Report of the Expert Advisory Panel on the Safety of COX-2 Selective Non-steroidal Anti-Inflammatory Drugs (NSAIDs)

2005-07-06

Help on accessing alternative formats, such as Portable Document Format (PDF), Microsoft Word and PowerPoint (PPT) files, can be obtained in the alternate format help section.

Contact Policy Bureau Enquiries

1. Mandate of the Panel

The panel was asked to address the following questions¹:

1. Do Celebrex (celeCOXib), Bextra (valdeCOXib) and Vioxx (rofeCOXib) significantly increase the risk of cardiovascular events such as heart attack and stroke?
2. What is the role of the concomitant use of low-dose aspirin in reducing the cardiovascular risk in patients treated with COX-2 inhibitors?
3. Do the overall benefits and risks of Celebrex, Bextra and Vioxx justify the marketing of these drugs in Canada?
4. Are there patient populations in which the potential risks of Celebrex, Bextra and Vioxx outweigh the potential benefits?
5. What actions does the Panel recommend that Health Canada consider implementing to ensure the most appropriate use of Celebrex, Bextra and Vioxx?
6. What additional clinical trials or other studies are essential to further evaluate the potential cardiovascular risk of all NSAIDs (conventional and COX-2 inhibitor)?

2. Membership of the Panel

The 13 member Panel consisted of three rheumatologists, two individuals with rheumatoid arthritis, two cardiologists, two general internists, two methodologists, one gastroenterologist and one family physician¹. Information about their potential conflicts of interest is available¹.

3. Panel deliberations

The panel met in Ottawa on June 9 and 10, 2005. Prior to the meeting, Health Canada provided the panel with summaries of relevant clinical trial data, information from post marketing data, and the manuscripts of published randomized trials and observational studies felt by Health Canada to be most relevant for the panel. Some of the clinical trial and post marketing information provided by Health Canada is publicly available and some is not. However, the confidential information provided by Health Canada did not lead the panel to different conclusions than had they only based their report on information currently in the public domain.

On June 9 the panel heard presentations from, and was able to ask questions of, the manufacturers of Celebrex, Bextra, Vioxx and lumiraCOXib, representatives of the public and interested groups, non-industry affiliated experts (Drs. Dubé and Baigent, academics who had completed systematic reviews of the gastrointestinal and cardiovascular effects of COX-2 inhibitors respectively), and Health Canada¹. The panel also considered statements from the public that were submitted to the Health Canada website¹. On June 10 the committee deliberated in private, and completed its report during the subsequent two weeks.

¹ http://www.hc-sc.gc.ca/english/protection/cox2/index-eng.php#hcpres

4. Non-steroidal anti-inflammatory drugs and the management of pain

Non-steroidal anti-inflammatory drugs are indicated for the treatment of various types of inflammatory arthritis such as rheumatoid arthritis. They are also beneficial in some patients with fever, osteoarthritis, non-specific musculoskeletal pain, and other pain syndromes - the latter are the conditions for which non-steroidal anti-inflammatory drugs are most frequently prescribed.

Non-steroidal anti-inflammatory drugs can block two cellular cyclo-oxygenase pathways. The earlier drugs (referred to as NSAIDs in this report) are relatively non-selective and block both COX-1 and COX-2, although there is considerable variability in their COX-selectivity. More recently, the so-called COX-2 inhibitors have been introduced, which selectively inhibit COX-2. The rationale for the development of selective COX-2 inhibitors was the observation that traditional non-selective NSAIDs, by inhibiting the COX-1 enzyme, decrease the formation of protective gastrointestinal prostaglandins, thereby favouring the development of upper gastrointestinal tract ulcers, an important source of morbidity and mortality. It was hypothesized that COX-2 inhibitors would be less likely to cause ulcers than traditional NSAIDs.

Aspirin is an anti-inflammatory drug with well-established cardioprotective properties, and is not considered an NSAID in this report.

