COX-2 Inhibitors: A CLASS Act or Just VIGORously Promoted

Abstract

Selective cyclo-oxygenase (COX)-2 inhibitors were developed with the hope of producing lesser gastrointestinal (GI) side effects as compared with the conventional nonsteroidal anti-inflammatory drugs (NSAIDs). Soon after their introduction into the market, the sales of celecoxib and rofecoxib went up considerably. Most of this was attributed to the results of the Celecoxib Long-term Arthritis Safety Study (CLASS) and Vioxx Gastrointestinal Outcome Research (VIGOR) trials. However, several discrepancies were noted in the presentation of the actual trial results submitted to the US Food and Drug Administration (FDA) and those used for the purpose of publication in scientific journals. These issues were discussed subsequently by the way of scientific communications. Moreover, with increasing use of these agents, evidence of their adverse effects is coming to light. The present review aims at discussing the above issues, with emphasis on the results of the CLASS and VIGOR trials.

Are the COX-2 inhibitors really superior in safety profile to the older NSAIDs?

Introduction

NSAIDs are among the most commonly used medications in the world.[1] They act by inhibiting COX, a key enzyme in arachidonic acid metabolism. The COX enzyme catalyzes the initial steps in the conversion of arachidonic acid to various eicosanoids, including prostaglandins (PGs) and thromboxanes. A major factor limiting their use is GI toxicity, ranging from mild dyspepsia to peptic ulcer to perforation and bleeding. This results from NSAID-induced disruption of the protective activities of PGE₂ and prostacyclin formed by COX in the gastric mucosa.

In 1990, Fu and colleagues[2] detected a novel COX protein in monocytes stimulated by interleukin, and a year later, Kujubu and colleagues[3] identified a gene with considerable homology to COX-1. Further research demonstrated that this novel COX-2 protein was an inducible enzyme with increased expression in inflammation. On the other hand, COX-1 was named a “housekeeping” enzyme because it was expressed constitutively, with relatively ubiquitous presence. It was also recognized as the main source of cytoprotective PGs in the gastric mucosa. Since the conventional NSAIDs inhibited both COX-1 and COX-2, it was postulated that the efficacy of NSAIDs (attributable to COX-2 inhibition) could be achieved without GI toxicity (due to COX-1 inhibition).

This realization rekindled the efforts of the pharmaceutical industry to produce a safe NSAID via selective inhibition of COX-2, and this class of agents (celecoxib and rofecoxib) was introduced in 1999.[4] By October 2000, celecoxib and rofecoxib had sales exceeding US$ 3 billion in the United States alone and a prescription volume in excess of 100 million for the 12-month period ending in July 2000.[5] Moreover, the sales of celecoxib alone increased from US$ 2623 million in 2000 to US$ 3114 million in 2001.[6] Most of the credit for this more than 80% increase in sales could be attributed to a widely distributed study — CLASS, published in JAMA in 2000.[7] The impact of the study can be gauged from the fact that about 30,000 reprints of CLASS were bought from the publisher, and it was cited more than 10 times as frequently as any other article published in the same issue.[8] No less influential was another trial, VIGOR, a double-blind trial conducted at 301 centers in 22 countries. Both of these trials concluded that COX-2 inhibitors were associated with significantly fewer adverse effects than the conventional NSAIDs.

Were these conclusions justified? Are the COX-2 inhibitors really superior in safety profile to the older NSAIDs? The current review summarizes the adverse effect profile of COX-2 inhibitors as more adverse drug reactions (ADRs) are being attributed to COX-2 inhibitors with their growing use.

Gastrointestinal Adverse Drug Reactions

NSAID-associated serious upper GI adverse events result in 103,000 hospitalizations and 16,500 deaths per year in the United States alone.[9,10] NSAID-induced GI adverse effects may be the commonest cause of drug-related events leading to emergency visits, 43% in an earlier study by us.[11] In this light, the COX-2 hypothesis, proposing that at comparable inhibitory doses, selective COX-2 inhibitors would be as effective as traditional NSAIDs and would spare the GI mucosa, seemed not only attractive but also plausible. The decade of the 1990s saw several in vitro and animal studies that seemed to prove this hypothesis being published, and this was the topic of several review articles as well.[12-14] Results of clinical trials[15-17] also supported the COX-2 hypothesis.[18]

