Unexpected Results of PRECISION Clarify the Relative Safety of Celecoxib, Ibuprofen, and Naproxen

Presenters noted that this was an important but unusual trial, because it examined drugs not in isolation but how they are used in combination.

At the 2016 American Heart Association Scientific Sessions, Steven E. Nissen, MD, MACC, chief of cardiology at the Cleveland Clinic, presented result of the prospective randomized controlled trial evaluating the cardiovascular effects of classical nonsteroidal anti-inflammatory drugs (NSAIDs) versus the cycloxygenase-2 (COX-2) inhibitor celecoxib. The trial, known by the acronym PRECISION (Prospective Randomized Evaluation of Celecoxib Integrated Safety versus Ibuprofen or Naproxen), was sponsored by Pfizer, the makers of celecoxib. Eliminating any concerns of bias, Nissen did not receive financial compensation for his work on the trial. Honoraria were donated directly to charity.

NSAIDs, which are among the most commonly prescribed class of medications in the world, reduce pain and inflammation through inhibition of inflammatory mediators known as prostaglandins. Although inhibiting production of prostaglandins has desirable therapeutic effects in pain relief, prostaglandins also have important vascular effects. In large part due to concerns of cardiovascular adverse events related to some COX-2 inhibitors (eg, rofecoxib), celecoxib is the only remaining COX-2 inhibitor available on the United States market.

The PRECISION trial was designed to demonstrate noninferiority between celecoxib and the widely used non-selective NSAIDs naproxen and ibuprofen on cardiovascular end points, as well as rates of all-cause mortality, gastrointestinal adverse events, and renal adverse events.

Patients included in the PRECISION trial were required to have established cardiovascular disease or increased cardiovascular risk, and were required to take an NSAID for 6 months or more for symptomatic relief of osteoarthritis or rheumatoid arthritis. Patients were randomized to receive celecoxib 100 mg twice daily, ibuprofen 600 mg 3 times daily, or naproxen 375 mg twice daily. Patients in all 3 arms also received esomeprazole 20 mg to 40 mg daily. Over the course of the trial, patients were allowed to receive a higher NSAID dosage to manage unrelieved symptoms. For this event-driven trial, investigators prespecified a minimum duration of follow-up of 18 months.

In the primary analysis, investigators evaluated a major cardiovascular adverse event (MACE) end point identical to that used by the Antiplatelet Trialists Collaboration (APTC). Events included in the MACE primary end point included cardiovascular death (including hemorrhagic death), nonfatal myocardial infarction, or nonfatal stroke. This study was powered to evaluate noninferiority on the primary end point.

In addition to a primary analysis powered to detect noninferiority, investigators assessed superiority on an enriched MACE end point, including all included in the primary end point plus additional end points of revascularization and hospitalization for unstable angina or transient ischemic attack. Researchers conducted other superiority assessments for major gastrointestinal events, major renal events, and hospitalizations for hypertension or congestive heart failure.

A total of 31,857 patients were screened, and 24,081 were randomized to receive treatment at 926 centers around the world. Over a mean drug exposure period of 20.3 months, and a mean follow-up period of 34.1 months, patients received average mean daily doses of 104 mg twice daily (celecoxib), 681 mg 3 times daily (ibuprofen), and 426 mg twice daily (naproxen).

Patients in all 3 groups had similar characteristics, with an average age of 63 years, approximately 64% of study participants of female gender, approximately 90% of patients in each group receiving NSAIDs for osteoarthritis, and the remaining 10% of patients receiving NSAIDs for rheumatoid arthritis.

In their evaluation for noninferiority, researchers conducted an intention-to-treat (ITT) analysis, which has the benefit of maintaining the integrity of randomization, and conservatively represents the benefits of therapies in medical trials. However, an on-treatment analysis was also performed, as ITT analyses carries forward data from patients who are no longer on therapy, which could lead to underrepresentation of adverse events. Given the importance of adverse event detection in this trial, Nissen and colleagues performed analyses using both methods.

On the primary end point of cardiovascular death (including hemorrhagic death), nonfatal myocardial infarction, or nonfatal stroke, researchers confirmed the noninferiority of ibuprofen, naproxen, and celecoxib, both in the ITT analysis (P <.001 for noninferiority of celecoxib vs ibuprofen, P <.001 for celecoxib vs naproxen, P <.02 for ibuprofen vs naproxen), and in the on-treatment analysis (P <.001 for noninferiority of celecoxib vs ibuprofen, P <.001 for celecoxib vs naproxen, P <.025 for ibuprofen vs naproxen).

Regarding secondary analyses of secondary end points, Nissen cautioned, “these analyses should be viewed as hypothesis-generating, rather than conclusive, as they are not adjusted for multiplicity.”

In the secondary analysis of time to occurrence of MACE, in the intent-to-treat analysis, researchers identified a borderline significant trend toward lower rates of MACE events in patients receiving celecoxib versus patients receiving ibuprofen (HR: 0.87 [95% CI: 0.75-1.01], P = .06). In the on-treatment analysis, this effect reached significance (HR: 0.82 [95% CI: 0.59-0.97]). Importantly, a P value was not quoted for this, or any other on-treatment analysis, as these are considered sensitivity analyses.