The first NSAID was marketed in Canada in the 1950s, and currently, many NSAIDs are available in Canada. All are prescription medications except for ibuprofen (Advil and Motrin), which is available "over-the-counter" for the short-term management of pain and fever. The first COX-2 inhibitor was marketed in Canada in 1998. In 2003, three selective COX-2 inhibitors were available in Canada - Celebrex, Bextra, and Vioxx. However, in September 2004 Vioxx was voluntarily withdrawn from the market by Merck because of concern about cardiovascular toxicity, and more recently sales of Bextra have been voluntarily suspended by the manufacturer because of concerns about skin toxicity. At the time of this report, Celebrex is the only COX-2 inhibitor that continues to be available in Canada. COX-2 inhibitors have also been evaluated in other disease processes such as Alzheimers Disease and pre-cancerous polyps of the large colon.

Numerous randomized trials have established the benefits of NSAIDs and COX-2 inhibitors for the acute and subacute management of pain; few studies have followed patients beyond weeks to months. In groups of patients studied in these clinical trials, no anti-inflammatory drug has clearly been shown to be more efficacious than another. However, despite lack of rigorous evidence, it is widely accepted in clinical practice that some patients respond to one anti-inflammatory drug and not another; that the effectiveness of one anti-inflammatory drug may diminish over time in an individual patient; and that the same patient may then respond to another non-steroidal anti-inflammatory drug. The panel heard considerable public and provider input (both at the June 9^th meeting and via Health Canada's website) supporting these assertions, and believes that a broad selection of anti-inflammatory drugs is desirable.

5. Rationale for NSAIDs with higher levels of COX-2 Inhibition

The most recognized side-effect of NSAIDs is gastrointestinal toxicity, manifest most frequently by nausea and abdominal pain (so-called "dyspepsia"), and less frequently by peptic ulcer disease (which can lead to complications such as severe bleeding, obstruction or perforation). Although COX-2 inhibitors have never been shown to be more effective for pain control than NSAIDs, they have become widely prescribed because numerous randomized trials and systematic reviews of randomized trials have shown that COX-2 inhibitors cause less nausea and abdominal pain, and fewer peptic ulcers than NSAIDs. For patients entered into randomized trials, who are generally elderly with relatively severe arthritis, it is estimated that the number needed to treat for one year to avoid one bleeding ulcer, perforation or obstruction is approximately 125, when a COX-2 inhibitor is compared to an NSAID. Of course, alternative gastroprotective strategies for those requiring non-steroidal anti-inflammatory drugs also exist, such as the concurrent use of misoprostol or proton pump inhibitors. Nevertheless, COX-2 inhibitors have been aggressively marketed, and adopted by practitioners, as "safe" alternatives to NSAIDs, so that within a year of licensing, there was a marked increase in the total number of Canadians on anti-inflammatory agents.

Many Canadians are taking these drugs - studies suggest that between 15 and 24% of Canadians 65 years of age or older have taken a prescription non-steroidal anti-inflammatory agent during the previous year. Typical usage can range from a few days for acute pain to continuous use over months or even years for chronic rheumatoid arthritis. The proportion of prescription anti-inflammatory drug consumption that is accounted for by COX-2 inhibitors varies by province, likely due at least in part to different reimbursement and formulary restriction policies - for example, for those over 65 years of age, in 2003 it was 70% of all non-steroidal anti-inflammatory use in Quebec and 4% in British Columbia. Because of the large number of individuals taking anti-inflammatory drugs in general, and COX-2 inhibitors in particular, the recent concern about the cardiovascular toxicity of COX-2 inhibitors has considerable population health implications.

6. Do Celebrex, Bextra and Vioxx significantly increase the risk of cardiovascular events such as heart attack and stroke?

1. COX-2 inhibitors compared with no anti-inflammatory drugs: Short-term randomized trials (12 weeks or less) consistently show that these three drugs increase the frequency of high blood pressure and peripheral edema, and cause a decrease in renal function. The Panel was convinced that all COX-2 inhibitors increase the risk of clinically important cardiovascular events (a constellation of vascular death, heart attack and stroke) compared with placebo. This conclusion was based upon the results of numerous individual randomized trials, and a systematic review of 138 randomized trials of at least four weeks duration involving 144,296 patients, which was presented to the Panel by Dr Colin Baigent (not yet available for public release because it is being considered for peer-reviewed publication). The review demonstrated a statistically significant 41% relative increase in clinically important cardiovascular events in patients allocated to COX-2 inhibitors versus placebo. The estimates of the increased risk associated with Vioxx, Celebrex and Bextra were similar, although there were considerably less data for Bextra than the other two drugs.
   
