GASTROENTEROLOGY
1990;98:1549-1557
A Report of Three Multiclinic Trials Evaluating Arbaprostil in Arthritic Patients With ASA/NSAID Gastric Mucosal Damage A. R. EULER, M. SAFDI, J. RAO, R. JASZEWSKI, J, WELSH, V. LE, J. RASKIN, R. FLEISCHMANN, M. RAZZAQUE, C. CHAMPION, G. GOTTLIEB, D. NASHEL, L. VAN DER VEER, J, LEVINE, M. BARREIRO, D. CHEATUM, H. MENA, E. BUDIMAN-MAK, S. ROTH, H. GASKILL, B. LEVINE, J. SWINEHART, C. McCLAIN, D. RIFF, D. GRAHAM, T. PHAN, M. BURNS, J. LOOKABAUGH, D. WOOD, and the coinvestigators of the UPJOHN COMPANY ARBAPROSTIL ASA/NSAID GASTRIC MUCOSAL DAMAGE TREATMENT STUDY GROUPS The Upjohn Company, Kalamazoo, Michigan
Three randomized, placebo-controlled multiclinic trials involving arbaprostil dosages of (a) 10 pg; (b) 25 fig; and (c) 10,25, or 50 pg orally for 4 wk in patients older than 18 yr with rheumatoid arthritis or osteoarthritis who had endoscopically documented nonsteroidal antiinflammatory drug-associated gastric mucosal damage were conducted in the United States. All patients continued taking the nonsteroidal antiinflammatory drugs and were reendoscoped after 4 wk of therapy. Success at that time was defined as comp:lete resolution of all gastric mucosal damage. Six hundred fifty-eight patients were enrolled in the three trials. Significantly more patients experienced healing in the arbaprostil treatment groups in all trials compared with those who received placebo. The healing rates in the various trials were 68% and 32% (10 pg vs. placebo; p = 0.007); 77% and 23% (25 pg vs. placebo; p < 0.001); and 529’0, 46%, 35%, and 16% (50, 25, and 10 pg vs. placebo; p < 0.001, ~0.001, and 0.002, respectively). Diarrhea, mostly of a mild nature, was the only arbaprostil-associated side effect and was found with the 25- and 50-pg dosages (33% and 59%, respectively). No exacerbation of arthritis signs or symptoms was found. Arbaprostil at doses with varying effects on gastric acid secretion (25 and 50 pg) was documented in these trials to be an effective and safe agent for healing gastric mucosal damage associated with aspirin or other nonsteroida1 antiinflammatory drugs in patients with either
rheumatoid arthritis or osteoarthritis versely affecting joint symptomatology.
without
ad-
G
astric mucosal damage is a major well-documented side effect of nonsteroidal antiinflammatory drug (NSAID) use. Significant morbidity and occasional mortality have been documented in patients with rheumatoid arthritis and osteoarthritis who must take large daily dosages on a chronic basis to control arthritic symptoms. Of 740 patients with small or large bowel hemorrhage or perforation, Morgan and Worrall found that approximately 25% of those with a perforation were taking NSAIDs at the time of admission (1). In a study in which the association between NSAID use and bleeding peptic ulcer was examined, Somerville et al. found that 80 of 230 elderly patients admitted with bleeding gastric or duodenal ulcers were taking NSAIDs (2). The Food and Drug Administration’s Arthritis Advisory Committee has also stated that a significant body of domestic and international data has documented a high frequency of gastrointestinal complications, including peptic ulcer, associated with NSAID use (3). The Abbreviations used in this paper: ASA, aspirin: NSAID, nonsteroidal antiinflammatory drug. o 1990 by the American Gastroenterological Association 0016-5065/90/$3.00
1550
GASTROENTEROLOGY
EULER ET AL.