However, the progress that we have made in science is because scientists tend to question everything, and not surprisingly, this hypothesis began to show cracks when in the late 1990s it was shown that within 40 minutes of oral challenge with acid, there was a marked upregulation of COX-2 in the rat stomach.[19] A subsequent study[20] demonstrated a crucial protective role for COX-2 in the so-called “adaptive cytoprotection” response of the stomach; that is, increase in resistance to injury observed following exposure to a mild irritant. A further functional role for COX-2 in mediating gastric epithelial proliferation was demonstrated by Sawaska and colleagues,[19] whereas Nakatsugi and colleagues[21] provided evidence that COX-2 contributes to mucosal defense in a rat model of stress ulcer. This realization led some researchers to postulate that inhibition of COX-2 may delay ulcer healing and cause exacerbation of inflammation.[22] More studies[23,24] showing worsening of gastric ulcers[23] or necrosis in the small intestine in the presence of altered gastric mucosa[23] and aggravation of ischaemia reperfusion injury[24] appeared in the literature.

As the debate[25] on the GI toxicity of COX-2 inhibitors intensified, in November 2001, JAMA published 2 letters[26,27] that drew attention to the fact that the conclusions given in CLASS differ from the complete information that was available to the FDA.[28]

The CLASS Trial

CLASS, funded by the manufacturers of celecoxib, Pharmacia, compared celecoxib 800 mg/day with ibuprofen 2400 mg/day and diclofenac 150 mg/day in osteoarthritis or rheumatoid arthritis. The 2 primary end points were the incidence rates for upper GI complications (bleeding, perforation, or obstruction) and symptomatic ulcers during the first 6 months of treatment. It was concluded that celecoxib was associated with a lower incidence of symptomatic ulcers and ulcer complications combined vs the traditional NSAIDs. CLASS was reported as a single trial whereas actually it was the combined analysis of 2 separate trials with protocols that differed markedly from each other in design, outcomes, duration of follow-up, and analysis; moreover, both the trials were of longer duration.[26-28] Two comparisons planned were celecoxib with ibuprofen and celecoxib with diclofenac, with the primary end point being ulcer-related complications (and not symptomatic ulcers as reported) and a follow up of 15 months (for complications) and 12 months (for ulcers) in both the trials. Analyzed accordingly, the number of ulcer-related complications were similar in both the groups, and almost all the ulcer complications that had occurred during the second half of the trials were in the celecoxib group.[29,30] Moreover, using a preplanned (by the FDA) definition of ulcer-related complications, a nonsignificant trend in favor of diclofenac was observed.[29,30] Even more disturbing is the fact that though these results seem to have been available at the time of submission of the manuscript, no mention was made of them in the paper.[7] And when the above-mentioned contradictions were pointed out,[25,26] the authors' replies[31] failed to refer to these results.

The VIGOR Trial[32]

VIGOR recruited rheumatoid arthritis patients to either 50 mg/day rofecoxib or 500 mg/day naproxen twice daily. The primary end point was confirmed clinical upper GI events (gastroduodenal perforation or obstruction), upper GI bleeding, and symptomatic gastroduodenal ulcers. After a median follow-up of 9 months, it was shown that rofecoxib was associated with significantly fewer clinically important upper GI events than naproxen. Like in the case of the CLASS trial, the FDA reviewed the VIGOR data and presented them on its Web site.[33] It was shown that when all serious adverse events are included (and not just GI events), naproxen-treated patients had fewer serious events as compared with rofecoxib (7.8% vs 9.3%, relative risk [RR] 0.81%, 95% confidence interval [CI] 0.62–0.97). The cumulative risk of developing serious cardiovascular thrombotic events (mainly myocardial infarction [MI]) was 1.7% in the rofecoxib group compared with 0.7% in the naproxen group. Moreover, there were significantly more withdrawals in the rofecoxib group vs the naproxen group, and they were due to hypertension, edema, hepatotoxicity, heart failure, or pathological laboratory findings. These safety data as well as their analyses were subject of the design and protocol of the VIGOR trial,[33] but were not provided in the publication[32]; only the GI data favoring rofecoxib were mentioned. The Editor's note to the publication[32] mentioned that “11 of the 13 principal authors have had financial associations with Merck, which sponsored the study…. The other two principal authors are…employees of Merck.”

The Investigator-Industry Alliance

The striking parallels of aggressive and perhaps misleading presentation of the VIGOR data and the troubling data intrigue in the CLASS trials provide insight into the large role of commercialization of the process of drug development and marketing. Quick, favorable results and rapid entry into the market seem to be a major driving force for pharmaceutical companies. This should not be surprising given that for each day's delay in gaining approval for marketing, the manufacturer may lose, on an average, US$ 1.3 million in revenue.[34] Unfortunately, examples such as these may not be uncommon. Stelfox and colleagues[35] found that authors whose work supported the safety of calcium channel antagonists had a higher frequency of financial relationships with the drugs' manufacturers than authors whose work did not support the safety of the medicines.

