Risks of Cardiovascular Disease and Beyond in Prescription of Nonsteroidal Anti-Inflammatory Drugs

Abstract

Introduction:

Nonsteroidal anti-inflammatory drugs (NSAIDs) include aspirin, naproxen, diclofenac, and ibuprofen, as well as selective cyclooxygenase 2 inhibitors such as celecoxib. Their use is common, as well as their side effects which cause 100 000 hospitalizations and 17 000 deaths annually. Recently, the US Food and Drug Administration strengthened its warning about the risks of cardiovascular disease (CVD) attributed to nonaspirin NSAIDs.

Methods:

When the sample size is large, randomization provides control of confounding not possible to achieve with any observational study. Further, observational studies and, especially, claims data have inherent confounding by indication larger than the small to moderate effects being sought.

Results:

While trials are necessary, they must be of sufficient size and duration and achieve high compliance and follow-up. Until then, clinicians should remain uncertain about benefits and risks of these drugs. Conclusions: Since the totality of evidence remains incomplete, health-care providers should consider all these aforementioned benefits and risks, both CVD and beyond, in deciding whether and, if so, which, NSAID to prescribe. The factors in the decision of whether and, if so, which NSAID to prescribe for relief of pain from inflammatory arthritis should not be limited to risks of CVD or gastrointestinal side effects but should also include potential benefits including improvements in overall quality of life resulting from decreases in pain or impairment from musculoskeletal pain syndromes. The judicious individual clinical decision-making about the prescription of NSAIDs to relieve pain based on all these considerations has the potential to do much more good than harm.

Nonsteroidal anti-inflammatory drugs (NSAIDs) are among the most commonly used drugs to treat pain among adults in the United States. Among over 29 million users, ¹ each year in the United States, 100 000 hospitalizations and 17 000 deaths are attributed annually to NSAIDs. Recently, the US Food and Drug Administration strengthened its warning about the risks of nonaspirin NSAIDs on myocardial infarction (MI) and stroke. ^2,3

Clinicians prescribing NSAIDs should be aware of the complexities of the decision-making. Specifically, clinical decision-making should include but not limited to protective and deleterious effects on cardiovascular disease (CVD). For NSAIDs, the non-CVD risks are primarily gastrointestinal (GI), but also renal, which should be considered by the clinician on an individual basis in deciding whether and, if so, which to prescribe for analgesia. In this commentary, we address the CVD as well as GI and renal issues of NSAIDs used for pain relief to aid health-care providers in their complex clinical decision-making. Nonsteroidal anti-inflammatory drugs are comprised of 2 subgroups including traditional NSAIDs (unsaid) such as ibuprofen, naproxen, and diclofenac, as well as selective cyclooxygenase 2 inhibitors (COXIBs), such as celecoxib. Traditional NSAIDs reversibly and nonspecifically inhibit both isoforms of the enzyme cyclooxygenase (COX-1 and COX-2), thereby blocking prostaglandin release which in turn mitigates downstream inflammation and pain, which in turn suggests a plausible explanation for their observed increased risks of CVD. ^2,3

Cyclooxygenase 2 inhibitors were developed primarily because of their favorable GI side effect profile relative to traditional NSAIDs. The Cyclooxygenase 2 and Non-Steroidal Anti-Inflammatory Drug Trialists (CNT) collaboration conducted a comprehensive meta-analysis exploring the risks of COXIBs and NSAIDs on CVD. The main outcomes were major vascular events, major coronary events, stroke, upper GI complications, and mortality. Both COX-2 inhibitors and diclofenac conferred significantly increased risks of major vascular events. Ibuprofen demonstrated a possible increased risk of major vascular events, which did not achieve statistical significance. Naproxen did not significantly increase either major vascular or major coronary events. ⁴

The Prospective Randomized Evaluation of Celecoxib Integrated Safety versus Ibuprofen or Naproxen (PRECISION) trial directly compared celecoxib, ibuprofen, and naproxen with respect to CVD. Patients with rheumatoid or osteoarthritis were assigned at random to celecoxib, ibuprofen, or naproxen. The primary outcome was the noninferiority of celecoxib on the primary prespecified combined outcome of (1) nonfatal MI, (2) nonfatal stroke, or (3) cardiovascular (CV) death. The authors concluded that celecoxib was noninferior to naproxen and ibuprofen on risks of CVD. With respect to GI side effects, as expected, celecoxib conferred significantly lower risks, which included gastroduodenal hemorrhage, gastric outlet obstruction, small or large bowel perforation or hemorrhage, and symptomatic gastric or duodenal ulcer. In PRECISION, 68.8% of the participants stopped taking their assigned study drug and 27.4% were lost to follow-up. Further, regulatory dosing restrictions precluded dose escalation for celecoxib in patients with osteoarthritis, which constituted 90% of the randomized participants. ⁵