economic consequences of these complications were compiled recently in a large multiclinic investigation in which it was estimated that nearly $1.8 billion of rheumatoid arthritis hospital costs and approximately 42% of the admissions in this arthritic population resulted from gastrointestinal complications associated with NSAID use (4). Numerous studies have shown that various prostaglandins, including arbaprostil, protect the stomach against the gastric mucosal damage usually seen after the ingestion of various noxious agents (5-9). Gilbert et al. showed that a lo-pug dose of arbaprostil protected the human stomach when subsequently challenged with aspirin (ASA) while Johansson et al. documented that a 4O-pg t.i.d. dosage of this prostaglandin reduced indomethacin-induced fecal blood loss in patients with rheumatoid arthritis (9,lO). In this we report the results of three placebocontrolled domestic multiclinic trials that evaluated oral arbaprostil dosages of 10 kg; 25 pg; and 10, 25, or 50 pg administered g.i.d. for 4 wk in patients with rheumatoid arthritis or osteoarthritis who were taking ASA or other NSAIDs and had endoscopically documented gastric mucosal damage. Materials and Methods The initial multiclinic trial compared an oral arbaprostil dosage of 10 pg g.i.d. for 4 wk with placebo. The two subsequent trials were also placebo-controlled and evaluated arbaprostil doses of either 25 fig or 10, 25, and 50 rg administered orally g.i.d. for 4 wk. All trials were randomized and double blinded. Except where noted, all of the following were identical in the three trials. The inclusion criteria were as follows: [a] male, or female and surgically sterile or postmenopausal; [b] age >18 yr; (c) endoscopically documented gastric mucosal damage more severe than erythema; and (d) ASA or other NSAID therapy for rheumatoid arthritis or osteoarthritis for at least z wk before entry. The exclusion criteria were a confirmed diagnosis or medical history of the following: (a) renal failure (serum creatinine >2.5 mg/dl); (b) significant cardiovascular disease; (c] hemoglobin concentration ~10 g/dl; (d) acute or chronic hepatic encephalopathy; (e) colitis or regional enteritis; [f) upper gastrointestinal surgery (except cholecystectomy, hiatal hernia repair, or oversewing of a previous ulcer); (g) upper gastrointestinal hemorrhage within the last 4 wk and a blood transfusion of >500 ml]; (h) malignant gastric ulcer; (i] hypochlorhydria (pH > 4.0); (j) psychiatric disorders or any other medical conditions that could preclude adherence to the protocol requirements; (k) gastric ulcer >20 mm in diameter; or (1) treatment with cimetidine, ranitidine, carbenoxolone, sucralfate, anticholinergics, trimipramine, or other tricyclic antidepressants within 1 wk before study drug administration, or antacids at therapeutic doses within 72 h before study entry. Patients meeting the entrance inclusion and exclusion criteria were then randomized to either the arbaprostil or placebo treatment groups in all the studies.
Vol. 98, No. 6
Esophagogastroduodenoscopy was performed before study entry, 4 wk posttreatment, or at the time of withdrawal from the study. The endoscopic scoring system shown in Table 1 was used for the esophagus, fundus, and body, antrum, and duodenum. The following laboratory parameters were obtained at pretreatment, wk 2, and wk 4 [or at time of withdrawal) to assess drug safety: hemoglobin, hematocrit, white blood count, platelet count, calcium, phosphorus, magnesium, triglycerides, cholesterol, uric acid, blood urea nitrogen, creatinine, alkaline phosphatase, serum glutamic oxaloacetic transaminase (SGOT), total protein, albumin, total bilirubin (fractions if elevated), urinalysis, pregnancy test (female at pretreatment only if indicated], and gastric aspirate pH [pretreatment only; measured by either litmus paper or titration at the discretion of the investigator). A history and physical examination were performed on each patient within 24-48 h before entry into the study. Patients were also seen during wk 2 of the study. The following specific data were collected at baseline, at z and 4 wk, or at time of withdrawal from the study: (a) ethanol and tobacco consumption; (b) gastrointestinal and arthritic symptoms; and (c) side effects. Global assessment of the response to treatment was performed by the investigator at the wk 4 visit. The patients maintained diaries, in which they listed the following: (a] time study medication was taken; (b) quantity of antacid tablets taken (antacid administration instructions are detailed later in this section); (c) meal times; [d) frequency and severity of any type of abdominal pain; and (e) incidence of other specific gastrointestinal symptoms and signs such as nausea, vomiting, or diarrhea. The diaries wese reviewed by the investigator at each visit. Each patient received a 2-wk supply of arbaprostil or placebo capsules at the baseline and 2-wk visits with instructions to take the medication 30 min before each meal and at bedtime. Each patient was also given a bottle of Amphojel tablets (Wyeth Laboratories, Philadelphia, Pa.) with instructions to take 2 tablets for relief of abdominal pain lasting more than 15 min. Antacids were not to be taken Table 1. Endosconic
Scoring Svstem Used
Lesion Erythema (severity)
Hemorrhage
(no. present)
Rating None Mild Moderate/severe 0 1 2-5 6-20 >20
Erosion (no. present)
Ulcer (longest diameter]”
0 1 2-5 6-20 rzo 54 mm >4 mm
Score 0 1
2 0 1 2 3 4 0 2 4 6 8 4 8
“This designation was based on the measurement of the ulcer’s longest diameter with either endoscopic biopsy forceps or an ACM1 measuring device.
