Esomeprazole 20 mg od, administered over 1 year, had a favourable safety profile and was well tolerated in Japanese NSAID users with a history of peptic ulcer. Such findings are consistent with another recent study of treatment with lansoprazole (15 mg od) in long-term NSAID users in Japan . Notably, the nature and frequency of AEs with esomeprazole in the present study was generally consistent throughout the 1-year treatment duration, and only a small proportion of patients experienced AEs that were assessed as possibly drug-related by the investigator. Moreover, few serious AEs were reported.
In terms of efficacy (secondary objective), esomeprazole proved efficacious in preventing the recurrence of peptic ulcers in this at-risk patient population, with an overall estimated ulcer-free rate after 1 year of 95.9%. Such findings compare favourably with a similar study of lansoprazole in Japanese long-term NSAID users and a history of peptic ulcer, in which the estimated ulcer-free rate after 1 year was somewhat lower at 87.3% . Moreover, there was a trend for a reduction in endoscopic GI lesions (modified LANZA score) in the present study, and few patients experienced dyspeptic symptoms during study treatment. These results, together with the favourable safety and tolerability findings, are consistent with previous long-term studies of esomeprazole for prevention of pre-specified GI symptoms and peptic ulcers in European and North American patients receiving NSAIDs [12, 13]. Indeed, the ulcer-free rate with esomeprazole 20 mg od in the present 1-year study (95.9%) is better than those seen in the placebo-controlled, 6-month VENUS and PLUTO studies (ulcer-free rates of 94.7% and 94.8%, respectively; life-table estimates) . A double-blind, randomised placebo-controlled trial conducted in Japan also showed superior efficacy of esomeprazole in preventing NSAID-related ulcer recurrence compared with Western studies .
In accordance with earlier studies , we found that esomeprazole seemed to be similarly efficacious for prevention of peptic ulcers irrespective of the type of NSAID that the patient was being treated with (non-selective or COX-2 selective), although patient numbers were small. Unfortunately, we were not able to evaluate the efficacy of esomeprazole according to the duration of NSAID treatment prior to study start, as such information was not collected during baseline assessments. This represents a potential study limitation, as duration of prior NSAID treatment could impact on the risk of ulcer recurrence. H. pylori status did not seem to substantially impact the therapeutic efficacy of esomeprazole (as also observed for lansoprazole in a similar Japanese population ), and this is consistent with findings showing that eradication of this pathogen does not prevent recurrent ulcers in NSAID users . Notably, our study population included a larger proportion of H. pylori-positive patients (43.8%) than comparable studies in Western populations (10–20%) [12, 13].
Esomeprazole is mainly metabolised by CYP2C19 , which is subject to genetic polymorphism. Consequently, we performed genetic testing of CYP2C19 genotype in order to determine whether this influenced long-term efficacy of esomeprazole for prevention of peptic ulcers in our at-risk Japanese population. Overall, CYP2C19 genotype did not indicate any major impact on the gastroprotective efficacy of esomeprazole, as the estimated ulcer-free rates were similar for all CYP2C19 genotypes. While the results are limited by the small number of patients, they are in agreement with previous findings that esomeprazole-mediated healing of reflux (erosive) esophagitis is unrelated to CYP2C19 genotype .
We observed a low incidence of dyspeptic symptoms during the present study, which is consistent with findings elsewhere that esomeprazole provides sustained relief from such symptoms in long-term NSAID users . However, it should be noted that few patients reported such symptoms at baseline; for example, only 18 patients (13.8%) reported epigastric pain with NSAID therapy. This may be somewhat discordant from studies in Western populations that have reported upper GI symptoms, such as dyspepsia, in as many as 40% of NSAID users . Nevertheless, we believe our study population to be representative of Japanese NSAID users, given that the baseline rate of dyspeptic symptoms was generally comparable to the reported incidence in a recent meta-analysis of NSAID treatment studies of osteoarthritis patients in Japan . This difference may be important when interpreting the dyspeptic symptom findings during long-term esomeprazole therapy.
A strength of the present study was that it used endoscopic endpoints rather than surrogate markers of peptic ulcer recurrence, such as symptom relapse. Frequent endoscopic assessment also served to ensure that the at-risk study population was closely monitored for prompt detection of ulcer recurrence. Another strength is that concomitant use of mucosal protectants such as gefarnate, which are frequently used in Japan to reduce the risk of peptic ulcers with NSAID therapy, was not permitted. This helped to ensure that the true gastroprotective effect of esomeprazole was observed, without confounding by the moderate efficacy of mucosal protection noted in comparable long-term studies . The lack of a placebo or active treatment comparator could be considered a study limitation, although use of a placebo arm would be unethical in a study population at high risk of ulcer recurrence. Moreover, the existence of an extensive published literature on esomeprazole in long-term NSAID users allows some historical comparisons to be made. Indeed, findings in the present Japanese population are in agreement with results from other populations [12, 13]. Another limitation is that compliance was measured by returned capsule counting and NSAID diaries. While this was the only feasible option in the current study, we have to accept the limitations of such an approach for evaluating compliance. The small size of the study also has to be considered, as this precludes the precise evaluation of rare AEs such as fracture. Indeed, there has been concern about the moderately elevated risk of fracture (hip, spine and wrist) with PPI long-term therapy based on a recent meta-analysis of observational studies . However, the underlying biological mechanism(s) behind this increased risk remain unclear, and only 3 patients experienced a fracture (hand finger, rib and upper arm) in the present long-term study; in all cases the event was not considered to be related to the study drug. Concerns have also been raised about the increased risk of pneumonia with long-term PPI therapy in some  but not all  studies, although only 1 patient experienced such an event that was considered possibly-related to study drug in the present investigation. Larger studies of PPI therapy in at-risk NSAID users will be required to explore the risk of these rare AEs in greater detail.