Effect of 15(R)-15-Methyl Prostaglandin E2 (Arbaprostil) on the Healing of Duodenal Ulcer

Transcription

1 GASTROENTEROLOGY 1982;83: ALIMENTARY TRACT Effect of 15(R)-15-Methyl Prostaglandin E2 (Arbaprostil) on the Healing of Duodenal Ulcer A Double-Blind Multicenter Study G. VANTRAPPEN, J. JANSSENS, T. POPIELA, J. KULIG, G. N. J. TYTGAT, K. HUIBREGTSE, R. LAMBERT, J. P. PAUCHARD, and A. ROBERT Department of Medicine, University of Leuven, Leuven, Belgium; L Klinik Chirugii, Narutowicz Hospital, Krakow, Poland; Department of Gastroenterology, Wilhelmina Gasthuis, Amsterdam, The Netherlands; Department of Gastroenterology, H6pital Edouard Herriot, Lyon, France; and Department of Experimental Medicine, The Upjohn Company, Kalamazoo, Michigan A multicenter study was conducted on 173 patients with active, endoscopically proven duodenal ulcers (158 men, 15 women). They were randomly assigned, in a double-blind manner, to two groups: those receiving placebo capsules (91 patients) and those receiving capsules containing 100 p,g of ls(r) ls-methyl prostaglandin E2 (arbaprostil) (82 patients). Each drug was ingested four times a day (1 h before meals and at bedtime) for 28 days. Endoscopy was performed on days 0, 14, and 28 after the trial began. At each examination, the ulcer size was measured and whether the ulcer had healed was recorded. Arbaprostil increased the incidence of ulcer healing to approximately the same degree as reported in most extensive studies with cimetidine. Received June 30, Accepted March 2, Address requests for reprints to: Dr. Andre Robert, Department of Experimental Sciences, The Upjohn Company, Kalamazoo, Michigan This study was made possible through the dedicated participation of the following individuals: P. Rutgeerts and L. Broeckaert, Leuven, Belgium; D. Karcz and W. Nowak, Krakow, Poland; M. Schneider, T. Vosmaer, and H. Stamer, Amsterdam, The Netherlands; and S. Beorchia and A. Martin, Lyon, France. We acknowledge the active participation of Dr. Dean 1. Griffith, Department of Experimental Medicine, The Upjohn Company, in the gathering and analysis of the data. We are also indebted to Dr. Arthur E. Hearron, Department of Biostatistics, The Up john Company, for the statistical evaluation of the data. Arbaprostil is the approved generic name (United States Adopted Names Council) for 15(R)-15-methyl prostaglandin E z. The subjects who participated in this study signed an informed consent release, and the study was approved by Institutional Review Boards at each clinical center by the American Gastroenterological Association /82/ $02.50 At 14 days, three times as many patients were totally healed in the arbaprostil-treated as in the placebotreated group (37% vs. 12%, p < 0.001). At 28 days, 67% of patients receiving arbaprostil were healed compared with 39% in the group receiving placebo (p < 0.001). Similarly, the ulcer size, measured endoscopically, was much smaller after arbaprostil administration than in the group receiving placebo after both 14 and 28 days (p < 0.001). Side effects attributable to treatment consisted primarily of loose stools and diarrhea (34%). Smoking retarded healing in the placebo-treated group (p < 0.05), but did not significantly retard healing in patients treated with arbaprostil. We conclude that arbaprostil markedly accelerates the healing rate of active duodenal ulcers. This effect may be due to inhibition of acid secretion as well as to gastric cytoprotection. Numerous reports have shown that prostaglandin Ez (PGEz) and certain methyl analogs of PGEz inhibit gastric acid secretion in animals (1,2) and humans (3,4), and that they prevent the formation of experimental