Effect of Risk Factors on Complicated and Uncomplicated Ulcers in the TARGET Lumiracoxib Outcomes Study

Transcription

1 GASTROENTEROLOGY 2007;133:57 64 Effect of Risk Factors on Complicated and Uncomplicated Ulcers in the TARGET Lumiracoxib Outcomes Study CHRISTOPHER J. HAWKEY,* WILFRED M. WEINSTEIN, WALTER SMALLEY, XAVIER GITTON, PETER SALLSTIG, KIRSTIN STRICKER, GERHARD KRAMMER, BERNHARD MELLEIN, DOMINIK RICHARD, and PATRICE MATCHABA *Wolfson Digestive Diseases Centre, University Hospital, Nottingham, United Kingdom, Division of Digestive Diseases, David Geffen School of Medicine at University of California Los Angeles, Los Angeles, California, Vanderbilt University, Nashville, Tennessee, Novartis Pharma AG, Basel, Switzerland, and Novartis Pharmaceuticals Corporation, East Hanover, New Jersey Background & Aims: Selective cyclooxygenase-2 inhibitors were developed to reduce the gastrointestinal risk associated with nonsteroidal anti-inflammatory drugs (NSAIDs). The Therapeutic Arthritis Research and Gastrointestinal Event Trial was the largest study to evaluate primarily the gastrointestinal safety outcomes of selective cyclooxygenase-2 inhibitors. Data from the Therapeutic Arthritis Research and Gastrointestinal Event Trial were used to identify risk factors and investigate the safety of lumiracoxib in subgroups. Methods: Patients with osteoarthritis (age, >50 y) were randomized to receive lumiracoxib 400 mg once daily, naproxen 500 mg twice daily, or ibuprofen 800 mg 3 times daily for 12 months. Events were categorized by a blinded adjudication committee. The primary end point was all definite or probable ulcer complications. Results: For patients taking NSAIDs, factors associated with an increased risk of ulcer complications were age 65 years or older (hazard ratio [HR], 2.30; 95% confidence interval [CI], ), previous history of gastrointestinal bleed or ulcer (HR, 3.61; 95% CI, ), non-caucasian racial origin (HR, 2.10; 95% CI, ), and male sex (HR, 1.70; 95% CI, ). With lumiracoxib, significant risk factors were age 65 years or older (HR, 3.18; 95% CI, ), male sex (HR, 2.60; 95% CI, ), non-caucasian racial origin (HR, 2.16; 95% CI, ), and concomitant aspirin use (HR, 2.89; 95% CI, ). Increased risks in patients age 65 years and older were increased further if other risk factors were present. Lumiracoxib maintained an advantage over NSAIDs across all subgroups except aspirin use. Conclusions: Lumiracoxib was associated with a reduced risk of ulcer complications compared with NSAIDs in all significant subgroups except aspirin users. Patients with chronic painful conditions, of which the most prevalent is osteoarthritis (OA), commonly use nonselective nonsteroidal anti-inflammatory drugs (NSAIDs). 1,2 The gastrointestinal (GI) toxicity of nonselective NSAIDs is well recognized 3 and is estimated to cause 1000 deaths every year in the United Kingdom. 4 Selective cyclooxygenase-2 (COX-2) inhibitors were developed to reduce the GI toxicity associated with nonselective NSAIDs. Lumiracoxib is a highly selective COX-2 inhibitor with a COX-2/COX-1 selectivity ratio of 515, compared with ratios of 37 for celecoxib, 141 for rofecoxib, and 265 for etoricoxib. 5 In addition, lumiracoxib has a short half-life ( 4 h), 6 and its rapid clearance from the circulation might allow recovery periods, which may aid its systemic tolerability. Despite its short half-life, lumiracoxib provides effective analgesia with once-daily dosing, possibly because its weakly acidic nature allows the drug to distribute preferentially into inflamed tissue 7 where concentrations persist beyond those in the circulation. 8 Large GI outcomes studies have been performed to assess the GI benefit of selective COX-2 inhibitors in comparison with controls. The Therapeutic Arthritis Research and Gastrointestinal Event Trial (TARGET; n 18,325) was designed to establish the GI safety of lumiracoxib compared with ibuprofen and naproxen. TARGET is the largest GI outcomes study in OA published to date. Patients 50 years and older with OA, including approximately 24% of patients receiving low-dose aspirin, received lumiracoxib 400 mg once daily (4 times the therapeutic dose in OA), naproxen 500 mg twice daily, or ibuprofen 800 mg 3 times daily for 1 year. 9 For the primary end point the incidence of definite or probable upper GI ulcer complications (bleeding, perforation, or obstruction) lumiracoxib showed a highly significant GI benefit compared with the nonselective NSAIDs naproxen and ibuprofen. However, it is important to identify those patients who are at highest risk and, indeed, to show benefits in them. Previous studies have suggested that increasing age, previous history of GI bleed or ulcer, aspirin use, and NSAID Abbreviations used in this paper: BMI, body mass index; CI, confidence interval; COX, cyclooxygenase; CV, cardiovascular; GI, gastrointestinal; HR, hazard ratio; NNT, number of patients needed to be treated; NSAIDs, nonsteroidal anti-inflammatory drugs; OA, osteoarthritis; TARGET, Therapeutic Arthritis and Gastrointestinal Event Trial; UGIT, upper gastrointestinal tract by the AGA Institute /07/$32.00 doi: /j.gastro