ARBAPROSTIL IN ARTHRITIC PATIENTS
June 1990
within 1 h before or l/z h after the arbaprostil or placebo capsules. In addition to the review of the patient’s diary, a count of both returned capsules and Amphojel tablets was made at each visit. Antacids, except as outlined above, were not allowed during the trial; if antacids were required, the patient was removed from the study. Throughout the study, all patients continued to receive the sarne NSAIDs they were taking at the time of study entry. The dosage could be increased, but if it was decreased the patients were removed from the trial. The lo-, 25-, and 50-Kg multiclinic trial differed from the 1O-pg and 25-pg trials in the following ways: Patients had total gastric mucosal lesion scores of at least 5. This change was made to restrict entry of patients with only minimal gastric mucosal damage. Administration of antimetabolite or immunosuppressive dru,gs such as methotrexate or azathioprine and antimalarial compounds, which had been allowed in the other two tria.ls. was not allowed because gastric mucosal healing was; a study objective and these compounds have documented adverse gastrointestinal effects. Although patients had received gold, corticosteroids, and penicillamine during the other trials, specific instructions were added to the lo-, 25-, or 50-pg trial protocol. If patients were receiving any of these drugs at the time of entry, they had to have been receiving a stable dose for the 3 mo before that time and were to have no abnormalities on physical examination or in laboratory evaluations that could be attributable to these compounds. Administration of compounds known or believed to be gastric irritants such as KCI, colchichine, or allopurinol were excluded. They were allowed in the other two trials. Success was defined as complete resolution of all gastric lesions or resolution to a point at which only minimal erythema was present. If a patient did not
complete the study or undergo a second endoscopy, that p,atients treatment was considered a failure. Statistical analyses in all three multiclinic trials included the use of linear models. Linear models with treatment and investigator effects in the model and the Cochran-Mantel-Haenszel test were used to compare healing rates. Linear models were also used to test the effects of other variables on healing rates. x2 tests were used for other categorical data. Two-factor analysis of variance on ranks with treatment, investigator, and interaction effects in the model were used to compare continuous data. All tests were two-sided and considered statistically significant at p < 0.05. All study protocols were reviewed and approved by the re.sponsible institutional review boards at the various sites. All patients gave written informed consent. Results Ten, 22, and 39 investigators participated in the lo-pg, 25-hg, and lo-, 25-, or 50-pg multiclinic trials, respectively. The treatment groups in all the trials were similar for the following population demograph-
1551
ics: sex, percentage of female patients (surgically sterilized and postmenopausal); race; age: type of arthritis: duration of arthritic disease; duration of morning stiffness: gastric acid pH; use of tobacco, alcohol, or caffeine; height; and weight except that the percentage of postmenopausal female patients entered in the arbaprostil treatment group was greater than in the placebo treatment group in the 25qg trial (66.7% vs. 39.5%). Ten-Microgram
Trial
One hundred ten patients (arbaprostil, 55; placebo, 55) ranging in age from 22-79 yr with median ages of 57 (arbaprostil group] and 59 (placebo group) yr were enrolled in the 1O-pg trial. Eight patients (4 in each group] withdrew from the trial and did not undergo follow-up endoscopies. The reasons for withdrawal were lack of efficacy (placebo, 1); possible side effects (arbaprostil, 2); surgery (placebo, 1); lost to follow-up (arbaprostil, 2; placebo, 1); and ineligibility for entry (placebo, 11. Twenty-five-Microgram
