NSAIDs: Side Effects and James J Hale FY1 Department of Anaesthetics
Introduction The non-steroidal anti-inflammatory drugs (NSAIDs) are a diverse group of drugs that have analgesic, antipyretic and anti-inflammatory actions. They are used for both short and long term pain relief, for example post-operatively, in traumatic injuries and for osteo- and rheumatoid arthritis. NSAIDs are widely prescribed by medical practioners and some are also available over-the-counter. Despite widespread use, NSAIDs may not be as safe as some practioners believe and are associated with an increased mortality in many circumstances. It is important that practioners are aware of the risks when prescribing NSAIDs. The COXIBs are a newer class of NSAID that have a different target profile to regular NSAIDs. 1. Mechanism of Action NSAIDs primary mode of action lies with their inhibition of the enzyme cyclooxygenase (COX). COX produces prostaglandins from arachadonic acid. The prostaglandins are a diverse group of molecules that have a number of functions in different areas of the body. There are two variants of COX: COX-1 is a constitutive enzyme which carries out a housekeeping role in many organs around the body and COX-2 is an inducible enzyme that has a role in inflammation. The NSAIDs vary in their degree of inhibition of COX-1 and COX-2. General perception is that the COXIBs inhibit COX-2 whilst NSAIDs are unselective inhibiting both COX enzymes. While this simplified view is commonly held, the reality is more complex. For example, not only do the COXIBs also inhibit COX-1 but different NSAIDs have different selectivities for the two COX enzymes. The diagram below demonstrates this relationship for a number of common NSAIDs and COXIBs. Figure 1 This graph uses data from Warner et al. (1). It logs the ratio of drug concentration required to achieve 80% inhibition of either COX-1 or COX-2. Positive values indicated greater COX-1 selectivity and vica-versa. Aspirin has also been plotted for interest. The important points to note are that, as well as the COXIBs, diclofenac and meclofenamate (similar to mefamenic acid) are also more COX-2 selective. Ketoprofen and indomethacin are the most COX-1 selective of the common NSAIDs. Ibuprofen and naproxen are COX-1 selective but are the most balanced of all the NSAIDs.
2. Gastrointestinal Side Effects COX-1 is constitutively expressed in the gastric mucosa where it has a role in gastric protection. The prostaglandins produced inhibit acid secretion and promote bicarbonate and mucus secretion. NSAIDs inhibit these protective mechanisms therefore promoting gastric erosions and ulceration. Observation and extrapolation suggests that UK wide, the NSAIDs are attributable for 12,000 emergency admissions a year for upper gastrointestinal bleeding, leading to over 2,200 deaths (2). This places upper gastrointestinal bleeding as the leading cause of NSAID related mortality. This is estimated to cost the UK around 251m per annum (3). All NSAIDs are associated with these adverse events; however some are worse than others. A meta-analysis in 1996 (4) demonstrated the following hierarchy: Ibuprofen < Diclofenac < Naproxen < Indomethacin < Ketoprofen < Azapropazone Ibuprofen being the safest and azapropazone the least safe. Azapropazone has now been withdrawn due to the extremely high rate of GI complications. Generally, GI toxicity increases with COX-1 selectivity and this is consistent with the mechanism relying on COX-1 inhibition. However, it is inconsistent that diclofenac, which is the most COX-2 selective of the above drugs, is not associated with the least GI side effects. The authors of the papers noted that the apparent low GI toxicity of ibuprofen may well be associated with the way it is used: notably in low doses. A number of the papers included in the meta-analysis involved the low dose use of ibuprofen. Accordingly, ibuprofen may appear above diclofenac erroneously. Despite this inconsistency, ibuprofen, especially low dose use, and diclofenac are associated with the least GI toxicity. COXIBs were developed on the premise that the GI side effects