2 58 HAWKEY ET AL GASTROENTEROLOGY Vol. 133, No. 1 dose are among the most important risk factors for NSAIDassociated ulceration The status of smoking, Helicobacter pylori, alcohol, sex, race, body mass index (BMI), and cardiovascular (CV) risk profile are all less certain. The Vioxx Gastrointestinal Outcomes Research (n 8076) study investigated risk factors, but did not include patients taking aspirin and was restricted to patients with rheumatoid arthritis, which represents a small proportion of patients using NSAIDs. 10 In addition, the Vioxx Gastrointestinal Outcomes Research study used the end point of clinical upper GI events (defined as complicated upper GI events and uncomplicated upper GI ulcers), often referred to as perforations, ulcers, and bleeds. In contrast, the TARGET study specifically assessed the incidence of ulcer complications (bleeds, perforations, or obstructions) as its primary end point. We therefore investigated the risk factors for complicated and uncomplicated ulcers in patients taking lumiracoxib compared with those taking nonselective NSAIDs (naproxen or ibuprofen) in the TARGET study, and we assessed whether the GI safety of lumiracoxib shown in the primary analyses was maintained across subgroups. Materials and Methods The TARGET methods have been published previously 9 and only a brief overview is provided here. TARGET was a randomized, double-blind, double-dummy, parallelgroup, active-controlled, international, multicenter study in patients with OA. TARGET consisted of 2 substudies of identical design but used different comparators. The study was undertaken in accordance with Good Clinical Practice Guidelines and the Declaration of Helsinki, and all patients provided informed consent at the time of enrollment. Patients The study enrolled men and women aged 50 years and older with a diagnosis of primary OA of the hip, knee, hand, or spine. Patients were required to have at least moderate pain (3 on a 5-point Likert scale) at baseline in a specified target joint and were anticipated to need treatment for at least 52 weeks to enter the study. Patients taking gastroprotective drugs including proton pump inhibitors, misoprostol, and full-dose H 2 antagonists were excluded, as were those with an active upper GI ulceration in the previous 30 days, upper GI bleeding in the previous year, or any history of gastroduodenal perforation or obstruction. Other exclusion criteria are detailed elsewhere. 9 By design, TARGET included patients at high CV risk (including cerebrovascular risk). Patients taking low-dose aspirin ( mg/day) for primary or secondary prevention of coronary heart disease were included in a controlled, stratified way. Study Design and Assessments Study treatments were lumiracoxib 400 mg once a day, naproxen 500 mg twice a day, or ibuprofen 800 mg 3 times a day for up to 52 weeks. For logistical and masking reasons, TARGET was divided into 2 substudies of identical design, one with naproxen as the comparator and the other with ibuprofen. The investigators prespecified that data from the 2 substudies would be pooled for analysis. Within each substudy, randomization was stratified by age and low-dose aspirin use. Patient visits to the study clinic were made at baseline and after 4, 13, 20, 26, 39, and 52 weeks or until withdrawal, with a follow-up telephone call 4 weeks after exit from the study. Patients underwent investigation by endoscopy or other procedure according to clinical need, as judged by the local investigator. Data from all endoscopies performed for any symptoms of suspected ulcers (complicated and uncomplicated), dyspeptic pain, vomiting, nausea, or weight loss were considered for analysis. Results from routine scheduled endoscopies (eg, if the patient was in a surveillance program) were adjudicated but not included in the analysis regardless of findings. H pylori status was checked serologically as patients left the study by a central laboratory (Pharmaceutical Product Development, Inc, Central Global Laboratory, Brussels, Belgium) using the H pylori IgG antibodies chemiluminescence assay (sensitivity, 95%; specificity, 90%) by Nichols Advantage (Nichols Institute Diagnostics, San Clemente, CA). Investigators were required to report all suspected occurrences of GI, CV, and hepatobiliary events, which then were forwarded to independent safety committees for adjudication. By using criteria reported elsewhere, 9 the GI Safety Committee adjudicated and categorized the reported clinical event, its likelihood, the likely underlying cause, and whether serious clinical ulcer bleeding had occurred. This adjudication was performed blinded to treatment. End Points The primary GI end point was definite or probable ulcer complications (bleeding, perforation, or