Trial
In the 25qg trial, 203 patients (arbaprostil, 108; placebo, 95) were entered. They ranged in age from 27-80 with median ages of 59 (arbaprostil group) and 57 (placebo group] yr. Twenty-four patients withdrew from this trial and did not undergo second endoscopies (arbaprostil, 13; placebo, 11). The reasons for withdrawal are shown in Table 2. Ten-,
Twenty-five-,
or Fifty-Microgram
Trial
In the lo-, 25-, or 50-pg trial, 345 patients were randomized to the four treatment groups: 10 pug,84; 25
pg, 86; 50 lug, 82; and placebo, 93. The age ranges were 28-82, 25-82, 33-79, and 30-89 yr in the lo-Ng, 25qg, 50-pg, and placebo treatment groups, respectively. The median ages in the lo-pg, 25qg, 5Oqg, and placebo groups were 59,59,60, and 58 yr, respectively. Forty-three patients were withdrawn from the trial and did not undergo second endoscopies. The reasons for withdrawal are shown in Table 3. Table 2. Reasons for Withdrawal Twenty-five-Microgram
From Trial
Arbaprostil Lack of efficacy Possible side effects Intercurrent illness Lost to follow-up Patient decision Ineligible Protocol violation
3 3 1
2 0
2 2
Placebo 1 2 1 4 2 1 0
1552
EULER
GASTROENTEROLOGY
ET AL
Table 5. Relationship of Initial Gastric MucosaJ Damage
Table 3. Reasons for Withdrawal From the Ten-, Twenty-five-, or Fifty-Microgram Trial
Possible side effects Patient decision Lost to follow-up Ineligible Protocol violation
10 rg
25 tig
56 pg
4 2 1 1 1
6 1 0 1 3
9 0 2 2 2
and Success Rate in Ten-Microgram Twenty-five-Microgram Trial
Placebo 4 0 2 1 1
Intent-to-treat 19 fig Success rate p Value vs. placebo
25 pg
Success
Initial gastric mucosal score l-3 4-10 >lO
The consumption of arbaprostil or placebo capsules was similar in all treatment groups in all three trials. Success in all the trials was defined as complete resolution of all gastric lesions or resolution to the point that only minimal erythema was present. In the 1O-pg trial, the overall success rate when an intent-totreat analysis was performed was 47% (X/55) in the arbaprostil treatment group and 22% (12/55) in the placebo treatment group (p = 0.01). If only evaluable patients were included in the analysis, the success rates were 68% (19/37) and 32% (g/39), respectively (p = 0.007). In the 25-pg trial, the success rate with an intent-to-treat analysis in the arbaprostil group was 49% (53/108) compared with 20% (19/95] in the placebo group (p c 0.001). When the analysis included only evaluable patients, the success rate was 770/o (46/76) in those receiving arbaprostil and 23% (14/64] in those who received placebo (p < 0.001). In the IO-, 25-, or 50-hg trial, the success rates are shown in Table 4. Ethanol ingestion, smoking status, and initial gastric pH were found not to affect the success rates. Success rates were also analyzed based on the sum of the initial gastric mucosal scores. In these analyses, the sums were grouped as l-3, 4-10, and >lO in the lo-Kg and 25-pg trials; in the IO-, 25-, or 50-pg trial, the groupings were 5-10, 11-15, and >15 because in the lo-, 25-, or 50-pg trial, those patients who did not have gastric mucosal scores of at least 5 were not eligible. In the IO-pg trial, the initial degree of mucosal damage had no effect on the success rate, although the arbaprostil success rate tended to be greater with increasing initial gastric mucosal damage, as shown Table 5 (p = 0.32). In the 25-pg trial, a statistically significant effect was seen, and arbaprostil had a more profound effect when more severe degrees of gastric damage were present at baseline (p = 0.01). In the lo-, 25-, or 50-pg trial, no statistically significant differences were found, although large numerical Table 4. Success Rates in Ten-, Twenty-five-,