gastric and duodenal ulcers (1,2). Arbaprostil [15(R)-15-methyl prostaglandin Ezl is one of the potent analogs of PGEz. Its mode of action requires conversion of the "(R)" configuration at carbon 15 to the "(S)" configuration. This epimerization is effected in an acid ph, such as found in the stomach, as shown by studies in dogs (5). Gastric ulcers healed faster after a 2-wk treatment with arbaprostil or its methyl ester than after placebo (6,7). Similarly, 15(S)-15-methyl PGEz, methyl ester, and 16,16-dimethyl PGEz, methyl ester, given orally for 2 wk, accelerated the healing of duodenal ulcers (8). Arba-

2 358 V ANTRAPPEN ET AL. GASTROENTEROLOGY Vol. 83, No.2 prostil also inhibited basal, pentagastrin, and mealstimulated gastric secretion in human volunteers (4,9) and duodenal ulcer patients (10,11). In the present multicenter study, 15(R)-15-methyl PGE 2 (arbaprostil) was administered orally for 4 wk to patients with duodenal ulcers. The study was double-blind, and the evolution of the ulcer was followed endoscopically. An abstract was published earlier (12). Methods Patients Of a total of 181 patients entered into the study, 173 completed the 4-wk treatment during which they received either placebo (91 patients: 82 men, 9 women) or arbaprostil (82 patients: 76 men, 6 women). The 8 patients who did not complete the treatment were excluded in the analysis of ulcer data, but were included in the evaluation of side effects. Of these, 1 had been assigned to receive placebo and 7 to receive arbaprostil. The reasons for not completing the study were as follows: (a) the patient who took the placebo was lost to follow-up after day 14 of examination (Lyon); (b) an initial ulcer (day 0) was seen only by spraying methylene blue, and because the endoscopy was normal on day 14, the initial presence of ulcer was uncertain (Lyon); (c) the patient did not follow the treatment schedule (Lyon); (d) the drug was discontinued on day 24 because of vomiting and diarrhea (Lyon); (e) the patient had watery diarrhea on day 2 (Leuven); (f) the patient was hospitalized after 7 days of treatment for vomiting due to gastric obstruction (Leuven); (g) the patient was hospitalized on day 1 because of upper gastrointestinal bleeding; (h) the patient had three ulcers at first endoscopy (Krakow). The conditions of the study and the patient characteristics are given in Table 1. The patients were predominantly men and in each group they were well matched for age, the number of smokers, and for the number of alcohol and coffee users. Inclusion Criteria The following criteria were used for patient inclusions: (a) Patients were between the ages of 18 and 70 yr, except 2 men, 1 74 yr old and the other 76 yr old, both of whom were assigned to placebo therapy. (b) The presence of an active, symptomatic ulcer was visualized by endoscopy. (c) Patients had no prior gastrointestinal surgery. (d) Patients had no overt gastrointestinal bleeding during the month preceding the entry into the study. (e) No patient Table 1. Patient Characteristics Center and Sex No. of Mean aged (yr) treatment patients M F M F Krakow Placebo (21-54) Arbaprostil (18-57) Leuven Placebo (28-76) (35-59) Arbaprostil (29-68) (47-49) Amsterdam Placebo (24-70) (34-62) Arbaprostil (28-62) (32-52) Lyon Placebo (23-56) Arbaprostil (26-70) Total Placebo (21-76) (34-62) Arbaprostil (18-70) (32-52) Initial Mean ulcer size, length of longest Cigarette Alcohol ulcer disease dimension smokers users historya (mm) Yes No Yes No (1-15) (7-14) (1-25) (6-15) (0-30) (4-13) (1-20) (4-10) (1-30) (4-12) (1-27) (3-10) (0-18) (3-20) (1-22) (4-15) (0-30) (3-20) (1-27) (3-15) a Numbers in parentheses denote range.

