Placebo-controlled, randomized, evaluator-blinded endoscopy study of risedronate vs. aspirin in healthy postmenopausal women
M. A.BlankDr Procter & Gamble Pharmaceuticals, Health Care Research Center, 8700 Mason-Montgomery Road, Mason, OH 45040, USA. E-mail: firstname.lastname@example.org
Bisphosphonates are effective treatments for osteoporosis. Since some primary amino bisphosphonates are associated with oesophageal injury, we conducted a study of the upper gastrointestinal effects of risedronate, a pyridinyl bisphosphonate.
Healthy, postmenopausal women received risedronate 5 mg (n=26), aspirin 2600 mg (n=27), or placebo (n=27) daily for 14 days and underwent endoscopy at baseline, Day 8 and Day 15.
Oesophageal erosions were noted in one subject in the aspirin group, two in the placebo group, and none in the risedronate group, and an ulcer in one aspirin-treated subject. Gastric erosions and ulcers were observed most frequently in the aspirin group. Gastric ulcers were noted in eight subjects in the aspirin group, one in the placebo group, and none in the risedronate group (P=0.010, placebo vs. aspirin; P=0.002, risedronate vs. aspirin). Duodenal erosions and ulcers were observed in the aspirin group only. Gastroduodenal erosion scores of three or more occurred more frequently in the aspirin than in the risedronate and placebo groups (P < 0.001).
Risedronate 5 mg was not associated with oesophageal or gastroduodenal ulcers in healthy, postmenopausal women, a population representative of patients who will receive risedronate in the clinical setting.
Nitrogen-containing bisphosphonates are selective inhibitors of bone resorption and have been shown to be effective in the treatment and prevention of postmenopausal osteoporosis. The characteristics of the different bisphosphonates are determined by substitutions on their amino side chains. Among nitrogen-containing bisphosphonates approved for use in humans, pamidronate (Aredia, Novartis Pharmaceutical Corporation, East Hanover, NJ) and alendronate (Fosamax, Merck & Co. Inc., West Point, PA) are configured with primary amino side chains, whereas risedronate (Actonel, Procter & Gamble Pharmaceuticals, Cincinnati, OH) has a pyridinyl side chain. The differences in side chains between these agents may be related to differences in their potential to cause gastrointestinal injury when administered orally. This injury has been attributed to topical irritation,1, 2 which may be correlated with the ability of bisphosphonates to compromise the surface hydrophobic phospholipid barrier of gastric tissue.3 The variability in the gastric toxicity between these compounds may be due to the degree to which they interfere with this barrier (pamidronate > alendronate > risedronate).3
In general, clinical trials have shown that bisphosphonates administered at doses approved for the treatment and prevention of postmenopausal and corticosteroid-induced osteoporosis are well tolerated, with adverse event profiles similar to those of placebo.4–11 However, with the widespread use of nitrogen-containing bisphosphonates in clinical practice, reports have emerged that a proportion of patients treated with these agents have experienced adverse gastrointestinal events, such as abdominal pain and nausea.12, 13 In addition, there has been a higher than expected frequency of severe oesophagitis and oesophageal ulceration;14–16 gastric ulceration and haemorrhage have also been reported.17 Although rare, such injuries may be quite serious.
Risedronate, a pyridinyl bisphosphonate, has been demonstrated in clinical trials to be well tolerated, with an overall gastrointestinal safety profile that is comparable to that in patients who received placebo.6, 7, 9–11 Since nitrogen-containing bisphosphonates have been associated with adverse gastrointestinal effects, a 2-week endoscopy study was conducted in healthy postmenopausal women to determine whether orally dosed risedronate would be associated with similar adverse effects. Short-term endoscopy studies are routinely used to assess the effects on the upper gastrointestinal mucosa of drugs with a potential to induce gastrointestinal irritation.18, 19 The effects on the gastrointestinal mucosa noted in these studies have correlated well with clinical experience.20
In this study, the end-point of primary interest was the incidence of gastric ulcers. The rationale for this choice was that, although mucosal haemorrhages and erosions may occur with gastrointestinal irritants, the development of ulcers is the most clinically relevant process, since acute superficial mucosal injury does not reliably predict the development of ulcers or complications.21 Aspirin, at a dose used by the investigator in a similar study,22 was included as a positive control.