Vol. 98. No. 6
16 Pg 2/3(67%) 14/32(44%) 7/15(47%)
Placebo
25 pg
2/3(67%) 7/27(26%) 2/23(8'S)
4/7(57%) 31/56(55%) 15/42(36%)
4/7(57%) 13/45(24'S) l/30(3%)
10.0001. Trial
analysis
29% (24/84) 41% (35/85) 44% (36/81)
1990;98:1549-1557
A Report of Three Multiclinic Trials Evaluating Arbaprostil in Arthritic Patients With ASA/NSAID Gastric Mucosal Damage A. R. EULER, M. SAFDI, J. RAO, R. JASZEWSKI, J, WELSH, V. LE, J. RASKIN, R. FLEISCHMANN, M. RAZZAQUE, C. CHAMPION, G. GOTTLIEB, D. NASHEL, L. VAN DER VEER, J, LEVINE, M. BARREIRO, D. CHEATUM, H. MENA, E. BUDIMAN-MAK, S. ROTH, H. GASKILL, B. LEVINE, J. SWINEHART, C. McCLAIN, D. RIFF, D. GRAHAM, T. PHAN, M. BURNS, J. LOOKABAUGH, D. WOOD, and the coinvestigators of the UPJOHN COMPANY ARBAPROSTIL ASA/NSAID GASTRIC MUCOSAL DAMAGE TREATMENT STUDY GROUPS The Upjohn Company, Kalamazoo, Michigan
Three randomized, placebo-controlled multiclinic trials involving arbaprostil dosages of (a) 10 pg; (b) 25 fig; and (c) 10,25, or 50 pg orally for 4 wk in patients older than 18 yr with rheumatoid arthritis or osteoarthritis who had endoscopically documented nonsteroidal antiinflammatory drug-associated gastric mucosal damage were conducted in the United States. All patients continued taking the nonsteroidal antiinflammatory drugs and were reendoscoped after 4 wk of therapy. Success at that time was defined as comp:lete resolution of all gastric mucosal damage. Six hundred fifty-eight patients were enrolled in the three trials. Significantly more patients experienced healing in the arbaprostil treatment groups in all trials compared with those who received placebo. The healing rates in the various trials were 68% and 32% (10 pg vs. placebo; p = 0.007); 77% and 23% (25 pg vs. placebo; p < 0.001); and 529’0, 46%, 35%, and 16% (50, 25, and 10 pg vs. placebo; p < 0.001, ~0.001, and 0.002, respectively). Diarrhea, mostly of a mild nature, was the only arbaprostil-associated side effect and was found with the 25- and 50-pg dosages (33% and 59%, respectively). No exacerbation of arthritis signs or symptoms was found. Arbaprostil at doses with varying effects on gastric acid secretion (25 and 50 pg) was documented in these trials to be an effective and safe agent for healing gastric mucosal damage associated with aspirin or other nonsteroida1 antiinflammatory drugs in patients with either
rheumatoid arthritis or osteoarthritis versely affecting joint symptomatology.
without
ad-
G
astric mucosal damage is a major well-documented side effect of nonsteroidal antiinflammatory drug (NSAID) use. Significant morbidity and occasional mortality have been documented in patients with rheumatoid arthritis and osteoarthritis who must take large daily dosages on a chronic basis to control arthritic symptoms. Of 740 patients with small or large bowel hemorrhage or perforation, Morgan and Worrall found that approximately 25% of those with a perforation were taking NSAIDs at the time of admission (1). In a study in which the association between NSAID use and bleeding peptic ulcer was examined, Somerville et al. found that 80 of 230 elderly patients admitted with bleeding gastric or duodenal ulcers were taking NSAIDs (2). The Food and Drug Administration’s Arthritis Advisory Committee has also stated that a significant body of domestic and international data has documented a high frequency of gastrointestinal complications, including peptic ulcer, associated with NSAID use (3). The Abbreviations used in this paper: ASA, aspirin: NSAID, nonsteroidal antiinflammatory drug. o 1990 by the American Gastroenterological Association 0016-5065/90/$3.00
1550
GASTROENTEROLOGY
EULER ET AL.
