Mechanism of Increase in Steatorrhea With Calcium and Magnesium in Exocrine Pancreatic Insufficiency: An Animal Model

Transcription

1 GASTROENTEROLOGY 1982;83: Mechanism of Increase in Steatorrhea With Calcium and Magnesium in Exocrine Pancreatic Insufficiency: An Animal Model DAVID Y. GRAHAM and JEFFREY W. SACKMAN Digestive Disease Section, Veterans Administration Medical Center, and Baylor College of Medicine, Houston, Texas We used a rat model to investigate the phenomenon of increased steatorrhea associated with administration of calcium or magnesium containing antacids in humans with pancreatic insufficiency. Adult male rats with bile and pancreatic duct ligation were fed test meals containing 56 /-Lmol [14C]triolein (0.5 ml), synthetic human bile (1.0 ml, 100 /-Lmol bile salts, 75% glycine and 25% taurine conjugates, and 14.5 /-Lmol lecithin), pancreatic enzymes (0.5 ml), and antacids (1.0 ml). The percent lipid malabsorbed when antacids were fed in addition to the test meal was: control 19.3 ± 1%, NaHC ± 1% (p < 0.05 vs. control), Al(OH) ± 2%, Mg(OH) ± 2% (p < vs. control), and CaC ± 1 % (p < vs. control). With NaCl, Al(OHh, and NaHC0 3 the malabsorbed fat was primarily triolein, whereas with Ca++ or Mg++ the majority of the lipid recovered was oleic acid. Calcium or magnesium administration was associated with precipitation of glycine-, but not taurine-, conjugated bile salts in the small intestine. When calcium was administered to animals in which the bile consisted entirely of glycine-conjugated bile salts, the lipid recovered (64.0 ± 3% malabsorption) was almost entirely triolein suggesting reduced lipolysis. These studies suggest that these divalent cations exert their deleterious effect on replacement enzyme therapy by formation of poorly soluble calcium or magnesium soaps and precipitation of glycine-conjugated bile salts. Received November 24, Accepted April Address requests for reprints to: David Y. Graham, M.D., Veterans Administration Medical Center, 2002 Holcombe Blvd. Room 612. Building la. Houston, Texas The authors thank Dr. David Via for his many helpful suggestions. and Dr. Barbara Segal for helping to initiate the project. This study was supported in part by the Research Service of the Veterans Administration by the American Gastroenterological Association /82/ $02.50 Antacids are used to improve the efficacy of administered pancreatic enzymes by preventing inactivation in the acid environment of the stomach. We recently reported that calcium and magnesium containing antacids increased steatorrhea in patients with pancreatic insufficiency by partially negating the beneficial effect of supplemental pancreatic enzymes (1). Patients with the highest gastric acid secretory rates had the least reduction in steatorrhea with enzymes. These same patients also had the greatest increase in steatorrhea associated with the administration of magnesium or calcium containing antacids. Therefore, those who needed the antacid most experienced the least benefit. The detrimental effect of the divalent cation antacids, calcium carbonate or magnesium hydroxide, was not due to either binding to or inactivation of the administered enzymes (1). The present study was designed to develop a reproducible animal model for studying the effects of antacids on enzyme therapy of fat malabsorption in pancreatic insufficiency and to determine the mechanism or mechanisms by which calcium and magnesium antacids reduce fat absorption. Materials and Methods Experimental Animals Male Sprague-Dawley rats ( g) were housed individually under controlled lighting (14 h of light, 10 h of dark) and temperature conditions (21 ± 1 C). Food and water were available ad libitum. To allow for acclimatization to the facility, experiments were not performed until >7 days after the rats arrived. In preparation for an experiment, both the bile and pancreatic secretions were prevented from entering the intestinal lumen. This was accomplished, under light ether anesthesia. by ligation of the common bile duct at its point of penetration through the duodenal wall. The animals were then fasted

