June 07, 2017
8 min read

An evidence-based look at the CV safety of NSAIDs

You've successfully added to your alerts. You will receive an email when new content is published.

Click Here to Manage Email Alerts

We were unable to process your request. Please try again later. If you continue to have this issue please contact customerservice@slackinc.com.

There has been significant controversy in the past several years surrounding the safe use of nonsteroidal anti-inflammatory drugs in patients with or at risk for CVD. Data from the PRECISION trial, comparing the selective COX-2 inhibitor celecoxib vs. the nonselective nonsteroidal anti-inflammatory drugs ibuprofen and naproxen, were presented at the American Heart Association Scientific Sessions in November 2016 and simultaneously published in The New England Journal of Medicine. The question remains: Does the PRECISION trial deliver on its promise to clarify this controversy or does it muddy the waters even more?

Widely used treatment

NSAIDs are a key treatment option for the management of acute and chronic pain. It is estimated that more than 42.5 million Americans use NSAIDs each day, and more than 100 million prescriptions are written annually in the United States.

Carol Heunisch, PharmD, BCPS
Carol Heunisch

Traditional NSAIDs, such as naproxen, ibuprofen and diclofenac, exert their effects through nonselective inhibition of the COX enzymes — namely, COX-1 and COX-2. COX-1 is expressed constitutively in the majority of tissues, whereas COX-2 is expressed constitutively in only a few tissues. In the presence of inflammation, COX-2 expression is upregulated in a variety of cells and tissues, including vascular endothelium. Inhibition of COX-1 and COX-2 results in reduced prostaglandin production, and a subsequent reduction of inflammation, pain and swelling. The inhibition of COX-2 by traditional NSAIDs accounts for the anti-inflammatory effect of the drugs, whereas the inhibition of COX-1 can lead to NSAID toxicity and associated adverse effects such as gastrointestinal ulceration and bleeding, hypertension and acute kidney injury.

To overcome some of the adverse effects associated with the nonselective COX blockade, agents have been developed with specificity for inhibiting the COX-2 enzyme, reducing the potential for gastrointestinal issues. However, COX-2 inhibition creates an environment of imbalance between prostacyclin suppression and a prothrombotic risk that results from thromboxane A2-mediated vasoconstriction and platelet aggregation. This imbalance may increase the risk for CV events in patients with pre-existing CVD or those at high risk for CVD.

Concerns, labeling, guidance

Concerns related to the heightened risk for CV events with COX-2 inhibitors began to surface in the early 2000s. In the VIGOR trial, rofecoxib (Vioxx, Merck), at therapeutic and higher doses, was associated with an increased rate of MI in high-risk patients. The CLASS study, however, demonstrated no difference in the incidence of CV events between use of NSAIDs or celecoxib. Rofecoxib was voluntarily removed from the U.S. market in September 2004, based on an interim safety analysis of the APPROVe trial. That study showed an increased risk for MI and stroke in the rofecoxib group compared with placebo. Valdecoxib (Bextra, Pfizer) exited the market in 2005 at the request of the FDA, citing that the increased risk for CV events outweighed the benefit.

In 2005, the FDA required revised labeling for celecoxib (Celebrex, Pfizer) to include a boxed warning to advise of increased risk for CV thrombotic events, stroke and MI. In addition, there is a medication guide for patients that informs of increased risk for CV events, and advises against use in high-risk patients. Within the same public health advisory, the FDA required similar labeling changes for prescription nonselective NSAIDs. For over-the-counter products, such as ibuprofen and naproxen, information was added regarding potential for CV risk; however, the FDA concluded that available data did not suggest increased risk when these agents were used at low doses and for a short period. Meta-analyses have suggested less CV risk with the nonselective NSAID naproxen, but a dose-dependent risk for ibuprofen. Diclofenac, a prescription-only semi-selective NSAID with COX-2 selectivity similar to celecoxib, has also been associated with increased risk for CV events including MI and stroke, and all-cause mortality.

A 2007 AHA scientific statement recommended a stepped-care approach to management of patients with musculoskeletal issues and known CVD or those with CV risk factors with naproxen as the preferred agent if lower-risk agents such as acetaminophen or nonacetylated salicylates fail. However, the document stressed the importance of risk vs. benefit assessments at each step. For high-risk patients, use of celecoxib should be reserved for treatment failure of nonselective NSAIDs and deemed to have no other alternatives.

The 2014 American College of Cardiology/AHA non-ST-elevation ACS guidelines stressed the importance of a predischarge assessment of need for treatment of chronic musculoskeletal discomfort in the setting of non-ST-elevation ACS. As in the 2007 statement, the stepped-care approach for pain management was recommended, starting with acetaminophen, nonacetylated salicylates, tramadol or small doses of narcotics prior to consideration of NSAIDs. If this initial therapy was unsuccessful, the guidelines suggest naproxen as a reasonable NSAID choice. The use of agents with increasing COX-2 selectivity was recommended only if pain was intolerable, and the use of lowest effective doses for the shortest possible time was encouraged. As a class III recommendation, NSAIDs with increasing degrees of COX-2 selectivity were not recommended for patients with non-ST-elevation ACS who have failed the stepped-care approach.


PRECISION was a randomized, multicenter, double blind, parallel-group, noninferiority trial. Researchers enrolled patients with established CVD or increased CV risk who required treatment with NSAIDs for osteoarthritis or rheumatoid arthritis. The trial aimed to assess the noninferiority of celecoxib on the primary composite outcome of CV death, nonfatal MI or nonfatal stroke. Secondary outcomes included clinically significant gastrointestinal events and pain control. Renal events, iron deficiency anemia, and hospitalization for hypertension or HF were tertiary outcomes.