economic consequences of these complications were compiled recently in a large multiclinic investigation in which it was estimated that nearly $1.8 billion of rheumatoid arthritis hospital costs and approximately 42% of the admissions in this arthritic population resulted from gastrointestinal complications associated with NSAID use (4). Numerous studies have shown that various prostaglandins, including arbaprostil, protect the stomach against the gastric mucosal damage usually seen after the ingestion of various noxious agents (5-9). Gilbert et al. showed that a lo-pug dose of arbaprostil protected the human stomach when subsequently challenged with aspirin (ASA) while Johansson et al. documented that a 4O-pg t.i.d. dosage of this prostaglandin reduced indomethacin-induced fecal blood loss in patients with rheumatoid arthritis (9,lO). In this we report the results of three placebocontrolled domestic multiclinic trials that evaluated oral arbaprostil dosages of 10 kg; 25 pg; and 10, 25, or 50 pg administered g.i.d. for 4 wk in patients with rheumatoid arthritis or osteoarthritis who were taking ASA or other NSAIDs and had endoscopically documented gastric mucosal damage. Materials and Methods The initial multiclinic trial compared an oral arbaprostil dosage of 10 pg g.i.d. for 4 wk with placebo. The two subsequent trials were also placebo-controlled and evaluated arbaprostil doses of either 25 fig or 10, 25, and 50 rg administered orally g.i.d. for 4 wk. All trials were randomized and double blinded. Except where noted, all of the following were identical in the three trials. The inclusion criteria were as follows: [a] male, or female and surgically sterile or postmenopausal; [b] age >18 yr; (c) endoscopically documented gastric mucosal damage more severe than erythema; and (d) ASA or other NSAID therapy for rheumatoid arthritis or osteoarthritis for at least z wk before entry. The exclusion criteria were a confirmed diagnosis or medical history of the following: (a) renal failure (serum creatinine >2.5 mg/dl); (b) significant cardiovascular disease; (c] hemoglobin concentration ~10 g/dl; (d) acute or chronic hepatic encephalopathy; (e) colitis or regional enteritis; [f) upper gastrointestinal surgery (except cholecystectomy, hiatal hernia repair, or oversewing of a previous ulcer); (g) upper gastrointestinal hemorrhage within the last 4 wk and a blood transfusion of >500 ml]; (h) malignant gastric ulcer; (i] hypochlorhydria (pH > 4.0); (j) psychiatric disorders or any other medical conditions that could preclude adherence to the protocol requirements; (k) gastric ulcer >20 mm in diameter; or (1) treatment with cimetidine, ranitidine, carbenoxolone, sucralfate, anticholinergics, trimipramine, or other tricyclic antidepressants within 1 wk before study drug administration, or antacids at therapeutic doses within 72 h before study entry. Patients meeting the entrance inclusion and exclusion criteria were then randomized to either the arbaprostil or placebo treatment groups in all the studies.
Vol. 98, No. 6
Esophagogastroduodenoscopy was performed before study entry, 4 wk posttreatment, or at the time of withdrawal from the study. The endoscopic scoring system shown in Table 1 was used for the esophagus, fundus, and body, antrum, and duodenum. The following laboratory parameters were obtained at pretreatment, wk 2, and wk 4 [or at time of withdrawal) to assess drug safety: hemoglobin, hematocrit, white blood count, platelet count, calcium, phosphorus, magnesium, triglycerides, cholesterol, uric acid, blood urea nitrogen, creatinine, alkaline phosphatase, serum glutamic oxaloacetic transaminase (SGOT), total protein, albumin, total bilirubin (fractions if elevated), urinalysis, pregnancy test (female at pretreatment only if indicated], and gastric aspirate pH [pretreatment only; measured by either litmus paper or titration at the discretion of the investigator). A history and physical examination were performed on each patient within 24-48 h before entry into the study. Patients were also seen during wk 2 of the study. The following specific data were collected at baseline, at z and 4 wk, or at time of withdrawal from the study: (a) ethanol and tobacco consumption; (b) gastrointestinal and arthritic symptoms; and (c) side effects. Global assessment of the response to treatment was performed by the investigator at the wk 4 visit. The patients maintained diaries, in which they listed the following: (a] time study medication was taken; (b) quantity of antacid tablets taken (antacid administration instructions are detailed later in this section); (c) meal times; [d) frequency and severity of any type of abdominal pain; and (e) incidence of other specific gastrointestinal symptoms and signs such as nausea, vomiting, or diarrhea. The diaries wese reviewed by the investigator at each visit. Each patient received a 2-wk supply of arbaprostil or placebo capsules at the baseline and 2-wk visits with instructions to take the medication 30 min before each meal and at bedtime. Each patient was also given a bottle of Amphojel tablets (Wyeth Laboratories, Philadelphia, Pa.) with instructions to take 2 tablets for relief of abdominal pain lasting more than 15 min. Antacids were not to be taken Table 1. Endosconic
Scoring Svstem Used
Lesion Erythema (severity)
Hemorrhage
(no. present)
Rating None Mild Moderate/severe 0 1 2-5 6-20 >20
Erosion (no. present)
Ulcer (longest diameter]”
0 1 2-5 6-20 rzo 54 mm >4 mm
Score 0 1
2 0 1 2 3 4 0 2 4 6 8 4 8
“This designation was based on the measurement of the ulcer’s longest diameter with either endoscopic biopsy forceps or an ACM1 measuring device.