2 September 1982 CALCIUM, MAGNESIUM, AND STEATORRHEA 639 except for free access to water. Adequacy of surgery was verified by visual inspection at the time of death. At least 6 animals were used in each experiment. Forty-eight hours after pancreatic bile duct ligation, the animals were fed a test meal by gavage (3 ml total volume), consisting of triolein, synthetic human bile, and pancreatic enzymes. In addition, an antacid or an equivalent volume of saline (control) was administered. The test lipid consisted of 56 j.lmol of triolein containing 1.5 j.lci of [carboxyl- 14 C)triolein (New England Nuclear, Boston, Mass.). Synthetic human bile consisted of 100 /Lmol of conjugated bile salts (1 ml of a 100-mM solution of 35% glycocholate, 25% glycochenodeoxycholate, 15% glycodeoxycholate, 13% taurocholate, 8% taurochenodeoxycholate, and 4% taurodeoxycholate) and 14.5 /Lmol L-a-phosphatidylcholine (Type VII-E, from egg yolk). Bile salts, triolein, and L-a-phosphatidylcholine were obtained from Sigma Chemicals, and were > 98% pure by thin-layer chromatography (TLC). The lipid and bile were homogenized in an Omni-Mixer (DuPont Instruments, Newtown, Conn.) at maximum speed for 2-3 min immediately before feeding the test meal. The resulting emulsion was stable for at least 1.0 h. Fresh pancreatic enzymes were prepared by mixing one pancreatic enzyme tablet (Ilozyme, Warren-Teed Pharmaceuticals, Columbus, Oh.) in 10 ml 0.9% saline; 0.5 ml of the enzyme solution was administered, which was equivalent to a lipase activity of 190 U as determined by the phstat method (1). The antacids were each administered in a volume of 1 ml and consisted of 150 /Lmol NaCI (control), 100 /Lmol CaC03, 150 /Lmol NaHC0 3, 150 /Lmol AI(OHh (Alternagel, Stuart Pharmaceuticals, Wilmington, Del.), or 137 /Lmol Mg(OHlz (Milk of Magnesia, Phillips Roxane Laboratories, Columbus, Oh.). The order of administration of the test meal was: the lipid-bile mixture immediately followed by the enzymes and antacid. Lipid Extraction Procedures Preliminary experiments revealed that all the intraluminal radioactivity reached the colon within 22 h. Each animal was killed 24 h after feeding, and its colon was removed. The colon, its contents, and 10 ml of 0.9% NaCI were homogenized at maximum speed for 2 min in an Omni-Mixer. The homogenized colon and contents were added to a flask containing 200 ml of a chloroform/methanol mixture (2:1 vol/vol). Any feces that had been passed during the 24 h after the gavage feeding were collected separately and solubilized using the method of Michaels et al. (2), which employs trypsin and quatenary ammonium hydroxide (Sigma Chemicals). This solution was then added to the colon extraction flask for the last 12 h of the 24-h extraction period. The chloroform-methanol extract was transferred to a separatory funnel, and 200 ml of 0.9% NaCI (ph ) added. The funnel was shaken intermittently for 24 h, after which the contents were allowed to settle and the chloroform layer removed. The saline-methanol solution remaining in the separatory funnel was washed twice with 50-ml volumes of chloroform that were added to the chloroform extract. Duplicate 250-/LI aliquots of the total chloroform extract were removed for liquid scintillation counting. The chloroform was then distilled (Rotavapor-R, Brinkman