6 The Open Rheumatology Journal, 2012, 6, 6-20 Open Access Efficacy of Etoricoxib, Celecoxib, Lumiracoxib, Non-Selective NSAIDs, and Acetaminophen in Osteoarthritis: A Mixed Treatment Comparison W.B. Stam1, J.P. Jansen2 and S.D. Taylor*,3 1Mapi Group, Houten, The Netherlands 2Mapi Group, Boston, MA, USA 3Merck & Co., Inc., Whitehouse Station, NJ, USA Abstract: Objective: To compare the efficacy of etoricoxib, lumiracoxib, celecoxib, non-selective (ns) NSAIDs and acetaminophen in the treatment of osteoarthritis (OA) Methods: Randomized placebo controlled trials investigating the effects of acetaminophen 4000mg, diclofenac 150mg, naproxen 1000mg, ibuprofen 2400mg, celecoxib 100-400mg, lumiracoxib 100-400mg, and etoricoxib 30-60mg with treatment duration of at least two weeks were identified with a systematic literature search. The endpoints of interest were pain, physical function and patient global assessment of disease status (PGADS). Pain and physical function reported on different scales (VAS or LIKERT) were translated into effect sizes (ES). An ES 0.2 - 0.5 was defined as a “small” treatment effect, whereas ES of 0.5 – 0.8 and > 0.8 were defined as “moderate” and “large”, respectively. A negative effect indicated superior effects of the treatment group compared to the control group. Results of all trials were analyzed simultaneously with a Bayesian mixed treatment comparison. Results: There is a >95% probability that etoricoxib (30 or 60mg) shows the greatest improvement in pain and physical function of all interventions compared. ESs of etoricoxib 30mg relative to placebo, celecoxib 200mg, ibuprofen 2400mg, and diclofenac 150mg were -0.66 (95% Credible Interval -0.83; -0.49), -0.32 (-0.50; -0.14), -0.25 (-0.53; 0.03), and -0.17 (-0.41; 0.08), respectively. Regarding physical functioning, ESs of etoricoxib 30mg relative to placebo, celecoxib 200mg, ibuprofen 2400mg, and diclofenac 150mg were -0.61 (-0.76; -0.46), -0.27 (-0.43; -0.10), -0.20 (-0.47; 0.07), and -0.09 (- 0.33; 0.14) respectively. The greatest improvements in PGADS were expected with either etoricoxib or diclofenac. Conclusion: The current study estimated the efficacy of acetaminophen, nsNSAIDs, and COX-2 selective NSAIDs in OA and found that etoricoxib 30 mg is likely to result in the greatest improvements in pain and physical function. Differences in PGADS between interventions were smaller. Keywords: Acetaminophen, Bayesian, celecoxib, etoricoxib, lumiracoxib, NSAIDs, osteoarthritis, meta-analysis. INTRODUCTION complications is a major factor limiting the use of nsNSAIDs at various doses [5]. The COX 2 selective class of NSAIDs Osteoarthritis (OA) is the most common arthritic were developed to decrease the risk of gastrointestinal tract condition in adults [1]. OA is extremely painful and causes injury; in OA the recommended daily doses are: celecoxib disability and a reduced quality of life, which poses a 200mg, etoricoxib 30-60mg and lumiracoxib 100mg. (At the substantial economic burden for society [2-4]. time of the writing of this manuscript (December 2007), the The principal treatment objectives in OA are to EMEA CHMP has recommended the withdrawal of the adequately control pain, improve function, and reduce marketing authorisations for all lumiracoxib-containing disability. In order to achieve this, analgesic medication medicines, because of the risk of serious side effects including acetaminophen, non-selective non-steroidal anti- affecting the liver. [7]) inflammatory drugs (nsNSAIDs) and more recently Medical decision making requires the assessment of the cyclooxygenase (COX) 2 selective NSAIDs are commonly relative value of an intervention versus relevant comparators prescribed in the treatment of OA [5]. Acetaminophen is (e.g., COX 2 selective inhibitors, NSAIDs and paracetamol). used at a maximum recommended dose of 4000 mg/day and A comprehensive relevant comparison in OA requires nsNSAIDs like naproxen at 1000 mg/day, diclofenac 150 considering all available best evidence (i.e. all RCTs) of the mg/day and ibuprofen 2400 mg/day. A Cochrane review above mentioned interventions. Ideally, we want a RCT that demonstrated that NSAIDs display superior efficacy relative compares all interventions of interest simultaneously. to acetaminophen [6]. However, concern for gastrointestinal However, such a study has not been performed. A mixed treatment comparison (MTC) is a valuable alternative to *Address correspondence to this author at the Merck & Co, Inc., One Merck synthesis evidence when the interest is to compare multiple Drive, WS2E-85, Whitehouse Station, NJ 08889, USA; Tel: +1 908-423- interventions of different RCTs [8]. MTC is an extension 3512; Fax: +1 908-735-1688; E-mail: [email protected] of traditional meta-analysis by