Both the CNT meta-analyses and the PRECISION trial have methodologic limitations that add to the complexities of the individual clinical decision-making of clinicians in choosing the most appropriate analgesic for their patients. In the CNT meta-analyses, the randomized trials included were not specifically designed a priori to test the CVD hypothesis. Such meta-analyses are usually far less reliable than data from a single large-scale, randomized trial designed a priori to do so. In PRECISION trial, however, the low adherence and follow-up rate are likely to have resulted in uncontrollable bias and confounding. Finally, average duration of treatment and follow-up was 8 months in the CNT meta-analysis and 20 months in PRECISION. Despite these limitations of both the CNT meta-analyses and PRECISION, their findings are generally consistent with the notable exception of naproxen. In the indirect comparisons, in the CNT meta-analysis, naproxen showed no increase in CVD, and the findings in PRECISION were compatible with a possible small but nonsignificant increased risk.

Based on their concerns about the risks of CVD and GI side effects of NSAIDs, the American College of Rheumatology ⁶ recommends the use of acetaminophen for pain relief in patients with mild to moderate osteoarthritis. It is important to note that acetaminophen has no clinically relevant anti-inflammatory properties. Further, acetaminophen is associated with hepatotoxicity. In a small randomized trial, acetaminophen at recommended doses was associated with frequent elevations of hepatic transaminases. ⁷ In addition, acetaminophen accounts for over 50% of overdose-related liver failure and about 20% of liver transplant cases. ⁸

Aspirin has beneficial effects on CVD that must be weighed against its GI side effects that are mainly dyspepsia and bleeding. These GI side effects increase with higher doses, which add no benefit on CVD. ⁹ In addition, aspirin has no material hepatotoxicity or renal toxicity. Finally, aspirin also has the additional benefit of clinically relevant anti-inflammatory properties as measured by high-sensitivity C-reactive protein, an accurate marker and sensitive predictor of CVD. ¹⁰

The benefits of aspirin in secondary prevention patients at high risk of a subsequent event have been conclusively demonstrated in both individual trials and their meta-analysis. Aspirin confers significant reductions of MI, stroke, and CV death. ¹¹ Further, in secondary prevention patients, the absolute benefits of aspirin far exceed the absolute risks of major extracranial bleeding and the very rare but catastrophic risk of hemorrhagic stroke. In primary prevention, aspirin confers significant and conclusive benefits on first MI, but the data on CVD on end points of death and stroke are inconclusive. Of far greater relevance in decision-making for clinicians, in primary prevention, the absolute risk reductions are significantly lower, with the average absolute risk just below 5%, whereas the absolute risks are about the same as in secondary prevention. A 10% relative risk reduction in a patient with a 10-year risk of 5% in a risk reduction of approximately 0.5%. If we assume that the risk of a bleeding event is the same as described for secondary prevention, the risk to benefit ratio is favorable only when the risk of a CV event begins to exceed 6% to 10% over a 10-year period. Not all patients share an equal risk of bleeding. Those with a prior history of GI bleed, history of ulcer disease, gastritis, and those on chronic NSAID therapy bear a disproportionate risk. ^12,13 It should also be noted that in primary prevention, patients are at increased risks of CVD, such as the 40% of the US population with metabolic syndrome. ¹⁴ Specifically, these overweight and obese primary prevention patients have risks of a first MI that are similar to the secondary prevention patients who have already sustained a prior occlusive vascular event. These higher risk primary prevention patients are likely to experience an absolute benefit of aspirin that exceeds the absolute risk. In primary prevention, any decision to prescribe aspirin should be an individual clinical judgment that carefully weighs the absolute benefits in reducing the risk of an occlusive vascular events against the overall risk of major extracranial bleeding, which is most commonly GI. ^{12 –14}

Aspirin is widely known to increase GI side effects, notably dyspepsia and GI bleeding. In addition, however, health-care providers should be aware that there are antisecretory medications which have been demonstrated to be effective in reducing the GI complications of aspirin. Two commonly used medications with gastroprotective benefit are H2 receptor antagonists (H2RA) and proton pump inhibitors (PPIs). The Famotidine for the Prevention of Peptic Ulcers in Users of Low-dose Aspirin (FAMOUS) trial demonstrated a risk reduction of peptic ulcers in patients using low-dose aspirin. ¹⁵ In contrast to H2RAs, PPIs dramatically decrease gastric acid production. A meta-analysis of randomized trials was conducted to test the relative benefits of H2RAs and PPIs in reducing GI complications in patients using aspirin. In this meta-analysis, H2RAs produced a statistically significant 2.5-fold increase in GI bleeding when compared to PPIs (relative risk = 2.55; 95% confidence interval: 1.19-5.49). ¹⁶

These data indicate that the addition of PPIs to NSAIDs decreases the risk of GI complications. In addition, however, there are adverse effects to using long-term PPIs.

Proton pump inhibitors also have relatively rare but significant adverse effects, but the evidence is neither cogent nor persuasive. The reported side effects include, in the short run, pneumonia, and in the longer term, dementia, MI, nutrient deficiencies, and kidney disease. In contrast, the quality and quantity of evidence for the use of long-term PPI in patients on NSAIDs to prevent upper GI bleeding are both cogent and persuasive. As a consequence, the American Gastroenterological Association recommends the use of PPI therapy in patients on long-term NSAIDs and at a high risk of GI adverse events to reduce the risk of upper GI bleeding. ¹⁷

Cogent and persuasive evidence was derived from the GI safety of celecoxib versus naproxen in patients with cardiothrombotic disease and arthritis after upper GI bleeding (CONCERN) trial. The CONCERN was a double-blind randomized trial conducted to test that the hypothesis of a COX-2 inhibitor plus a PPI is superior to a nonselective NSAID plus a PPI in the prevention of recurrent upper GI bleeding in concomitant users of aspirin who have high risks of CVD as well as upper GI bleeding. ¹⁸ Specifically, 514 patients were assigned at random to take either celecoxib 100 mg twice per day plus esomeprazole 20 mg once per day or naproxen 500 mg twice per day plus esomeprazole 20 mg once per day. All patients were already taking 80 mg aspirin and continued this medication throughout the trial. The duration of the trial was 18 months. The primary end point was upper GI bleeding, and secondary end points included CV events (nonfatal MI, nonfatal stroke, or death from a vascular cause). With respect to upper GI bleeding, 14 (5.6%) cases occurred in patients in the celecoxib group compared with 31 (12.3%) in the naproxen group. As regards serious CV events, 11 (4.4%) occurred in the celecoxib group and 14 (5.5%) in the naproxen group. ¹⁸ Although a combination of celecoxib, aspirin, and a PPI did not completely eliminate the risk of upper GI bleeding, this strategy seemed to offer a superior alternative compared to naproxen, aspirin, and a PPI for decreasing upper GI bleeding in patients with arthritis and at high risk of CV events and upper GI bleeding. Although the trial was not designed to test hypotheses on CV event, there was no apparent increase in the celecoxib group. The CNT meta-analysis suggested that celecoxib conferred about a 41% increased risk of CV events, whereas PRECISION noted noninferiority in CV effects.

With respect to renal adverse events, NSAIDs alter renal function due to their effects on renal prostaglandins. The use of NSAIDs has been associated with acute renal failure, sodium retention, peripheral edema, congestive heart failure, and hyperkalemia. ^19,20 Cyclooxygenase 1 plays a role in controlling renal hemodynamics and glomerular filtration rate (GFR), and COX-2 affects salt and water excretion. Inhibition of these enzymes may alter renal function in patients who are at an increased risk of renal syndromes. ^19,20 Prostaglandins dilate the renal vascular bed, lower renal vascular resistance, and increase renal perfusion. ¹⁹ Normally these enzymes play a minor role in mainlining renal perfusion; however, in one who has a compromised renal function, the prostaglandins play a major role in maintaining GFR. In clinical settings where renal perfusion depends on prostaglandins, NSAIDs can significantly decrease renal blood flow, resulting in acute renal failure. Patients who are on diuretics or have underlying chronic kidney disease, hepatic dysfunction, proteinuria, diabetes, hypertension, congestive heart failure, or increased age are at an increased risk of renal compromise due to NSAID use. ^19,20

With respect to the benefits and risks of NSAIDS, the totality of evidence remains incomplete. Based on the available evidence, health-care providers should consider all these aforementioned benefits and risks, both CVD and beyond, in deciding whether and, if so, which NSAID to prescribe. The factors in the decision of whether and, if so, which NSAID to prescribe for relief of pain from inflammatory arthritis should not be limited to risks of CVD or GI side effects but should also include potential benefits including improvements in overall quality of life resulting from decrease in pain or impairment from musculoskeletal pain syndromes. The judicious individual clinical decision-making about the prescription of NSAIDs to relieve pain based on all these considerations has the potential to do much more good than harm.