obstruction). Secondary GI safety end points included all ulcers (complicated and uncomplicated). All patients spent a fixed 52-week period in the study unless they were withdrawn prematurely for reasons such as adverse effects or lack of efficacy. Statistical Analysis Statistical analyses were performed on the safety population, defined as all randomized patients who received at least one dose of study drug. The primary end point of TARGET was the difference in the time-to-event distribution of definite or probable upper GI ulcer complications in patients not taking low-dose aspirin, analyzed on a modified intention-to-treat basis, using all data up to the time of study discontinuation. In the analysis presented here, demographic and baseline characteristic variables were considered as potential prognostic factors for developing ulcers and their complications. Because there were a priori reasons to consider

3 July 2007 EFFECT OF RISK FACTORS IN THE TARGET STUDY 59 that risk factors would interact differently with lumiracoxib and the nonselective NSAIDs combined, separate analyses were conducted. Interactions between the benefit of lumiracoxib vs naproxen and lumiracoxib vs ibuprofen with both definite or probable ulcer complications and all ulcers were investigated. Similarly, formal tests for interaction between treatment (lumiracoxib vs NSAIDs) and prognostic factors were performed. There was no significant heterogeneity in the relationship between lumiracoxib and NSAIDs by substudy for both end points, ulcer complications (P.7771), and all ulcers (P.9436); therefore, data for ibuprofen and naproxen were combined. Fourteen categoric factors were entered into initial multivariate Cox proportional hazards models: age ( 65 or 65 y), previous history of GI bleed or ulcer, sex, use of low-dose aspirin, H pylori status (positive, negative, or not measured), race (Caucasian or non- Caucasian); current smoker, alcohol consumption ( 1or 1 drink/day), previous NSAID use, BMI ( 27 or 27 kg/m 2 ), history of diabetes mellitus, history of hypertension, history of CV disease, and history of dyslipidemia. Backward elimination, with a significance level of.1 for removing an explanatory variable from the model, was used to identify final models. Hazard ratios (HRs) and crude incidence rates for lumiracoxib vs all NSAIDs were determined for various subgroups of interest, and elderly patients (age, 65 y) in particular. The number of patients needed to be treated (NNT) with lumiracoxib to avoid one event compared with NSAIDs were calculated from crude incidence rates. Data on the site of ulcers also were examined and are presented descriptively. The study power calculation had been made for the entire nonaspirin population and the entire overall population, thus the study had not been powered for lumiracoxib vs NSAID comparisons in the different risk factor groups. Results Patient Disposition and Baseline Characteristics Of the 21,787 patients who were screened, 18,325 were randomized and 18,244 received at least 1 dose of study medication: lumiracoxib 400 mg once a day (n 9117), naproxen 500 mg twice a day (n 4730), or ibuprofen 800 mg 3 times a day (n 4397). A patient flow-diagram has been presented elsewhere. 13 Their mean age was 63.5 years, 76.4% were women, and 23.7% were taking aspirin. Serology for H pylori was positive in 44.2%, negative in 44.3%, and not measured in 11.5% of patients. Patient demographics and baseline characteristics were well matched in comparator treatment groups. 9 Incidence of GI Events Reported for Adjudication There were 1011 patients with suspected upper GI tract (UGIT) events reported for adjudication: 386 (4.2%) in the lumiracoxib group vs 625 (6.8%) in the NSAID group. In total, 10 (0.1%) events in the lumiracoxib group and 38 (0.4%) events in the NSAID group were adjudicated as definite, 20 (0.2%) in the lumiracoxib group and 46 (0.5%) in the NSAID group were adjudicated as probable, and 24 (0.3%) events in the lumiracoxib group compared with 67 (0.7%) events in the NSAID group were adjudicated as possible ulcer complications. In addition, there were 58 (0.6%) UGIT uncomplicated ulcers in the lumiracoxib group and 103 (1.1%) in the NSAID group. Effect of Risk Factors on UGIT Ulcer Complications Because risk factors for ulceration on NSAIDs and COX-2 inhibitors may differ, risk was assessed within treatment groups. In patients taking NSAIDs, 4 factors were identified as significantly increasing the risk of ulcer complications: age 65 years and older, previous history of GI bleed or ulcer, male sex, and non-caucasian racial origin (Table 1). Age 65 years and older, male sex, and non-caucasian racial origin were identified similarly as risk factors in patients taking lumiracoxib; low-dose aspirin use was also a risk factor in this group (Table 1). Positive serology for H pylori was not associated with increased risk in either group of patients. However, data on