including multiple different 1874-3129/12 2012 Bentham Open Efficacy of COX-2 Selective NSAIDs, Non-Selective NSAIDs and Acetaminophen in OA The Open Rheumatology Journal, 2012, Volume 6 7 pair-wise comparisons across a range of interventions [8-11]. Data Extraction With MTC the relative efficacy of a particular intervention For each selected study, details were extracted on design, versus competing interventions can be obtained in the selection criteria, study population characteristics, absence of head-to-head comparisons; an indirect interventions, outcome measures, length of follow-up, and comparison of two interventions is made via a common comparator. A Bayesian approach to a MTC can be results, which were subsequently checked by a second reviewer. With regard to etoricoxib the original company considered the method of choice because it allows for a trial reports were available to extract outcome data. The probabilistic interpretation and therefore leads naturally into outcome measures of interest were the change from baseline the decision making context. The ranking of interventions (CFB) in pain, physical functioning, and patient global regarding their ability of providing greatest outcomes assessment of disease status (PGADS) reported at the last becomes particularly useful. available follow-up measurement of the double-blind The objective of the current study was to evaluate the randomized period of the RCTs. To compare VAS and likert efficacy of acetaminophen 4000mg, the nsNSAIDs scales for pain and physical function across studies and diclofenac 150mg, naproxen 1000mg, and ibuprofen combine them in the MTC, CFB values were translated into 2400mg, and the COX-2 selective NSAIDs celecoxib and effect sizes (ES) [13]. Relative ES were calculated as the lumiracoxib at doses ranging from 100 to 400mg and difference in CFB between two interventions divided by the etoricoxib 30mg and 60mg in the treatment of OA. corresponding standard deviation. If the standard deviation was not reported, the ES was calculated by conversion of the MATERIALS AND METHODS reported P value to a Z-statistic according to: ES = z (1/n1 Identification and Selection of Studies +1/n2), where n1 and n2 are the number of patients in the groups that are compared. For studies that report cut-off P In order to identify relevant publications a systematic values, the quoted value (e.g., 0.05) was used to estimate the literature review was performed. Computerised bibliographic ES. databases (MEDLINE 1966 – 14th December 2006 November, EMBASE 1980 – 23rd November 2006 and Analysis Cochrane Library issue 4 (central register of clinical trials th The results of the different regimens in the included and systematic reviews, until 14 December 2006) were studies were combined by means of a Bayesian MTC and as searched. The following search terms were used: celecoxib; a result, estimates of relative efficacy between each of the Celebrex; lumiracoxib; Prexige; etoricoxib; Arcoxia; possible pair-wise comparisons were obtained [8-11]. The acetaminophen; paracetamol; diclofenac; naproxen; ibuprofen, randomized method; randomized trial; interventions were acetaminophen 4000 mg/day, diclofenac 150 mg/day, naproxen 1000 mg/day, ibuprofen 2400 randomised method; randomised controlled trial; controlled mg/day, dexibuprofen 800mg/day, celecoxib 100, 200 or 400 trial; controlled study; controlled studies; clinical trial; mg/day, etoricoxib 30 mg/day and 60 mg/day and clinical study; double blind. In addition, trials from the lumiracoxib 100, 200 or 400 mg/day. Analyses were etoricoxib development program were included. The performed for pain, physical function, and PGADS. identified studies were included according to the following predetermined conditions: MTC within the Bayesian framework involve data, a likelihood distribution, a model with parameters, and prior Type of design –Randomised controlled trials with a distributions for these parameters. The model relates the data double blind period. Only full-published reports were from the individual studies to basic parameters reflecting the considered; letters and abstracts were excluded. (pooled) relative treatment effect of each intervention compared Intervention – The interventions included acetaminophen to placebo. Based on these basic parameters, the relative 4000mg/day, ibuprofen 2400mg/day, naproxen 1000mg/day, efficacy between each of the competing interventions was diclofenac 150mg/day, celecoxib 100, 200 or 400 mg/day, estimated. For the 3 outcomes of interest, linear models with etoricoxib 30 and 60 mg/day. The duration of the normal likelihood distributions were used. A MTC relies on the intervention was at least 2 weeks. assumption that there are no differences in the distribution of Comparison