3 August 1982 PROSTAGLANDIN IN DUODENAL ULCER 359 had serious abnormalities detectable by a thorough physical examination, electrocardiogram, and routine blood and urine laboratory tests, which were performed within 1 wk of entry into the study. (f) Patients had not taken antisecretory drugs (e.g., anticholinergics, cimetidine) or antiinflammatory drugs (including aspirin) during the week before entry into the study. (g) Women were included only if they were not of child-bearing potential or if they were taking oral contraceptives. Medication The structure of arbaprostil is shown in Figure 1. The dose of arbaprostil was 100 J.Lg, contained in a softelastic capsule. Matching placebo capsules were prepared. Design of the Study The patients were randomized within each center in a double-blind manner. Only the hospital pharmacist, who distributed the drug, was in possession of the code. The patient, the endoscopist, and the attending clinician were not aware of the treatment given until all 173 patients had completed the treatment. The subjects were outpatients, and they were told to follow their usual diets. Ninety-one patients received the placebo (82 men, 9 women) and 82 received arbaprostil (76 men, 6 women). The patients were given a bottle containing capsules of either placebo or the prostaglandin in quantities enough to last 2 wks. They were instructed to take one capsule four times a day (1 h before meals and at bedtime). They were also given an unlabeled bottle of antacid tablets, each containing 200 mg of magnesium hydroxide and 200 mg of aluminum hydroxide. These tablets were to be taken only for the relief of severe pain. A diary was given to each patient to record pain frequency and severity, number of antacid tablets consumed, side effects, and number and consistency of stools per day. Diarrhea was defined as watery stools for 6 days or more, consecutive or not, during the 28 days of the clinical trial. Endoscopic Examination The first endoscopic examination was done within 48 h before the start of treatment. It was repeated on days 14 and 28. Each time, the ulcer was measured either with a biopsy forceps (Leuven, Lyon), or with a measuring device (American Cytoscope Makers, Inc., Stamford, Conn.) (Amo ~ \\" _... \\ ~ -~ '" COOH sterdam, Krakow). The longest and shortest diameters were recorded on special forms. For almost every patient, a color photograph of the ulcer was made at each endoscopic examination. An ulcer was diagnosed as healed when the crater had disappeared and the original lesion was covered with what was considered to be new epithelium. Possible Factors Influencing Ulcer Healing On the basis of other studies, the possible influence of certain habits were assessed. In particular, the patients were asked whether they were habitual smokers, alcohol users, or coffee drinkers. The incidence of healing in these subgroups was compared to that of nonusers. Assessment of Results Most patients were seen at weekly intervals; all were seen on days 14 and 28. At each visit, the diary was examined. Any uncertainty was discussed and, whenever possible, clarified. In addition to recording the number of antacid tablets taken each day, the patients were asked to bring back the medication capsules as well as the antacid bottles at each visit so that the number of remaining tablets could be counted. The initial ulcer size was compared with the size measured at days 14 and 28. Statistical Methods Initial ulcer sizes were compared by analysis of variance with treatment, country, and treatment by country effects in the model (13). The mean ulcer size on days 14 and 28 was determined, and the percent difference from day a was calculated. Both groups were compared by using the Student's t-test (14). A similar log-linear model was used to compare ulcer healing rates among countries (15). Results from this latter method are reported as likelihood ratio X 2 statistics. The Fisher's exact test (16) was used to compare dichotomous variables, including the overall incidence of ulcer healing. Continuous variables such as age and laboratory assays were analyzed by Student's t-test (14). The episodes of " ulcer pain" were evaluated statistically in terms of frequency and severity (minimal, moderate, severe) for each week, as well as for the status immediately before the start of the study. The physician who saw the patients at weekly intervals made a final assessment of pain on the basis of the diary and the interviews; the physician was unaware of the treatment given. All tests were two-sided and the results considered statistically significant for p < The effect of the duration of the disease (in years) was compared between patients who healed and those who did not. For this purpose, the Wilcoxon rank-sum test was used (17). HO 15(!!)-15-methyl PGE2 Figure 1. Structure of 15(R)-15-methyl PGEz (arbaprostil). Results Total Healing Arbaprostil markedly accelerated the incidence of ulcer healing. At 14 days, 11 of 91 patients