   The panel felt that this increase in clinically important cardiovascular events is associated with both short-term and long-term use of COX-2 inhibitors. The absolute magnitude of the increased risk depends upon the length of use (greater absolute risk with longer use), the patient's underlying risk of cardiovascular disease (greater absolute risk in those with a history of, or risk factors for, cardiovascular disease), and perhaps the dosage used. Dr Baigent found an absolute increased risk of 0.3% per year in cardiovascular events when comparing COX-2 inhibitors with placebo. This was based upon a systematic review of randomized trials that were mostly of short duration. Thus, the absolute increased risk with use beyond one year is not known, although the panel felt that the absolute risk likely increases with prolonged use.
2. COX-2 inhibitors compared with NSAIDs: Randomized trials have consistently shown that NSAIDs also increase the risk of hypertension and edema compared with placebo. Professor Baigent did not compare the cardiovascular outcomes of patients receiving COX-2 inhibitors with all NSAIDs, but instead separately compared patients receiving COX-2 inhibitors with patients receiving non-naproxen NSAIDs, and those receiving COX-2 inhibitors with naproxen (because of a pre-specified hypothesis that naproxen would have a lower risk of cardiovascular events than other NSAIDs). He found a statistically insignificant relative 12% decrease in clinically important cardiovascular events in patients treated with COX-2 inhibitors compared with non-naproxen NSAIDs. He also found a statistically significant 57% relative increase in clinically important cardiovascular events in patients treated with COX-2 inhibitors compared with naproxen. The panel was not in complete agreement about whether this analysis establishes that naproxen has a better cardiovascular safety profile than other NSAIDs, or whether this should be interpreted as a suggestive finding that needs further confirmation. However, there was agreement that this analysis suggests that COX-2 inhibitors and non-naproxen NSAIDs have a similar cardiovascular risk profile.
3. Summary: The panel believes that, as a group, selective COX-2 inhibitors are associated with an increased risk of clinically important cardiovascular events compared with placebo, and that this increased risk is similar to the risk associated with most NSAIDs. This increased risk is present for all patients taking anti-inflammatory agents (with the possible exception of naproxen), but the absolute risk likely increases with longer-term use and in the presence of risk factors for, or a history of, cardiovascular disease. Naproxen may be associated with a lower risk than other anti-inflammatory agents.

7. What is the role of the concomitant use of low-dose aspirin in reducing the cardiovascular risk in patients treated with COX-2 inhibitors?

It is theoretically possible that the concomitant use of aspirin with COX-2 inhibitors could protect against the increased risk of cardiovascular events. On the other hand, it is also possible that the concomitant use of aspirin could negate the protective effect of COX-2 inhibitors on peptic ulcer disease (compared with NSAIDs).

Available evidence, based upon systematic reviews of randomized trials, suggests that the risk of peptic ulcer disease in patients who take both a COX-2 inhibitor and aspirin is similar to the risk of patients taking NSAIDs - thus, the addition of aspirin markedly reduces or eliminates the decreased incidence of gastrointestinal side-effects demonstrated with COX-2 inhibitors compared with NSAIDs.

The panel was disappointed that a systematic review of the effect of the addition of aspirin to a COX-2 inhibitor upon cardiovascular events has not been performed. The panel was not presented with any convincing data that the addition of aspirin to COX-2 inhibitors protects against the increased risk of cardiovascular disease.

8. Do the overall benefits and risks of Celebrex, Bextra and Vioxx justify the marketing of these drugs in Canada?

All three of these drugs are effective anti-inflammatory agents, and are associated with a decrease in the frequency of both gastrointestinal intolerance and clinically important peptic ulcer disease compared with NSAIDs. At the same time, all three drugs increase the frequency of hypertension, edema, renal disease and clinically important cardiovascular events, which seems similar in magnitude to that associated with NSAIDs. The decision about whether it is justified to market these drugs in Canada depends upon one's interpretation of their risks and benefits, and the amount of information available for each individual drug.

It is the panel's opinion that the available information justifies marketing Celebrex in Canada (Vote: 13 in favour, 0 against). This was based upon:
a) the increased risk of cardiovascular disease caused by Celebrex appears similar to that of most NSAIDs, b) the risk of gastrointestinal harm caused by Celebrex appears less than most NSAIDs, and c) patients benefit from having a variety of drugs to choose from for pain relief.

