Cardiovascular risks of cyclooxygenase-2 inhibitors and traditional anti-inflammatory drugs: necessary but not sufficient for clinical decision making

Abstract

Cyclooxygenase-2 inhibitors were initially developed, then received regulatory approval and were subsequently widely marketed to achieve effective pain relief in patients with inflammatory conditions while decreasing gastrointestinal complications. Gastrointestinal symptoms as well as signs had been a major concern with the use of traditional non-steroidal anti-inflammatory drugs. Individual clinical judgements about the prescription of cyclooxygenase-2 inhibitors and non-steroidal anti-inflammatory drugs for relief of pain from inflammatory arthritis should not be limited to risks of cardiovascular disease but should also consider gastrointestinal complications, symptoms as well as signs, and other benefits which include, but are not limited to improvements in quality of life resulting from decreases in pain or impairment from musculoskeletal pain syndromes.

Keywords::

• cardiovascular disease
• COXIBs
• gastrointestinal complications
• NSAIDs
• relief of pain or impairment

Cyclooxygenase-2 inhibitors (COXIBs) were initially developed, then received regulatory approval, and subsequently, were widely marketed. This new and novel class of drugs achieved effective pain relief in patients with inflammatory conditions while decreasing gastrointestinal complications. Gastrointestinal symptoms and signs that accompanied the widespread use of traditional NSAIDs had been a major concern. In addition, patients with inflammatory conditions tend to have higher risks of cardiovascular disease (CVD), and so they are more likely to require prophylactic aspirin, which also increases gastrointestinal side effects [1].

With respect to CVD, major clinical and public health concerns have been raised about the COXIBs. The initially reported high risks, derived primarily from claims and observational data as well as randomized trials not designed to test the hypothesis, led to major controversies about the true magnitude of the increased risks of CVD attributable to COXIBs [1]. In summary, basic research had provided plausible mechanisms that offered explanations about why COXIBs as well as NSAIDs might increase risks of CVD. Descriptive studies and claims databases, which are useful only to formulate but not test hypotheses, suggested large increased risks of CVD. In addition, some, but not all, observational analytical studies, both case–control and cohort studies, indicated that patients prescribed COXIBs and NSAIDs had moderate-to-large increased risks. Nonetheless, all such studies are subject to confounding by indication. Thus, for small-to-moderate risks and benefits, the amount of uncontrolled and uncontrollable data inherent in all observational studies, no matter how well designed, conducted and analyzed, is about as large as the effect sizes being sought, and so all such observations should be considered only hypothesis-formulating, not hypothesis-testing. Thus, in the absence of reliable data from randomized trials designed a priori to test the hypothesis, it was not possible to discern whether, and if so, how much, of the observed increases in risks of CVD attributable to COXIBs, were real [2]. Recent data have been published from a comprehensive, worldwide meta-analysis of individual patient data from all randomized trials, which include both CVD and gastrointestinal effects, and so this new information markedly influences the totality of available evidence [3]. The purpose of this editorial is to add to the totality of evidence the most recent randomized data in order to facilitate the most rational individual clinical decisions that can be safely made for individual patients as well as policy decisions for the health of the general public.

The early randomized trials were small in sample size and not designed a priori to test CVD hypotheses [3]. In addition, many were terminated early based on data-dependent emphasis on possible but unreliable evidence of increased risks [4]. These circumstances created considerable confusion for healthcare providers and patients.

At present, the most reliable evidence concerning risks of NSAIDs and COXIBs derives from the recently published comprehensive worldwide meta-analyses using individual participant data from 280 trials of NSAID versus placebo (124,513 participants, 68,342 person-years) and 474 trials of NSAID versus another NSAID (229,296 participants, 165,456 person-years) [5]. In these analyses, major vascular events were increased by about one-third by COXIBs (rate ratio [RR]: 1.37; 95% CI: 1.14–1.66; p = 0.0009). The findings were similar for diclofenac (RR: 1.41; 95% CI: 1.12–1.78; p = 0.0036). These observations were chiefly due to an increase in major coronary events (COXIBs: RR: 1.76; 95% CI: 1.31–2.37; p = 0.0001; diclofenac: RR: 1.70; 95% CI: 1.19–2.41; p = 0.0032). Ibuprofen also significantly increased major coronary events (RR: 2.22; 95% CI: 1.10–4.48; p = 0.0253), but a possible increase in major vascular events did not achieve statistical significance (RR: 1.44; 95% CI: 0.89–2.33). As compared with placebo, among 1000 patients allocated to a COXIB or diclofenac for a year, 3 experienced major vascular events, with 1 fatal. High-dose naproxen did not significantly increase major vascular events (RR: 0.93; 95% CI: 0.69–1.27). All the regimens studied have well-described increases in blood pressure, so it remains unexplained why there were no significant increases in risks of stroke. Vascular death was significantly increased by COXIBs (RR: 1.58; 99% CI: 1.00–2.49; p = 0.0103) or diclofenac (RR: 1.65; 99% CI: 0.95–2.85; p = 0.0187), but a possible increase by ibuprofen did not achieve statistical significance (RR: 1.90; 99% CI: 0.56–6.41) and naproxen was not associated with any apparent increase (RR: 1.08; 99% CI: 0.48–2.47). The proportional effects on major vascular events were independent of baseline characteristics, including vascular risk. Specifically, the proportional effects of COXIBs and NSAIDs appeared similar irrespective of baseline characteristics at all levels of risk of major vascular events (<5%, 5–10%, >10% over 5 years). For high-risk individuals (about 40% of whom were taking aspirin), however, for every 1000 patients allocated to a year of treatment with a COXIB or diclofenac regimen, about 7 or 8 would experience a major vascular event, of which, 2 would be fatal. High-dose ibuprofen may be associated with a similar risk, and, in addition, is also likely to yield a higher risk of upper gastrointestinal complications than either a COXIB or diclofenac.

