The Safety of COX-2 Inhibitors

March 2005

Deliberations from the February 16-18, 2005, FDA Meeting

A joint meeting of the Arthritis Advisory Committee and the Drug Safety and Risk Management Advisory Committee was convened on February 16-18, 2005 to review the safety of COX-2 inhibitors. Originally planned as an Arthritis Advisory Committee meeting, attendance was expanded to include members of the FDA Drug Safety committee following a November Congressional hearing on the rofecoxib withdrawal. The meeting was chaired by Dr. Alastair Wood and comprised 32 voting panelists, eight rheumatologists, 19 physicians, eight epidemiologists/biostatisticians, one ethicist and several patient and industry representatives. During this three-day meeting, both FDA reviewers and pharmaceutical manufacturers gave detailed presentations on the cardiovascular safety of rofecoxib, celecoxib, valdecoxib, etoricoxib, lumiracoxib and naproxen. The evidence addressing any potential association between these agents and cardiovascular (CV) and thromboembolic events was presented and analyzed. The FDA put forth several key questions about each drug to the committee: 1) does the agent pose a significant risk for cardiovascular events?; 2) does the risk versus benefit profile (when used for the approved indications in accord with product labeling) of the drug support its marketing in the U.S. ?; and, 3) if continued marketing is supported, what actions are recommended to ensure its safe use?

Several types of trials were analyzed, including placebo-controlled and active-controlled trials and observational studies. Most of the available data on CV events was accrued during the course of phase II and III drug development trials, many of which used active comparators (naproxen, diclofenac, ibuprofen); only a few were placebo-controlled. The committee recognized the limitations of observational studies and stated the signals arising from these should prompt additional prospective controlled trials. In the controlled trials, few CV adverse events were observed. The ability to interpret such data is limited, as these trials were not performed with safety events as a primary endpoint and were underpowered to detect subtle differences in outcomes such as CV events. Indications of a significant safety concern can only be concluded if found repeatedly and with substantial weight, especially in controlled trials primarily designed to show efficacy. Conclusive evidence therefore will take further time and study.

Following on results from early trials (e.g., the VIGOR study, that suggested a potential CV risk for rofecoxib), prospective trials were designed to examine cardiovascular outcomes in patients on COX-2 drugs. However, to date there have been only three such trials (Table 1A). This includes the APPROVE colon polyp prevention trial and two trials in patients undergoing coronary bypass graft (CABG) surgery (these were actually designed to test the role of high dose COX-2 inhibitors in post-op pain control). While the smaller CABG-I did not show more CV events, it did show more strokes. The larger CABG-II trial showed a higher incidence of CV events in patients receiving sequential parecoxib and valdecoxib. Additional data concerning CV events were obtained from large placebo-controlled NSAID trials (Table 1B) and active comparator NSAID trials (Table 2).

Celecoxib

The results of multiple controlled trials indicate that low doses of celecoxib (200 mg qd) do not appear to be associated with an increased CV risk. It should be noted that the majority of these trials were of relatively limited exposure (4-12 weeks). Longer duration of drug exposure and higher doses may be associated with a greater risk. The celecoxib in the adenoma prevention (APC) trial, which included 2035 patients, retrospectively examined CV events among those randomized to receive placebo or celecoxib at doses of 200 mg bid or 400 mg bid for three years. This trial showed a dose related increase in CV events (Table 1B). A much smaller trial of celecoxib in Alzheimer's prevention showed a trend towards more events but was too small to draw conclusions. When controlled trials used diclofenac and ibuprofen as an active comparator, celecoxib yielded comparable rates of CV events. These data do not confirm the safety of the coxib, as these nonselective NSAIDs may also share an increased CV risk.

Valdecoxib

Data on the safety of valdecoxib is limited because of the smaller number of patients enrolled in clinical trials. In the recently published CABG-II trial, high risk CV patients (~544 per group) who recently underwent CABG were randomized to receive either placebo, parecoxib (a prodrug of valdecoxib) 40 mg IV for three days followed by oral valdecoxib 40 mg daily for another seven days, or IV placebo for three days followed by oral valdecoxib. Patients were treated for 10 days (Table 1A). In the placebo group there were three CV events, the sequential placebo/valdecoxib group had six CV events (three occurred while on placebo) and 11 CV events were seen in the parecoxib/valdecoxib group (RR: 3.7 vs. placebo). Two earlier short term trials (post-Surgical study; CABG-1) failed to show an increase in CV events for those patients exposed to high dose valdecoxib (or parecoxib) for 7-10 days post-operatively. Nonetheless, the parecoxib plus valdecoxib group did have a higher rate of cerebrovascular accidents and renal failure (9.3% vs. 4.6%) in the CABG-1 study. Meta-analyses of many short term trials (N >12,000 patients) for arthritis or pain failed to show any enhanced CV risk when valdecoxib was compared to other NSAIDs, but did show higher rates of edema and hypertension.