Instruments, Westbury, N.Y.); the temperature was maintained at 35 C to prevent lipid hydrolysis. The lipid remaining was resuspended in 5.0 ml of chloroform. Thin-layer chromatography was performed with duplicate 5-pJ aliquots on 20 x 20 cm, 500-/Lm-thick, Silica gel H plates (Applied Science Labs, State College, Pa.). The solvent system employed was hexanes/ethyl ether/acetic acid (80:20:1, vol/vol/vol). The lipid fractions were visualized in an iodine vapor tank. Each spot was scraped into a scintillation vial for liquid scintillation counting. Standards of triolein (RF = 0.75), diolein (RF = ), monolein (RF = )' and oleic acid (RF = 0.44) (Sigma, 97%-99% purity) were run simultaneously on Elach TLC plate. The saline-methanol fraction remaining in the separatory funnel was acidified to ph 1.3 with concentrated HCI and then reextracted with chloroform as described above. Bile Salt Recovery Experiments To investigate the possible interaction between calcium and glycine or taurine-conjugated bile salts, we modified the test meal by changing the bile salt constituent of the synthetic human bile. Meals were administered with L-a-phosphatidylcholine, and a mixture of glycocholate (75 /Lmol) and taurocholate (25 j.lmol). Labeled bile salt was added ([3H)taurocholate, generally labeled, New England Nuclear, Boston, Mass., and [1-14C)glycocholate, Amersham Radionuclides, Arlington Heights, Ill.) Preliminary experiments revealed that the optimum time to kill the animals, in order to recover the lipids in the small bowel before reaching the colon, was between 2 and 3 h after feeding. Earlier and later time-points were unsatisfactory because at 1 h most of the test meal remained in the stomach and at 4 h some of the test meal had reached the colon. Immediately after death, the small bowel from each animal was isolated by ligatures at the pylorus and ileocecal valve. The small bowel was removed from the mesentery and the contents were manually expressed into preweighed 1.5-ml polypropylene centrifuge tubes (Brinkman Instruments). The tubes were reweighed to obtain the volume collected and then centrifuged for 30 min at 12,500 rpm (Eppendorf Microcentrifuge, Brinkman Instruments). Aliquots of the supernatant were assayed in duplicate for radioactivity. Radiolabel in the precipitate was determined by digesting the material in each centrifuge tube with a quatenary base, hyamine hydroxide, for 48 h. The digestate was transferred to scintillation vials and 400 /LI of glacial acetic acid was added to eliminate chemiluminescence, and radioactivity was determined. Solubility of Calcium Oleate in Bile Salt Calcium [14C)oleate was synthesized from triolein by using the methods of van de Kamer et al. (3) and Harrison (4). Briefly, 20 g of triolein was dissolved in 20 ml

3 640 GRAHAM AND SACKMAN GASTROENTEROLOGY Vol. 83, No.3 of 95% ethanol. Twenty-four milliliters of 10 M NaOH was added followed by 40 ml of 95% ethanol with 0.4% isoamyl alcohol, and the mixture was heated under a reflux condensor for 60 min. The mixture was cooled on ice, 68 ml of 6.8 M HCI was added, followed by 200 ml of petroleum ether. The ether layer was separated and the ether evaporated in a water bath (56 C). The amount of fatty acid produced was measured by titration to ph 7 with 0.1 M NaOH. The fatty acids were resuspended in 145 ml of M NaOH. Eleven grams of anhydrous calcium chloride were added and the mixture was refluxed for 60 min. The mixture was cooled, the calcium oleate separated on a Buchner funnel, and washed with 200 ml of 95% ethanol. The specific activity (2 x 10 4 dpm/mmol) was determined after hydrolysis at ph 1.3 and extraction in petroleum ether. The ratio of calcium to oleic acid in the synthesized soap was found