H pylori status were not measured in 2106 patients (11.5%), including 25 (22%) of the 114 patients with ulcer complications. The heterogeneity of HRs of UGIT ulcer complications across subgroups was assessed and no treatment-by-subgroup interaction was observed (age, P.3429; sex, P.3351; race, P.8663; history of previous GI ulcers and bleeds, P.6286; BMI, P.6372; smoking, P.7565; or H pylori status, P.4151 for positive vs negative/not measured), with the exception of low-dose aspirin use (P.0050). Effect of Risk Factors on All Ulcers When all ulcers, both uncomplicated and complicated, were considered together, a broadly similar picture emerged (Table 2). Age, previous history, a non-caucasian racial origin, and a BMI of 27 kg/m 2 or less were associated with an increased risk of NSAID-associated ulceration. For patients on lumiracoxib, a previous history, aspirin use, current smoking, and H pylori positive status were significant risk factors. Risk Reduction With Lumiracoxib As reported elsewhere, 13 in patients not taking low-dose aspirin, an almost 4-fold reduction in the incidence of definite or probable ulcer complications was seen in patients taking lumiracoxib compared with NSAIDs. The incidence was 0.20% in the lumiracoxib group and 0.92% in the NSAID group (HR, 0.21; 95% confidence interval [CI], ; P.0001). 13 When

4 60 HAWKEY ET AL GASTROENTEROLOGY Vol. 133, No. 1 Table 1. Effect of Risk Factors on Definite or Probable UGIT Ulcer Complications (Aspirin and Nonaspirin Users) NSAIDs Lumiracoxib Factor (n 9127) Unadjusted HR Adjusted HR (n 9117) Unadjusted HR Adjusted HR Age, 65 y 3959 (43.4) 2.25 ( ) 2.30 ( ) 3980 (43.7) 3.02 ( ) 3.18 ( ) Previous history of GI 345 (3.8) 3.49 ( ) 3.61 ( ) 273 (3.0) 2.82 ( ) bleed or ulcer Male sex 2157 (23.6) 1.77 ( ) 1.70 ( ) 2154 (23.6) 2.63 ( ) 2.60 ( ) Non-Caucasian racial 2281 (25.0) 2.00 ( ) 2.10 ( ) 2225 (24.4) 2.21 ( ) 2.16 ( ) origin Low-dose aspirin use 2159 (23.7) 0.77 ( ) 2167 (23.8) 2.66 ( ) 2.89 ( ) H pylori positive 4028 (44.1) 1.09 ( ) 4029 (44.2) 0.95 ( ) H pylori not 1062 (11.6) 2.24 ( ) 2.18 ( ) 1044 (11.5) 5.00 ( ) 5.06 ( ) measured Current smoker 886 (9.7) 0.94 ( ) 929 (10.2) 0.84 ( ) 1 units alcohol/day 972 (10.6) 0.64 ( ) 970 (10.6) 1.07 ( ) Prior NSAID use 7005 (76.8) 0.81 ( ) 6989 (76.7) 0.81 ( ) BMI 27 kg/m (37.6) 1.18 ( ) 3281 (36.0) 1.03 ( ) History of diabetes 675 (7.4) 1.32 ( ) 744 (8.2) 1.18 ( ) mellitus History of 4061 (44.5) 0.99 ( ) 4219 (46.3) 1.58 ( ) hypertension CV/cerebrovascular 899 (9.8) 1.24 ( ) 981 (10.8) 0.98 ( ) history History of dyslipidemia 1834 (20.1) 1.26 ( ) 1829 (20.1) 0.85 ( ) NOTE. All potential risk factors put in model. Factors that did not significantly alter the risk of definite or probable UGIT ulcer complications were not included in the second model to calculate the final HRs. the overall population was considered, the incidence of definite or probable ulcer complications was reduced significantly by about 3-fold with lumiracoxib compared with NSAIDs. 13 Fewer uncomplicated ulcers arose with lumiracoxib (58; 0.64%) than with NSAIDs (103; 1.13%) (HR, 0.55; 95% CI, ; P.0003). 13 As a result, the incidence of the combined end point of all ulcers (complicated or uncomplicated) in patients not taking low-dose aspirin was reduced significantly for lumiracoxib vs NSAIDs (HR, 0.38; 95% CI, ; P.0001). 13 In the overall population, occurrence of this combined end point was significantly lower for lumiracoxib vs NSAIDs (HR, 0.46; 95% CI, ; P.0001). 13 Risk reduction with lumiracoxib in subgroups. As shown in Figure 1A, lumiracoxib maintained a significant advantage over NSAIDs across high-risk subgroups (age 65 y, male sex, and non-caucasian racial origin) for the incidence of definite or probable ulcer complications. A similar pattern was shown for the incidence of all ulcers (Figure 1B). Although numeric reductions in the incidence of ulcers and their complications with lumira- Table 2. Effect of Risk Factors on All Ulcers (Definite or Probable UGIT Ulcer Complications and Uncomplicated Ulcers) NSAIDs Lumiracoxib Factor (n 9127) Unadjusted HR Adjusted HR (n 9117) Unadjusted HR Adjusted HR Age, 65 y 3959 (43.4) 1.79 ( ) 1.81 ( ) 3980 (43.7) 1.26 ( ) Previous history of GI 345 (3.8) 3.91 ( ) 3.90 ( ) 273 (3.0) 4.45 ( ) 4.29 ( ) bleed or ulcer Non-Caucasian racial 2281 (25.0) 1.38 ( ) 1.41 ( ) 2225 (24.4) 1.45 ( ) origin Low-dose aspirin use 2159 (23.7) 0.87 ( ) 2167 (23.8) 2.20 ( ) 2.07 ( ) H pylori positive 4028 (44.1) 1.22 ( ) 4029 (44.2) 1.64 ( ) 1.71 ( ) H pylori not measured 1062 (11.6) 1.65 ( ) 1.50 ( ) 1044 (11.5) 2.28 ( ) 2.26 ( ) Current smoker 886 (9.7) 1.18 ( ) 929 (10.2) 2.28 ( ) 2.15 ( ) BMI 27 kg/m (37.6) 1.38 ( ) 1.36 ( ) 3281 (36.0) 0.88 ( ) NOTE. Only risk factors included in the final model (either treatment) are shown. Factors that did not significantly alter the risk of definite or probable UGIT ulcer complications were not included in the second model to calculate the final HRs.