MATERIALS AND METHODS
Postmenopausal women at least 40 years of age and in good general health were eligible for the study. Subjects were required to have normal oesophageal and gastroduodenal mucosa at baseline oesophago-gastroduodenoscopy (EGD). They were also required to be nonsmokers, with no history of smoking during the previous 12 months, and to abstain from using tobacco products during the study. In addition, they were to abstain from ingesting alcohol or vitamin/mineral supplements. Subjects were excluded from participation in the study if they had a history of peptic ulcer disease, oesophagitis, gastrointestinal bleeding or surgery, or any significant illness that contraindicated EGD or administration of the study drug. Additional exclusion criteria were treatment with any bisphosphonate drug within the previous 6 months, or current treatment with H2-receptor antagonists, proton pump inhibitors, aspirin or other nonsteroidal anti-inflammatory drugs (NSAIDs), or any other agent that would confound interpretation of study findings. All subjects gave written informed consent.
Study design and procedures
This was a single-centre, evaluator-blinded, randomized, placebo-controlled gastrointestinal endoscopy study. The protocol was reviewed and approved by the appropriate Institutional Review Board (Research Consultants Review Committee, Austin, Texas). Within 14 days of the start of the study, subjects underwent a baseline EGD and were screened according to inclusion and exclusion criteria. Eligible subjects were randomly assigned to receive one of the following three treatments for 14 consecutive days: risedronate 5 mg/day, aspirin 2600 mg/day, or matching risedronate placebo. EGDs were repeated after 7 (Day 8) and 14 (Day 15) days of treatment.
Study drugs and dosing
Risedronate was supplied as cellulose-film-coated 5 mg tablets (the commercial dosage form) and aspirin as 325 mg tablets. Placebo was supplied as cellulose-film-coated tablets identical in appearance to the risedronate tablets. Subjects in the placebo and risedronate groups did not know which of these two treatments they had been assigned to. All study medications were dispensed by a member of the research team who was not involved with subject evaluations. Subjects in the risedronate and placebo groups were instructed to take one tablet once daily with 180–240 mL or more of water at least 30 min before the first food or drink of the day other than water. They were to take the medication in an upright position and not to lie down for 30 min after taking the tablet. Subjects in the aspirin group were to take two aspirin tablets approximately 30 min before the morning, mid-day, and evening meals and at bedtime (with a snack if desired).
Each subject received a diary card with written dosing instructions and a place to record when the study medication was taken. Compliance was determined by tablet count and by a review of the diary cards. Subjects in all three treatment groups were considered compliant if they were 100% compliant on days 6, 7, 13 and 14 and at least 80% compliant on days 1 through 5 and 8 through 12.
The EGDs on days 8 and 15 for any given subject were performed by the same endoscopist, who was blinded to treatment. The oesophageal, gastric and duodenal mucosae were examined separately. The condition of the mucosa was assessed using adaptations of scales developed and validated by Hetzel23 for the oesophagus and Lanza22 for the stomach and duodenum (Table 1). In these evaluations, mucosal haemorrhage was defined as a red spot with no mucosal defect, erosion as an erythematous superficial mucosal defect that disrupted the epithelium and was not an ulcer, and ulcer as a break in the mucosa at least 3 mm in diameter with unequivocal depth observed by an experienced endoscopist indicating extension through the muscularis mucosa. Biopsy forceps were used to measure ulcer size.
