TRANSPARENCY COMMITTEE OPINION. 1 April 2009

Transcription

1 The legally binding text is the original French version TRANSPARENCY COMMITTEE OPINION 1 April 2009 CYCLADOL 20 mg, scored tablet Box of 14 (CIP: ) CYCLADOL 20 mg, effervescent tablet Box of 14 (CIP: ) Applicant: CHIESI Piroxicam ATC: M01AC01 List I Date of Marketing Authorisation: - CYCLADOL scored tablet: 22 March 1993 CYCLADOL effervescent tablet: 25 November Most recent revision: 21 May 2008 (Therapeutic indications, Posology and method of administration, Contraindications, Special warnings and precautions for use, Interactions with other medicinal products and other forms of interaction and Adverse effects). Reason for examination: Reassessment of the actual benefit of piroxicam-based proprietary medicinal products for systemic use under article R of the Social Security Code. This reassessment follows the EMEA arbitration procedure on piroxicam-based proprietary medicinal products. Medical, Economic and Public Health Assessment Division 1

2 1 CHARACTERISTICS OF THE MEDICINAL PRODUCT 1.1. Active ingredient piroxicam 1.2. Indications Old text: They are based on the anti-inflammatory properties of piroxicam, the severity of the signs of intolerance caused by the medicinal product, and its role in the panoply of anti-inflammatory products currently available. They are limited to the following forms of treatment in adults and children aged 15 and over: long-term symptomatic treatment of: - chronic forms of inflammatory rheumatism, particularly rheumatoid polyarthritis, ankylosing spondylitis (or related syndromes, such as Fiessinger-Leroy-Reiter syndrome and psoriatic rheumatism), - some forms of painful and disabling osteoarthritis. short-term symptomatic treatment of acute exacerbations of: - forms of abarticular rheumatism, such as scapulohumeral periarthritis, tendinitis, bursitis, - acute post-traumatic conditions of the locomotor system, - microcrystalline arthritis, - osteoarthritis, - radiculalgia. New text: Piroxicam is indicated in the symptomatic treatment of osteoarthritis, rheumatoid polyarthritis and ankylosing spondylitis. In view of its safety profile (see sections 4.2, 4.3 and ), piroxicam must not be used as first-line treatment where NSAID treatment is indicated. The decision to prescribe a proprietary medicinal product containing piroxicam must be based on an assessment of all the risks to which the patient in question may be exposed (see sections 4.3 and ).» 1.3. Dosage Proprietary medicinal products containing piroxicam must be initially prescribed by doctors who are experienced in the diagnosis and treatment of patients suffering from inflammatory or degenerative rheumatoid conditions. This medicinal product may only be administered to adults and children aged 15 or over. The maximum recommended daily dose is 20 mg. The occurrence of adverse effects can be minimised by administering the lowest possible dose needed to relieve symptoms, given for the shortest possible period. The benefits and safety of the treatment must be reassessed after 14 days. If continued treatment is necessary, this must be accompanied by frequent reassessments. 1 of the SPC. 2

3 As piroxicam has been associated with an increased risk of gastrointestinal complications, practitioners must seriously consider prescribing treatment to protect the gastric mucosa (such as misoprostol or proton pump inhibitors), especially for elderly patients. 2 SIMILAR MEDICINAL PRODUCTS 2.1. ATC Classification (2009) M : Musculoskeletal system M01 : Antiinflammatory and antirheumatic products M01A : Antiinflammatory and antirheumatic products, non-steroids M01AC : oxicams M01AC01 : piroxicam 2.2. Medicines in the same therapeutic category Strictly comparable medicines in the same therapeutic category Other piroxicam-based systemic proprietary medicinal products and their generics where the indications have been restricted to symptomatic treatment of osteoarthritis, rheumatoid polyarthritis or ankylosing spondylitis as second-line treatment where NSAID treatment needs to be prescribed: - BREXIN 20 mg, scored tablet and effervescent tablet - FELDENE 10 mg, capsule - FELDENE 20 mg, capsule, dispersible tablet and suppository - FELDENE 20 mg/1 ml, solution for injection - PROXALYOC 10 and 20 mg, oral lyophilisate Not strictly comparable medicines in the same therapeutic category Other proprietary products and their generics in the oxicam class for systemic use: - meloxicam: MOBIC 7.5 and 15 mg tablet, 7.5 and 15 mg capsule, 7.5 mg/ml oral suspension, 7.5 and 15 mg suppository and 15 mg/1.5 ml solution for injection - tenoxicam: TILCOTIL 20 mg tablet, effervescent tablet, suppository and lyophilisate and solution for parenteral administration 2.3. Medicines with a similar therapeutic aim Other NSAID proprietary medicinal products (indoles and derivatives, arylcarboxylics, fenamates, pyrazoles, selective COX-2 inhibitors and others) and steroidal antiinflammatories. 3