4 360 VANTRAPPEN ET AL. GASTROENTEROLOGY Vol. 83. No. 2 c=j PLACEBO _15 (R)-15-METHYL PGE 2 ULCERS HEALED (%) P<O.OOl Ulcer Size Changes in ulcer size in both groups after 14 and 28 days are shown in Figure 3. At 14 days, the mean ulcer size was reduced by 32% in patients receiving placebo and by 56% in patients receiving arbaprostil (p < 0.01). At 28 days, the corresponding values were 48% for placebo therapy and 83% for arbaprostil therapy (p < 0.001). Effect of Duration of Ulcer Disease In each group (placebo and arbaprostil), the duration of the ulcer disease did not influence the percentage of patients who were totally healed, except at 14 days, in the arbaprostil-treated group. In these patients, those who had healed (n = 30) had a mean duration of 7.5 yr of ulcer disease whereas those who had not healed (n = 49) had a mean duration of 4.0 yr of disease (p < 0.02) Symptoms Ulcer symptoms were rated "moderate-to-severe" at the start in both groups. After 1 wk, they were rated "minimal-to-moderate" with no statistical difference between groups. Antacid Use Figure 2. Percent of patients completely healed. (12.1%) receiving placebo were healed, compared with 30 of 82 patients (36.6%) receiving arbaprostil (Figure 2). At 28 days, complete healing was seen In 35 of 91 patients (38.5%) receiving placebo and in 55 of 82 patients (67.1%) receiving arbaprostil. On each of these 2 days, the differences were statistically significant (Fisher's exact test, p < 0.001) (Figure 2). Variations existed among locations, especially at 14 days (Table 2). Regardless of the treatment given, fewer patients were completely healed in Krakow than in other centers, whereas the highest percentage of healed patients was in Leuven. In both groups, the number of patients using antacids decreased gradually during the 4-wk of treatment. For each week, fewer patients receiving arbaprostil used antacids than those receiving placebo, although the difference was not statistically significant. In both groups, those who used antacids consumed approximately the same number of tablets per day. Smokers Versus Nonsmokers At day 28, the number of patients who had taken placebo and whose ulcer was healed was much higher among nonsmokers (17 out of 26 or 65.4%) than among smokers (18 out of 65 or 27.7%); Table 2. Number and Percentage of Patients Healed 14 Days 28 Days Placebo 15 (R) PGE 2 Placebo 15 (R) PGE 2 Krakow 3/33 (9.1%) Leuven 6/22 (27.3%) Amsterdam 0/21 (0%) Lyon 2/15 (13.3%) 7/34 (20.6%) 14/16 (87.5%) 5/19 (26.3%) 4/13 (30.8%) 6/33 (18.2%) 15/22 (68.2%) 7/21 (33.3%) 7/15 (46.7%) 19/34 (55.9% ) 15/16 (93.8%) 13/19 (68.4%) 8/13 (61.5%) Total 11/91 (12.1%) P 30/82 (36.6%) < /91 (38.5%) 55/82 (67.1%) <0.001

5 August 1982 PROSTAGLANDIN IN DUODENAL ULCER 361 = '> c[ Q 60 u.. o I- as 50 c..;i cc L.I.I Q.. W 40!::::! en cc ~...J :::::I D PLACEBO 15 (R)-15-METHYl PGE2 (-32"10) P<O.OOl ( -56"10 (-48"10) or arbaprostil. No difference was detected between groups. The details were reported separately (18). Side Effects Side effects were reported by 11 patients (12%) receiving placebo and by 37 patients (41%) receiving arbaprostil. In the placebo-treated group, 4 patients (4%) experienced diarrhea; in the arbaprostil-treated group there were 31 patients (34%). Two of these discontinued treatment because of diarrhea. Patients in the placebo-treated group averaged 1.3 stools/day, compared with 1.8 in the arbaprostiltreated group. One patient receiving arbaprostil had abdominal discomfort, 2 had nausea, and 1 patient was hospitalized after 7 days of treatment because of vomiting attributed to duodenal obstruction. The side effects reported by the placebo-treated group were varied: fatigue, dizziness, semisolid stools, neck and leg pains, and an unspecified skin lesion of both legs. In both groups, no significant changes were seen in laboratory assays. 