5 July 2007 EFFECT OF RISK FACTORS IN THE TARGET STUDY 61 a previous history of gastric ulceration, and in duodenal ulcers and ulcer complications if they had a previous history of duodenal ulceration. For lumiracoxib, these influences were less obvious: the proportion of patients developing uncomplicated duodenal ulcers, but not duodenal ulcer complications, appeared to be increased with a previous history of duodenal ulceration. Results in Patients 65 Years and Older Personal and demographic features. There were 3980 patients treated with lumiracoxib and 3959 patients treated with NSAIDs who were 65 years and older. Demographic variables and baseline characteristics generally were comparable across treatment groups and within the 2 substudies for this subgroup of patients (Table 3). There was a higher use of aspirin in patients 65 years and older compared with those younger than 65 (31.1% vs 18.1% on lumiracoxib and 30.7% vs 18.2% on NSAIDs). Effect of risk factors. Because older patients are at increased risk of ulcers and their complications we analyzed which factors further increase risk in these patients. In patients aged 65 years and older taking NSAIDs, a previous history of an ulcer was the main additional risk factor. In patients 65 years and older who were on NSAIDs, the rate of ulcer complications increased from 1.21% (46 of 3788) in those without a past history to 4.09% (7 of 171) (HR, 3.69; 95% CI, ) Figure 1. (A) HR, incidence rate, and NNT by risk factor: definite or probable UGIT ulcer complications. (B) HR, incidence rate, and NNT by risk factor: all ulcers (definite or probable UGIT ulcer complications and uncomplicated ulcers). coxib were maintained, for the small subgroup of patients with a previous history of GI bleed or ulcer and those taking low-dose aspirin, the differences were not statistically significant. Thus, in patients taking aspirin there were 33 ulcers (18 uncomplicated, 15 complicated) on lumiracoxib compared with 45 (26 uncomplicated, 19 complicated) in patients taking NSAIDs (HR, 0.73; 95% CI, ; P.1706). In the primary analysis of patients not using aspirin, the NNT with lumiracoxib was 140 to avoid an ulcer complication and 81 to avoid any ulcer. The NNT in subgroups ranged from 47 (patients with a past history of ulcer or bleed) to 533 (patients on aspirin) for definite or probable ulcer complications and from 30 (patients with a past history of ulcer or bleed) to 237 (current smokers) for all ulcers (Figure 1A and B). Ulcer Site Figure 2 shows the site of current ulceration in patients with a previous ulcer at a known site compared with those with no past ulcer or one at an unknown site. In patients taking NSAIDs, there was an increase in gastric ulcers and gastric ulcer complications if they had Figure 2. (A) Event frequency at the gastric site by site of previous ulcer. (B) Event frequency at the duodenal site by site of previous ulcer.

6 62 HAWKEY ET AL GASTROENTEROLOGY Vol. 133, No. 1 Table 3. Patient Characteristics for the Population 65 Years and Older Lumiracoxib 400 mg once a day (n 3980) NSAIDs (n 3959) Lumiracoxib 400 mg once a day (n 2131) Naproxen 500 mg twice a day (n 2098) Lumiracoxib 400 mg once a day (n 1849) Ibuprofen 800 mg 3 times a day (n 1861) Previous history of 12 (0.3) 17 (0.4) 6 (0.3) 9 (0.4) 6 (0.3) 8 (0.4) GI bleed, n (%) Previous history of 109 (2.7) 158 (4.0) 52 (2.4) 90 (4.3) 57 (3.1) 68 (3.7) GI ulcer, n (%) Female, n (%) 3013 (75.7) 2932 (74.1) 1628 (76.4) 1557 (74.2) 1385 (74.9) 1375 (73.9) White/Caucasian, 3078 (77.3) 3063 (77.4) 1618 (75.9) 1593 (75.9) 1460 (79.0) 1470 (79.0) Low-dose aspirin 1239 (31.1) 1216 (30.7) 678 (31.8) 656 (31.3) 561 (30.3) 560 (30.1) use, n (%) Mean BMI SD, kg/m in those with a past history. For all ulcers, risk increased from 2.46% (93 of 3788) to 8.19% (14 of 171) (HR, 3.74; 95% CI, ). The effect of other risk factors was either not significant or inconsistent in patients age 65 years and older, reflecting the smaller numbers of patients in this subgroup. For patients taking lumiracoxib, the use of aspirin was the main additional risk factor, increasing the risk of ulcer complications from 0.29% (8 of 2741) to 1.13% (14 of 1239) (HR, 4.33; 95% CI, ), and of all ulcers from 0.69% (19 of 2741) to 1.86% (23 of 1239) (HR, 3.01; 95% CI, ). Risk reductions with lumiracoxib. In patients age 65 years and older who were not taking low-dose aspirin, there was a significant reduction (P.0001; HR, 0.20; 95% CI, ; NNT, 89) in definite or probable ulcer complications in individuals taking lumiracoxib (0.29%) vs patients taking NSAIDs (1.42%; Figure 3). Similar results were seen when all ulcers were