al (17) using this method were small and larger studies using the acid steatocrit have not been as promising, with a lower sensitivity and a weak correlation between the acid steatocrit and the 72-hour fat balance method (18). Recently, work using dysprosium chloride as a nonabsorbable marker with stably labeled triglycerides as a method to assess fat absorption has been reported (19). This work has demonstrated that a single marked stool with brilliant blue in which the fractional excretion of a nonabsorbable marker, dysprosium, is assessed and can be compared with the presence of a labeled fatty acid. A high correlation was demonstrated between the fractional excretion of dysprosium chloride and [13C] in stool, indicating that this was a promising method to assess fat absorption on a single marked stool. Unfortunately, the measurement of dysprosium chloride and [13C] requires a mass spectrometry that is not routinely available, and material costs are expensive.
There are few data examining the effects of timing of pancreatic enzyme supplementation in relation to meals. There is anecdotal evidence that administering enzyme supplements during the meal rather than before the meal may reduce gastrointestinal symptoms and potentially enhance fat absorption (P. Campbell, personal communication). There has been no controlled study examining this question. Our results do not suggest a relation between timing of PERT administration. Based upon the small differences observed in fat absorption in relation to timing of PERT administration, the calculated large sample size needed to assess this question makes the performance of such a study in the future unlikely.
The reason for the suboptimal correlation between the fecal behenate method and the 72-hour fat balance method is not clear. There are a number of possibilities:
1. Absorption from a single meal, on which the fecal behenate method is based, may not be representative of the average fat absorption in the 3-day period on which the 72-hour fat balance method is based. There are few data on meal-to-meal and day-to-day variation in fat excretion in normal or diseased humans even though assays for fecal fat have been reported for more than a century (20). Early work on fat absorption does suggest there is day-to-day variation in fat excretion in adults with jaundice reported secondary to cirrhosis or hepatitis (21). The results of additional studies that we have performed examining the effect of Xenical on fat absorption in 10 participants comparing the fecal behenate method with the 72-hour fat balance method using nuclear magnetic resonance analysis had correlations between 0.51 and 0.80 (Heubi and Jandacek, unpublished results).
2. There may be intrinsic variability in fat absorption in CF as suggested by the failure to show robust correlations between the 13C- triolein breath test and the 72-hour fat balance method (22); however, studies by Jongorbani et al suggest that the technique that uses stable-labeled triglyceride with a nonabsorbable marker, dysprosium, given as a single meal has a strong correlation with the 72-hour fat balance method in patients with CF (18).
3. Although the 72-hour fecal fat method is considered the gold standard for assessing fat absorption, there is considerable variability of results in normal subjects and those with CF. In normal subjects, the variability on test-retest can be −8.1 to 5.9%, whereas in CF it may be larger with test-retest values of −19.7% to 42.8% (23). Additional unpublished results indicate large test-retest variability in patients with CF on PERT (−30% to 58%) or on placebo (42 to −42%) (D. Borowitz, personal communication). Additional factors may play a role, although they are to be minor. Conditions causing interruption of the enterohepatic circulation of bile acids, such as ileal resection, lead to intraluminal bile acid concentrations falling below the critical micellar concentration as the day progresses (24). Although patients with CF have mild perturbations in the enterohepatic circulation and none of our participants had a previous ileal resection, it seems unlikely that they have sufficient reductions in intraluminal bile acids with meals during the day to explain the discrepencies (25–27). It is possible that dietary fat containing lauric acid may have compromised the results because the fecal behenate method is dependent upon knowing the dietary lauric acid at and around the time of the test meal. This seems unlikely because lauric acid is contained in only a limited number of foods and the participants' diets were carefully controlled in a hospital setting.
4. Half of the subjects were receiving some type of acid suppression therapy. This may have had some impact on their absorption of dietary fats, especially if there was any variability in their compliance with the medication.
5. Additional factors in CF may have some impact on fat absorption that have been incompletely explored including bacterial overgrowth, variability in release of exogenous pancreatic enzymes during the course of the day, and recently recognized mucosal abnormalities in absorption.
Despite the somewhat discouraging results of the present study, there are some encouraging findings. The behenate test does correlate with fat absorption and, with modification, may ultimately prove useful in this regard because of its ease of sample collection and analytical requirements. Investigators should be encouraged to continue to pursue development of facile methods for assessing fat malabsorption that may be used in adult and children with diarrhea for diagnosis and management of their diseases.
The authors acknowledge the contributions of the research participants and their families; Churee Pardee, BS, RN; and the staff of the GCRCs at Cincinnati Children's Hospital Medical Center and The Children's Hospital; and Procter and Gamble for the gift of sucrose polybehenate.
1. Schmitz, J. Maldigestion and malabsorption. In: Walker WA, Durie PR, Hamilton JR, et al, eds. Pediatric Gastrointestinal Disease: Pathophysiology, Diagnosis, Management. Toronto: BC Decker; 2000, pp 48–55.