14 DAYS 28 DAYS Figure 3. Mean ulcer size on days 14 and 28, expressed as percent of size measured on day 0 (before treatment). this difference was significant (p < 0.05). In the group treated with arbaprostil, the difference in ulcer healing between the two subgroups was not statistically significant (11 out of 14 or 78.5% among nonsmokers compared with 44 out of 68 or 64.7% among smokers). Alcohol Users Versus Nonusers In the placebo-treated group, the incidence of ulcer healing was approximately the same in alcohol users (5 out of 38 or 13.2%) as in nonusers (6 out of 53 or 11.3%). In the arbaprostil-treated group, however, more nonusers were healed at day 14 (24 out of 52 or 46.2%) than users (6 out of 30 or 20%) (p < 0.05). The difference between alcohol users and nonusers was not significant at day 28. Coffee Users Versus Nonusers No difference in ulcer healing was noted relative to habitual coffee drinking. Serum Gastrin In one center (Amsterdam), basal and mealstimulated serum gastrin was determined before treatment and after 4-wk of treatment with placebo Compliance The compliance to treatment was estimated in two centers by counting the number of capsules returned to the clinic at the end of the trial. Compliance was 92% in Lyon and 84% in Amsterdam. Discussion This study demonstrates that oral treatment with 15(R)-15-methyl PGE 2 (arbaprostil) accelerates the incidence of healing of active, endoscopically proven duodenal ulcers. The benefit of arbaprostil was particularly evident during the first 2 wk of treatment. Complete healing after 2 wk occurred in three times as many patients as in the placebotreated group (36.6% vs. 12.1%). The ulcer was also significantly reduced in size in the group receiving arbaprostil. Similar increases in the incidence of ulcer healing were reported in most large studies with cimetidine given for 2 and 4 wk (19-26). The lower healing rate in Krakow can be ascribed to two factors. First, the ulcers were initially larger than in other countries (Table 1). Second, it was discovered, after the study was completed, that all patients in Krakow that had been selected for the study had previously failed to respond to medical therapy and had in fact been referred for surgery. They can be classified as intractable. Differences in healing rates existed also between Amsterdam and Leuven, for which an explanation is not apparent. Although ulcer pain is difficult to quantitate, pain appeared to be relieved about equally in both groups. However, as an indirect assessment of pain, the

6 362 VANTRAPPEN ET AL. GASTROENTEROLOGY Vol. 83, No. 2 number of patients requmng antacid tablets was consistently less in the arbaprostil-treated group than in the placebo-treated group. This difference was maintained for each of the 4 wk of the trial. The duration of ulcer disease had little effect on the incidence of healing. A statistical difference was found only after 14 days of treatment with arbaprostil (higher incidence of healing in patients with a longer duration of disease). Binder et al. (19Lreported that patients with a recent onset of ulcer disease (2-3 yr) appeared to have more frequent healing regardless of therapy (cimetidine or placebo). Smoking retarded healing significantly in the placebo-treated group, but not significantly in patients treated with arbaprostil. A similar effect of smoking, also limited to patients receiving placebo, was noted in clinical trials with antacids (27) and cimetidine (22). The reverse was true for alcohol users; alcohol consumption did not influence healing in the placebo-treated group, but in the group receiving arbaprostil fewer users were healed on day 14 than nonusers (20% vs. 46.2%). The reason for this is not apparent. Coffee drinking did not appear to influence ulcer healing. Side effects attributable to treatment were primarily loose stools and diarrhea. Two patients were discontinued from the study because of this side effect. The enhanced healing in arbaprostil-treated patients cannot be explained by changes in serum gastrin levels because there was no statistical difference between both groups. The antiulcer effect of arbaprostil may be due to a dual mechanism, namely, to inhibition of acid secretion and to cytoprotection. This latter property was demonstrated for PGEz in animals and for a variety of PGEz analogs, including several methyl analogs of PGEz (28,29). This property was also demonstrated in humans. Prostaglandin Ez prevented gastric bleeding produced by aspirin (30) and indomethacin (31). Arbaprostil prevented development of gastric mucosal injury produced by aspirin as assessed endoscopically (32). Prostaglandin Ez, given orally, prevented gastric cellular desquamation, measured as deoxyribonucleic acid in gastric washings, produced by intragastric administration of 40% ethanol (33). Because cytoprotection is usually manifest after very low doses of prostaglandins, arbaprostil might still accelerate ulcer healing at a dose lower than the one used in the present study (100 p.,g q.i.d.). A lower dose would not cause loose stools while retaining antiulcer activity. References 1. Robert A, Nezamis JE, Phillips JP. Effect of prostaglandin E2 on gastric secretion and ulcer formation in the rat. Gastroenterology 1968;55: Robert A, Schultz JR, Nezamis JE, et al. Gastric antisecretory and antiulcer properties of PGE2, 15-methyl PGE2, and 