This editorial refers to ‘Effects of Celecoxib On Restenosis after Coronary Intervention and Evolution of Atherosclerosis (Mini-COREA) Trial: celecoxib, a double-edged sword for patients with angina’†, by H.-J. Kang et al., on page 2653
We are now in the second decade of controversy regarding the cardiovascular effects of cycloxygenase-2 (COX-2) inhibitors. The introduction of this class of drugs occurred on 31 December 1998, when the US Food and Drug Administration (FDA) approved the sale of celecoxib, followed shortly by the approval of rofecoxib on 20 May 1999. The market acceptance of these two drugs was extraordinary, and both rapidly achieved more than a billion dollars in annual sales. The science leading to the discovery of COX-2 inhibitors is a fascinating story and illustrates the creativity of the pharmaceutical industry in medicinal chemistry. By targeting the COX-2 enzyme, these drugs offered the hope for agents that would preserved the clinical benefits of non-steroidal anti-inflammatory drugs (NSAIDs) without the gastrointestinal (GI) adverse effects attributable to inhibition of the COX-1 isoform.
By any reasonable assessment, this targeted therapeutic approach was initially a tremendous success. Studies showed that both COX-2 inhibitors possessed anti-inflammatory and analgesic effects comparable with conventional NSAIDs, but produced fewer symptomatic GI adverse effects. Then, in November 2000, a pivotal study, the VIGOR trial, showed that rofecoxib, compared with naproxen, also produced fewer complicated upper GI adverse events including severe bleeding, ulceration, and perforation.1 However, artfully concealed within the VIGOR manuscript was a dark secret about rofecoxib—although this drug enhanced GI safety, there were significantly more adverse thrombotic cardiovascular events, particularly myocardial infarctions. This safety issue received little public attention until we published an independent analysis of the VIGOR trial in August 2001.2
The rest of the story is now a matter of history and has been fully described in both the medical literature and the lay press. In the autumn of 2004, following a decision by the data monitoring committee to recommend termination of a study of rofecoxib to suppress adenomatous colon polyps, the maker of rofecoxib, Merck, withdrew the drug from the market worldwide.3 At the time of withdrawal, >100 million prescriptions for rofecoxib had been written and >20 million patients had received this drug in the USA alone. Shortly thereafter, one of two trials studying celecoxib for prevention of colorectal adenomas also showed excess cardiovascular thrombotic events, but only at the supratherapeutic dosage of 800 mg daily.4 After the withdrawal of rofecoxib and the concerns about celecoxib, the public reaction was extraordinary and included multiple Congressional investigations, liability lawsuits, and media stories.
But how much do we really know about NSAIDs, COX inhibitors, and cardiovascular risk? There remain several large gaps in medical knowledge that prevent rational decision-making in choosing anti-inflammatory analgesics for patients with arthritis. Ten years into this controversy, we simply do not know very much. In fact, there exist no reliable data about the cardiovascular safety of conventional NSAIDs, some of which are more active inhibitors of COX-2 than COX-1 (e.g. meloxicam).5 Although these drugs are widely available without a prescription in most countries, they have never been systematically studied to determine their cardiovascular safety. There are good reasons to suspect that all NSAIDs are capable of inducing adverse cardiovascular events. All of these drugs increase blood pressure, impair salt excretion, and may induce renal insufficiency or heart failure.
To attempt to answer the questions about the cardiovascular safety of NSAIDs and COX-2 inhibitors, many physician-scientists have undertaken research efforts. Innumerable observational studies examining larger and larger administrative databases have sought to answer these critical questions.6 However, like all observational studies using administrative databases, such studies are critically flawed. Adjustment for known and unknown confounders in observational studies is never adequate, and we are left with results that can only be assessed as ‘hypothesis-generating’. There have been few, if any, randomized controlled trials (RCTs), and none has been adequate in size to provide robust statistical power to compare the cardiovascular effects of this class of drugs. The VIGOR trial adjudicated only 22 myocardial infarctions (18 in the rofecoxib treatment group and four in the naproxen group). The APC trial that showed excess events with celecoxib reported only 21 myocardial infarctions, three in the placebo group and nine in each of two celecoxib dosage groups.4 The APPROVe Trial that resulted in the withdrawal of rofecoxib had only 35 myocardial infarctions, 23 in the rofecoxib group and 12 in the placebo group.7
A decade after this controversy emerged, we still have lots of passionate arguments (heat), but few high quality RCTs (light). In this context, the Mini-COREA trial should be complimented and appreciated.8 Kang et al. have undertaken a systematic effort to study the sole remaining COX-2 inhibitor currently available in the USA in patients undergoing coronary intervention, and the results of the study are welcomed. The primary endpoint, lumen loss after stenting, was slightly reduced by 400 mg of celecoxib daily for 3 months, a difference of only 0.09 mm with a marginally significant P-value of 0.02. In parallel, there was a weak trend toward less clinically driven target lesion revascularization (TLR), 3.2% vs. 5.7%, P = 0.09. However, the composite endpoint of myocardial infarction and cardiovascular death (not the primary endpoint) showed a worse outcome with celecoxib, 1.6% vs. 0.2%, P = 0.03.