considered and also for both end points in each substudy (Figure 3). Trends in the GI safety advantage for lumiracoxib compared with NSAIDs, in higher risk subpopulations, generally were consistent between all ages and for patients 65 years and older. Patient numbers were smaller in this subanalysis; thus, despite the consistent numeric advantage, statistical significance was not always reached in analyses split by the presence of risk factors (Figure 4A and B). In patients 65 years and older taking low-dose aspirin, 23 (1.86%) developed an ulcer or an ulcer complication on lumiracoxib compared with 35 (2.88%) receiving an NSAID (P.096). These numeric differences were owing Figure 3. Incidence of all ulcers in patients 65 years and older (nonaspirin population) as split by definite or probable ulcer complications and uncomplicated ulcers. Figure 4. (A) Definite or probable UGIT ulcer complications in patients 65 years and older. (B) All ulcers (definite or probable UGIT ulcer complications and uncomplicated ulcers) in patients 65 years and older.

7 July 2007 EFFECT OF RISK FACTORS IN THE TARGET STUDY 63 to differences in uncomplicated ulcer rates. In patients treated with aspirin who were age 65 years and older there were significantly fewer uncomplicated ulcers with lumiracoxib (n 4; 0.59%) compared with naproxen (n 14; 2.13%) (P.017). In the ibuprofen substudy, no significant differences were observed between treatments for uncomplicated ulcers in patients age 65 years and older taking low-dose aspirin. Discussion TARGET is the largest randomized, prospective outcomes study in OA patients to date in which risk factors for the occurrence of ulcer complications and all ulcers have been investigated. TARGET included large patient populations with and without concomitant aspirin use. In contrast to other large GI outcomes studies, 14,15 TARGET had the stringent primary end point of ulcer complications (perforations, obstructions, and bleeds), with the broader end point of all ulcers (perforations, ulcers, and bleeds) measured as a secondary outcome. To ensure statistical stability, this required a considerable increase in patient numbers compared with previous studies. 9 The study was not powered for withinsubgroups comparisons. Fourteen risk factors were included in the initial model. When considering the incidence of ulcer complications, 5 factors emerged as significant and remained in the final model. Age 65 years and older, male sex, and non-caucasian racial origin were common risk factors in both NSAID and lumiracoxib groups. In addition, previous history of GI bleed or ulcer was a significant risk factor for patients taking NSAIDs, whereas taking lowdose aspirin emerged as a significant risk factor in the lumiracoxib group. When all ulcers were considered, there were 4 significant risk factors for the incidence of all ulcers in patients taking NSAIDs: age 65 years and older, previous history of GI bleed or ulcer, non-caucasian race, and BMI of 27 kg/m 2 or less. In patients taking lumiracoxib a previous history of GI bleed or ulcer, lowdose aspirin use, H pylori positive, and current smoking all were identified as significant risk factors. Lumiracoxib maintained a significant advantage over NSAIDs across high-risk subgroups: patients 65 years and older, male sex, and those of non-caucasian race. Compared with NSAIDs, the NNT with lumiracoxib to avoid an ulcer complication ranged from 47 to 533. Cost effectiveness will be evaluated in a future publication. Despite the size of the study, the number of patients who had a history of GI ulcer or bleed was low (n 618). Therefore, there was insufficient information to determine whether or not there was an advantage in such patients. A broadly similar picture emerged for the incidence of all ulcers, except that the advantage for lumiracoxib compared with NSAIDs in patients with a history of a GI bleed or ulcer reached statistical significance. Increasing age was identified as a significant risk factor and was analyzed further. The risk of upper GI ulcer complications increased progressively with advancing age. Lumiracoxib maintained its GI safety advantage for both the incidence of ulcer complications and all ulcers in patients who were at increased GI risk because of their age ( 65 y). The advantage shown in the combined NSAID analysis of patients 65 years and older was consistent between the substudies. There was also a fairly consistent increase in risk with non-caucasian ethnicity. Because this was seen with both lumiracoxib and NSAIDs, a drug-independent effect via host phenotype or geography seems likely. Patients with a previous history of peptic ulceration are also at increased risk of ulcers and ulcer complications. Previous work has shown that the ulcer recurrence in NSAID users is site-preferential, suggesting re-activation of the original ulcer rather than a more general effect. 