ARBAPROSTIL IN ARTHRITIC PATIENTS
June 1990
within 1 h before or l/z h after the arbaprostil or placebo capsules. In addition to the review of the patient’s diary, a count of both returned capsules and Amphojel tablets was made at each visit. Antacids, except as outlined above, were not allowed during the trial; if antacids were required, the patient was removed from the study. Throughout the study, all patients continued to receive the sarne NSAIDs they were taking at the time of study entry. The dosage could be increased, but if it was decreased the patients were removed from the trial. The lo-, 25-, and 50-Kg multiclinic trial differed from the 1O-pg and 25-pg trials in the following ways: Patients had total gastric mucosal lesion scores of at least 5. This change was made to restrict entry of patients with only minimal gastric mucosal damage. Administration of antimetabolite or immunosuppressive dru,gs such as methotrexate or azathioprine and antimalarial compounds, which had been allowed in the other two tria.ls. was not allowed because gastric mucosal healing was; a study objective and these compounds have documented adverse gastrointestinal effects. Although patients had received gold, corticosteroids, and penicillamine during the other trials, specific instructions were added to the lo-, 25-, or 50-pg trial protocol. If patients were receiving any of these drugs at the time of entry, they had to have been receiving a stable dose for the 3 mo before that time and were to have no abnormalities on physical examination or in laboratory evaluations that could be attributable to these compounds. Administration of compounds known or believed to be gastric irritants such as KCI, colchichine, or allopurinol were excluded. They were allowed in the other two trials. Success was defined as complete resolution of all gastric lesions or resolution to a point at which only minimal erythema was present. If a patient did not
complete the study or undergo a second endoscopy, that p,atients treatment was considered a failure. Statistical analyses in all three multiclinic trials included the use of linear models. Linear models with treatment and investigator effects in the model and the Cochran-Mantel-Haenszel test were used to compare healing rates. Linear models were also used to test the effects of other variables on healing rates. x2 tests were used for other categorical data. Two-factor analysis of variance on ranks with treatment, investigator, and interaction effects in the model were used to compare continuous data. All tests were two-sided and considered statistically significant at p < 0.05. All study protocols were reviewed and approved by the re.sponsible institutional review boards at the various sites. All patients gave written informed consent. Results Ten, 22, and 39 investigators participated in the lo-pg, 25-hg, and lo-, 25-, or 50-pg multiclinic trials, respectively. The treatment groups in all the trials were similar for the following population demograph-
1551
ics: sex, percentage of female patients (surgically sterilized and postmenopausal); race; age: type of arthritis: duration of arthritic disease; duration of morning stiffness: gastric acid pH; use of tobacco, alcohol, or caffeine; height; and weight except that the percentage of postmenopausal female patients entered in the arbaprostil treatment group was greater than in the placebo treatment group in the 25qg trial (66.7% vs. 39.5%). Ten-Microgram
Trial
One hundred ten patients (arbaprostil, 55; placebo, 55) ranging in age from 22-79 yr with median ages of 57 (arbaprostil group] and 59 (placebo group) yr were enrolled in the 1O-pg trial. Eight patients (4 in each group] withdrew from the trial and did not undergo follow-up endoscopies. The reasons for withdrawal were lack of efficacy (placebo, 1); possible side effects (arbaprostil, 2); surgery (placebo, 1); lost to follow-up (arbaprostil, 2; placebo, 1); and ineligibility for entry (placebo, 11. Twenty-five-Microgram