How should we interpret these findings? Unfortunately, although intriguing, the results of Mini-COREA must be viewed as hypothesis-generating at best. First, the primary endpoint was a surrogate measure, lumen loss, not the more clinically important outcome of adverse cardiovascular events. The differences between groups (0.09 mm) was statistically significant, but probably not biologically significant. In any RCT, secondary endpoints should always be viewed as hypothesis-generating and potentially unreliable. For these secondary outcome measures, we have mixed results. TLR trended toward fewer events with celecoxib, but myocardial infarction and cardiovascular death showed an excess in the celecoxib treatment group (1.6% vs. 0.2%), a finding of marginal statistical significance (P = 0.03). While this may seem like a significant hazard ratio, the number of events was very small, with only five myocardial infarctions and three deaths in both treatment groups combined. In fact, a single additional myocardial infarction or death in the control group would render this result non-significant.
There are several other concerns that make interpretation of Mini-COREA difficult. The authors measured multiple secondary endpoints including minimum lumen diameter (MLD) in the stented segment, MLD in the stent, percentage diameter stenosis in the segment, percentage diameter stenosis in the stent, and binary restenosis. In none of these secondary endpoints was there a statistically significant difference between treatment groups. Accordingly, the finding that celecoxib reduces lumen loss must be viewed as fragile and potentially represents the play of chance, rather than a real effect. Similarly, the ‘excess’ in adverse cardiovascular events represents a finding of questionable robustness. In fact, the totality of adverse cardiovascular events was completely neutral (table 4), with rates of 8.6% in the control group and 7.7% in the celecoxib group.
So, while the authors of the Mini-COREA study should be complimented for their efforts, we are left with the same dilemma that has existed for the last decade. Are COX-2 inhibitors or conventional NSAIDs safe for patients at high cardiovascular risk? Are there important differences between agents that inhibit COX-2 selectively and drugs that inhibit both the COX-2 and COX-1 isoforms? The answers to these questions have been elusive. Only a large-scale RCT comparing NSAIDs and COX-2 inhibitors can accurately define the risks and benefits of these agents. Avoidance of these drugs in individuals at high cardiovascular risk is not an option for most patients. Arthritic disorders are painful and significantly compromise quality of life. NSAIDs and COX inhibitors relieve the pain and inflammation of arthritic disorders and will probably remain the mainstay of treatment for many years to come. Accordingly, it is essential for the makers of these drugs and the medical community to determine their safety through high quality medical evidence. We will not answer these questions definitively through observational studies or short-term, small RCTs.
Fortunately, an effort to resolve the critical question of NSAID safety is underway, but it has taken many years to accumulate sufficient cardiovascular events. The PRECISION Trial (Prospective Randomized Evalaution of Celecoxib Integrated Safety vs. Ibuprofen or Naproxen) was launched in 2006.9 I serve as the study chairman. The trial will require ∼20 000 patients to have adequate statistical power to determine whether celecoxib is non-inferior to ibuprofen or naproxen. The study is well along, but has recruited slowly, primarily because European authorities have regrettably not permitted the inclusion of patients with pre-existing cardiovascular disease because these regulatory authorities believe that this scientific question is already answered. We vehemently disagree. When PRECISION is concluded, we will have carefully adjudicated cardiovascular outcome data in a large population at high cardiovascular risk and will have adjudicated >500 major adverse cardiovascular events. When completed, the PRECISION trial will hopefully replace heated debate with sunlight.
The PRECISION trial is funded by Pfizer.
Conflict of interest: S.E.N. is the Study Chairman for the Precision Trial, an ongoing cardiovascular outcome trial comparing celecoxib, ibuprofen, and naproxen..