Endoscopic grading scales used in evaluating the condition of the mucosa in the oesophagus (Reprinted from Ref. 23 with permission from the publisher) and stomach and duodenum (Reprinted from Ref. 22 with permission from the American College of Gastroenterology)
Adverse event reporting
During each study visit, subjects were asked whether they had had any changes in their well-being. They were also asked whether they had been hospitalized, had any accidents, used any new medications, or changed concomitant medication regimens (prescription and over-the-counter medications).
Evaluation of data
All data from subjects who were compliant with the treatment and had completed the study were included in the summary of EGD scores. If a subject was compliant during the first week of treatment but not during the second, only data from the first week were included in the evaluation. If a subject was noncompliant during the first week of treatment, then that subject was excluded from the evaluation of data collected at both weeks. The groups were compared with respect to demographic characteristics and history at baseline using one-way ANOVA models (age, weight) and the Mantel–Haenszel test (race, tobacco use, alcohol use). Fisher’s Exact test was used to compare the proportion of subjects with ulcers of the oesophagus, stomach, or duodenum and the proportion of subjects with a gastric and/or duodenal EGD score of 3 or more between the risedronate and placebo groups, the risedronate and aspirin groups, and the aspirin and placebo groups. All statistical tests were two sided, and significance was defined as P ≤ 0.05.
Eighty women were randomly assigned to receive risedronate 5 mg (n=26), aspirin (n=27), or placebo (n=27). Of these, two subjects in the aspirin group were withdrawn at day 8 because they were not compliant with the study treatment. (One of these subjects had a gastric EGD score of 3 at day 8; all other day 8 EGD scores for these subjects were zero.) Data from two additional subjects were excluded from the evaluation of day 15 data. One subject in the risedronate 5-mg group was excluded because the EGD was performed 1 day early, and one subject in the placebo group was excluded because of noncompliance with the study treatment between days 8 and 15. (The subject in the risedronate 5-mg group had EGD scores of zero at all sites on day 15. The subject in the placebo group had an oesophageal EGD score of 2 at day 15 and gastric and duodenal scores of zero.) The treatment groups were similar with respect to age, weight, race and alcohol use (P ≥ 0.193) (Table 2). The groups differed significantly with respect to history of tobacco use (P=0.038). Vital signs at baseline were similar among the treatment groups. All subjects had normal oesophageal and gastroduodenal mucosa at screening.
Demographic characteristics and history of the study groups
No oesophageal lesions were observed in the risedronate 5-mg group during the treatment period. In contrast, oesophageal erosions were noted in one subject in the aspirin group and in two in the placebo group. In addition, one subject in the aspirin group developed an oesophageal ulcer (Table 3).
Number (%) of subjects who had ulcers of the oesophagus, stomach, or duodenum*
Gastric lesions were observed more frequently in the aspirin group than in the other two treatment groups. In the risedronate 5-mg group, gastric erosions were noted in eight (31%) subjects at day 8 and in four (16%) subjects at day 15. In the aspirin group, erosions were observed in 23 (92%) subjects at day 8 and in 24 (96%) at day 15. In the placebo group, gastric erosions were observed in one (4%) subject at each timepoint. No gastric ulcers were noted in the risedronate group, whereas gastric ulcers were observed in eight (32%) subjects in the aspirin group and one (4%) subject in the placebo group (Table 3). The differences in the percentages of subjects with gastric ulcers at day 8 or 15 were significant between the aspirin and placebo groups (P=0.010) and the aspirin and risedronate groups (P=0.002), but not between the placebo and risedronate groups (P=1.00).
Duodenal erosions and ulcers were observed in the aspirin group only. One subject in the aspirin group had a duodenal ulcer at day 8 (Table 3).
In the risedronate 5 mg group, oesophageal EGD scores were zero for all subjects at days 8 and 15, reflecting normal oesophageal mucosa with no lesions. In the aspirin group, EGD scores for the oesophagus were zero for all subjects at day 8. At day 15, two subjects had scores of 2. One of these had erosions and the other had a superficial oesophageal ulcer. In the placebo group, one subject had a score of 2 at days 8 and 15, and one other had a score of 3 at day 15.