were mediated through the COX-1 enzyme. Therefore, it follows that a highly selective COX-2 inhibitor may induce fewer GI side effects. This theory was correct and the COXIBS are associated with notably reduced GI side effects. Randomised controlled trials (RCTs), such as the CLASS study (5) suggest that the relative risk of gastric ulcer complications for COXIBS compared to standard NSAIDs, such as ibuprofen or diclofenac, is 0.52 (P > 95%). Other papers have further highlighted that COXIB used alone, still performs better than diclofenac when used in conjunction with standard gastric protection strategies. (6) Gastric protective strategies have been proven to reduce the rate of GI complications. Proton pump inhibitors (PPIs) and E. pylori eradication therapy have been the two most intensely studied strategies in recent years. Meta-analysis of a number of RCTs has demonstrated that, of the two options, a PPI is the better (7). In patients with active peptic ulcer disease, all NSAIDs and COXIBS should be avoided. In those that are high risk for peptic ulcer disease, including patients who are over 65yrs, with symptomatic or treated dyspepsia and previous peptic ulcer disease, the COXIBs are the preferred option. Class 1 Evidence (5, 6) Whenever a COXIB or NSAID is prescribed long term, adequate gastric protection should also be prescribed especially in those at higher risk. The best prophylaxis has been shown to be a standard dose PPI such as omeprazole 40mg OD. Class 1 Evidence (7)
3. Thrombotic Events As well as being expressed in the stomach, COX is also expressed in vascular endothelial cells. Here it has a role in producing prostaglandin I 2 (prostacycline). Prostacycline inhibits thrombus formation via inhibiting platelet activation and organisation. The inhibition of this endothelial COX by NSAIDs is thought to increase the tendency to form thrombi and hence increases the risk of myocardial infarction and stroke, aspirin notwithstanding. An increased risk of thrombotic events was first noticed with COXIBs during post-marketing research. This increased risk was associated with rofecoxib, one of the first COXIBs, and this was published in the NEJM (8). As a result, rofecoxib was withdrawn from the market and it was assumed that this would be a class effect associated with all the COXIBs. Accordingly, the COXIBs received a degree of adverse publicity, but was this justified? A recent network meta-analysis published in the BMJ (9) looked at a number of RCTs regarding thrombotic risk. It combined 31 trials with over 115,000 patients to look at the risks of stroke, myocardial infarction and thrombosis related death. The following figure shows the relative risk associated with some common NSAIDs and COXIBs for these endpoints. Figure 2 Data from Trelle et al. (9) Relative risk of various NSAIDs/COXIBs versus placebo for myocardial infarction (yellow), stroke (red) and total thrombus related deaths (black). Despite the high number of follow-up years, most of the error bars overlap with a ratio of 1. Very few NSAIDs are associated with a statistically significant increased risk. Rofecoxib is statistically associated with more myocardial infarctions and diclofenac with more strokes and total thrombus related deaths. For all other NSAIDs, there is no statistically significant increased risk. Importantly, this analysis suggests that there is no class effect associated with the COXIBs. The risk of MIs and strokes varies greatly with individual drugs. The risks are compound specific not class specific. Despite the large error bars, Trelle et al. (9) still retain that naproxen is most likely associated with the lowest number of thrombosis related deaths. NSAIDs and COXIBs should be avoided where possible if there is an increased risk of thrombotic events. Naproxen is associated with the lowest risk and is the preferred choice if an NSAID is required. Firm Evidence still required. Not all COXIBs are associated with an increased risk of thrombotic events. Celecoxib is no worse than most of the commonly prescribed NSAIDs and should not be treated differently. Class 1 Evidence (9).