Table 1A. Placebo-controlled trials: Prospective CV Endpoint*
Trial	N	Population	COX-2, dose/day (Rx duration)		Control	Results
APPROVE**	2586	Polyp prevention	Rofecoxib 25 mg (3 yr)		Placebo	Cardiac events 31 v 12; HR 2.8 (CI, 1.4-5.4)
CABG-II**	1671	High risk CABG	Valdecoxib 40 mg Parecoxib 40 mg (10 d)		Placebo	CV/thromboembolic 11 v 3; RR 3.7 (CI, 1-13.5)
CABG-I (035)**	462	CABG	Parecoxib IV 80mg Valdecoxib 80 mg (10 days)		Placebo	No increase in MI; Increased CVA in parecoxib/ valdecoxib (2.9% v 0.7%)
Table 1B. Placebo-controlled trials: Sporadic, Observed CV Adverse Events
APC	2035	Polyp prevention		Celecoxib 400 mg Celecoxib 800 mg (>2.8 yrs)	Placebo	HR 400mg RR 3.0 (0.3-28.6) 800 mg RR 6.1 (0.7-50.3)
Alzheimers-001	425	Alzheimer's prevention		Celecoxib 400 mg (1 yr)	Placebo	MI 2 v 0; CV deaths 2.4% v 1.4%
Pre-SAP	1561	Polyp prevention		Celecoxib 400 mg	Placebo	No CV increase
ADAPT +	2463	Alzheimer's prevention		Celecoxib 400 mg (~2 yrs)	Placebo Naproxen	No CV increase
Alzheimer/MCI	2091	Alzheimer's prevention		Rofecoxib 25 mg (1 yr)	Placebo	No CV increase
VICTOR $	1976 PY	Polyp prevention		Rofecoxib (3 yr)	Placebo	RR 3.14 (1-9.75)

* Abbreviations: RR Relative Risk; HR Hazard Ratio; CV Cardiovascular event; CVA Cerebrovascular accident; MI Myocardial infarction; PY Patient-years; v versus; CI 95% Confidence Intervals

+ Manuscript in preparation. ** Trials that permitted background aspirin < 325 mg qd. $ Interim analysis

Rofecoxib

Rofecoxib use was shown to be associated with an increased CV risk in the VIGOR, APPROVE and VICTOR trials, as well as in several observational databases. This risk was greatest at the 50 mg dose, but was infrequent at the 25 mg and not observed at the 12.5 mg dose. In the APPROVE trial, 2586 patients with colonic adenomas received either placebo or rofecoxib 25 mg daily and were followed for three years. Initially there were no differences, however, after 18 months, more cardiac events (26 vs. 46), myocardial infarctions (9 vs. 21) and cerebrovascular events (mostly ischemic stroke; 6 vs. 11), occurred in the rofecoxib group (Table1A). The presence of background low dose aspirin therapy did not change these results.

Lumiracoxib and Etoricoxib

These agents have been studied and also shown some safety concern for CV outcomes. The results of the two part TARGET study have been published. When lumiracoxib was compared to ibuprofen, no differences in CV risk were observed (Table 2B). However, when this drug was compared to naproxen, more CV events were noted in the lumiracoxib group. Moreover, of the 25% of patients who also took low dose aspirin, variable and inconsistent effects on CV outcomes and stroke were seen.

In pooled analyses, etoricoxib (given for arthritis) was not shown to be different from non-naproxen NSAIDs (data not shown), but was shown to yield more CV events when compared to patients receiving naproxen (Table 2A). Several large, noninferiority trials are currently in progress: EDGE (7100 OA patients); EDGE II (4090 RA patients); and MEDAL (23450 OA and RA). These trials will examine the long term (over 16-40 mo.) CV risk with etoricoxib versus diclofenac therapy in OA and RA patients, some of whom have CV risk factors and take low dose ASA.

Table 2A. Naproxen active comparator trials: Observed Adverse events
Trial	N	Population	COX-2, dose/day	Comparator	Results
VIGOR	8076	RA	Rofecoxib 50 mg (10 mos)	Naproxen	MI: 20 v 4 @ 8mos RR 2.38 (1.39-4.0)
Etoricoxib	3457	Meta-analyses	Etoricoxib	Naproxen	RR 1.70 (0.9-3.18)
TARGET (0117)**	9471	OA	Lumiracoxib 400 mg (1 yr)	Naproxen	HR 1.77 (0.82-3.84)
Table 2B. NSAID active comparator trials: Observed Adverse events
CLASS**	5968	OA, RA	Celebrex 800 mg (9-15mos)	Diclofenac Ibuprofen	No CV increase
SUCCESS-1	13194	OA	Celebrex 200, 400 mg	Naproxen Diclofenac	No CV increase
EDGE**	7111	OA/GI tolerability	Etoricoxib (9-16 mos)	Diclofenac	No CV increase Small ↑ HTN
TARGET (2332)**	8773	OA	Lumiracoxib 400 mg (1 yr)	Ibuprofen	No CV increase

** Trials that permitted background aspirin < 325 mg qd

Naproxen and other NSAIDs

Among the traditional NSAIDs, naproxen has received particular attention. A number of studies have shown that treatment with naproxen resulted in fewer CV events in head to head trials compared to COX-2 inhibitors. Several meta-analyses have suggested that naproxen (500 mg bid) either had no increased risk or a lower risk of CV events when compared to placebo. In December 2004, an increase in CV events was reported in Alzheimer's patients receiving naproxen (220 mg bid) in the ADAPT trial. However, the data safety monitoring board analysis never judged there to be an increase CV risk for naproxen (when compared to placebo and celecoxib). The ADAPT study was halted because of media attention and not because of an unacceptable safety signal. That decision now appears ill-advised.

Data suggesting CV risks with the remaining NSAIDs is incomplete, but concerning. Unpublished observational studies reported by Dr. David Graham using the Kaiser HMO and Medi-Cal databases suggest an increase CV risk with many NSAIDs (indomethacin, meloxicam, sulindac); risk was highest for high dose rofecoxib ( see ACR Hotline "Cardiovascular Complications Related to COX-2 Inhibitors" September 2004). Preliminary data from a Norwegian study + (Sudbo J, et al) was presented. This nested case-controlled study examined causes of death in patients with oral cancer from 1975-2003 and showed increased CV deaths with CV hazard ratios for traditional NSAIDs, ranging from 1.70 for naproxen to 2.86 for ibuprofen.

Recommendations and voting

The committee voted unanimously (32-0) that all of the COX-2 inhibitors currently (or previously) available in the United States (celecoxib, valdecoxib, rofecoxib) significantly increase the risk of CV events in users of these drugs. Considering potential benefits as well as risks and their magnitude, the committee voted unanimously (32-0) in favor of keeping celecoxib on the market for its current indicated uses. There was less agreement in favor of keeping valdecoxib (17 yes, 13 no, 2 abstained) and rofecoxib (17 yes, 15 no) on the market. The committee unanimously agreed that a “black box” warning should be added to the label of each of the COX-2 inhibitors indicating potential cardiovascular risk and encouraging avoidance in high risk individuals. Many panelists were in favor of withdrawing the 50 mg form of rofecoxib from future use, but many also favored reintroduction of rofecoxib liquid suspension for limited use in pediatric populations. All were in favor of education measures in this area directed at physicians and patients. Most panelists favored restrictions on direct-to-consumer advertising of COX-2 inhibitors. There was unanimous agreement that product labels for all currently marketed traditional NSAIDs should also be revised with regard to cardiovascular risk, with most panelists favoring a “warning,” rather than a black box warning. All voted in favor of requiring future agents in this class (both COX-2 and nonselective NSAIDs) to perform cardiovascular safety studies prior to market introduction.

What's next?

The FDA will review the findings of the panel; it is expected that the agency will expediently adopt the recommendations of the Advisory Panel. The first changes noticeable to clinicians may be in product labeling, particularly risks and warnings, and how to best advise patients. More complex measures directed at education, advertising, and future research will probably take a year or more to enact. The future availability of rofecoxib, etoricoxib and lumiracoxib was not discussed. Each will have to conform to appropriate product labeling modifications and some may have to perform additional clinical trials focused on safety outcomes (cardiovascular events). It is not known whether Merck will petition the FDA to reintroduce rofecoxib to the U.S. market.

The Bottom Line

• The risk versus benefit of the specific COX-2 inhibitors favors their continued use in the U.S., although patients should be counseled about potential cardiovascular risks.
• While all NSAIDs may impose similar cardiovascular risks, some agents (naproxen 500 mg bid, celecoxib 200 mg qd) appear to be safer than others (e.g., rofecoxib, valdecoxib, diclofenac, and ibuprofen).
• Cardiovascular risks of the COX-2 inhibitors may be greater with higher doses, longer durations of therapy, and when used in high risk individuals.
• The use of low dose aspirin does not consistently abrogate the potential CV risk of a COX-2 inhibitor. Patients who require the cardioprotective effects of aspirin may not be ideal candidates for COX-2 inhibitor or NSAID therapy.
• The use of COX-2 inhibitors (and other NSAIDs thought to increase CV risk) should be avoided in high risk individuals.

Hotline Authors: John J. Cush, MD, FDA Arthritis Advisory Committee Member; Arthur Kavanaugh , MD, Hotline Editor; Eric L. Matteson, MD, MPH, Hotline Co-Editor.

Disclosures: Dr. Cush is not a consultant, advisor or speaker for any of the COX-2 manufacturers. Prior to 2004, he has advised or been a speaker for the NSAID manufacturers. Drs. Kavanaugh and Matteson are not advisors, consultants, or speakers for any of the NSAID or COX-2 manufacturers.

Hotline is provided by the ACR as a service to members. This Hotline reflects the views of the authors and does not represent a position statement of the College.

References

Arthritis advisory committee.

Solomon SD, et al. NEJM , E. Pub Feb. 15, 2005 (APC trial)

Nussmeier NA, et al. NEJM, E. Pub Feb. 15, 2005 (CABG trial)

Bresalier RS, et al. NEJM, E. Pub Feb 15, 2005 (APPROVE trial)

 

Previous | Index | Next