to be 1:2 (moles/moles). Calcium was determined by atomic absorption spectrometry and oleic acid by titration after hydrolysis. Calcium oleate, 29.4 ± 1 mg (98.2 ± 4 /Lmol), was added to centrifuge tubes containing 1.5 ml of glycocholate, taurocholate, or a mixture (0-46 mm) in 0.1 M Sorensen's glycine-sodium hydroxide buffer containing 0.1 M NaCl, ph The tubes were capped, incubated for 2 h at 37 C, and centrifuged at 12,500 rpm for 15 min. Radioactivity was measured in 1 ml of supernatant. A similar experiment was performed in which a glycocholate-taurocholate solution was used at concentrations found in the experiment of bile salt solubility with calcium administration (1.1:2.8, respectively, with mm L-a-phosphatidylcholine). Radioactivity Determinations Liquid scintillation spectroscopy was performed with a refrigerated Packard 8801 liquid scintillation counter using Aquasol 2 (New England Nuclear), a toluene-base liquid scintillation fluor. Results Fat malabsorption was measured by determining the percentage of the ingested lipid recovered in the colon and feces. When the standard test meal containing triolein, bile, pancreatic enzymes, and saline was fed to sham-operated animals, 3.8 ± 1.0% of the ingested triolein was malabsorbed. This contrasts markedly with 83.4 ± 2.8% of the ingested lipid that was malabsorbed when triolein alone was fed to animals with pancreatic and bile duct ligation. Fat malabsorption decreased to 19.3 ± 1.3% when bile and pancreatic eilzymes were replaced orally in animals with pancreatic and bile duct ligation. The addition of antacids to the test meal did not result in a uniform reduction of steatorrhea (Table 1). In fact, only the addition of sodium bicarbonate resulted in a significant decrease (20.7% decrease vs. control) in the amount of fat malabsorbed (p < 0.05 compared with control). Both calcium carbonate and magnesium hydroxide antacids resulted in increased steatorrhea (130.0% and 97.9% increase, as compared with control, respectively) (p < 0.001), whereas aluminum hydroxide did not have a significant effect. Calcium chloride (100 p,mol) was found to give similar results to calcium carbonate (100 p,mol) (42.4 ± 1.0% vs ± 1.3% fat malabsorbed, respectively). Calcium chloride was more convenient to use than CaC0 3 because it is highly soluble in water, which allows calculation of the amount of ionized calcium administered. A dose-response curve was constructed to evaluate the effect of calcium ( p,mol) on fat absorption (Figure 1). The amount of calcium that produced the half-maximal increase in steatorrhea (31.8% fat malabsorbed) was 13.0 p,mol. Coadministration of sodium bicarbonate (50 p,mol) with calcium chloride or carbonate (100 p,mol) abolished the detrimental effect of calcium on fat absorption (18.3 ± 1.2% and 19.9 ± 0.5% fat malabsorbed, respectively) confirming that calcium carbonate per se was not detrimental, but that the ionized calcium produced by the interaction of calcium carbonate with gastric acid was responsible for the increased steatorrhea. Analysis of the products of triolein hydrolysis provided additional insight into the cause of increased steatorrhea (Figure 2). With saline, sodium bicarbonate, and aluminum hydroxide, no addition- Table 1. Effect of Antacids on Enzyme Therapy of Pancreatic Insufficient Rats Test meal fed No. of Group animals Triolein Bile Enzymes Antacid % Malabsorbed 1. Sham-operated Saline 3.8 ± Insufficient 9 + Saline 83.4 ± Insufficient Saline 19.3 ± Insufficient NaHC ± 1.2 b 5. Insufficient AI (OHh 18.3 ± Insufficient Mg (OH), 38.2 ± Insufficient CaC ± o Significantly different from group 3 (saline control) (p < 0.001). b Significantly different from group 3 (saline control) (p < 0.01).

4 September 1982 CALCIUM, MAGNESIUM, AND STEATORRHEA 641 :: c L---L_-'-_'----L_-'-_'----L_-'-_'----L o CALCIUM CHLORIDE ADMINISTERED (,AlIIDI'1 Figure 1. Dose-response curve showing the relationship between the amount of calcium administered and the increase in steatorrhea. The line was constructed by a seconddegree polynomial equation. al lipid was recovered after acidification of the extraction mixtures. With sodium bicarbonate and aluminum hydroxide antacids, there was less triolein and more monolein and diolein recovered than with the saline control (p < 0.001), which suggests that more lipase escaped destruction in the stomach and led to increased hydrolysis. Before acidification of the extraction mixture, the lipid recovered from the colon in the calcium and magnesium experiments was also primarily triglyceride. Acidification of these extracts, however, resulted in the recovery of a large amount of additional lipid that in each instance exceeded the total obtained before acidification. The majority of the lipid co 30 c... c w co ;;; * = p<0.05 * recovered was oleic acid suggesting that it was present as calcium or magnesium soaps. We administered 45Ca to trace its relationship to ingested triolein. When 100 /-Lmol of calcium chloride were administered, 5.3 ;± 1.0 /-Lmol of calcium (15.3% of the total recovered) was recovered in the aqueous layer of the extraction flask. Acidification of the extraction mixture resulted in the appearance of an additional 29.4 ;± 2.1 /-Lmol of calcium in the aqueous layer, and 68.0 /-Lmol were not recovered. The 45Ca was never recovered from the chloroform layer. The lipid recovered after acidification consisted of a mixture of products ranging from triolein to oleic acid, which were insoluble in chloroform until acidification and which were possibly present as a complex mixture. The ratio of calcium to oleic acid recovered after acidification was 1: 1.2, which deviates from the expected ratio of 1: 2. This calculation included any sodium or potassium soap trapped or bound in the insoluble layer before acidification. We did not attempt to isolate the calcium fatty acid soaps [Le., by extraction with diethyl ether (5)], but rather we calculated the ratio from the total amount of oleic acid recovered after acidification. Interaction Between Bile Salts and Calcium The bile salts in synthetic human bile were substituted with 100 /-Lmol of either glycocholate or taurocholate. The percent fat malabsorbed increased with glycocholate compared with mixed synthetic human bile (25.7 ;± 1.9% vs ;± 1.3%, respectively, p < 0.02) (Figure 3). The addition of calcium chloride (100 /-Lmol) resulted in increased steator- I *'P<O.OI co 50 CI++ CI++ :: ! '" _ D OLEIC ACIO MONOLEIN DIOLEIN TRIOLEIN Figure 2. The total amount and type of lipid recovered upon acidification of the homogenized colon and feces collected after feeding test meals with antacids (calcium carbonate or magnesium hydroxide) or saline. See Materials and Methods for details of experiment. Asterisks denote a statistical difference from the saline control, p < MIXED TAUROCHOLATE GLYCOCHOLATE c::::::j OLEIC ACID M.IOLEII OIOLEIN TRIOLEIN Figure 3. The effect of calcium administration on fat malabsorption when the standard mixed bile salt solution was replaced by glycocholate or taurocholate. Each pair of bars represent saline control compared with feeding calcium with the meal. Asterisks denote a statistical difference from the saline control, p < 0.01.