It is the panel's opinion that the available information justifies marketing Vioxx in Canada (Vote: 12 in favour, 1 against). The rationale for the majority position was similar to that for Celebrex. They felt that no additional studies of Vioxx were mandatory before Vioxx is marketed again, but felt strongly that the studies mentioned in section 11b) should immediately be undertaken by an independent group supported by the manufacturers of all coxibs. The individual who voted against allowing Vioxx to be marketed felt that the bulk of randomized and observational studies suggested that Vioxx caused a greater increased risk of cardiovascular harm than Celebrex (especially at higher doses) and that this risk was sufficiently high to justify continuing not to market Vioxx, especially since alternatives are available.

It is the panel's opinion that the available information does not justify marketing Bextra in Canada (Vote: 8 in support of this opinion, 5 against). The majority who voted that Bextra should not be marketed in Canada at the present time felt that

1. there is very little information about the long-term cardiovascular risk of Bextra (although the cardiovascular effects of COX-2 inhibitors may represent a "class effect", some COX-2 inhibitors may have a greater cardiovascular effect than others),
2. there is concern about the possible increased risk of severe adverse skin reactions (although the absolute frequency of these reactions appears small, the consequences of these reactions can be severe and sometimes fatal), and c) a number of other anti-inflammatory agents with more complete information about benefits and harms are already on the market. They were concerned that should Bextra be allowed back on the market it might be used as a "first line" anti-inflammatory agent, even if drug information contained warnings about skin toxicity. The minority who voted that Bextra should be marketed felt that a) the increased risk of cardiovascular disease caused by Bextra is likely to be similar to that of other COX-2 inhibitors, b) the number of patients suffering severe adverse skin reactions is very small and similar to many other drugs still on the market (such as some antibiotics and anti-convulsants) and
3. some patients will benefit from having another anti-inflammatory agent available. Those who voted for the marketing of Bextra felt that it should only be marketed as a third-line anti-inflammatory agent to be used if others have failed.

9. Are there patient populations in which the potential risks of Celebrex, Bextra and Vioxx outweigh the potential benefits?

Because of the panel's vote against marketing Bextra, the comments in this section only apply to Celebrex and Vioxx.

As discussed in Section 6, all patients who consume COX-2 inhibitors and non-naproxen NSAIDs are exposed to an increased risk of cardiovascular events. However, the magnitude of that risk varies enormously depending upon the circumstance - it is extremely low in a healthy 20 year old taking one of these drugs for a month, and much higher for a patient immediately after coronary artery bypass surgery, or a patient with heart disease who requires the drug continuously over many months or years.

Just as there is a gradation of cardiovascular risk with these drugs, there is a gradation of benefit. Some patients are at high risk of developing a peptic ulcer on an anti-inflammatory drug, and in such patients COX-2 inhibitors are associated with a substantial benefit compared with NSAIDs. Without anti-inflammatory drugs, some patients with severe inflammatory arthritis can barely function because of pain and stiffness, while other patients with minor aches can be adequately treated with rest or acetaminophen. Again, the panel notes that the vast majority of use of any type of non-steroidal anti-inflammatory drug is not for chronic inflammatory arthritis.

Thus, the appropriate use of these drugs must, as with all drugs, be influenced by the individual circumstance of each patient, and how they value the benefits and harms of the drugs. In the end, this decision should be left to informed patients, advised by their well-educated healthcare providers.

10. What actions does the Panel recommend that Health Canada consider implementing to ensure the most appropriate use of Celebrex, Bextra and Vioxx?

Because of the panel's vote against marketing Bextra, the comments in this section only apply to Celebrex and Vioxx.

1. Introduction: Although COX-2 inhibitors are effective anti-inflammatory agents and have less gastrointestinal toxicity than NSAIDs, most of the twofold increase in the use of anti-inflammatory agents in Canada that occurred immediately after the introduction of COX-2 inhibitors cannot be explained by their use in patients with severe inflammatory arthritis who were previously under-treated. Most panel members felt that much of this increase was due to what now appears to be overuse or clinical "creep", fuelled by aggressive industry marketing and under-recognition of the totality of risks associated with these drugs. One of the ironies and unexpected benefits of the studies of COX-2 inhibitors is that they have for the first time convincingly demonstrated the adverse cardiovascular effects of all NSAIDs, both COX-2 selective and non-selective.
   
   Ensuring the appropriate use of anti-inflammatory drugs will require a concerted effort of regulators, patients, physicians, pharmacists and others.
2. Regulations and labeling: The panel discussed the apparent contradiction in having one NSAID (ibuprofen) available as an over-the-counter medication, while the other anti-inflammatory medications are prescription drugs. The panel recognized that the indication for over-the-counter prescribing of ibuprofen is for the short-term relief of pain and fever only, but felt that in reality, the drug was frequently being used chronically and at a high dose, despite the fact that the over-the-counter product is available in only a relatively low dose pill. Health Canada should consider that ibuprofen only be sold after discussion with a pharmacist, and must ensure that the risks of cardiovascular events are prominently displayed in material that individuals receive at the time they purchase the drug as well as any package inserts.
   
   The panel supports ensuring that appropriate warnings about the risks of COX-2 inhibitors are added to the product monograph and material given to the patient (including adding a "black box" warning). Although there may be a small group of patients in whom COX-2 inhibitors are contraindicated (e.g. immediately after cardiac bypass surgery), the panel felt that the absolute risks and harms of COX-2 inhibitors will vary depending upon a patient's cardiovascular and gastrointestinal risks, and the type of musculoskeletal disorder the patient has. Thus, establishing "absolute" contraindications will be difficult, and the ultimate decision about the benefits and harms of these drugs in each patient must be considered on an individual basis. The panel felt strongly that the information about COX-2 inhibitors available to patients must be simple, and should briefly and succinctly describe the most important benefits and risks of these drugs (this applies to all material, not just material from Health Canada).
3. Unbiased information for the public: The public should have access to balanced and unbiased information about these drugs, which they can access through a number of avenues: the world-wide web, material dispensed at their pharmacies, material from groups such as the Arthritis Society, and material from professional organizations such as the College of Family Physicians. The panel encourages all of these groups to join together as soon as possible to develop and harmonize the contents of such materials. The ever-increasing sources of information currently available to patients and the public are both a blessing and a curse - a blessing because patients and the public now have access to an unprecedented amount of information; a curse because it is often not clear how comprehensive, up-to-date and unbiased the various sources of information are. It is important that the groups mentioned above develop and "brand" one source of information as unbiased - a source to which Canadians can turn with confidence. These evidence-based consumer summaries and decision aids should be disseminated through a variety of sources.
4. Pharmaceutical industry advertising: The panel strongly supports the continued ban on direct-to-consumer advertising by those who profit financially from these drugs, recognizing that there is already advertising "creep" across the 49^th parallel. COX-2 inhibitors are among the most highly advertised drugs in history. The panel believes that the degree of advertising was not justified by their therapeutic equivalence related to pain relief (compared with NSAIDs), their incremental benefit with respect to gastrointestinal safety, and their cardiovascular toxicity. The future advertising and detailing (to physicians and pharmacists) of all COX-2 inhibitors and NSAIDs should prominently mention the increased risk of hypertension, edema, renal disease and cardiovascular events, and the therapeutic equivalence in terms of pain relief among all drugs in this class.
5. Clinical Practice Guidelines: Guidelines for physicians and pharmacists should be developed and disseminated. They should describe the benefits and risks of COX-2 inhibitors and NSAIDs, as well as the indications for their use, in a clear and unbiased manner. It should be remembered that the vast majority of non-steroidal anti-inflammatory drugs are prescribed by family physicians, not specialists. The guidelines must be updated regularly, given the quickly changing landscape of trials and data.

11. What additional clinical trials or other studies are essential to further evaluate the potential cardiovascular risk of all NSAIDs (conventional and COX-2 inhibitor)?

1. Making existing data publicly available: The panel unanimously felt that all data related to a drug that is licensed in Canada should be publicly available in an easy-to-access format, so that experts, researchers and the public can independently assess the efficacy and safety of the drug. At the June 9 public meeting, the panel was told by Merck and Pfizer that data from all of their trials were publicly available. However, a search of the web sites of Merck and Pfizer and of an industry-sponsored website ( http://www.clinicalstudyresults.com) on June 10 yielded some data from some published studies, but did not contain information about all randomized trials of these drugs.
   