4 3 ANALYSIS OF AVAILABLE DATA Background: Following a request from the European Commission, the Committee for Medicinal Products for Human Use (CHMP) reviewed the safety data on non-selective NSAIDs. During this review, the CHMP issued recommendations aimed at jointly revising the summary of product characteristics of non-selective NSAIDs and decided that the risk/benefit ratio of some NSAIDs, including piroxicam, should be assessed in greater detail. This assessment, which was completed in June 2006, found that there could be a greater risk of gastrointestinal and skin reactions (including potentially fatal bullous eruption) with piroxicam compared to other non-selective NSAIDs. The European Commission decided that the risks to public health were such that a new assessment of the risk/benefit ratio of piroxicam needed to be carried out, and asked the CHMP to produce an opinion on whether the marketing authorisations for medicinal products containing piroxicam should be maintained, amended, suspended or withdrawn. On 21 June 2007 the CHMP concluded that systemic formulations of piroxicam were associated with a greater risk of serious gastrointestinal toxicity than other NSAIDs, and a greater risk of serious acute skin reaction than other non-oxicam NSAIDs. However, the CHMP was of the opinion that the risk/benefit ratio of piroxicam was still positive in the symptomatic treatment of osteoarthritis, rheumatoid polyarthritis and ankylosing spondylitis, provided that the conditions of use of these medicinal products were revised. The risk/benefit ratio for the treatment of other pathologies in the acute phase was judged to be negative in view of piroxicam's long half-life (50 hours) and its safety profile. No restriction was thought to be necessary for topical piroxicam formulations because systemic exposure is lower than with systemic formulations. The following changes have been made to the summary of product characteristics for proprietary medicinal products containing piroxicam: - Indication: piroxicam is restricted to the symptomatic treatment of osteoarthritis, rheumatoid polyarthritis and ankylosing spondylitis. However, it must not be used as firstline treatment in these indications if an NSAID is indicated. - As with all NSAIDs, treatment must be administered at the lowest possible dose (not more than 20 mg/day) and for the shortest possible period. - The product must be initially prescribed by a doctor experienced in the management of patients suffering from inflammatory or degenerative rheumatoid conditions. Treatment must always be reviewed after 14 days. - Doctors must consider prescribing piroxicam in combination with a gastroprotective drug (such as misoprostol or a proton pump inhibitor), especially in the case of elderly patients. - Medicinal products containing piroxicam must not be prescribed for patients with a history of gastrointestinal bleeding or skin reactions to any other medicinal product. - Medicinal products containing piroxicam must not be prescribed in combination with another NSAID or an anticoagulant. The data presented below is the data on which the CHMP based its opinions during the arbitration procedure which led to the reassessment of the risk/benefit ratio of piroxicambased proprietary medicinal products for systemic use (opinion dated 3 July 2007). 4

5 3.1. Efficacy Gaffney (1998) 2 : Meta-analysis of 38 randomised published clinical studies assessing the efficacy and digestive safety of piroxicam 20 mg compared to various NSAIDs (naproxen, diclofenac, nabumetone, ibuprofen, tenoxicam, indomethacin, etodolac and oxaproxin) in patients suffering from chronic (osteoarthritis and rheumatoid polyarthritis) and acute conditions. Of the 11,705 patients included in the efficacy analysis, 9,041 were being treated for a chronic condition (2,599 with piroxicam) and 2,664 for an acute condition (1,333 with piroxicam). In the osteoarthritis and rheumatoid polyarthritis indications, the response rates (criterion not specified) were 53.3% with piroxicam 20 mg and 54.6% with the other comparators (RR = 0.97, difference not significant). These results need to be interpreted with caution as this study has not been published but was merely presented as a poster at a conference Adverse effects Gastrointestinal safety Data from clinical studies Three meta-analyses (2 of which have not been published but have been presented at conferences) and one combined analysis of studies were taken into consideration by the CHMP (see table 1 in the appendix). Richy (2004) 3 : Thirty-two randomised, controlled clinical studies comparing various NSAIDs to a placebo group or a group of patients not exposed to NSAIDs were considered in this meta-analysis. The results showed an increased risk of gastrointestinal complications for all NSAIDs compared to the risk for the inactive control group: - indomethacin: RR = 2.25; 95%CI = [1.00; 5.07] - naproxen: RR = 1.83; 95%CI = [1.25; 2.68] - diclofenac: RR = 1.73; 95%CI = [1.21; 2.46] - piroxicam: RR = 1.66; 95%CI = [1.14; 2.44] - tenoxicam: RR = 1.43; 95%CI = [0.40; 5.14] - meloxicam: RR = 1.24; 95%CI = [1.00; 5.07] - ibuprofen: RR = 1.19; 95%CI = [1.00; 5.07] This study does not show piroxicam to be associated with any greater risk of gastrointestinal complications than other NSAIDs. It should be noted that the numbers of patients included for the various NSAIDs varied, ranging from 284 (three studies) for piroxicam to 1,495 for meloxicam. Singh (2004) 4 : The results of 28 clinical studies 5, selected primarily on the basis that they included a meloxicam group, were compared in this study in order to calculate the accumulated risk of 2 Results presented in the form of a poster: Gaffney M, Curiale C and Cimmino AA. Piroxicam risk/benefit ratio: a meta-analysis of 38 published papers. Meeting of the OARS, October 15-17, 1998, Florence, Italy. 3 Richy F et al. Time dependant risk of gastrointestinal complications induced by non-steroidal anti-inflammatory drug use : a consensus statement using a meta-analytic approach, Annals of the rheumatic diseases, 2004, 63: Singh S et al. Risk of serious upper gastrointestinal and cardiovascular thromboembolic complications with meloxicam. The American journal of medicine, 2004, 117:

6 serious gastrointestinal adverse effects for treatments lasting up to two months. This risk was higher with piroxicam than with meloxicam 15 mg/d (0.9% versus 0.2%, p = 0.03). No statistically significant difference between piroxicam and diclofenac or between piroxicam and naproxen was observed. These results need to be interpreted with caution as the work involved a combined analysis of several studies and does not meet the methodological requirements of a meta-analysis of studies in terms of analysis of survival (in particular, there was no individual description of the studies, no test for heterogeneity of results, no selection bias analysis, and no account was taken of the multiplicity of analysis and inflation of the significance level). Gaffney (1998) 6 : In this meta-analysis of 38 studies described above 5 (total of 15,208 patients analysed for safety), no statistically significant difference was shown between piroxicam and its NSAID comparators (the type of NSAID was not specified) in respect of the occurrence of gastrointestinal adverse events for either chronic or acute pathologies. The incidence of gastrointestinal adverse events taking the form of bleeding or haemorrhage was too low to allow for a statistical analysis of the difference between piroxicam and other NSAIDs (15/3,020 with piroxicam and 20/7,221 with the comparators). No bleeding or ulcer adverse effect was observed among patients with acute pathologies. These results need to be interpreted with caution as we do not have details of this study, which was presented at a conference. Richy (2007) 7 : This meta-analysis included 75 clinical studies 5, 45 of which analysed major gastrointestinal adverse events in patients taking piroxicam versus the other NSAIDs. No statistically significant difference was shown between piroxicam and the other NSAIDs in terms of the risk of major gastrointestinal events (combined OR = 1.33; 95%CI = [0,.96; 1.84]). These results need to be interpreted with caution as this study has not been published but was merely presented verbally at a conference. Data from epidemiological studies The assessment of epidemiological data on the risk of gastrointestinal complications was based primarily on the meta-analysis carried out by Hernandez-Diaz (2000) 8 (see table 2 in the appendix). Hernandez-Diaz (2000): This meta-analysis of 15 case-control studies and 3 observational studies revealed an increased risk of severe upper gastrointestinal complications among patients taking piroxicam or other NSAIDs compared to patients not taking NSAIDs. No statistical analysis of the comparisons between piroxicam and the other NSAIDs was carried out. However, it was noted that the lower limit of the 95% confidence interval of the relative risk of piroxicam versus patients not taking NSAIDs was above the upper limit of that of the other NSAIDs apart from ketoprofen: - piroxicam: RR = 6.3; 95%CI = [5.5; 7.2] - ketoprofen: RR = 4.6; 95%CI = [3.3; 6.4] 5 It was not made clear whether these studies were randomised and/or controlled. 6 Gaffney M et al. Piroxicam risk/benefit ratio: a meta-analysis of 38 published papers. Meeting of the OARS, 1998, Florence, Italy (poster) 7 Richy F et al. Meta-analysis of 75 RCT with Piroxicam 2007, (ECCE07: 7th European Congress on Clinical and Economic Aspects of Osteoporosis and Osteoarthritis), 8 Hernandez-Diaz S et al. Association between nonsteroidal anti-inflammatory drugs and upper gastrointestinal tract bleeding/perforation. Archives of internal medicine, 2000, 160:

7 - indomethacin: RR = 4.6; 95%CI = [5.5; 5.5] - naproxen: RR = 4.0; 95%CI = [3.5; 4.6] - sulindac: RR = 3.8; 95%CI = [3.6; 4.1] - diclofenac: RR = 3.6; 95%CI = [2.8; 4.7] - ibuprofen: RR = 3.3; 95%CI = [2.8; 3.9] Five other studies which were not included in these meta-analyses have been published since 2001 (see table 3 in the appendix): 2 cohort studies, Gallerani (2004) 9 and Mellemkjaer (2002) 10 and 3 case-control studies, Laporte (2004) 11, Lanas (2003) 12 and Lanas (2006) 13. A similar result was observed in these studies, with the risk of upper gastrointestinal complications (haemorrhage) appearing to be greater with piroxicam than with the other NSAIDs apart from ketoprofen in one study and ketorolac in two studies. In these studies, the odds ratio of the risk with piroxicam was 4.76 (95%CI = [2.98; 7.60]), 5.00 (95%CI = [3.3; 7.2]), 15.5 (95%CI = [10.0; 24.2]), 18.5 (95%CI = [9.2; 36.9]) and 12.6 (95%CI = [7.8; 20.3]) Skin safety The assessment of the skin safety of piroxicam is based on two case-control studies: Roujeau (1993) 14 and Mokenhaupt (2003) 15. These studies showed that oxicam-based NSAIDs, including piroxicam, were associated with a greater risk of serious skin reactions (mainly Stevens-Johnson syndrome, toxic epidermal necrolysis or Lyell s syndrome, and polymorphic erythema). However, among oxicam drugs, the greatest risk is found with tenoxicam. In the Mockenhaupt study, which is an update of the Roujeau study (1993), the relative risks observed are as follows: - oxicams: RR = 34; 95%CI = [11; 105] - tenoxicam: RR = 43; 95%CI = [12; 145] - piroxicam: RR = 20; 95%CI = [6; 67] - propionic acid derivatives: Adjusted RR = 1.9; 95%CI = [0.7; 5.0] - diclofenac: Adjusted RR = 4.1; 95%CI = [1.5; 11.0] These results must be interpreted with caution because cases are so rare, which is reflected in very broad confidence intervals. This study also showed that the risk of serious skin reactions with NSAIDs is greater during the first two months of treatment: - oxicams 2 months: RR = 72; 95%CI = [25; 209] - oxicams > 2 months: RR = 0; 95%CI = [0; 47] - propionic acid derivatives 2 months: RR = 5.4; 95%CI = [2.4; 12] - propionic acid derivatives > 2 months: RR = 2.4; 95%CI = [0.4; 13] The CHMP decided that the risk of serious skin reactions associated with piroxicam, although not very high in absolute terms, could not be dismissed given the severity of the 9 Gallerani M et al. Risk of hospitalization for upper gastrointestinal tract bleeding. Journal of Clinical Epidemiology, 2004, 57: Mellemkjaer L et al. Upper gastrointestinal bleeding among users of NSAIDs : a population-based cohort study in Denmark. British journal of clinical pharmacology, 2002, Laporte JR et al. Upper gastrointestinal bleeding associated with the use of NSAIDs : newer versus older agents. Drug Safety, 2004, 27 (6): Lanas A et al. Risk of upper gastrointestinal bleeding associated with non-aspirin cardiovascular drugs, analgesics and nonsteroidal anti-inflammatory drugs. European Journal of Gastroenterology and Hepatology, 2003, 15: Lanas A et al. Risk of upper gastrointestinal ulcer bleeding associated with selective cyclo-oxygenase-2 inhibitors, traditional non-aspirin non-steroidal anti-inflammatory drugs, aspirin and combinations. Gut, 2006, 55: Roujeau JC et al. Medication use and the risk of Stevens-Johnson syndrome or toxic epidermal necrolysis. The new England journal of medicine, 1995, 333: Mockenhaupt M. The risk of Stevens - Johnson syndrome and Toxic Epidermal Necrolysis associated with nonsteroidal antiinflammatory Drugs: a multinational perspective. The journal of rheumatology, 2003, 30:

8 conditions in question and the risk of a fatal outcome (mortality rate around 40% for the severe forms) Cardiovascular safety The assessment of piroxicam s cardiovascular safety is based primarily on the meta-analysis conducted by McGettingan (2006) 16 which included 23 observational studies ( ), 17 of which were case-control studies (86,193 cases and at least 528,000 controls) and 6 were observational studies (75,520 patients on coxib, 375,619 on non-coxib NSAID and 594,720 not exposed to NSAID). The aim of this meta-analysis was to compare the risks of serious cardiovascular complications with the various NSAIDs. In the case-control studies, the cases had had an infarction or sudden cardiovascular death. One study also included cases of unstable angina. The main criterion used in the observational studies was infarction and sudden cardiovascular death. No additional risk of cardiovascular complication for piroxicam versus the other NSAIDs was found Pharmacovigilance data The pharmacovigilance data for piroxicam shows that severe gastrointestinal adverse effects occur at the recommended doses (94% of cases) and after the first few weeks of treatment (59%). Most of the spontaneous reports of serious skin reactions also related to cases which occurred at the recommended doses (20 mg) and within the first 14 days of treatment (68% of cases). The CHMP emphasises that although there may be a bias towards events occurring shortly after starting to take piroxicam being reported, the data suggests that this risk of short-term serious gastrointestinal and skin reactions must not be ignored Conclusion A meta-analysis of 38 randomised clinical studies confirmed that piroxicam has a level of efficacy comparable to that of other NSAIDs. The findings of epidemiological studies (observational studies and case-control studies) suggest a higher risk of serious gastrointestinal complications (haemorrhage, ulcers, perforations) and serious skin reactions (Lyell s syndrome, Stevens-Johnson syndrome and polymorphic erythema) with piroxicam than with other NSAIDs. The vast majority of these serious adverse effects occur at the recommended doses and in the first few weeks of treatment. Serious skin adverse effects are rare in absolute terms, but this risk must not be dismissed since 40% of cases prove fatal. This increased risk is shared with other active ingredients in the oxicam class. The highest risk appears to be associated with tenoxicam. 16 McGettingan P. et al. Cardiovascular risk and inhibition of cyclooxygenase : a systematic review of the observational studies of selective and non selective inhibitors of cyclooxygenase. JAMA, 2006;vol 296, n 13:

9 Epidemiological data did not cause the CHMP to change its views as to the cardiovascular safety of NSAIDs, and piroxicam does not appear to be associated with a greater cardiovascular risk than other NSAIDs. 4 TRANSPARENCY COMMITTEE CONCLUSIONS 4.1. Reassessment of actual benefit Rheumatoid polyarthritis is a chronic inflammatory rheumatic condition. It is potentially serious and disabling, and affects women more than men. Ankylosing spondylitis is a chronic inflammatory disease which usually affects the vertebral column and the sacroiliac joints (axial form) and the peripheral entheses (especially the heel). This condition, which usually starts between the ages of 15 and 40, affects three times as many men as women. Osteoarthritis is a joint disease which becomes more common with increasing age. It can cause pain, and is characterised by the possible development of disability and/or a marked deterioration in the quality of life. Disabling forms, especially those affecting the hip and knee, can be treated surgically. These proprietary medicinal products are intended for use as part of symptomatic therapy. Public health benefit: Osteoarthritis (disabling osteoarthritis of the knee and hip) represents a major public health burden. Rheumatoid polyarthritis and ankylosing spondylitis represent a significant public health burden. The therapeutic requirements of most patients suffering from osteoarthritis, rheumatoid polyarthritis or ankylosing spondylitis are theoretically met by the use of other NSAIDs. The data available for the therapeutic indications, and in particular data relating to the risk of adverse effects, does not indicate that CYCLADOL has any positive impact on morbidity or mortality. Consequently, CYCLADOL is not expected to benefit public health. The adverse effects associated with piroxicam are the classic adverse effects of NSAIDs: gastrointestinal disorders, hypersensitivity reactions (dermatological, respiratory or general), effects on the central nervous system, skin and mucous membrane reactions. However, the epidemiological data available has shown an increased risk of serious adverse effects on the gastrointestinal system (ulcers with haemorrhagic complications, perforations) and the skin (Lyell s syndrome, Stevens- Johnson syndrome and polymorphic erythema) with piroxicam compared to other NSAIDs. The efficacy/adverse effects ratio in rheumatoid polyarthritis and ankylosing spondylitis is moderate. The efficacy/adverse effects ratio in symptomatic osteoarthritis of the hip and knee is low. These are second-line treatments when an NSAID is indicated. 9

10 The actual benefit of the CYCLADOL proprietary medicinal products is moderate in secondline use where an NSAID is indicated, in rheumatoid polyarthritis, ankylosing spondylitis and disabling symptomatic osteoarthritis of the hip and knee. It is inadequate in other types of osteoarthritis Treatment strategy Rheumatoid polyarthritis: Treatment for rheumatoid polyarthritis aims at relieving pain, improving the functional disability and maintaining joint function. Drug treatment involves immediate symptomatic treatment (NSAIDs, low-dose corticoids and/or analgesics) and disease-modifying treatment aimed at achieving remission and slowing joint destruction. Anti-TNFs are prescribed for forms which do not respond to conventional disease-modifying medication, such as methotrexate, and for severe, active and progressing forms of the condition. Ankylosing spondylitis: The main aim of the treatment is to relieve pain, improve the functional disability and to prevent stiffening and deformation. NSAIDs are used as first-line treatment. Disease-modifying medication is useful in forms in which the peripheries of the body are affected. Rehabilitation is carried out to prevent ankylosis (joint amplitudes and thoracic enlargement) and deformation. Anti-TNFs (infliximab, etanercept) are indicated in patients with severe forms that are not sufficiently well controlled by NSAIDs prescribed at the maximum safe dose. Osteoarthritis of the hip and knee: The medical management of patients suffering from osteoarthritis is based on: - non-drug treatments: loss of excessive weight, functional rehabilitation, use of sticks, etc. - drug treatments (including painkillers) when pain is experienced. Paracetamol is the first-choice painkiller and should be used over the long term if it is effective. NSAIDs are used as second-line treatment (if paracetamol fails) for the shortest period required and at the lowest effective dose Transparency Committee recommendations The Transparency Committee recommends inclusion on the list of medicines reimbursed by National Health Insurance and on the list of medicines approved for use by hospitals and various public services in the following indications and at the dosages in the marketing authorisation: symptomatic treatment of rheumatoid polyarthritis, ankylosing spondylitis and symptomatic osteoarthritis of the hip and knee, as second-line treatment where an NSAID is indicated Packaging: Appropriate for the prescription and dispensing conditions Reimbursement rate: 35% 10

11 APPENDICES Table 1: Summary of the meta-analyses of clinical studies examining the gastrointestinal safety of piroxicam Ref.: Study type Effect investigated Comparators Results Richy (2004) Richy (2007) - Meta-analysis of 32 randomised controlled clinical studies and 13 observational controlled studies ( ) - 3 randomised clinical trials assessing piroxicam, controlled versus placebo or non-nsaid users (284 patients) - Aim: to assess the risk of GI complications of the main NSAIDs, especially in respect of the length of exposure. - Meta-analysis of 75 published clinical studies ( ) - 45 studies (17,165 patients) assessing piroxicam - Aim: to reassess the safety/efficacy profile of piroxicam compared to 7 other NSAIDs 2 types of assessment criteria depending on the studies: - minor GI event (abdominal pain, nausea, constipation, diarrhoea and dyspepsia). - major GI event (gastric, duodenal or intestinal ulcer, haemorrhage, perforation, hospitalisation and related death). Assessment criteria of the 3 piroxicam studies: minor (29 patients), major (229 patients), or severe (26 patients) GI complication Major GI events (perforation, ulcer and haemorrhage) indomethacin, naproxen, diclofenac, piroxicam, tenoxicam, meloxicam, ibuprofen Comparison with an inactive control: placebo or group not exposed to NSAIDs 7 other NSAIDs (not specified in the CHMP report) Risk of GI complication among patients on NSAIDs compared to patients not exposed to NSAIDs or on placebo Combined RR values based on clinical trials (95%CI) Indomethacin RR = 2.25 [1.00; 5.07] (n=1.357) Naproxen (n=792) RR =1.83 [1.25; 2.68] Diclofenac (n=1132) RR = 1.73 [1.21; 2.46] Piroxicam (284) RR = 1.66 [1.14; 2.44] Tenoxicam (n=310) RR = 1.43 [0.40; 5.14] Meloxicam (n=1495) RR = 1.24 [0.98; 1.56] Ibuprofen (n=1384) RR = 1.19 [0.93; 1.54] NSAIDs (n=6754) RR = 1.55 [1.40; 1.72] Combined RR values based on cohorts NSAIDs RR = 2.22 [1.70; 2.90] According to length of exposure: 1 year RR = 2.17 [1.76; 2.67] 1 to 2 years RR = 1.45 [1.19; 1.75] Combined OR of the risk of major GI events for patients exposed to piroxicam compared to other NSAIDs: OR = 1.33 [0.96; 1.84] 11