Other Forms of GI Injury

Several other types of GI injuries have lately been found to be associated with COX-2 inhibitors. A case report linking rofecoxib with hemorrhagic colitis was recently published.[36] We have also demonstrated that celecoxib aggravated experimental colitis induced by trinitrobenzensulfone in rats.[37] Two patients receiving rofecoxib were hospitalized with GI bleeds in another case report.[38] Cholestatic hepatitis, reversible upon discontinuation, has also been observed with celecoxib.[39] Gastrotoxic effects of COX-2 inhibitors in ischaemia-reperfusion injury to the rat stomach have also been observed.[24,37]

In conclusion, the earlier promise shown by the COX-2 inhibitors may only be partially fulfilled as it is increasingly being recognized that these drugs are not as GI-safe as believed. With time, more reports of GI ADRs are likely to emerge from postmarketing surveillance studies. Moreover, as recommended by Juni and colleagues,[8] industry-independent,[10] individual patient meta-analysis[12] of all long-term COX-2 inhibitor trials, including both published and unpublished data, may provide a reliable, unbiased answer.

Cardiovascular Adverse Effects

Thrombotic Cardiovascular Events

It was postulated even before their introduction in 1999 that COX-2 inhibitors may decrease vascular prostacyclin (PGI₂) production and may affect the balance between prothrombotic and antithrombotic eicosanoids.[40] Unlike the platelet inhibition afforded by COX-1 inhibitors, COX-2 inhibitors do not have this property. In contrast, by decreasing vasodilatory and antiaggregatory PGI₂ production, COX-2 antagonists may tip the balance in favor of prothrombotic eicosanoids (thromboxane A₂) and may lead to increased cardiovascular thrombotic events.[41]

It was therefore not unexpected when within less than 1 year of their marketing, 4 cases of ischemic complications in patients receiving COX-2 inhibitors were reported.[42] Moreover, as predicted, urinary levels of a metabolite of thromboxane A₂ were markedly elevated. However, the patients had connective tissue diseases, which predispose to thrombosis, and the generalizability of these results was problematic. One year later, a meta-analysis of 23 trials of COX-2 inhibitors failed to show an excess of cardiovascular thrombotic events, relative to conventional NSAIDs or placebo.[43]

Similarly, the VIGOR trial[32] did not show any difference in the rate of death from cardiovascular causes and the overall mortality rate, although it was reported that the incidence of MI was higher in the COX-2 inhibitor (rofecoxib) group as compared with the NSAID (naproxen) group (0.4% vs 0.1%). It was also reported that this increase in MI incidence was confined to a subgroup of patients who may have benefited from aspirin due to their cardiovascular condition, but were not allowed to take aspirin according to the study protocol. This was attributed to the cardioprotective effect of naproxen, suggesting that rofecoxib is without any cardiovascular risks in patients not needing aspirin.

There are several discrepancies in this form of reporting of the VIGOR data that may lead to misleading conclusions. For instance, during the trial, concern was raised by the higher mortality and the higher rate of serious cardiovascular events in the rofecoxib vs the naproxen group and additional safety analyses were planned. Serious thrombotic cardiovascular effects were found to be twice as high with rofecoxib as with naproxen (RR 2.38; 95% CI 0.39–4.00).[4]

Although these results were reported to the FDA[33] in June and October 2000, only the increased rate of MI was mentioned in the VIGOR publication[32] and not the overall increase in cardiovascular events. Moreover, the increased MI rate reported was in a subgroup of patients only, which was defined retrospectively. Also, FDA data show that even that statement was not true, as increased risk in serious thrombotic cardiovascular events was not only found in the subgroup of patients in whom aspirin was indicated (RR 4.89; 95% CI 1.41–16.88), but also in patients not needing aspirin (RR 1.89; 95% CI 1.03–3.45).