5 642 GRAHAM AND SACKMAN GASTROENTEROLOGY Vol. 83, No.3 rhea with each bile salt preparation: glycocholate, taurocholate, and the mixed synthetic human bile salt solution. The pattern of triolein hydrolysis with glycocholate was different from that observed with either taurocholate or mixed bile salts. The decrease in fat absorption with glycocholate was primarily related to an increased recovery of triolein and a concomitant decrease in oleic acid recovery (i.e., diminished hydrolysis) (p < compared with taurocholate or mixed bile salts). The effect of calcium on triolein hydrolysis and malabsorption was compared for each bile salt. With mixed bile salts there was a significant increase in oleic acid, monoglyceride, and total fat recovered (p < 0.04). With taurocholate there were significant increases in oleic acid, diglyceride, and total triolein recovered, and with glycocholate there were significant increases in each fraction: oleic acid, monolein, diolein, triolein, and total fat recovered (p < 0.04). Two-way analysis of variance revealed a significant effect on total fat malabsorbed of calcium (p < 0.001), bile salt (p < 0.035), and an interaction between calcium and bile salts (p < 0.001). To directly evaluate the effect of calcium on bile salt solubility, we collected postprandial small intestinal contents after ingestion of the test meal with glycocholate-taurocholate bile mixture (75:25). The volume of intestinal contents recovered in calcium chloride experiments was significantly greater than that recovered in control (NaCI), with both glycocholate (p < 0.001) and with taurocholate (p < 0.001) (Table 2). The amount of soluble taurocholate was similar in both control and calcium chloride experiments. However, the amount of glycocholate was markedly reduced compared with control animals (p < 0.001) (Table 2). The ph of the intestinal contents recovered in the calcium experiments was similar to that of the saline controls (ph> 5.0) eliminating the possibility that glycocholate precipitated as a result of the ph being below the pka of the bile salt. The effects of divalent cations on bile salt solubility were repeated with calcium carbonate and magnesium hydroxide and yielded similar results (data not shown). Solubility of Calcium Oleate in Bile Salt Solutions Calcium oleate was uniformly more soluble in glycocholate than taurocholate at bile salt concentrations between 0.5 and 20.0 mm (Figure 4). Each calcium oleate molecule theoretically contains two fatty-acid molecules, and therefore the amount of the oleic acid in solution is twice as great as the calcium oleate value. For example, the solubility of calcium oleate in 20 mm glycocholate is 2.5 mm and it would contain the same amount of oleic acid as a 5.0 mm oleic acid solution. The solubility of calcium oleate in 20 mm taurocholate is 1.9 mm, which is equal to a 3.8-mM oleic acid solution. An additional experiment was performed using glycocholate-taurocholate solutions at concentrations observed in the small bowel after calcium treatment (1.1: 2.8 mm respectively, with mm phospholipid). Calcium oleate solubility was 2.3 ± 0.3 film. In our synthetic human mixed bile salt solution (bile salt concentrations of 5, 10, or 15 mm) containing phospholipid, calcium oleate solubility was 2.0, 2.6, and 3.2 mm, respectively. Discussion Lipase is irreversibly inactivated at ph 4 or below (6). The major barrier to complete resolution of steatorrhea in patients with pancreatic insufficiency is the destruction of administered enzymes in the acid environment of the stomach. We have previously demonstrated (7), and Dimagno et al. (8) have confirmed, that the reduction in steatorrhea in these patients is related to the duration the intragastric ph remains above 4 after the ingestion of enzymes. Antacids should enhance the effectiveness of enzyme therapy by increasing the time the intragastric ph remains above 4. In humans with pancreatic insufficiency, and in this animal model, calcium- or magnesium-containing antacids exacerbated steatorrhea. We used an animal model to study the pathogenesis of the enhanced steatorrhea, and we demonstrated at least two detrimental events: the formation of Table 2. Effect of Calcium on Bile Salt Solubility In Vivo Supernatant Precipitate Glycocholate Taurocholate Glycocholate Taurocholate Antacid Volume I-tmol mm I-tmol mm I-tmol I-tmol Control ±0.1 ±3.4 ±1.7 ±1.2 ±1.4 ±2.3 ±0.8 Calcium 5.1 b 5.3 b 1.1 b b ±0.2 ±1.8 ±0.2 ±1.4 ±0.2 ±3.5 ±0.3 Q P < b P < determined by Student's t-test, 2-tailed, compared with saline control.