In these analyses, the vascular risks of high-dose diclofenac are comparable with COXIBs, while high-dose naproxen is associated with less vascular risk than other NSAIDs. These observations are compatible with basic research concerning the reversible inhibition of platelets by NSAIDs, which could, at least in theory, result in aspirin-like benefits for naproxen whose time of action is about 12 h in contrast to about 4 h for ibuprofen. Nonetheless, the conclusion that naproxen is not associated with an increased risk of major vascular events must be interpreted with caution. First, it remains uncertain whether this would be true among patients treated with aspirin, in whom naproxen will not result in any additional inhibition of COX-1 and may actually interfere with the antiplatelet effect of low-dose aspirin. Second, the effects of lower naproxen doses, such as those typically used in over-the-counter preparations (e.g., 220 mg twice daily), are uncertain since they do not mimic the aspirin-like effect of 500 mg twice daily. Third, the apparent advantage of naproxen regimens may not be preserved after longer-term use, since basic research as well as clinical investigations has suggested that extended COX-2 inhibition may result in hazards through other mechanisms. Finally, non-naproxen NSAIDs appeared to have effects on CVD that were similar in magnitude to COXIBs [5].

In this meta-analysis, all COXIBs and NSAIDs were associated with an approximate doubling of the risk of hospital admissions for heart failure. In addition, all these drugs increase salt and fluid retention, leading to edema, in a dose-dependent manner. These findings are consistent with a COX-2-dependent hazard that is unrelated to variable platelet inhibition. It remains unclear from these analyses whether particular NSAIDs increase vascular risk within 6 months of starting treatment. Further, since the average trial duration was less than 1 year, these analyses do not provide much useful information about whether the risks of NSAIDs persist with prolonged treatment.

In the meta-analysis, all regimens also increased upper gastrointestinal complications by about two- to fourfold but COXIBs yielded the lowest risks. The relative risks and 95% CIs for each regimen were as follows: COXIBs 1.81, 1.17–2.81, diclofenac 1.89, 1.16–3.09, ibuprofen 3.97, 2.22–7.10 and naproxen 4.22, 2.71–6.56 [5]. Finally, it should be noted that all the drugs used in these randomized trials were at prescription-strength, not over-the-counter doses.

Although NSAIDs increase vascular and gastrointestinal risks, their magnitude can be predicted, which may help guide clinical decision making. The current totality of evidence suggests that, for the relief of pain from inflammatory arthritis, naproxen may have the best benefit-to risk-ratio on CVD, but is associated with a high risk of gastrointestinal bleeding. In contrast to naproxen, other traditional NSAIDs as well as COXIBs confer similar moderately increased risks of CVD, while COXIBs produce fewer gastrointestinal complications. Further, the use of proton pump inhibitors with traditional NSAIDs will decrease their gastrointestinal complications [6]. Finally, morbidity and mortality from gastrointestinal complications are likely to be lower than those resulting from occlusive vascular disease events. Given the high prevalence of pain due to arthritis and the benefit–to-risk ratios, based on the totality of evidence, it is reasonable to recommend that NSAIDs and COXIBs be considered to be prescribed more widely. The current underutilization may be due, at least in part, to recommendations that they should ‘be limited to patients for whom there are no appropriate alternatives’ [7]. These recommendations, however, were made before the availability of the current totality of evidence.

Thus, individual clinical judgments about the prescription of COXIBs and NSAIDs for relief of pain from inflammatory arthritis should not be limited to risks of CVD but should include gastrointestinal side effects as well as other benefits which include, but are not limited to improvements in quality of life resulting from decreases in pain or impairment from musculoskeletal pain syndromes.

Financial & competing interests disclosure

CH Hennekens has received investigator initiated research grant support from Bayer to the Charles E Schmidt College of Medicine at Florida Atlantic University; serves as an independent scientist either as Chair or member of the Data and Safety Monitoring Boards or as an advisor to: Actelion, Amgen, Anthera, AstraZeneca, Bayer, Bristol-Myers Squibb, British Heart Foundation, Canadian Institutes of Health Research, Children's Services Council of Palm Beach County, Food and Drug Administration, Legal counsels for GlaxoSmithKline and Stryker, Lilly, National Institutes of Health, Sunovion and UpToDate. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

No writing assistance was utilized in the production of this manuscript.