Patients on antithrombotics after an acute myocardial infarction (MI) will face a greater risk for bleeding and secondary cardiovascular (CV) events if they start taking any non-aspirin nonsteroidal anti-inflammatory drug (NSAID), confirms a large observational study.
Like other research before it, the new study suggests those risks will be much lower for some non-aspirin NSAIDs than others. But it may also challenge at least some conventional thinking about the safety of these drugs, and is based solely on a large cohort in South Korea, a group for which such NSAID data has been in short supply.
“It was intriguing that our study presented better safety profiles with celecoxib and meloxicam vs other subtypes of NSAIDs,” notes the report, published online July 27 in the Journal of the American College of Cardiology.
Most of the NSAIDs included in the analysis, “including naproxen, conferred a significantly higher risk for cardiovascular and bleeding events compared with celecoxib and meloxicam,” write the authors, led by Dong Oh Kang, MD, Korea University Guro Hospital and College of Medicine, Seoul, South Korea.
A main contribution of the study “is the thorough and comprehensive evaluation of the Korean population by use of the nationwide prescription claims database that reflects real-world clinical practice,” senior author Cheol Ung Choi, MD, PhD, of the same institutions, told theheart.org | Medscape Cardiology.
“Because we included the largest number of patients of any comparable clinical studies on NSAID treatment after MI thus far, our study may allow the generalizability of the adverse events of NSAIDs to all patients by constituting global evidence encompassing different population groups,” Choi said.
The analysis has limitations along with its strengths, the authors acknowledge, including its observational design and potential for confounding not addressed in statistical adjustments.
Observers of the study concur, but some cite evidence pointing to such confounding that is serious enough to question the entire study’s validity.
Among the cohort of more than 100,000 patients followed for an average of about 2.3 years after their MI, the adjusted risk of thromboembolic CV events went up almost 7 times for those who took any NSAID for at least 4 consecutive weeks compared with those who didn’t take NSAIDs, based on prescription records.
Their adjusted risk of bleeding events — which included gastrointestinal, intracranial, respiratory, or urinary tract bleeding or post-hemorrhagic anemia, the group writes — was increased fourfold.
There was wide variance in the adjusted hazard ratios (HR) for outcomes by type of NSAID. The risk of CV events climbed from a low of about 3 with meloxicam and almost 5 for celecoxib to more than 10 and 12 for naproxen and dexibuprofen, respectively.
The hazard ratios for bleeding ranged from about 3 for both meloxicam and celecoxib to more than 6 for naproxen.
Of note, celecoxib and meloxicam both preferentially target the cyclooxygenase type 2 (COX-2) pathway, and naproxen among NSAIDs once had a reputation for relative cardiac safety, although subsequent studies have challenged that notion.
“On the basis of the contemporary guidelines, NSAID treatment should be limited as much as possible after MI; however, our data suggest that celecoxib and meloxicam could be considered possible alternative choices in patients with MI when NSAID prescription is unavoidable,” the group writes.
They acknowledge some limitations of the analysis, including an observational design and the possibility of unidentified confounders; that mortality outcomes were not available from the National Health Insurance Service database used in the study; and that the 2009 to 2013 span for the data didn’t allow consideration of more contemporary antiplatelet agents and direct oral anticoagulants.
Also, NSAID use was based on prescriptions without regard to over-the-counter usage, they note. Although use of over-the-counter NSAIDs is common in Korea, “most MI patients in Korea are prescribed most medications, including NSAIDs, in the hospital. So I think that usage of over-the-counter NSAIDs did not change the results,” Choi told theheart.org | Medscape Cardiology.
“This study breaks new ground by demonstrating cardiovascular safety of meloxicam (and not only of celecoxib), probably because of its higher COX-2 selectivity,” write the authors of an accompanying editorial, Juan J. Badimon, PhD, and Carlos G. Santos-Gallego, MD, both of the Icahn School of Medicine at Mount Sinai, New York City.
Notably, “this paper rejects the cardiovascular safety of naproxen, which had been suggested classically and in the previous Danish data, but that was not evident in this study.” The finding is consistent with the PRECISION trial, in which both bleeding and CV risk were increased with naproxen vs other NSAIDs, observe Badimon and Santos-Gallego.
They agree with the authors in recommending that, “although NSAID treatment should be avoided in patients with MI, if the use of NSAIDs is inevitable due to comorbidities, the prescription of celecoxib and meloxicam could be considered as alternative options.”
But, “as no study is perfect, this article also presents some limitations,” the editorial agrees, citing some of the same issues noted by Kang and associates, along with potential confounding by indication and the lack of “clinical information to adjust (e.g., angiographic features, left ventricular function).”
“There’s undoubtedly residual confounding,” James M. Brophy, MD, PhD, a pharmacoepidemiologist at McGill University, Montreal, Canada, said for theheart.org | Medscape Cardiology.
The fivefold to 10-fold relative risks for CV events “are just too far in left field compared to everything else we know,” he said. “There has never been a class of drugs that have shown this sort of magnitude of effect for adverse events.”
Even in PRECISION with its more-than 24,000 high-coronary-risk patients randomized and followed for 5 years, Brophy observed, relative risks for the different NSAIDs varied by an order of magnitude of only 1 to 2.
“You should be interpreting things in the context of what is already known,” Brophy said. “The only conclusion I would draw is the paper is fatally flawed.”
The registry included 108,232 primarily male patients followed from their first diagnosed MI for CV and bleeding events. About 1.9% were prescribed at least one NSAID for 4 or more consecutive weeks during the follow-up period averaging 2.3 years, the group reports.
The most frequently prescribed NSAID was diclofenac, at about 72% of prescribed NSAIDs in the analysis for CV events and about 69% in the bleeding-event analysis.
Table 1. Hazard Ratio (HR) for CV Events for NSAIDs vs No NSAIDs Across Entire Cohort
HR (95% CI) *
* P < .001 for all. Adjusted for age, sex, “baseline medical comorbidities, and concomitant medications.”
Adding any NSAID to post-MI antithrombotic therapy led to an adjusted HR of 6.96 (P < .001) for CV events and 4.08 (P < .001) for bleeding events compared with no NSAID treatment.
Table 2. Hazard Ratio* (HR) for Bleeding Events for NSAIDs vs No NSAIDs Across Entire Cohort
HR (95% CI)
*Adjusted for age, sex, “baseline medical comorbidities, and concomitant medications.”
Further studies are needed to confirm the results “and ensure their generalizability to other populations,” Choi said for theheart.org | Medscape Cardiology. They should be validated especially using the claims data bases of countries near Korea, “such as Japan and Taiwan, to examine the reproducibility of the results in similar ethnic populations.”
That the study focused on a cohort in Korea is a strength, contend the authors as well as Badimon and Santos-Gallego — given, the editorial states — “that most data about NSAIDs were extracted from Western populations, but the risk of thrombosis/bleeding post-MI varies according to ethnicity.”
Brophy agreed, but doubts that ethnic differences are responsible for variation in relative risks between the current results and other studies. “There are pharmacogenomic differences between different ethnicities as to how they activate these drugs. But I suspect that sort of difference is really minor. Maybe it leads to a 2% or a 5% difference in risks.”
Kang and associates, Badimon, Santos-Gallego, and Brophy have disclosed no relevant financial relationships.