6 September 1982 CALCIUM, MAGNESIUM, AND STEATORRHEA E iii 2.0 => -' c:> en... t- oe... -' -- r 't 1.0.Ii'J"" o O - -L-_----'- L -'--_---..I o BILE SALT CONCENTRATION '.. I Figure 4. Solubility of calcium oleate in taurocholate (open circles) and glycocholate (closed circles) solutions in 0.1 M Sorensen's glycine-sodium hydroxide buffer containing 0.1 M NaCl, ph , 37 C for 2 h. calcium soaps and the precipitation of glycine-conjugated bile salts. The administration of calcium has been associated with a decrease in fat absorption in animals (9-11), healthy adult men (minimal change) (12), and infants (large increase in steatorrhea) (5,13-15). The majority of the fat present in the stools of these experiments has been in the form of calcium soaps.. Previous workers have noted that the production of calcium soaps is directly related to the amount of calcium in the diet and to the solubility of the calcium compound (16). Both calcium carbonate and magnesium hydroxide are largely insoluble in the large and small intestines, although they are excellent antacids. Gastric acid increases calcium or magnesium bioavailability for either absorption or for interaction with fatty acids. Thus, the beneficial effect-enhancing lipase activity by preventing lipase inactivation-may be associated with a detrimental effect-reducing fatty acid absorption in the small bowel. The amount of calcium fatty-acid soap appearing in the colon and feces is the sum of a number of related events. The important variables include: the availability of ionized calcium or magnesium, the solubility of the calcium soap in the contents of the small bowel, the concentration and type of bile salt present in luminal contents, and the length and integrity of the mucosal surface of the small bowel. Bosworth et al. (14) were probably the first to conclusively demonstrate calcium soaps in stools of children. They showed that the less-soluble longchain saturated fatty acids were the most likely to be malabsorbed when calcium was administered. After the administration of calcium or magnesium, the types of fatty acids found in stools are predictable if one knows the composition of the fats in the diet and the position of the poorly soluble fatty acids on the glycerol backbone (2 position allows absorption as monoglyceride) (17,18). The long-chain unsaturated fatty acids are better absorbed if the other liberated fatty acid (lor 3 position) is unsaturated. Calcium soaps are often referred to as insoluble even though their solubility in bile was demonstrated as early as the 1920s (19-21). Bosworth et al. (14) showed that the calcium soap of the volatile fatty acids were soluble in bile, and they presented indirect evidence suggesting that calcium oleate was more soluble in bile than calcium palmitate or calcium stearate. They concluded that the presence of ionized calcium in the intestine determined the extent of soap formation, although the amount ultimately excreted in the feces depended upon the solubility of the soap formed. Similar conclusions concerning the effects of magnesium or calcium were reached by Cheng et al. (11), Telfer (16), Boyd et al. (10), and Givens (22). Two experimental approaches were employed to test whether bile salt precipitation was important in our model. If calcium or magnesium precipitated glycine-conjugated bile salts, we expected increased steatorrhea in animals fed glycocholate compared with those fed either taurocholate or a synthetic human bile preparation. In fact, steatorrhea was 52% greater in animals whose bile contained only glycine-conjugated bile salts. There was also a difference in the pattern of lipid recovered from the stools of these animals. In the animals with mixed or taurine-conjugated bile, fecal lipid was primarily oleic acid soaps, whereas it was primarily triolein in the animals with glycine-conjugated bile salts. This marked reduction in lipid hydrolysis was also consistent with precipitation of glycocholate. In order to confirm precipitation of glycocholate, we measured the amount of glycocholate and taurocholate in the small bowel contents of rats after calcium or magnesium administration. Glycocholate, but not taurocholate, was precipitated, suggesting that the precipitation of glycine-conjugated bile salts may be responsible in part for the deleterious effects of divalent cations on fat absorption. The precipitation of glycine-conjugated bile salts by divalent cations, while of interest, is of unknown clinical importance. After the ingestion of calcium, infants may experience defective intraluminallipolysis as shown by the recovery of both calcium soaps and triolein (5). This suggests that precipitation of glycine-conjugated bile salts may playa role in this interaction, especially as infants have a reduced bile salt pool and a limited mucosal surface (23). The fact that steatorrhea did not improve when bile without glycine conjugates was administered suggests that the precipitation of glycine-conjugated bile salts did not playa major role in the increased