16,17 In our descriptive analyses we observed the influence of previous ulcer site on current ulceration site. Our data are compatible with data from previous studies. There has been concern that selective COX-2 inhibitors might increase the risk of recurrence in such patients by inhibiting COX-2 expressed in the rim of recurrent ulcers 18 and thereby impairing healing. 19 However, our data show that risks are reduced in such patients taking lumiracoxib compared with NSAIDs. In patients receiving lumiracoxib, low-dose aspirin use was a significant risk factor for ulcer complications. Aspirin did not increase risk in patients taking NSAIDs. The reasons for this observation are unclear, but it may be that the GI risk associated with NSAIDs was sufficiently high that a low dose of aspirin (maximum, 100 mg) would not have a significant additional effect. In patients taking low-dose aspirin, numeric differences between lumiracoxib and NSAIDs were not significant, so we do not know whether or not there is some, albeit reduced, benefit in these patients, as suggested for another COX-2 inhibitor. 20,21 How aspirin interacts with NSAIDs and with selective COX-2 inhibitors, and the relative contribution of ulcerogenicity and impaired hemostasis in aspirin users are subjects of intense interest. The interaction between NSAIDs and H pylori is also of considerable interest. H pylori status was tested at study exit, so there was a greater chance that this would not be performed if the exit visit was not a routine one conducted by a study investigator. This includes study exit because of development of an ulcer complication. Therefore, caution has to be exercised regarding assumptions with respect to the influence of H pylori as a risk factor for GI complications. Although previous descriptive studies have suggested that H pylori increases the risk of ulcers and ulcer complications in patients taking NSAIDs, this large, prospective, randomized study does not support that conclusion. Similarly, in the Vioxx Gastrointestinal Outcomes Research study of patients with rheumatoid

8 64 HAWKEY ET AL GASTROENTEROLOGY Vol. 133, No. 1 arthritis, H pylori did not increase the risk of perforations, ulcers, or bleeds. 10 Our analyses presented here thus identify a number of risk factors for NSAID-associated GI toxicity. In addition, they show that even in patients at increased risk of GI events owing to the presence of one or more of these factors (eg, age 65 y), lumiracoxib maintains the GI safety benefit shown previously. Lumiracoxib is a selective COX-2 inhibitor, which is distributed preferentially into inflamed tissue. This unique pharmacokinetic profile may confer important safety advantages when compared with other pain-relief treatment options. In suitable patients, lumiracoxib represents an important therapeutic treatment option for the relief of pain associated with OA. References 1. American College of Rheumatology. Recommendations for the medical management of osteoarthritis of the hip and knee: 2000 update. American College of Rheumatology Subcommittee on Osteoarthritis Guidelines. Arthritis Rheum 2000;43: Pendleton A, Arden N, Dougados M, Doherty M, Bannwarth B, Bijlsma JW, Cluzeau F, Cooper C, Dieppe PA, Gunther KP, Hauselmann HJ, Herrero-Beaumont G, Kaklamanis PM, Leeb B, Lequesne M, Lohmander S, Mazieres B, Mola EM, Pavelka K, Serni U, Swoboda B, Verbruggen AA, Weseloh G, Zimmerman-Gorska I. EULAR recommendations for the management of knee osteoarthritis: report of a task force of the Standing Committee for International Clinical Studies Including Therapeutic trials (ESCISIT). Ann Rheum Dis 2000;59: Singh G, Ramey DR, Morfeld D, Shi H, Hatoum HT, Fries JF. Gastrointestinal tract complications of nonsteroidal anti-inflammatory drug treatment in rheumatoid arthritis. A prospective observational cohort study. Arch Intern Med 1996;156: Hawkey CJ, Langman MJ. Non-steroidal anti-inflammatory drugs: overall risks and management. Complementary roles for COX-2 inhibitors and proton pump inhibitors. Gut 2003;52: Esser R, Berry C, Du Z, Dawson J, Fox A, Fujimoto RA, Haston W, Kimble EF, Koehler J, Peppard J, Quadros E, Quintavalla J, Toscano K, Urban L, van Duzer J, Zhang X, Zhou S, Marshall PJ. Preclinical pharmacology of lumiracoxib: a novel selective inhibitor of cyclooxygenase-2. Br J Pharmacol 2005;144: Rordorf CM, Choi L, Marshall P, Mangold JB. Clinical pharmacology