A cautionary flag about the risk of cardiovascular events with COX-2 inhibitors was thus raised[4] and a need for surveillance recognized.[44] More attempts to study the exact mechanism of why COX-2 inhibitors lead to cardiovascular events have recently been made. In an elegant study, Cheng and colleagues[45] found that mice without the PGI₂ receptors (which mimics the clinical effects of taking COX-2 inhibitors) have an enhanced vascular response to injury and showed increased TXA₂ formation and platelet activation. Moreover, the relative increase in thromboxane, coupled with elimination in prostacyclin, occurring in heart muscle, can lead to the development of thrombotic cardiovascular events.[46]

Hypertension

It is established that a 2-mm Hg reduction in diastolic blood pressure (BP) results in about a 40% reduction the rate of stroke and a 25% reduction in the rate of MI.[47] COX-2 inhibitors have been shown to increase BP,[48] and more patients in the VIGOR trial[33] developed hypertension in the rofecoxib group compared with the naproxen group. There was a 4.6-mm Hg increase in the mean systolic BP and a 1.7-mm Hg increase in the diastolic BP in the rofecoxib group compared with 1.0- and 0.1-mm Hg increases in systolic and diastolic BP, respectively, with naproxen.

Several factors may explain these BP raising effects: alterations in the renin angiotensin pathway, changes in sodium and water retention by the kidneys, inhibition of vasodilating PGs, and production of various vasoconstricting factors, including endothelin-1 and P450-mediated metabolites of arachidonic acid.[49] There may be some differences in the individual COX-2 inhibitors, and they may differ from the older NSAIDs in this regard. However, only head-to-head comparisons among these agents can provide a definitive answer. Until then, it seems prudent to be on the lookout for this ADR and report it to the concerned parties.

Other Cardiovascular Events

COX-2 inhibitors have also been reported to cause edema,[49] worsening of heart failure,[50] fatal allergic vasculitis,[51] and aggravation of doxorubicin-mediated cardiac injury.[52] Most of these possible ADRs may be due to inhibition of COX-2, though vasculitis is an allergic reaction due to the chemical structure.

Renal Adverse Effects

The constitutive COX-1 enzyme has been considered the physiologically important isoform for PG synthesis in the normal kidney. It has, therefore, been suggested that selective COX-2 inhibitors of the inducible isoform may be free from renal adverse effects. Recent evidence suggests that this may not be so.

Last year, Woywedt and colleagues[53] reported 2 cases of acute renal failure with COX-2 inhibitors and recommended that these agents be used with caution in renal transplant recipients and in patients with salt depletion. Decline in glomerular filtration rate was observed with a single dose of rofecoxib, which was similar to that with indomethacin, in a randomized trial in elderly patients receiving a low-salt diet.[54] These data suggest that COX-2 plays an important role in human renal function, and inhibition of COX-2 may have deleterious consequences. Consequently, the same precautions regarding renal risk that are followed for nonselective NSAIDs should be used when selective COX-2 inhibitors are administered.[55,56]

Other Adverse Effects Associated With COX-2 Inhibitors

COX-2 inhibitors have been recently shown to cause urticaria and angioedema,[57,58] hypersensitivity reactions,[59] and aggravation of seizures (in an animal model).[60] As clinical experience with COX-2 inhibitors becomes more widespread, reports of more ADRs are likely to emerge. It is not something new; it is well known that many times premarketing trials fail to detect some ADRs that are detected by postmarketing surveillance.

Celecoxib has subsequently been reported to have lesser risk of upper GI hemorrhage and better GI safety than nonselective NSAIDs.[61-63] In the observational study done by Mamdani and colleagues,[61] subjects older than 66 years of age who were taking either nonselective NSAIDs, diclofenac plus misoprostol, rofecoxib, or celecoxib were evaluated for short-term risk of upper GI hemorrhage. Relative to controls, an increased short-term risk of upper GI hemorrhage was observed in all groups except that of celecoxib. MacDonald and colleagues[62] have observed that newer NSAIDs were being prescribed to patients with more risk factors for GI hemorrhage. However, adjusting for risk factors also, it was shown that exposure to celecoxib and rofecoxib was associated with a significantly lower risk of GI hemorrhage.

In a meta-analysis done by Deeks and colleagues,[63] celecoxib was shown to have improved GI safety and tolerability compared with traditional NSAIDs. However, it was later pointed out that in the analysis of the results, instead of 15 months of data, the 6-month data from the CLASS trial were included.[64] By inclusion of the 15-month data, the reduction in GI adverse events with celecoxib as compared with conventional NSAIDs was still present but was appreciably less than suggested.[65]

Though the subsequent studies did show superior GI tolerability of celecoxib as compared with conventional NSAIDs, this article highlights the aggressive manner in which results of clinical trials can be used for marketing purposes. Termed “aspirin of the new millennium,”[66] COX-2 inhibitors need to continue to be evaluated critically.