On both days 8 and 15, the percentages of subjects with gastric or duodenal EGD scores of 3 or greater were significantly higher in the aspirin group than in the risedronate 5-mg and placebo groups (Table 4). The percentages of subjects with gastroduodenal EGD scores of 3 or greater were higher in the risedronate 5-mg group than in the placebo group, but the differences were not statistically significant. Importantly, the percentage of risedronate-treated subjects with a gastric or duodenal erosion score of 3 or greater at day 15 (1/25 [4.0%]) was only one third of that at day 8 (3/26 [11.5%]).
Number (%) of subjects with gastric or duodenal oesophago-gastroduodenoscopy (EGD) scores of 3 or greater
Gastroduodenal EGD scores of 4, reflecting 10 or more erosions or an ulcer of the stomach and/or duodenum, were recorded on day 8 for one subject in the risedronate 5-mg group and 14 subjects in the aspirin group. The risedronate-treated subject had multiple erosions, but no ulcers. By day 15, fewer erosions were observed in this subject and her EGD score improved accordingly. Of the 14 subjects in the aspirin group, nine had multiple erosions, and five had multiple erosions and one or more ulcers. Four of these subjects had gastroduodenal EGD scores that improved by day 15. At this point, 16 subjects in the aspirin group and one in the placebo group had gastroduodenal EGD scores of 4. Of these 16 subjects in the aspirin group, 12 had multiple erosions, and four had multiple erosions and one or more ulcers. The placebo-treated subject had multiple erosions and two ulcers.
Evaluation of the endoscopic data from the intention-to-treat population of all subjects in this study produced similar results.
Dyspepsia and abdominal pain were the only upper gastrointestinal adverse events observed in this study. Upper gastrointestinal adverse events were experienced by four subjects in the risedronate 5-mg group, 14 in the aspirin group, and six in the placebo group. Differences in the percentages of subjects reporting upper gastrointestinal adverse events were statistically significant between the aspirin and placebo groups (P=0.047) and between the aspirin and risedronate 5-mg groups (P=0.008), but not between the risedronate 5-mg and placebo groups (P=0.728). All of the upper gastrointestinal adverse events experienced by the risedronate-treated subjects were mild. In the aspirin group, upper gastrointestinal adverse events were mild in all but two subjects (both had moderate dyspepsia). In the placebo group, upper gastrointestinal adverse events were mild in all but one subject (who had severe dyspepsia). There was no apparent correlation between the occurrence of upper gastrointestinal adverse events and maximum EGD scores among subjects treated with placebo or risedronate 5 mg. In contrast, aspirin-treated subjects who reported upper gastrointestinal adverse events tended to have higher maximum EGD scores (scores of 3 and 4).
In this study in healthy postmenopausal women with normal oesophageal and gastroduodenal mucosa, risedronate, at a dose for the treatment and prevention of postmenopausal and corticosteroid-induced osteoporosis, was not associated with ulcer formation in any region investigated. These findings are consistent with the results of Phase III clinical trials in 15 066 patients, which have indicated that risedronate has a gastrointestinal safety profile similar to that of placebo.6, 7, 9–11 This profile was based on both reports of upper gastrointestinal adverse events and the results of endoscopic evaluations in the 497 patients who had upper gastrointestinal adverse events and underwent endoscopy. In contrast, oesophageal and duodenal ulcers were observed in one subject each in the aspirin group, and gastric ulcers were observed in eight subjects in the aspirin group and one subject in the placebo group. The absence of an association between risedronate and ulcer formation in this study differs from the findings of endoscopy studies of alendronate, a primary amino bisphosphonate, in which gastric ulceration has occurred consistently in 8% to 10% of the subjects treated.2, 20, 22, 24 In a head-to-head comparison of risedronate vs. alendronate at doses for the treatment of osteoporosis, risedronate was associated with a significantly lower incidence of gastric ulcers than alendronate.25
Despite the development of oesophageal mucosal erosions in the subjects in both the aspirin- and placebo-treated groups in this study, such erosions were not observed among the subjects treated with risedronate. The absence of oesophageal damage after risedronate treatment may be related to the rapid oesophageal transit of risedronate tablets.26 In our study, subjects in the risedronate group received cellulose-film-coated tablets. These tablets, the commercial dosage form, were developed to optimize oesophageal transit and thereby minimize contact between the tablet and the mucosa. In scintigraphic studies of oesophageal transit, no dysmotility or prolonged stasis was observed in any of 25 subjects who received risedronate film-coated tablets.26
Gastric erosions in the risedronate 5-mg group occurred less frequently than in the aspirin group at days 8 and 15 but more frequently than in the placebo group on both days. Importantly, the number of subjects with erosions in the risedronate 5-mg group was lower at day 15 than at day 8, indicating that healing had occurred during continued treatment with risedronate. As noted above, none of the subjects in the risedronate 5-mg group developed oesophageal erosions, whereas one subject in the aspirin group and two subjects in the placebo group developed erosions. Duodenal erosions were observed in the aspirin group only.