of lumiracoxib: a selective cyclo-oxygenase-2 inhibitor. Clin Pharmacokinet 2005;44: Weaver ML, Flood DJ, Kimble EF, Fujimoto RA. Lumiracoxib demonstrates preferential distribution to inflamed tissue in the rat following a single oral dose: an effect not seen with other cyclooxygenase-2 inhibitors (abstr). Ann Rheum Dis 2003;62(Suppl I): Scott G, Rordorf C, Reynolds C, Kalbag J, Looby M, Milosavljev S, Weaver M, Huff JP, Ruff DA. Pharmacokinetics of lumiracoxib in plasma and synovial fluid. Clin Pharmacokinet 2004;43: Hawkey CJ, Farkouh M, Gitton X, Ehrsam E, Huels J, Richardson P. Therapeutic Arthritis Research and Gastrointestinal Event Trial of lumiracoxib study design and patient demographics. Aliment Pharmacol Ther 2004;20: Laine L, Bombardier C, Hawkey CJ, Davis B, Shapiro D, Brett C, Reicin A. Stratifying the risk of NSAID-related upper gastrointestinal clinical events: results of a double blind outcomes study in patients with rheumatoid arthritis. Gastroenterology 2002;123: Garcia Rodriguez LA, Jick H. Risk of upper gastrointestinal bleeding and perforation associated with individual non-steroidal antiinflammatory drugs. Lancet 1994;343: Hansen JM, Hallas J, Lauritsen JM, Bytzer P. Non-steroidal antiinflammatory drugs and ulcer complications: a risk factor analysis for clinical decision-making. Scand J Gastroenterol 1996;31: Schnitzer TJ, Burmester GR, Mysler E, Hochberg MC, Doherty M, Ehrsam E, Gitton X, Krammer G, Mellein B, Matchaba P, Gimona A, Hawkey CJ, TARGET Study Group. Comparison of lumiracoxib with naproxen and ibuprofen in the Therapeutic Arthritis Research and Gastrointestinal Event Trial (TARGET), reduction in ulcer complications: randomised controlled trial. Lancet 2004;364: Bombardier C, Laine L, Reicin A, Shapiro D, Burgos-Vargas R, Davis B, Day R, Ferraz MB, Hawkey CJ, Hochberg MC, Kvien TK, Schnitzer TJ, VIGOR Study Group. Comparison of upper gastrointestinal toxicity of rofecoxib and naproxen in patients with rheumatoid arthritis. N Engl J Med 2000;343: Silverstein FE, Faich G, Goldstein JL, Simon LS, Pincus T, Whelton A, Makuch R, Eisen G, Agrawal NM, Stenson WF, Burr AM, Zhao WW, Kent JD, Lefkowith JB, Verburg KM, Geis GS. Gastrointestinal toxicity with celecoxib vs nonsteroidal anti-inflammatory drugs for osteoarthritis and rheumatoid arthritis: the CLASS study: a randomized controlled trial. JAMA 2000;284: Hawkey CJ, Naesdal J, Wilson I, Langstrom G, Swannell AJ, Peacock RA, Yeomans ND. Relative contribution of mucosal injury and Helicobacter pylori in the development of gastroduodenal lesions in patients taking non-steroidal anti-inflammatory drugs. Gut 2002;3: Hawkey CJ, Wilson I, Naesdal J, Langstrom G, Swannell AJ, Yeomans ND. Influence of sex and Helicobacter pylori on development and healing of gastroduodenal lesions in non-steroidal anti-inflammatory drug users. Gut 2002;51: Jackson LM, Wu KC, Mahida YR, Jenkins D, Hawkey CJ. Cyclooxygenase (COX) 1 and 2 in normal, inflamed and ulcerated human gastric mucosa. Gut 2000;47: Poonam D, Vinay CS, Gautam P. Cyclo-oxygenase-2 expression and prostaglandin E2 production in experimental chronic gastric ulcer healing. Eur J Pharmacol 2005;519: Rahme E, Bardou M, Dasgupta K, Toubouti Y, Ghosn J, Barkun AN. Hospitalization for gastrointestinal bleeding associated with non-steroidal anti-inflammatory drugs among elderly patients using low-dose aspirin: a retrospective cohort study. Rheumatology (Oxford) 2007;46: [Epub ahead of print]. 21. Goldstein JL, Lowry SC, Lanza FL, Schwartz HI, Dodge WE. The impact of low-dose aspirin on endoscopic gastric and duodenal ulcer rates in users of a non-selective non-steroidal anti-inflammatory drug or a cyclo-oxygenase-2-selective inhibitor. Aliment Pharmacol Ther 2006;23: Received November 15, Accepted April 12, Address requests for reprints to: Professor Christopher J. Hawkey, Wolfson Digestive Diseases Centre, University Hospital, Nottingham NG7 2UH, United Kingdom. cj.hawkey@nottingham.ac.uk; fax: (0044) The academic authors wrote the first draft of the article, which was then developed by all the authors. The authors would like to thank Dr Rebecca Douglas, a professional medical writer with ACUMED, for her assistance with this process. Supported by Novartis Pharma AG, Basel, Switzerland. C.J.H. has received research funding and/or honoraria from AstraZeneca, Glaxo- SmithKline, Merck, Novartis, Pfizer, Takeda, Serono International SA, and Wyeth. W.M.W. and W.S. have served as consultants to Novartis. X.G., P.S., K.S., G.K., B.M., and D.R. are employees of Novartis Pharma AG, Basel, Switzerland. P.M. is an employee of Novartis Pharmaceuticals Corporation, East Hanover, New Jersey.