   Officials at Health Canada have indicated to panel members that current Canadian commercial privacy laws prohibit them from publicly releasing the data from clinical trials and adverse drug reports (e.g. the data concerning the severe adverse skin reactions associated with Bextra) submitted to Health Canada by the manufacturers, as this is considered proprietary information, unless the manufacturer agrees.
   
   The panel strongly recommends that when a drug is being considered for marketing by Health Canada, pharmaceutical manufacturers must publicly release all data (published and unpublished) from completed randomized trials and other studies of relevance to the safety and efficacy of their drugs, register all ongoing trials publicly, and that Canadian regulations should be modified to ensure that all material submitted to Health Canada in support of a request to license a drug (as well as Health Canada's assessment of that drug) is made publicly available.
2. Further analyses of existing data: Five important analyses should immediately be undertaken with existing data from all of the completed randomized trials (published and unpublished):1) a systematic review using individual patient data from all randomized trials of coxibs versus placebo and NSAIDs, to determine whether the addition of aspirin decreases the risk of clinically important cardiovascular events in patients on coxibs and NSAIDs, 2) a time-to-event analysis using individual patient data to provide more information about whether COX-2 inhibitors and NSAIDs produce a smoothly increasing absolute risk difference for cardiovascular events that commences as soon as patients consume the drugs, or whether excess risk occurs only after prolonged use, 3) an individual patient analysis to determine the important risks factors for the development of cardiovascular events (including drug dose), and 4) individual patient analyses to identify risk factors for developing complicated peptic ulcer disease in patients on NSAIDs and COX-2 inhibitors, and 5) a systematic review, similar to that performed by Dr Baigent of COX-2 inhibitors, comparing the risk of clinically important cardiovascular events in patients on any NSAID compared with placebo, in the trials of COX-2 inhibitors that had NSAID and placebo comparisons (this would provide the best available evidence about the cardiovascular risks of NSAIDs compared with placebo). It should be possible to start these studies immediately, since the data are currently available. This analysis should be conducted by a methodologically sophisticated, clinically credible, independent group, with unrestricted logistical and financial support from the COX-2 manufacturers.
3. Future trials: The panel feels that future studies of COX-2 inhibitors should be of similar size to the TARGET study of lumiracoxib (a recently published efficacy trial comparing this new agent (not yet licensed in Canada) to both diclofenac and naproxen), and should be powered to identify small but clinically important increases in cardiovascular events, so that gastrointestinal safety and cardiovascular risk can be examined at the same time. As much attention should be given to the identification and adjudication of cardiovascular events as is given to gastrointestinal events, and information about congestive heart failure and hypertension should be carefully collected. All analyses should be presented as intention-to-treat and on-treatment analyses.
   
   Other issues that some panel members felt were particularly important were: a) designing studies to assess the overall gastrointestinal and cardiovascular effects of adding low-dose aspirin (e.g. 75 to 81mg daily) to COX-2 inhibitors and NSAIDs in patients at high risk of cardiovascular disease, b) using a non-NSAID control arm (e.g. acetaminophen) whenever possible, and c) comparing a COX-2 inhibitor with naproxen (with or without a gastroprotective agent) to compare their effect upon both gastrointestinal and cardiovascular outcomes.
4. Post-marketing evaluation: Although randomized trials clearly provide the most unbiased assessment of the efficacy and harms of anti-inflammatory drugs, post-marketing evaluation, using other study designs can provide important information about drug utilitzation and outcomes in the "real world" and rare side-effects. For example, many trials exclude patients at high risk of cardiovascular events, and carefully conducted observational studies using administrative data can provide information about the real-world risks and benefits of these drugs. These studies can also be used to monitor the real-world use of anti-inflammatory drugs, which can be fed back to patients, physicians and pharmacists, if it appears that their use might be "inappropriate".
   
   Adverse event reporting or active surveillance for adverse events, can also provide useful signals about the possibility of rare and unusual adverse events (e.g. severe adverse skin reactions). However, the panel felt that adverse event reporting was not useful for assessing the cardiovascular risks of anti-inflammatory drugs, because of the relatively high frequency of underlying cardiovascular disease in the patient population taking these drugs.