4. Cardiorenal Side Effects The inhibition of COX is also believed to directly affect kidney regulation. Prostaglandins are synthesised in response to low salt concentrations and have an effect of the renin system causing vasodilatation of the afferent arteriole. Inhibiting COX, leads to vasoconstriction and a reduced glomerular blood flow. This is thought to be the mechanism for the well recognised link between NSAIDs and renal failure. The reduced glomerular filtration rate also leads to salt and water retention which can contribute to hypertension and heart failure. There is a well established link between the NSAIDs and renal failure, both acute and chronic. Huerta et al. 2005 (10) carried out a case control trial which placed the relative risk of developing acute renal failure (ARF) at 3.2(CI 95%). This was a dose dependant effect with higher doses associated with an increased risk. As well as long term use, there was also an increased risk associated with short term use. Interestingly, the relative risk of developing ARF, if NSAIDs were used conjointly with diuretics, was 11.6 (CI 95%). The authors also noted an increased risk associated with a number of pre-morbid conditions including heart failure, diabetes and hypertension. Due to these adverse renal effects, the use of NSAIDs and COXIBs is contraindicated in all renal failure. It is also contraindicated in patients with an increased risk of renal failure, including those with hypertension, diabetes and heart failure. As well as renal failure, hypertension is a recognised side effect of NSAIDs. NICE suggests the following hierarchy based on a number of sources of data (11). Risk High Low Drug Etoricoxib, Ibuprofen Celecoxib, Diclofenac Figure 3 Data from Clinical Knowledge Summary (11) Stratified risk of hypertension associated with different NSAIDs and COXIBs. A significant association has been documented between NSAIDs and heart failure. Huerta et al. 2006 (12) completed a case-controlled trial of patients admitted to hospital with acute heart failure. They determined that the relative risk of a first admission with heart failure associated with NSAIDs was 1.3 (CI 95%). However, the relative risk if the patient already had pre-existing heart failure was 8.6 (CI 95%). Based on this evidence, it is advised that the use of NSAIDs is contraindicated in all patients with heart failure. Until recently, it was thought that the COXIBs were more detrimental to patients in heart failure than standard NSAIDs. This was based on data collected using rofecoxib and means that regular NSAIDs are the advised option in heart failure. However, recent data suggests that, as with thrombotic events, there is considerable variation between different drugs in the same class. Some COXIBs, such as celecoxib, are no worse in terms of heart failure than the regular NSAIDs (13). This is not reflected in current national guidelines. All NSAIDs and COXIBs should be avoided in patients with renal or heart failure. Class 2(a) Evidence (10, 12) Care should be taken with patients at increased risk of cardiorenal disease, including those with hypertension and diabetes as there is an increased risk of complications in these patients. Class 2(a) evidence (10) Patients on long term NSAID therapy should be monitored for cardiorenal complications (e.g. BP measurements and U&Es). 5. Interaction with Aspirin
Aspirin is a drug which has many mechanisms of action, some of which are still not fully understood. One of these actions is the inhibition of COX. However, aspirin is not associated with an increased thrombotic risk, infact there is a reduced risk. The effect is thought to be different to the NSAIDs as it has a short but irreversible action on COX. It inhibits COX in a number of locations but the effect lasts longer in platelets as they are unable to synthesise new COX enzymes. This means that, unlike NSAIDs which are associated with a prothrombotic risk, aspirin is associated with an anti-thrombotic cardioprotective effect. When aspirin is given with another NSAID this cardioprotection can be negated. Kurth et al (14) carried out a 5 year RCT and concluded that aspirin confers a 44% reduction in the risk of myocardial infarctions. At the same time, an observational study was carried out on the effect of NSAIDs on this cardioprotection. Short term use (<60 days per year) was not associated with a difference but long term use (>60 days per year) was. NSAIDs completely negated the cardioprotective effect of aspirin. Aspirin has a similar GI profile to NSAIDs increasing the risk of upper GI ulceration. When an NSAID is combined with aspirin the GI side effects are summative. Avoid NSAID use in patients who are taking aspirin, especially long term NSAID use which would completely negate the cardioprotective effect. Class 2(b) evidence (14) 6. Bleeding Risk In addition to being associated with thrombotic events, NSAIDs can also be associated with an increased risk of bleeding. This is due to the inhibition of COX in platelets. This inhibition decreases thromboxane production, which plays a role in platelet activation, organisation and arteriole vasoconstriction. One cohort study suggested that NSAIDs are associated with increased intra-operative and post-operative bleeding in the range of 157 to 208% (15). With aspirin, the increased risk of bleeding is acceptable due to the cardioprotective benefits. With NSAIDs, the benefits of analgesia may not out-weigh the bleeding risks. It takes roughly 3 days for platelet function to normalise after ceasing an NSAID. If NSAIDs are not required for pain relief they should be stopped 3 days prior to surgery. If high blood loss is expected, it is strongly advised that NSAIDs are stopped pre-op. 7. Summary NSAIDs are amongst the most commonly prescribed drugs, some of which are also available over the counter without prescription. This does not indicate they should be considered safe. NSAIDs are associated with a number of serious side effects and their use should be carefully considered. If an alternative is available, and not contraindicated, it should be used. There are many such alternatives including paracetamol, tramadol, colchicine, opiates and non-pharmacological methods. The COXIBs are a newer class of drugs that have received adverse press. Evidence now suggests this was unfounded and their side effect profile is no worse than the standard NSAIDs. 8. References
1. Warner et al. Nonsteroidal drug selectivities for cyclo-oxygenase-1 rather than cyclooxygenase-2 are associated with human gastrointestinal toxicity: A full in vitro analysis. Proc. Natl. Acad. Sci. USA. 1999 Vol. 96 p7563 2. Blower et al. Emergency admissions for upper gastrointestinal disease and their relation to NSAID use. Alimentary Pharmacology & Therapeutics 1997 Vol. 11(2) p283 291 3. Moore and Phillips. Cost of NSAID adverse effects to the UK National Health Service. Journal of Medical Economics 1999 Vol. 2 p45-55 4. Henry et al. Variability in risk of gastrointestinal complications with individual nonsteroidal anti-inflammatory drugs: results of a collaborative meta-analysis. BMJ 1996 Vol. 312 p7046 5. Silverstein et al. Gastrointestinal Toxicity with Celecoxib vs. Nonsteroidal Antiinflammatory Drugs for Osteoarthritis and Rheumatoid Arthritis. The CLASS Study. Journal of the American Medical Association 2000 Vol. 284(10) p1247 6. Francis et al. Celecoxib versus Diclofenac and Omeprazole in reducing the risk of recurrent Ulcer Bleeding in patients with Arthritis. NEJM 2002 Vol. 347(26) p2104 7. Vergara et al. Meta-analysis: role of Helicobacter pylori eradication in the prevention of peptic ulcer in NSAID users. Alimentary Pharmacology Therapeutics 2005 vol. 21 p1411 8. 8. Bombardier et al. Comparison of upper Gastrointestinal Toxicity of Rofecoxib and Naproxen in patients with Rheumatoid Arthritis. NEJM 2000 Vol. 343 p1520 9. Trelle et al. Cardiovascular safety of non-steroidal anti-inflammatory drugs: network meta-analysis. BMJ 2011 Vol. 342 p7806 10. Huerta et al. Nonsteroidal anti-inflammatory drugs and the risk of ARF in the general population. American Journal of Kidney Disease 2005 Vol. 45(3) p531 11. Clinical Knowledge Summary NSAIDs Prescribing Issues - http://www.cks.nhs.uk/nsaids_prescribing_issues 12. Huerta et al. Non-steroidal anti-inflammatory drugs and risk of first hospital admission for heart failure in the general population. Heart 2006 92 Vol. 92 p1610 13. Hudson et al. Differences in outcomes of patients with congestive heart failure prescribed celecoxib, rofecoxib, or non-steroidal anti-inflammatory drugs: population based study. BMJ 2005 Vol. 330 p 1370 14. Kurth et al. Inhibition of Clinical Benefits of Aspirin on First Myocardial Infarction by Nonsteroidal Anti-inflammatory Drugs. Circulation 2003 Vol. 108 p1191 15. Robinson et al. Nonsteroidal anti-inflammatory drugs, perioperative blood loss, and transfusion requirements in elective hip arthroplasty. The Journal of Arthroplasty 1993 Vol. 8(6) p607