12 Table 1 (continued) Ref.: Study type Effect investigated Comparators Results Combined analysis of 28 clinical studies selected according to the following criteria: - at least one meloxicam treatment group - treatment duration 21 days - cohort 20 per treatment arm - publication prior to patients at risk of gastrointestinal bleeding included Serious GI complication: - gastric or duodenal perforation - pyloro-duodenal stenosis - GI haemorrhage with significant haemodynamic impact - Meloxicam* (15 mg/d) - Diclofenac ( mg/d) - Piroxicam (20 mg/d) - Naproxen (1000 mg/d) Total risk of serious GI complication for an exposure period < 60 days Number of NSAID Total risk events Piroxicam 0.9% 15 Naproxen 0.5% 1 Meloxicam 15 mg/d 0.2% 5 Diclofenac 0.1% 7 Singh (2004) 24,196 patients (13,118 on meloxicam, 5,371 on piroxicam, 5,464 on diclofenac and 243 on naproxen) Kaplan-Meier survival analysis of serious GI adverse events and thromboembolic complications Aim: to assess the risk of serious GI adverse effects with piroxicam at the recommended doses. Two-by-two comparison of treatments for an exposure period < 60 days«p «Substances compared meloxicam vs diclofenac 0.9 meloxicam vs piroxicam 0.03 meloxicam vs naproxen 0.5 diclofenac vs piroxicam 0.09 piroxicam vs naproxen 0.7 p: log rank test There was a significant difference in the occurrence of serious GI complications between meloxicam and piroxicam.»in contrast, there was no significant difference between diclofenac and piroxicam or between piroxicam and naproxen. «12

13 Gaffney (1998) - Meta-analysis of 38 randomised studies (15,208 patients) - Aim: to assess the efficacy and GI toxicity of piroxicam 20 mg compared to other NSAIDs in chronic and acute pathologies. 2 measurements of GI toxicity: - proportion of patients with GI adverse effects - proportion of patients with a GI haemorrhage or GI ulcer 8 other NSAIDs: naproxen, diclofenac, nabumetone, ibuprofen, tenoxicam, indomethacin, etodolac and oxaprozin - No increase in the overall frequency of GI events for piroxicam when used for either chronic or acute treatment. - There were not enough cases of haemorrhage or ulcer to reach the significance threshold (15/3,020 for piroxicam and 20/7,221 for the NSAIDs). - The gross RR value for piroxicam vs. other NSAIDs was 1.79 with a 95%CI = [0.92; 3.50], which is not significant. - No haemorrhage or GI ulcer in acute pathologies. *: Two doses of meloxicam were studied: 7.5 and 15 mg. The results presented relate only to the 15 mg dose as this corresponds to the dose recommended for first-line treatment of RP and AS. Table 2: Summary of the meta-analyses of observational studies assessing the gastrointestinal safety of piroxicam Ref.: Study type Effect investigated Hernandez-Diaz (2000) - Meta-analysis of 18 studies ( ): 15 case-control studies and 3 observational studies - Aim: to assess the risk of upper GI complications according to the type of NSAID or individual risk factors Perforation or upper GI haemorrhage, or upper GI event leading to hospitalisation or consultation with a specialist Comparators 8 NSAIDs: Azapropazone, piroxicam, ketoprofen, indomethacin, naproxen, ibuprofen, sulindac, diclofenac Results The GI complications risk factors: age, GI history, gender, recent introduction of treatment, dose NSAID (number of studies involved) Pooled RR value for upper GI complications for NSAID users vs. nonusers Azapropazone (2) 27.5 [12.0; 62.9] Piroxicam (12) 6.3 [5.5; 7.2] Ketoprofen (6) 4.6 [3.3; 6.4] Indomethacin (10) 4.6 [3.8; 5.5] Naproxen (12) 4.0 [3.5; 4.6] Sulindac (6) 3.6 [2.8; 4.7] Diclofenac (11) 3.3 [2.8; 3.9] Ibuprofen (9) 1.9 [1.6; 2.2] All NSAIDs (18) 3.8 [3.6; 4.1] 13