Trial
In the 25qg trial, 203 patients (arbaprostil, 108; placebo, 95) were entered. They ranged in age from 27-80 with median ages of 59 (arbaprostil group) and 57 (placebo group] yr. Twenty-four patients withdrew from this trial and did not undergo second endoscopies (arbaprostil, 13; placebo, 11). The reasons for withdrawal are shown in Table 2. Ten-,
Twenty-five-,
or Fifty-Microgram
Trial
In the lo-, 25-, or 50-pg trial, 345 patients were randomized to the four treatment groups: 10 pug,84; 25
pg, 86; 50 lug, 82; and placebo, 93. The age ranges were 28-82, 25-82, 33-79, and 30-89 yr in the lo-Ng, 25qg, 50-pg, and placebo treatment groups, respectively. The median ages in the lo-pg, 25qg, 5Oqg, and placebo groups were 59,59,60, and 58 yr, respectively. Forty-three patients were withdrawn from the trial and did not undergo second endoscopies. The reasons for withdrawal are shown in Table 3. Table 2. Reasons for Withdrawal Twenty-five-Microgram
From Trial
Arbaprostil Lack of efficacy Possible side effects Intercurrent illness Lost to follow-up Patient decision Ineligible Protocol violation
3 3 1
2 0
2 2
Placebo 1 2 1 4 2 1 0
1552
EULER
GASTROENTEROLOGY
ET AL
Table 5. Relationship of Initial Gastric MucosaJ Damage
Table 3. Reasons for Withdrawal From the Ten-, Twenty-five-, or Fifty-Microgram Trial
Possible side effects Patient decision Lost to follow-up Ineligible Protocol violation
10 rg
25 tig
56 pg
4 2 1 1 1
6 1 0 1 3
9 0 2 2 2
and Success Rate in Ten-Microgram Twenty-five-Microgram Trial
Placebo 4 0 2 1 1
Intent-to-treat 19 fig Success rate p Value vs. placebo
25 pg
Success
Initial gastric mucosal score l-3 4-10 >lO
The consumption of arbaprostil or placebo capsules was similar in all treatment groups in all three trials. Success in all the trials was defined as complete resolution of all gastric lesions or resolution to the point that only minimal erythema was present. In the 1O-pg trial, the overall success rate when an intent-totreat analysis was performed was 47% (X/55) in the arbaprostil treatment group and 22% (12/55) in the placebo treatment group (p = 0.01). If only evaluable patients were included in the analysis, the success rates were 68% (19/37) and 32% (g/39), respectively (p = 0.007). In the 25-pg trial, the success rate with an intent-to-treat analysis in the arbaprostil group was 49% (53/108) compared with 20% (19/95] in the placebo group (p c 0.001). When the analysis included only evaluable patients, the success rate was 770/o (46/76) in those receiving arbaprostil and 23% (14/64] in those who received placebo (p < 0.001). In the IO-, 25-, or 50-hg trial, the success rates are shown in Table 4. Ethanol ingestion, smoking status, and initial gastric pH were found not to affect the success rates. Success rates were also analyzed based on the sum of the initial gastric mucosal scores. In these analyses, the sums were grouped as l-3, 4-10, and >lO in the lo-Kg and 25-pg trials; in the IO-, 25-, or 50-pg trial, the groupings were 5-10, 11-15, and >15 because in the lo-, 25-, or 50-pg trial, those patients who did not have gastric mucosal scores of at least 5 were not eligible. In the IO-pg trial, the initial degree of mucosal damage had no effect on the success rate, although the arbaprostil success rate tended to be greater with increasing initial gastric mucosal damage, as shown Table 5 (p = 0.32). In the 25-pg trial, a statistically significant effect was seen, and arbaprostil had a more profound effect when more severe degrees of gastric damage were present at baseline (p = 0.01). In the lo-, 25-, or 50-pg trial, no statistically significant differences were found, although large numerical Table 4. Success Rates in Ten-, Twenty-five-,
Vol. 98. No. 6
16 Pg 2/3(67%) 14/32(44%) 7/15(47%)
Placebo
25 pg
2/3(67%) 7/27(26%) 2/23(8'S)
4/7(57%) 31/56(55%) 15/42(36%)
4/7(57%) 13/45(24'S) l/30(3%)
10.0001. Trial
analysis
29% (24/84) 41% (35/85) 44% (36/81)