16,16- dimethyl PGE2. Intravenous, oral and intrajejunal administration. Gastroenterology 1976;70: Wilson DE, Phillips C, Levine RA. Inhibition of gastric secretion in man by prostaglandin A,. Gastroenterology 1971 ;61 : Karim SMM, Carter DC, Bhana D, et al. Effect of orally and intravenously administered prostaglandin 15(R)-15-methyl E2 on gastric secretion in man. Adv Biosci 1973;9: Robert A, Yankee EW. Gastric antisecretory effect of 15(R)-15- methyl PGE2, methyl ester and 15(S)-15-methyl PGE2, methyl ester. Proc Soc Exp BioI Med 1975;148: Fung WP, Karim SMM, Tye CY. Effect of 15(R)-15-methyl prostaglandin E2 methyl ester on healing of gastric ulcers. Controlled endoscopic study. Lancet 1974;2: Fung WP, Karim SMM. Effect of 15(R)-15-methyl prostaglandin E2 on the healing of gastric ulcers-double blind endoscopic study. Med J Aust 1976;2: Gibinski K, Rybicka I. Mikos E, et al. Double-blind clinical trial on gastroduodenal ulcer healing with prostaglandin E2 analogues. Gut 1977;18: Konturek SJ, Kwiecien N, Swierczek I. et al. Comparison of methylated prostaglandin E2 analogues given orally in the inhibition of gastric responses to pentagastrin and peptone meal in man. Gastroenterology 1976;70: Chen FWK, Teck HS, Karim SMM. The effect of 15(R)-15- methyl prostaglandin E2 on gastric acid secretion in duodenal ulcer patients. Prostaglandins 1977;13: Peterson W, Feldman M, Taylor I, et al. The effect of 15(R)15- methyl prostaglandin E2 on meal-stimulated gastric acid secretion, serum gastrin, and pancreatic polypeptide in duodenal ulcer patients. Dig Dis Sci 1979;24: Vantrappen G, Popiela T, Tytgat GN, et al. A multicenter trial of 15(R)-15-methyl prostaglandin E2 in duodt;mal ulcer (abstr). Gastroenterology 1980;78: Winer BJ. Statistical principles in experimental design. New York: McGraw-Hill, 1962: Snedecor GW, Cochran WG. Statistical methods. 6th ed. Ames, Iowa: Iowa State University Press, 1967: Bishop WMM, et al. Discrete multivariate analysis. Cambridge, Mass.: MIT Press, 1975: Siegel S. Nonparametric statistics. New York: McGraw-Hill, 1956: Conover WJ. Practical nonparametric statistics. 2nd ed. New York: John Wiley and Sons, 1980: Tytgat GNJ, Huibregtse K. The effect of 15(R)-15-methyl prostaglandin E2 on basal and meal-stimulated serum gastrin in duodenal ulcer patients. Prostaglandins 1981;21: Binder HI. Cocco A, Crossley RJ, et al. Cimetidine in the treatment of duodenal ulcer. A multicenter double blind study. Gastroenterology 1978;74: Ippoliti AF, Sturdevant RAL, Isenberg JI, et al. Cimetidine versus intensive antacid therapy for duodenal ulcer. A multicenter trial. Gastroenterology 1978;74: Gillies RR, Archambault A, Kinnear DG, et al. Controlled comparison of two dosage regimens of cimetidine in duodenal ulcer. A multicenter study. Gastroenterology 1978;74: Bardhan KD, Saul DM, Edwards JL, et al. Comparison of two doses of cimetidine and placebo in the treatment of duodenal ulcer: a multicenter trial. Gut 1979;20: Fedeli G, Anti M, Rapaccini GL, et al. A controlled study comparing cimetidine treatment to an intensive antacid regimen in the therapy of uncomplicated duodenal ulcer. Dig Dis Sci 1979;24: Collen MJ, Hanan MR, Maher JA, et al. Cimetidine vs. placebo in duodenal ulcer therapy. Six-week controlled double-blind

7 August 1982 PROSTAGLANDIN IN DUODENAL ULCER 363 investigation without any antacid therapy. Dig Dis Sci 1980;25: Martin DF. May SJ, Tweedle DEF. e t al. Difference in relapse rates of duodenal ulcer after healing with cimetidine or tripotassium dicitrato bismuthate. Lancet 1981;1: Vantrappen G. Rutgeerts p. Broeckaert L. et al. Randomized open controlled trial of colloidal bismuth subcitrate tablets and cimetidine in the treatment of duodenal ulcer. Gut 1980;21 : Peterson WL. Sturdevant RAL. Franke HD. et al. Healing of duodenal ulcer with an antacid regimen. N Eng\ I Med 1977;297: Robert A. Nezamis IE, Lancaster C, e t al. Cytoprotection by prostaglandins in rats. Prevention of gastric necrosis produced by alcohol, HCl, NaOH, hypertonic NaC!, and thermal injury. Gastroenterology 1979;77: Robert A. Cytoprotection by prostaglandins. Gastroenterology 1979;77: Cohen MM. Mucosal cytoprotection by prostaglandin E2. Lancet 1978;2: Johansson C, Kollberg B, Nordeman R, et al. Protective effect of prostaglandin E2 in the gastrointestinal tract during indomethacin treatment of rheumatic diseases. Gastroenterology 1980;78: Gilbert DA, Feld AD, Silverstein FE, et al. 15(R)-15-methyl prostaglandin E2 (PGE2) on cytoprotection in aspirin-induced gastric mucosal injury-an endoscopic study (abstr). Gastroenterology 1981;80: Ruppin H, Person B. Domschke W, et al. Zytoprotective wirkungen von prostaglandin E2 auf die magenschleimhaut beim menschen. Dstch Med Wochenschr 1979;104: