Background: Highly selective inhibitors of the inducible cyclooxygenase-2 enzyme (coxibs) have been associated with less gastrotoxicity than nonselective NSAIDs in clinical studies. Aim: To evaluate the influence of risk factors for NSAID-induced gastrotoxicity on endoscopic and clinical ulcers in patients taking rofecoxib or ibuprofen. Methods: We analysed pooled data from two identical double-blind, randomized, 12-week endoscopy studies which compared the gastroduodenal toxicity of placebo (n=371), rofecoxib 25 mg (n=390), rofecoxib 50 mg (n=379), and ibuprofen 2400 mg daily (n=376) in patients with osteoarthritis. The potential risk factors evaluated were: age (<65, <greater than or equal to>65 years), sex, race (white, nonwhite), Helicobacter pylori status, presence of gastroduodenal erosions at baseline, a history of upper gastrointestinal disease, prior NSAID use within 30 days of study entry, and smoking. We also evaluated these factors for possible association with the development of clinically-evident gastrointestinal perforations, ulcers or bleeds over 12 weeks. Results: Across all treatment groups, the likelihood of detecting endoscopic ulcers, or of clinical presentation with a bleed, over 12 weeks was increased approximately 4-5-fold in patients with previous upper gastrointestinal disease (relative risk [95% confidence interval] of 4.2 [2.5, 7.1] for endoscopic ulcers; 3.8 [1.4, 10.6] for bleeds), or those with gastroduodenal erosions at baseline endoscopy (relative risk of 4.4 [2.6, 7.5] for endoscopic ulcers; 5.0 [1.9, 13.5] for bleeds). H. pylori infection did not increase the risk of endoscopic ulcers or bleeds (relative risk of 1.1 [0.8, 1.6] for endoscopic ulcers; 0.3 [0.1, 0.9] for bleeds). The risk factor subgroup effects were constant across all treatment groups, and the significantly higher incidence of ulcers with ibuprofen as compared to rofecoxib, and placebo was maintained in all risk factor subgroups. Conclusions: Gastroduodenal erosions at baseline and a clinical history of upper gastrointestinal disease, but not H. pylori colonization, increased the risk for endoscopically-detected ulcers and clinical bleeds. Rofecoxib did not magnify the risk in any of the patient subgroups studied.