14 Table 3: Summary of observational studies which assessed gastrointestinal safety but were not included in the aforementioned meta-analyses Ref. Study type Effect investigated Results Gallerani (2004) - Transversal study - 32,388 patients (average age: 70+/- 15) Data obtained from a register (GIFA: Italian pharmacovigilance group for elderly subjects) - Aim: to assess links between clinical parameters, medication at home, and the risk of hospitalisation for upper GI haemorrhage Upper GI haemorrhage: -Diagnosis: haemorrhage of the oesophagus, stomach or duodenum, or haemorrhagic peptic ulcer, -Haematemesis, melaena, anaemia or shock with confirmation of acute oesophago/gi disease Adjusted OR associated with the risk of upper GI haemorrhage Substance Number of cases Adjusted OR (% of cohort) Piroxicam 22 (2.3%) 4.76 [2.98; 7.60] Indomethacin 10 (1.1%) 3.12 [1.63; 5.38] Ketorolac 15 (1.6%) Naproxen 6 (0.6%) 2.45 [1.44; 4.15] 2.15 [0.92; 4.89] Indobufen 19 (2.0%) Diclofenac 40 (4.3%) 1.81 [0.62; 5.25] 1.80 [1.29; 2.51] Nimesulide 6 (0.6%) 1.41 [0.61; 3.25] All NSAIDs 95 (10.1%) 2.06 [1.65; 2.85] Mellemkjaer (2002) - Historical cohort study - 156,138 subjects exposed to NSAIDs - Aim: to assess the association between and excess risk of upper GI haemorrhage and NSAIDs, by comparing the frequency with which haemorrhage occurs in the population studied (O) and that expected in the general population (E), by calculating an O/E ratio. Upper GI haemorrhage: oesophagitis, gastritis, gastric, duodenal, gastroduodenal or gastrojejunal ulcer haematemesis, melaena or unspecified GI haemorrhage O/E ratio of risk of upper GI haemorrhage Number of cases O/E observed Ketoprofen [4.5; 8.5] Piroxicam [3.3; 7.2] Diclofenac [3.5; 6.6] Indomethacin [2.9; 6.0] Naproxen [2.1; 4.2] Ibuprofen [2.0; 2.9] 14

15 Table 3 (continued): Ref. Laporte, (2004) Lanas (2003) - Case-control study ( ) Study type Effect investigated Results - 2,813 cases, hospital controls - Aim: to assess the risk of upper GI haemorrhage associated with the use of painkillers and NSAIDs, paying particular attention to new substances - Case-control study ( ) - 1,122 cases, 2,231 controls (general population and hospital patients) - Main aim: to assess the effect of nitrate vasodilators on the risk of upper GI haemorrhage associated with NSAIDs and lowdose aspirin Upper GI haemorrhage: all cases admitted with a diagnosis of acute upper GI haemorrhage caused by a gastric or duodenal ulcer, an acute lesion of the gastric mucous membrane, an erosive duodenitis, or mixed lesions. Upper GI haemorrhage: all patients admitted to hospital with a bleeding peptic lesion (including ulcers, erosions and acute mucosal lesions) OR associated with the risk of upper GI haemorrhage cases controls OR Ketorolac [8.0; 77.0] Piroxicam [10.0; 24.2] Ketoprofen [3.9; 25.8] Indomethacin [4.4; 22.6] Naproxen [5.7; 17.6] Aspirin [6.7; 9.6] Rofecoxib [2.3; 23.0] Meloxicam [2.2; 15.0] Dexketoprofen [1.7; 13.9] Diclofenac [2.6; 5.4] Nimesulide [1.9; 5.6] Ibuprofen [2.0; 4.9] Aceclofenac [0.6; 3.3] OR associated with the risk of upper GI haemorrhage cas cont Gross OR OR adjusted for es rols risk factors Ketorolac [3.1; 185.6] 56.7 [7.2; 444] Piroxicam [4.9; 18.5] 18.5 [9.2; 36.9] Aspirin [2.7; 3.8] 6.6 [5.2; 8.2] Naproxen [1.3; 5.1] 5.3 [2.6; 10.8] Diclofenac [1.4; 3.6] 5.1 [3.1; 8.4] Paracetamol [0.52; 1.10] 0.6 [0.4; 1.0] 15

16 Table 3 (continued): Ref. Study type Effect investigated Results Lanas (2006) - Case-control study ( ) - 2,777 cases, 5,532 hospital controls Aim: to quantify the risk of upper GI haemorrhage associated with coxib treatment in their everyday use. Secondary aims: to quantify and compare the risks of upper GI haemorrhage with conventional NSAIDs, aspirin, platelet antiaggregants and anticoagulants. Upper GI haemorrhage: patient hospitalised following upper GI bleeding (haematemesis or melaena) confirmed by endoscopy showing a peptic ulcer lesion. OR associated with the risk of upper GI haemorrhage cases controls OR adjusted for risk factors Non-NSAID 2,017 4,693 Reference users Ketorolac [5.2; 39.9] Piroxicam [7.8; 20.3] Meloxicam [4.0; 23.8] Indomethacin [3.9; 20.7] Ketoprofen [2.5; 29.2] Lormoxicam [2.4; 24.4] Naproxen [4.7; 11.4] Ibuprofen [3.1; 5.3] Diclofenac [2.3; 4.2] Aceclofenac [1.5; 4.6] 16