The end-point of primary interest in this study was the incidence of gastric ulcers. Although mucosal haemorrhage, erosions and ulcers are often associated with gastrointestinal irritants,27 ulcers are considered the only clinically meaningful injury21 and the surrogate markers for ulcer complications.28, 29 Subepithelial haemorrhages and erosions are confined to the mucosa and therefore do not generally lead to major upper gastrointestinal haemorrhage.30, 31 Ulcers, on the other hand, can lead to serious upper gastrointestinal bleeding.30, 31 The detection of ulcerogenic potential is therefore important in the evaluation of bisphosphonate toxicity.
In addition to evaluating ulcer formation, we compared the percentages of subjects in each group with gastroduodenal EGD scores of 3 or greater. A cut off of 3 was chosen since investigators generally agree that EGD scores of 3 or 4 are indicative of clinically significant injury, whereas scores of 0–2, which consist predominantly of petechial haemorrhages or fewer than five erosions, are not clinically significant.19 Such lesions are common in the general population and do not appear to progress to more serious abnormalities. The percentages of subjects with gastroduodenal EGD scores of 3 or greater were significantly higher in the aspirin group than in the risedronate or placebo groups at both days 8 and 15.
There was no correlation in this study between the occurrence of upper gastrointestinal adverse events and maximum EGD scores in the risedronate 5 mg group. A lack of correlation between the presence of gastric ulcers and upper gastrointestinal symptoms was previously noted in an endoscopic study of alendronate.32 Indeed, there is a frequent dissociation between peptic ulcers and symptoms, and most patients who die from ulcer complications have no symptoms to warn them of the severity of their condition.33 Thus, even though ulcers may be asymptomatic, they are of great clinical significance.
In this study, the instructions for treatment administration for subjects in the aspirin group differed from those for the subjects in the risedronate and placebo groups. It is not known what effect, if any, this difference in ingestion techniques had on the outcome of the study.
In conclusion, at a dose for the treatment and prevention of postmenopausal and corticosteroid-induced osteoporosis, risedronate was not associated with the development of oesophageal, gastric, or duodenal ulcers in healthy, postmenopausal women. These findings contrast with the results of recent studies of another nitrogen-containing bisphosphonate in which the gastric ulceration rate ranged from 8% to 10%. The findings of this risedronate study are consistent with those of large-scale, double-blind, placebo-controlled trials of risedronate in osteoporosis, in which risedronate was shown to have a safety profile that was similar to that of placebo, even in patients with ongoing gastrointestinal disease at baseline and those receiving concomitant treatment with NSAIDs or aspirin.6, 7, 9–11
This study was supported by Procter & Gamble Pharmaceuticals, Cincinnati, Ohio, USA and Aventis Pharmaceuticals, Bridgewater, NJ, USA. The authors acknowledge Mary G. Royer for assisting in the preparation of the manuscript.