2. Van de Kamer JH, Huinink H, Weyers HA. Rapid method for the determination of fat in feces. J Biol Chem 1949; 177:349–355.
3. Jandacek RJ, Heubi JE, Tso P. A novel and non-invasive method of measuring intestinal fat absorption. Gastroenterology 2004; 127:139–144.
4. Metcalfe LD, Schmitz AA, Pelka JR. Rapid preparation of fatty acid esters from lipids for gas chromatographic analysis. Anal Chem 1966; 38:514–515.
5. Schneider MU, Demling L, Jones SA, et al. NMR spectrometry. A new method for total stool fat quantification in chronic pancreatitis. Dig Dis Sci 1987; 32:494–499.
6. Korpi-Steiner NL, Ward JN, Kumar Vivek, et al. Comparative analysis of fecal fat quantitation via nuclear magnetic resonance spectroscopy (1H-NMR) and gravimetry. Clin Chim Acta 2009; 400:33–36.
7. Drummey GD, Benson JA, Jones CM. Microscopical examination of the stool for steatorrhea. New Engl J Med 1961; 264:85–87.
8. Walters MP, Kelleher J, Gilbert J, et al. Clinical monitoring of steatorrhea in cystic fibrosis. Arch Dis Child 1990; 65:99–102.
9. Moore JG, Englert E, Adelbert H, et al. Simple fecal tests of absorption. Dig Dis 1971; 16:97–105.
10. Khouri MR, Huang G, Shiau YF. Sudan stain of fecal fat: new insight into an old test. Gastroenterology 1989; 96:421–427.
11. Benini L, Scuro LA, Menini E, et al. Is the 14C-triolein breath test useful in the assessment of malabsorption in clinical practice. Digestion 1984; 29:91–97.
12. Newcomer AD, Hofmann AF, DiMagno EP, et al. Triolein breath test: a sensitive and specific test for fat malabsorption. Gastroenterology 1979; 76:6–13.
13. Pedersen NT, Jorgensen BB, Rannem T. The 14C-triolein breath test is not valid as a test of fat absorption. Scand J Clin Lab Invest 1991; 51:699–703.
14. Phuapradit P, Narang A, Mendonca P, et al. The steatocrit: a simple method for estimating fat content in newborn infants. Arch Dis Child 1981; 56:725–727.
15. Tran M, Forget P, Van den Neucker A, et al. The acid steatocrit: a much improved method. J Pediat Gastroenterol Nutr 1994; 19:299–303.
16. Tran M, Forget P, Van den Neucker A, et al. Improved steatocrit results obtained by acidification of fecal homogenates are due to improved fat extraction. J Pediatr Gastroenterol Nutr 1996; 22:157–160.
17. Van den Neucker A, Pestel N, Tran T, et al. Clinical use of acid steatocrit. Acta Paediatr 1997; 86:466–469.
18. Wagner MH, Bowser EK, Sherman JM, et al. Comparison of steatocrit and fat absorption in persons with cystic fibrosis. J Pediatr Gastroenterol Nutr 2002; 35:202–205.
19. Schuette SA, Janghorbani M, Cohen MB, et al. Dysprosium chloride as a nonabsorbable gastrointestinal marker for studies of stable isotope-labeled triglyceride excretion in man. J Am Coll Nutr 2003; 22:379–387.
20. Ascoli G. Zur Pathologic der Lebercirrhose. Dtsch Arch Klin Med 1901; 71:387.
21. Gross JB, Comfort MW, Wallaeger EE, et al. Total solids, fat and nitrogen in the feces: V. A study of patients with primary parenchymatous hepatic disease. Gastroenterology 1950; 16:140–150.
22. Kalivianakis M, Minich DM, Bijleveld CMA, et al. Fat malabsorption in cystic fibrosis patients receiving enzyme replacement therapy is due to impaired intestinal uptake of long-chain fatty acids. Am J Clin Nutr 1999; 69:127–134.
23. Borowitz D, Konstan MW, O'Rourke A, et al. Coefficients of fat and nitrogen absorption in healthy subjects and individuals with cystic fibrosis. J Pediatr Pharmacol Ther 2007; 12:477–552.
24. Hofmann AF, Poley JR. Role of bile acid malabsorption in pathogenesis of diarrhea and steatorrhea in patients with ileal resection. Gastroenterology 1972; 62:918–934.
25. Weber AM, Roy CC, Morin CL, et al. Malabsorption of bile acids in children with cystic fibrosis. New Engl J Med 1973; 289:1001–1005.
26. Watkins JB, Tercyak AM, Szczepanik P, et al. Bile salt kinetics in cystic fibrosis: Influence of pancreatic enzyme replacement. Gastroenterology 1977; 73:1023–1038.
27. Strandvik B, Einarsson K, Lindblad A, et al. Bile acid kinetics and biliary lipid composition in cystic fibrosis. J Hepatol 1996; 25:43–48.
© 2010 Lippincott Williams & Wilkins, Inc.