In total, 24,081 patients were enrolled at 926 centers in 13 countries from October 2006 to June 2014. Patients were randomly assigned in a 1:1:1 fashion to receive celecoxib 100 mg twice daily (n = 8,072), ibuprofen 600 mg three times daily (n = 8,040) or naproxen 375 mg twice daily (n = 7,969) with matching placebo. Patients with rheumatoid arthritis could have their drug doses increased at subsequent study visits as follows: celecoxib dose could be increased to 200 mg twice daily, ibuprofen to 800 mg three times daily or naproxen to 500 mg twice daily. In the osteoarthritis group, only ibuprofen or naproxen doses could be increased, but not celecoxib, based on current product labeling. All patients received esomeprazole for gastric protection, and patients who were taking daily doses of aspirin of 325 mg or less were permitted to continue therapy.

The primary composite outcome was the first occurrence of a major CV event that met Antiplatelet Trialists Collaboration (APTC) criteria (death from CV causes including hemorrhagic death, nonfatal MI or nonfatal stroke). A secondary composite outcome of major CV events included the components of the primary outcome plus coronary revascularization or hospitalization for unstable angina or transient ischemic attack. Requirements specified to prove noninferiority were 580 events in the intention-to-treat population and 420 events in the on-treatment group to reach 80% power with the HR not to exceed 1.12, with an upper 97.5% confidence limit of 1.33 or less in the intention-to-treat population and 1.4 or lower in the on-treatment population. The protocol was amended to these parameters based on lower observed event rates, higher discontinuation rates and slower-than-anticipated enrollment. The mean duration of follow-up was 34 months with mean treatment duration of 20 months.


Three-quarters of the patients enrolled were classified as being in a primary prevention CV risk category, and one-quarter in a secondary prevention risk category. More than three-quarters of patients had a history of hypertension, 35% diabetes and 62% dyslipidemia. Fifty-four percent of patients were taking statins and 46% were taking aspirin at baseline.

In both the intention-to-treat and the on-treatment groups, celecoxib was noninferior to naproxen and ibuprofen, and ibuprofen was noninferior to naproxen (see Table below). The researchers reported no interaction between the primary outcome and the patient CV risk subgroup, as no differences were seen between the NSAIDs studied in the primary and secondary CV risk patients. Similarly, there were no differences observed in the subgroup of patients taking or not taking aspirin.

When considering the composite secondary outcome of CV events with hospitalization or revascularization, as well as death from any cause, no significant differences were found between groups, with the exception of a higher rate of nonfatal MI in the ibuprofen group vs. the naproxen group (HR = 1.39; 95% CI, 1.01-1.91). Specific results in the subgroups of patients with a history of MI, stroke or peripheral artery disease were not provided. Gastrointestinal event rates were lower in the celecoxib group vs. the naproxen or ibuprofen groups. Serious renal events occurred at a significantly lower rate in the celecoxib group vs. the ibuprofen group (HR = 0.61; 95% CI 0.44-0.85), but was insignificant between the celecoxib and naproxen groups (HR = 0.79; 95% CI, 0.56-1.12). Changes in pain scores were statistically significant in favor of naproxen, but of small and questionable clinical significance.

The PRECISION investigators concluded that the trial provides evidence that celecoxib, at moderate doses, is not associated with greater CV risk compared with nonselective NSAIDs. The trial design was intended to provide insight into the safety of the drugs studied, since prespecified criteria to establish noninferiority needed to be met in both the intention-to-treat and on-treatment populations. This design accounts for differences between groups when discontinuation rates are high (68.8% of patients stopped taking study drug). Despite enrolling patients with significant CVD risk factors, the event rates in the study were low, the treatment duration relatively short (less than 2 years with 68% discontinuing treatment) and 27% of patients were lost to follow-up. Subgroup results for patients taking statins vs. not taking statins at baseline were not reported. In addition, the low doses of celecoxib and the nonselective NSAIDs used in the study may have provided safety advantages. Therefore, the results cannot be extrapolated to higher doses that may be used clinically. With respect to the effects of aspirin on NSAID safety, the trial was not designed to assess that, nor were patients randomly assigned to receive or not receive aspirin. Another important consideration with PRECISION compared with prior NSAID CV risk studies was the high percentage of patients using statins and aspirin for CV risk reduction.

Table 1. Adjudicated primary outcome in the PRECISION trial

Implications for practice

Sarah A. Spinler, PharmD, FCCP, FAHA, FASHP, AACC, BCPS (AQ Cardiology)
Sarah A. Spinler

Has the long-awaited PRECISION trial delivered and provided clinicians with clear guidance on the use of nonselective NSAIDs or COX-2 inhibitors in patients with established CVD or those at risk? In short, the study demonstrated that in a low-risk population — most of whom were receiving CV risk-reducing treatments — the short-term CV risks of the three drugs studied were similar. The answer to the question that has been open for years regarding which NSAID should be used and the safety over longer periods of time, especially in patients with established CVD, remains unanswered.

At this time, the guidance put forth by the 2007 AHA statement and reiterated in the 2014 non-ST-elevation ACS guidelines regarding a stepped-care approach to the patient with chronic musculoskeletal discomfort appears to be prudent. Utilization of acetaminophen, nonacetylated salicylates, tramadol or small doses of narcotics should be exhausted before considering NSAIDs. Although the guidelines suggest addition of naproxen in the setting of failure of the initial stepped-care approach, the similarity in event rates for ibuprofen and naproxen observed in the PRECISION trial does not seem to clearly define a “safe” nonselective NSAID for use in high-risk patients.

Disclosure: Heunisch reports no relevant financial disclosures.