Cardiovascular death rates over time were not statistically different in the ITT analysis, but an on-treatment analysis indicated lower rates of cardiovascular death in patients receiving celecoxib versus patients receiving ibuprofen (HR: 0.64 [95% CI: 0.42-0.99]). Similarly, an ITT analysis of all-cause mortality rates indicated a borderline significant trend toward lower rates of mortality with celecoxib versus naproxen (HR: 0.80 [95% CI: 0.63-1.00], P = .052). The on-treatment analyses indicated lower rates of all-cause mortality in patients receiving celecoxib versus patients receiving naproxen ibuprofen (HR: 0.68 [95% CI: 0.48-0.97]), or naproxen (HR: 0.65 [95% CI: 0.46-0.92]).

Secondary analyses of major gastrointestinal events over time favored celecoxib over either comparator in both ITT (celecoxib vs ibuprofen, HR: 0.65 [95% CI: 0.50-0.85], P = .002; celecoxib vs naproxen, HR: 0.71 [95% CI: 0.54-0.93], P = .01) and on-treatment analyses (celecoxib vs ibuprofen, HR: 0.44 [95% CI: 0.32-0.61]; celecoxib vs naproxen, HR: 0.45 [95% CI: 0.33-0.63]). Similarly, secondary analyses of serious renal events over time favored celecoxib over ibuprofen in the ITT analysis (0.61 [95% CI: 0.44-0.85], P = .004), and favored celecoxib versus either comparator in the on-treatment analyses (celecoxib vs ibuprofen, HR: 0.54 [95% CI: 0.37-0.80]; celecoxib vs naproxen, HR: 0.66 [95% CI: 0.44-0.97]).

An additional post hoc analysis evaluating rates of any adjudicated cardiovascular, gastrointestinal, or renal event favored celecoxib over ibuprofen (HR: 0.78 [95% CI: 0.69-0.87], P <.001), and celecoxib over naproxen (HR: 0.87 [95% CI: 0.77-0.99], P = .03). These findings were consistent with the on-treatment analysis, which also favored the safety celecoxib over ibuprofen (HR: 0.69 [95% CI: 0.51-0.79]), and celecoxib over naproxen (HR: 0.78 [95% CI: 0.68-0.90]). Of note, in the ITT analysis, ibuprofen showed higher rates of adjudicated cardiovascular, gastrointestinal, or renal events than naproxen (HR: 1.13 [95% CI: 1.01-1.26], P = .04), although this effect was not replicated in the on-treatment analysis.

To contextualize the post hoc analysis, Nissen expressed findings as a number-needed-to-harm (NNH) statistic. The 28% higher rate of any adjudicated cardiovascular, gastrointestinal, or renal event observed with ibuprofen translates to a NNH of 59 for ibuprofen versus celecoxib, and a NNH of 117 for naproxen versus celecoxib. In other words, 1 additional cardiovascular, gastrointestinal, or renal adverse event would be expected for every 59 patients receiving ibuprofen rather than celecoxib and for every 114 patients receiving naproxen rather than celecoxib at the doses compared in this trial.

An additional analysis addressed the issue of a potential interaction between aspirin and NSAIDs. Because NSAIDs may interfere with aspirin-induced inactivation of the COX-1 enzyme, which may interfere with the cardiovascular benefits of low-dose aspirin regimens, Nissen and colleagues also analyzed subgroups of patients who took low-dose aspirin or did not take low-dose aspirin. No significant differences between rates of the primary end point were identified.

Several adverse event rates were significantly higher with ibuprofen than with celecoxib, including anemia, increased blood pressure, hypertension, increased creatinine, constipation, diarrhea, and dyspnea. Rates of several adverse events were significantly more likely to occur with naproxen than with celecoxib. However, unexpectedly, rates of adverse events with ibuprofen were also significantly higher than those observed with naproxen.

Results of the PRECISION trial challenge the perception that naproxen is a safer-to-use alternative to other NSAIDs. It is important to recognize, however, that the dosage of celecoxib used in this study was 100 mg twice daily—substantially lower than the 800-mg daily supratherapeutic doses that were previously associated with safety signals. Even so, the doses used in PRECISION were consistent with the maximum dose restrictions on celecoxib present in US labeling for the medication.

Elliott Antman, MD, PRECISION trial discussant, noted the importance of this study, stating “We don’t see a lot of trials like this, but we need this information. This is extremely important — this is not a comparison of drugs. This is a comparison of drug regimens.” By striving for a study designed to reflect real-world practice, and by allowing dose adjustment, the PRECISION investigators have provided much-needed clarity that may help clinicians minimizing the risks and maximize the benefits of common NSAID regimens.

Results were simultaneously published in the New England Journal of Medicine.


Nissen SE, Yeomans ND, Solomon DH et al. Cardiovascular safety of celecoxib, naproxen, or ibuprofen for arthritis [published online November 13, 2016}. N Engl J Med. DOI: 10.1056/NEJMoa1611593.