7 644 GRAHAM AND SACKMAN GASTROENTEROLOGY Vol. 83. No.3 steatorrhea associated with calcium or magnesium administration in our animals. A possible explanation is that a combination of events occurred, including the reduction of the bile salt and lipase concentration because of dilution, and the formation of divalent soaps of fatty acids that are of limited solubility in luminal contents. The reduced concentration of lipase and bile salts would retard lipolysis and require more intestinal surface to accomplish absorption of dietary fat (24). This would be further complicated by the presence of poorly soluble calcium (or magnesium) soaps and calcium soap-lipid complexes. References 1. Graham DY. Pancreatic enzyme replacement: the effect of antacids or cimetidine. Dig Dis Sci 1982;(in press). 2. Michaels EB, Hahn EC, Kenyon AJ. An improved procedure for solubilization and assay of blood and feces in liquid scintillation counting. Anal Biochem 1979;99: van de Kamer JH, Huinink HB, Weyers HA. Rapid method for the determination of fat in feces. J BioI Chem 1949;177: Harrison GA. A note on the solubilities of calcium soaps. Biochem J 1924;18: Watkins JB, Bliss CM, Donaldson FM, Jr, Lester R. Characterization of newborn fecal lipid. Pediatrics 1974;53: Heizer WD, Cleaveland CR, Iber R1. Gastric inactivation of pancreatic supplements. Bull Johns Hopkins Hospital 1965;116: Graham DY. Enzyme replacement therapy of exocrine pancreatic insufficiency in man: relation between in vitro enzyme activities and in vivo potency in commercial pancreatic extracts. N Engl J Med 1977;296: Dimagno EP. Medical treatment of pancreatic insufficiency. Mayo Clin Proc 1979;54: Tadayyon B. Lutwak 1. Interrelationship of triglycerides with calcium. magnesium, and phoshorus in the rat. J Nutr 1968;97: Boyd OF. Crum CL, Lyman JL. The absorption of calcium soaps and the relation of dietary fat to calcium utilization in the white rat. J BioI Chern 1932;95: Cheng ALS, Morehouse MG. Deuel HJ Jr. The effect of the level of dietary calcium and magnesium on the digestibility of fatty acids. simple triglycerides. and some natural and hydrogenated fats. J Nutr 1949;37: Drenick EJ. The influence of ingestion of calcium and other soap-forming substances on fecal fat. Gastroenterology 1961;41: Katz L, Hamilton JR. Fat absorption in infants of birth weight less than 1,300 gm. J Pediatr 1974;85: Bosworth AW. Bowditch HI, Giblin LA. Studies of infant feeding. X. The digestion and absorption of fats. I. Calcium in its relation to the absorption of fatty acids. Am J Dis Child 1918;15: Hanna FM. Navarrete DA. Hsu FA. Calcium-fatty acid absorption in term infants fed human milk and prepared formulas simulating human milk. Pediatrics 1970;45: Telfer SV. Studies in calcium and phosphorus metabolism. III. Absorption of calcium and phosphorus and their fixation in the skeleton. Q J Med 1923; Mattson FH, Nolen GA, Webb MR. The absorbability by rats of various triglycerides of stearic and oleic acid and the effect of dietary calcium and magnesium. J Nutr 1978;109: Jensen GR, Hagerty GJ, McMahon KE. Lipids of human milk and infant formulas: a review. Am J Clin Nutr 1978;31: Adler E. Uber die einwirkung von gallensauren auf die losung von kalkseifen im stuhl. Arch Fuer Verbauungskr 1927; 40: Oelsner A, Klinke K. Beitrag zur frage der kalkseifenstuhle. Jahrb Kinderheilk 1928;122: Gacs G, Barltrop D. Significance of Ca-soap formation for calcium absorption in the rat. Gut 1977;18: Givens MH. Studies in calcium and magnesium metabolism. III. The effect of fat and fatty acid derivatives. J BioI Chern 1917;31: Grand RJ, Sutphen JL. Montgomery RK. The immature intestine: implications for nutrition of the neonate. In: Development of mammalian absorptive processes, Ciba Foundation Symposium 70. Amsterdam: Excerpta Medica, 1979: Ockner RK, Pittmen JP, Yager JL. Differences in the intestinal absorption of saturated and unsaturated long chain fatty acids. Gastroenterology 1972;62: