Adverse Effects of Acetaminophen and NSAIDs

GI Risk Factors

The greatest risk factor for NSAID-associated GI events is the presence of prior ulcer disease with ulcer complications (Chan, Graham, 2004). The AGS lists current peptic ulcer disease as an absolute contraindication and history of peptic ulcer disease as a relative contraindication to the use of NSAIDs in older adults (AGS, 2009). Other risk factors include age at or older than 60 years, CV disease and other co-morbidities, severe rheumatoid arthritis (RA), and concomitant treatment with corticosteroids or anticoagulants (Bhatt, Scheiman, Abraham, et al., 2008; Chan, Graham, 2004; Laine, 2001; Simon, 2007; Wilcox, Allison, Benzuly, et al., 2006). Helicobactor pylori (H. pylori) infection, excessive alcohol consumption, and cigarette smoking generally are considered independent and modifiable risk factors, though the magnitude of their effects is unclear (Burke, Smyth, Fitzgerald, 2006; Wilcox, Allison, Benzuly, et al., 2006). In 2009, the U.S. FDA required label changes for NSAIDs to reflect an increased risk of gastric bleeding, particularly in certain populations, including older adults and individuals taking anticoagulants, steroids, or other NSAID-containing products, or those consuming three or more alcoholic drinks daily (U.S. FDA, 2009). Table 6-1 provides a summary of risk factors for the development of NSAID-induced GI adverse events.

The First International Working Party on GI and CV Effects of NSAIDs and Anti-platelet Agents used a comprehensive series of clinical vignettes and possible scenarios to rate the appropriateness of NSAIDs. This group predefined high GI risk as age at or older than 70 years, prior upper GI event, and concomitant use of aspirin, corticosteroids, anticoagulants, or other antiplatelet drugs (Chan, Abraham, Scheiman, et al., 2008). In addition, risk was determined to be higher with specific NSAIDs and with the combination of an NSAID and low-dose aspirin (Laine, 2001; Simon, 2007; Silverstein, Faich, Goldstein, et al., 2000; Vardeny, Solomon, 2008). In a consensus guideline, the American College of Cardiology Foundation (ACCF), American College of Gastroenterology (ACG), and the American Heart Association (AHA) stated that the use of cardioprotective aspirin (81 mg) is associated with a 2- to 4-fold increase in upper GI adverse events (Bhatt, Scheiman, Abraham, et al., 2008) (see pp. 204-205 for more on cardioprotective aspirin). It is also important to note that any dose of aspirin can cause upper GI adverse events, and because the primary mechanism underlying GI toxicity is systemic rather than local, buffered and enteric-coated formulations do not decrease the incidence (Bhatt, Scheiman, Abraham, et al., 2008; Laine, 2001).

The risk of GI toxicity and death from a GI bleed increases with age at a rate of about 4% per year, and age is an extremely important consideration when starting NSAID therapy (APS, 2003; Wilcox, Allison, Benzuly, et al., 2006). At age 45 to 64 years, 15 in 10,000 individuals will have a serious GI bleed during long-term therapy, and 2 in 10,000 will die from a GI bleed; however, at age 65 to 74 years, 17 in 10,000 will have a serious GI bleed and 3 in 10,000 will die. The highest risk is in individuals age 75 and older; at this age, 91 in 10,000 will experience a GI bleed and 15 in 10,000 will die (AHRQ, 2007). This increased risk may be due to the combined effect of other risk factors associated with aging, such as medical comorbidities and concomitant use of drugs such as aspirin or anticoagulants, as well as an age-related decrease in protective GI prostaglandin concentrations (Wilcox, Allison, Benzuly, et al., 2006).

H. pylori is a gram-negative bacterium that can reside chronically in the stomach and elicit a local inflammatory response. NSAIDs produce a similar destructive effect on the gastric mucosa through COX-1 inhibition of prostaglandins in the protective lining of the gut. Although the data are conflicting, there is growing evidence of an interaction between the presence of H. pylori infection and risk of GI complications in NSAID users (Chan, To, Wu, et al., 2002; Huang, Sridhar, Hunt, 2002; Laine, 2001; Wilcox, Allison, Benzuly, et al., 2006). This interaction was suggested in a meta-analysis of 463 studies, which revealed that one-third of patients receiving long-term NSAID therapy had gastric or duodenal ulcers irrespective of the presence of H. pylori infection, but peptic ulcer disease was more common in H. pylori-infected NSAID users than in noninfected NSAID users (Huang, Sridhar, Hunt, 2002). A 2- to 4-fold increase in risk of upper GI complications in regular NSAID users who are H. pylori infected has been reported elsewhere (Chan, Graham, 2004). Eradication of H. pylori with antibiotic therapy decreases the incidence of peptic ulcers in patients who begin taking NSAIDs, but this does not seem to extend to patients with a previous history of ulceration (Wilcox, Allison, Benzuly, et al., 2006).

These data support the conclusion that H. pylori infection is an independent and modifiable risk factor for GI complications in long-term NSAID users (Chan, To, Wu, et al., 2002; Huang, Sridhar, Hunt, 2002; Kurata, Nogawa, 1997; Laine, 2001; Wilcox, Allison, Benzuly, et al., 2006). Although it may be difficult to justify routinely ruling out H. pylori infection prior to initiating NSAID therapy in patients with low risk, testing should be done routinely in patients with high GI risk (Chan, To, Wu, et al., 2002; Chan, Graham, 2004; Wilcox, Allison, Benzuly, et al., 2006). If H. pylori infection is present, it should be eradicated prior to initiating NSAID therapy (Chan, Graham, 2004; Wilcox, Allison, Benzuly, et al., 2006) (see Table 6-1).

NSAID Selection with Consideration of GI Risk

The expert panel of the First International Working Party on GI and CV Effects of NSAIDs and Anti-platelet Agents endorsed risk stratification based on a thorough patient assessment as a key step in the selection of NSAID therapy. This panel viewed the selection of a nonselective NSAID as appropriate in patients with average GI risk, including those younger than 70 years of age with no prior GI event and no concurrent use of corticosteroids, antithrombotic agents, or anticoagulants (Chan, Abraham, Scheiman, et al., 2008). The American Gastroenterological Association (AGA) (Wilcox, Allison, Benzuly, et al., 2006) and other researchers (Chan, Graham, 2004) also have provided recommendations that guide decision making with regard to NSAID selection and patient management during NSAID therapy. Table 6-2 provides a summary of NSAID treatment strategies correlated with GI risk.

Avoiding NSAIDs altogether is the best way to prevent an NSAID-related complication (Chan, Graham, 2004), and when NSAID therapy is indicated, the lowest effective NSAID should be used for the shortest time needed (APS, 2003; U.S. FDA, 2007). If pain is mild to moderate, a trial of acetaminophen should be tried. Patients with more severe pain who are at relatively high risk for NSAID complications should be considered for a trial of an opioid or another centrally-acting analgesic. The AGS specifically noted that opioids may be safer than NSAIDs in some older patients (e.g., those with high GI and CV risk factors) (AGS, 2002; AGS, 2009), an observation that may be underappreciated by clinicians.

NSAID-induced GI toxicity, like antiinflammatory activity, is believed to correlate with COX-1 inhibition (Chan, Graham, 2004; Simon, 2007). Some NSAIDs (e.g., piroxicam [Feldene], indomethacin [Indocin]) achieve adequate analgesia at doses associated with high antiinflammatory activity, and appear to have a relatively greater risk of GI toxicity (Chan, Graham, 2004). These drugs generally are not first-line for this reason. In addition, NSAIDs with longer half-lives (e.g., sulindac [Clinoril], piroxicam, indomethacin) may expose GI mucosa to the drug for longer periods at a higher concentration and also have been linked to greater GI toxicity (Wilcox, Allison, Benzuly, et al., 2006). These are the reasons cited for avoiding full-dose piroxicam, oxaprozin, and naproxen in older adults (Fick, Cooper, Wade, et al., 2003; Hanlon, Backonja, Weiner, et al., 2009)

Some of the nonselective NSAIDs exhibit relatively selective COX-2 inhibition and less COX-1 inhibition, particularly at lower doses; these drugs are considered safer in terms of GI toxicity and often are preferred as a result (Simon, 2007). For example, ibuprofen, which is one of the most commonly used NSAIDs in the United States, is considered a relatively safe choice from a GI perspective when doses are kept below full antiinflammatory effect (less than 2.4 gm/day) (Chan, Graham, 2004).

In general, a lower risk of GI toxicity characterizes all nonacetylated salicylates, including salsalate (Disalcid) and magnesium choline trisalicylate (Trilisate), and several of the nonselective, nonsalicylate NSAIDs. The latter group includes etodolac (Lodine), meloxicam (Mobic), nabumetone, and low doses of ibuprofen or naproxen (Laine, 2001; Simon, 2007; Wilcox, Allison, Benzuly, et al., 2006). It is reasonable to consider these drugs as first-line options for NSAID therapy, although, as mentioned, full-dose naproxen is not recommended in older adults because of its long half-life (Fick, Cooper, Wade, et al., 2003; Hanlon, Backonja, Weiner, et al., 2009). As mentioned, nabumetone also has the advantage of relatively low lipophilicity, leading to less penetration of the gastric mucosal barrier (Bannwarth, 2008; Hedner, Samulesson, Wahrborg, et al., 2004; Simon, 2007). Table 6-3 shows the various NSAIDs and their associated risk for GI events.

The so-called COX-2 selective NSAIDs (e.g., celecoxib [Celebrex]) also carry a relatively lower risk of GI adverse effects, at least when taken by patients who are not also using cardioprotective aspirin therapy (Chan, Graham, 2004; Dajani, Islam, 2008; Jacobsen, Phillips, 2004; Laine, White, Rostom, et al., 2008; Silverstein, Faich, Goldstein, et al., 2000; Simon, 2007; Singh, Fort, Goldstein, et al., 2006). The COX-2 selective NSAIDS, like all NSAIDs, vary in the degree to which they affect COX-2 over COX-1. The ratio of inhibition associated with the best safety profile overall is not known. At the present time, only one drug in this category is available in the United States—celecoxib.

According to one review (Laine, White, Rostom, et al., 2008), COX-2 selective NSAIDs are associated with a 75% relative risk reduction for gastroduodenal ulcers and a 61% relative risk reduction for ulcer complications compared with nonselective NSAIDs. These benefits decline or disappear during concurrent therapy with cardioprotective aspirin (Vardeny, Solomon, 2008). Although the combination of a COX-2 selective NSAID with cardioprotective aspirin may be somewhat safer than the combination of aspirin and a nonselective NSAID, the risks are still substantially higher than during treatment with a COX-2 selective NSAID alone (Simon, Fox, 2005). Nevertheless, a cohort study using government databases of patients age 65 and older concluded that celecoxib may be safer from a GI perspective than other NSAIDs in older patients receiving cardioprotective aspirin (Rahme, Bardou, Dasgupta, et al., 2007).

Although the results have not been uniform, most randomized trials support a relatively better GI risk profile for the COX-2 selective NSAIDs. One study showed no significant difference in gastroduodenal ulcers between celecoxib and naproxen (Simon, Weaver, Graham, et al., 1999), but another showed celecoxib produced a lower incidence of upper GI ulcers and adverse effects compared with diclofenac (Emery, Zeidler, Kvien, et al., 1999). The Celecoxib Long-term Arthritis Safety Study (CLASS), which compared high-dose celecoxib (400 mg twice daily), ibuprofen (800 mg 3 times daily), and diclofenac (75 mg twice daily) in patients with osteoarthritis (OA) or RA observed that celecoxib caused significantly fewer symptomatic ulcers and ulcer complications (Silverstein, Faich, Goldstein, et al., 2000).

As expected, newly developed COX-2 selective NSAIDs, such as lumiracoxib (Prexige) and etoricoxib (Arcoxia), also have a relatively lower risk of GI toxicity. The Therapeutic Arthritis Research and Gastrointestinal Event Trial (TARGET), which evaluated over 18,000 patients, showed a 3- to 4-fold reduction in ulcer complications with the use of lumiracoxib compared with naproxen or ibuprofen (Schnitzer, Burmester, Mysler, et al., 2004). A later study of healthy volunteers who were randomized to receive lumiracoxib, naproxen plus omeprazole (Prilosec), or placebo revealed that lumiracoxib yielded an incidence of small bowel mucosal breaks that was lower than naproxen plus omeprazole and similar to placebo (Hawkey, Ell, Simon, et al., 2008).

The Multinational Etoricoxib and Diclofenac Arthritis Long-term (MEDAL) program demonstrated fewer uncomplicated upper GI adverse events with etoricoxib than diclofenac; however, there was no difference in complicated events (perforation, obstruction, and complicated bleeding) (Laine, Curtis, Cryer, et al., 2007; Laine, Curtis, Langman, et al., 2008). Prior GI event and age 65 years and older were identified as significant risk factors for GI complications. Another study showed similar efficacy and lower GI adverse effects with etoricoxib compared with diclofenac in patients with OA over a 52-week evaluation period (Curtis, Bockow, Fisher, et al., 2005). An international multicenter study of 997 patients with OA compared long-term use of etoricoxib and naproxen and found the drugs had similarly satisfactory efficacy and tolerability, but there was a lower incidence of GI adverse effects and higher incidence of CV adverse effects in patients who took etoricoxib compared with those who took naproxen (Reginster, Malmstrom, Mehta, et al., 2007).

The latter study highlights the observation that diminished GI risk should not be interpreted as lower risk overall; CV risk remains a concern and continues to be evaluated in the newer COX-2 selective NSAIDs (Medscape Medical News, 2005; Topol, Falk, 2004; Wood, 2007) (see pp. 197-204 this chapter for discussion of CV risk). An assessment of risk factors (risk-benefit analysis) prior to NSAID therapy must include factors that go beyond GI risk (Tannenbaum, Bombardier, Davis, et al., 2006; Wilcox, Allison, Benzuly, et al., 2006). If the primary concern is GI risk, selection of an NSAID should focus either on a COX-2 selective drug or a nonselective drug with low GI risk (Wilcox, Allison, Benzuly, 2006) (see p. 204 for discussion of NSAID selection in the presence of both CV and GI risk).

Gastroprotective Co-Therapy

Gastroprotective co-therapy can reduce GI risk during NSAID therapy (Steen, Nurmohamed, Visman, et al., 2008; Wilcox, Allison, Benzuly, et al., 2006). The ACCF/ACG/AHA consensus guideline recommends gastroprotective co-therapy, preferably proton pump inhibitors (PPPs), in individuals who are taking a nonselective or COX-2 selective NSAID in conjunction with low-dose aspirin (Bhatt, Scheiman, Abraham, et al., 2008). The expert panel of the First International Working Party on GI and CV Effects of NSAIDs and Anti-platelet Agents also recommended co-therapy when NSAIDs are used in individuals with increased GI risk (Chan, Abraham, Scheiman, et al., 2008).

Three classes of gastroprotective agents are suggested as options for co-therapy: (1) the prostaglandin analogues such as misoprostol (Cytotec); (2) histamine receptor type-2 (H₂) antagonists such as cimetidine (Tagamet), famotidine (Pepcid), and ranitidine (Zantac); and (3) PPIs such as esomeprazole (Nexium), lansoprazole (Prevacid), and omeprazole (Prilosec). A Cochrane Collaboration Review of 33 clinical trials concluded that all three classes of drugs reduced the incidence of both gastric and duodenal NSAID-induced ulcers (Rostom, Dube, Wells, et al., 2002); however, their effectiveness varies. Following is an overview of key research related to the three classes of gastroprotective agents.

Misoprostol is a synthetic prostaglandin that has been studied extensively in trials of ulcer prevention (Wilcox, Allison, Benzuly, et al., 2006). A Cochrane Collaboration Review concluded that doses of 800 mcg/day of misoprostol were superior to 400 mcg/day for prevention of gastric ulcers; however, no dose-related response was noted for duodenal ulcers (Rostom, Dube, Wells, et al., 2002). A prospective, double-blind study of long-term NSAID users with a history of gastric ulcer found patients who received 800 mcg/day of misoprostol remained ulcer free longer than those who received placebo or 15 or 30 mg of the PPI lansoprazole; however, a higher number of patients in the misoprostol group reported adverse effects and withdrew early from the study (Graham, Agrawal, Campbell, et al., 2002). The high incidence of adverse effects and study withdrawals led the researchers to conclude that misoprostol is clinically equivalent to PPIs. All doses of misoprostol can produce a high incidence of adverse effects, such as nausea, abdominal cramps, and diarrhea, and may not be well tolerated, particularly by the older patient (Chan, Graham, 2004; Rostom, Dube, Wells, et al., 2002; Wilcox, Allison, Benzuly, et al., 2006). Adherence to the regimen can be another barrier due to misoprostol dosing requirements that include taking the drug 4 times daily with food.

Whereas standard doses of H₂-antagonists are reported to be effective in prevention of duodenal but not gastric ulcers in patients taking NSAIDs, double doses have been shown to be effective against both types of ulcers (Rostom, Dube, Wells, et al., 2002). However, a review of nearly 40 years of randomized controlled trials revealed that misoprostol and PPIs were more effective than H₂-antagonists in reducing the risk of clinically significant GI adverse effects associated with NSAID use (Jacobsen, Phillips, 2004). Further, it has been suggested that any major benefit from H₂-antagonists may be limited to patients with H. pylori infection (Chan, Graham, 2004).

The ASTRONAUT study of 541 NSAID users showed that the PPI omeprazole prevented and healed all types of ulcers more effectively than the H₂-antagonist ranitidine (Yeomans, Tulassay, Juhasz, et al., 1998). A larger double-blind study (N = 935), called the OMNIUM study, randomly assigned NSAID users with a history of ulcers (gastric or duodenal or both) to receive omeprazole or misoprostol for 4 weeks or, in the absence of healing, 8 weeks (Hawkey, Karrasch, Szczepanski, et al., 1998). Rates of successful ulcer treatment were similar between the two drugs; however, omeprazole was associated with a lower relapse rate and was better tolerated than misoprostol. Other PPIs have shown similar results. A prospective, double-blind study of 353 chronic NSAID users with active gastric ulcers found better healing rates after 8 weeks of treatment with lansoprazole than with the H₂-antagonist ranitidine (Agrawal, Campbell, Safdi, et al., 2000). AGA consensus guidelines recommend misoprostol if tolerated or a PPI in patients at high risk for GI complications (Wilcox, Allison, Benzuly, et al., 2006) (see Table 6-2). The AGS recommends a PPI or misoprostol in older adults taking either a nonselective NSAID or a COX-2 selective NSAID with low-dose aspirin (AGS, 2009).

The use of a COX-2 selective NSAID may be as effective as gastroprotective co-therapy in some populations (Chan, Graham, 2004; Jacobsen, Phillips, 2004; Schnitzer, Burmester, Mysler, et al., 2004; Laine, White, Rostom, et al., 2008; Silverstein, Faich, Goldstein, et al., 2000; Simon, 2007; Simon, Fox, 2005). Healthy, lesion-free subjects were randomized to receive celecoxib 200 mg twice daily or naproxen 500 mg twice daily plus omeprazole 20 mg once daily, or placebo for 2 weeks (Goldstein, Eisen, Lewis, et al., 2005). Those taking celecoxib exhibited significantly fewer small bowel mucosal breaks than those who took naproxen plus omeprazole as determined by video capsule endoscopy; those taking placebo had the fewest breaks. A later study by these researchers reported similar findings—celecoxib 200 mg twice daily was associated with significantly fewer small bowel mucosal breaks than ibuprofen 800 mg three times daily plus omeprazole 20 mg once daily, or placebo for 2 weeks (Goldstein, Eisen, Lewis, et al., 2007). There were no significant differences between celecoxib and placebo in this study.

A COX-2 selective NSAID with gastroprotective co-therapy may be most beneficial to patients with a high or very high risk for GI complications. A prospective, double- blind trial (N = 441) recruited long-term nonselective NSAID users with arthritis admitted to the hospital with upper GI bleeding. After ulcer healing, all patients were given 200 mg of celecoxib twice daily and were randomly assigned to receive the PPI esomeprazole or placebo twice daily for 12 months. Combination therapy with the PPI was found to be more effective in preventing GI adverse effects than celecoxib alone, leading the researchers to recommend a COX-2 selective NSAID and a PPI in patients at high risk for recurrent GI bleeding who need to take an NSAID (Chan, Wong, Suen, et al., 2007).

Greater attention to identification of those who might benefit from gastroprotective therapy and surveillance of patient adherence to gastroprotective therapy are warranted. A review of good-quality meta-analyses and large observational studies found that patients with GI risk factors often do not receive a prescription for gastroprotection and that those who do are often nonadherent to the treatment plan (Moore, Derry, Phillips, et al., 2006).

CV Risk Factors

All patients who are being considered for ongoing NSAID therapy should be carefully assessed for preexisting risk factors for vascular disease and closely monitored for CV adverse events during NSAID use, regardless of the type of NSAID administered (Andersohn, Suissa, Garbe, 2006; Brophy, Levesque, Zhang 2007; Crofford, Oates, McCune, et al., 2000; FitzGerald, Patrono, 2001; Gislason, Jacobsen, Rasmussen, et al., 2006). For example, Solomon and colleagues (2008) found that individuals with prior CV disease, hypertension, RA, chronic renal disease, chronic obstructive pulmonary disease, and those ≥ 80 years old all are at increased risk of CV events when taking NSAIDs. Assessment of these and other factors, and interventions to reduce risk and monitor over time, are important recommendations for NSAID therapy (Antman, Bennett, Daugherty, et al., 2007; Chan, Abraham, Scheiman, et al., 2008).

The First International Working Party on GI and CV Effects of NSAIDs and Anti-platelet Agents predefined high CV risk as the presence of established CV disease (e.g., prior myocardial infarction (MI), stroke, or angina) or an estimated 10-year CV risk of greater than 20% in patients without CV disease (Chan, Abraham, Scheiman, et al., 2008). The AHA and the American Stroke Association categorize CV risk factors as nonmodifiable, modifiable, or potentially modifiable (Goldstein, Adams, Alberts, et al., 2006). This provides a good format for consideration of CV risk prior to initiating NSAID therapy (see Table 6-4).

Nonmodifiable risks are those beyond the person’s ability to change, such as age, sex, and race. Advanced age is a major contributor, with the risk of stroke doubling with each decade after the age of 55 years. It is expected that the aging of the “baby boomer” population will become the biggest factor boosting the prevalence of heart failure (Schoken, Benjamin, Fonarow, et al., 2008). The aging process also brings other conditions that increase CV risk, including hypertension, diabetes, and renal insufficiency (McDermott, Fried, Simonsick, et al., 2000; Schoken, Benjamin, Fonarow, et al., 2008) (see renal effects later in this chapter). The AGS lists heart failure as an absolute contraindication and hypertension as a relative contraindication to the use of NSAIDs in older adults (AGS, 2009).

Other nonmodifiable risk factors, like age, are strongly influenced by related conditions and type of CV event. For example, stroke is more common in men than in women (Goldstein, Adams, Alberts, et al., 2006), whereas mortality from coronary heart disease is higher among women (Jousilahti, Vartiainen, Tuomilehto, et al., 1999). Genetic factors, including genetic disorders such as sickle cell anemia (James, 2000) also are important, both in terms of risk factors and CV events themselves. Although nonmodifiable risk factors cannot be changed, they help identify individuals who are at highest risk and who might benefit from aggressive disease prevention or treatment, as well as close monitoring during NSAID therapy (Goldstein, Adams, Alberts, et al., 2006).

Modifiable risk factors are those that are well documented as contributing to adverse CV events and are within the individual’s ability to change or control effectively enough to significantly reduce risk. Preexisting CV disease, hypertension, diabetes, and dyslipidemia are among the modifiable factors (Goldstein, Adams, Alberts, et al., 2006). Obesity, sedentary lifestyle, and diet interact with other physiologic abnormalities such as insulin resistance and a pro-inflammatory state to produce the metabolic syndrome, which has been separately identified as an important risk factor for CV events (Goldstein, Adams, Alberts, et al., 2006). Frank diabetes with poorly controlled glycemia is a potentially major factor (Fox, Coady, Sorlie, et al., 2007), as is hypertension alone (Schoken, Benjamin, Fonarow, et al., 2008). Hypertension, for example, causes a 2- to 3-fold increase in risk of heart failure, as well as an increased risk of coronary artery disease, and early detection and control of BP is perceived to be a key strategy for prevention of CV disease and catastrophic CV events (Pearson, Blair, Daniels, et al., 2002).

The role of hypertension as a mediator of both heart failure and coronary artery disease underscores the need for a careful assessment of CV risk factors in those considered for NSAID therapy. NSAIDs increase the risk of hypertension or worsen preexisting hypertension (Bombardier, Laine, Reicin, et al., 2000; Curhan, Willett, Rosner, et al., 2002; Dedier, Stampfer, Hankinson, et al., 2002; Simon, 2007; Whelton, Lefkowith, West, et al., 2006; Whelton, White, Bello, et al., 2002; White, Kent, Taylor, et al., 2002), and BP monitoring, with adjustment of antihypertensive therapy if needed, is essential to reduce NSAID-associated CV risk that is mediated through BP abnormalities (see pp. 205-206 for more discussion about NSAIDs and hypertension).

Cigarette smoking is another extremely well-documented, potent modifiable CV risk factor; both first-hand and second-hand smoke increase risk (Goldstein, Adams, Alberts, et al., 2006; Hermann, Krum, Ruschitzka, 2005; McAdam, Byrne, Morrow, et al., 2005). NSAID users who smoke have a higher incidence of CV adverse events than NSAID users who do not smoke (Chan, Manson, Albert, et al., 2006).

Some potentially modifiable factors are less well documented in terms of their association with adverse CV events, including obstructive sleep apnea and migraine headache (Goldstein, Adams, Alberts, et al., 2006). In addition to NSAIDs, several drugs have been implicated in heart disease, such as some chemotherapies and hormonal agents, including oral contraceptives; IV drug abuse and alcohol abuse also may predispose to heart disease (Schoken, Benjamin, Fonarow, et al., 2008).

Unfortunately, diseases with an inflammatory component that may benefit from NSAID therapy, such as RA and systemic lupus erythematosus (SLE), have also been shown to produce elevated CV risk (D’Cruz, 2006; de Carvalho, Bonfa, Borba, 2008; Solomon, Karlson, Rimm, et al., 2003). A large cohort study of over 114,000 women who participated in the Nurses’ Health Study (Colditz, 1995) found that those with RA had an increased risk for MI (Solomon, Karlson, Rimm, et al., 2003). This led the researchers to recommend aggressive CV disease-prevention strategies in individuals with RA. CV risk factors for individuals with SLE are many and include dyslipoproteinemia with low HDL (high-density lipoprotein) cholesterol and increased triglycerides (de Carvalho, Bonfa, Borba, 2008; Thiagarajan, 2001) as well as an association with autoantibodies to oxidized LDL (low-density lipoprotein) cholesterol (Svenungsson, Jensen-Urstad, Heimburger, et al., 2001).

The link between CV risk and inflammatory diseases is explained, at least in part, by evidence that shows inflammation is a prominent feature of atherosclerotic plaque formation (Solomon, Karlson, Rimm, et al., 2003; Svenungsson, Jensen-Urstad, Heimburger, et al., 2001; Vasan, Sullivan, Roubenoff, et al., 2003), such that atherosclerosis has been called an inflammatory condition of the vessel wall (Thiagarajan, 2001). A variety of inflammatory factors associated with increased CV risk, such as elevated C-reactive protein concentration, are also elevated in RA and SLE (Svenungsson, Jensen-Urstad, Heimburger, et al., 2001).

Potentially modifiable CV risk factors commonly co-exist and may potentiate each other, further highlighting the need to have a comprehensive medical assessment as part of safe NSAID prescribing. For example, women who smoke and use oral contraceptives are at 1.3 times greater CV risk than nonsmoking women who use oral contraceptives (Goldstein, Adams, Alberts, et al., 2006). Alcohol intake may increase CV risk by increasing BP (Schoken, Benjamin, Fonarow, et al., 2008). Obese individuals often experience multiple comorbidities that represent elevated risk including sleep-disordered breathing, physical inactivity, and diabetes (Schoken, Benjamin, Fonarow, et al., 2008).

CV Effects of COX-2 Selective NSAIDs

Shortly after the release of rofecoxib in 1999, the 12-month Vioxx Gastrointestinal Outcomes Research (VIGOR) trial, which was designed to compare GI adverse effects associated with naproxen and rofecoxib, showed a 50% reduction in serious GI events in the subjects who were given rofecoxib (Bombardier, Laine, Reicin, et al., 2000). However, an unexpected and disturbing finding was a 5-fold increase in thromboembolic events, such as acute MI, in the rofecoxib group. The VIGOR trial did not have a placebo control group, which lent plausibility to the frequently-offered explanation that naproxen, a nonselective NSAID, was cardioprotective and thus produced fewer CV events than rofecoxib (Graham, 2006; Graham, Campen, Hui, et al., 2005). Later analysis, however, suggested that naproxen is not cardioprotective (Graham, Campen, Hui, et al., 2005; McGettigan, Henry, 2006; Ray, Stein, Hall, et al., 2002; Salpeter, Gregor, Ormiston, et al., 2006), at least at typical doses (Chan, Abraham, Scheiman, et al., 2008). Graham (2006) proposed that a more accurate interpretation of the VIGOR findings would have been that rofecoxib increased CV risk by 5-fold rather than that the CV risk associated with naproxen was one-fifth that of rofecoxib.

The Adenomatous Polyp Prevention On Vioxx (APPROVe) study was undertaken to determine if long-term (3 years) rofecoxib treatment would reduce the risk of recurrent adenomatous polyps in patients with a history of colorectal adenomas (Bresalier, Sandler, Quan, et al., 2005). Over 2500 patients were randomized to receive either rofecoxib or placebo. During the first 18 months of treatment, the CV adverse event rates were similar among the two groups; however, after 18 months, data suggested a greater number of MIs and ischemic cerebrovascular events occurred in the rofecoxib group. A later systematic review contradicted the suggestion that vascular risk with rofecoxib occurred only after months of treatment (McGettigan, Henry, 2006). Increasing evidence of elevated CV risk prompted the withdrawal of rofecoxib from the worldwide market in September 2004 (Graham, 2006).

Valdecoxib, which is the active metabolite of the parenteral COX-2 selective NSAID parecoxib (Dynastat) available in Europe, was withdrawn from the U.S. market in 2005 following research showing CV events similar to rofecoxib (Vardeny, Solomon, 2008). Studies showing an increase in CV events in patients who were given parecoxib and valdecoxib perioperatively (Nussmeier, Whelton, Brown, et al., 2005; Ott, Nussmeier, Duke, et al., 2003) led to recommendations against the use of any COX-2 selective NSAIDs following high-risk open heart surgery (U.S. FDA, 2007). (See Chapter 8 for a discussion of perioperative NSAID use.) Valecoxib was withdrawn from the market on the basis of the CV risk as well as reported cases of cutaneous hypersensitivity (see pp. 207-208 for discussion of nonopioid hypersensitivity).

Celecoxib currently is the only COX-2 selective NSAID in the United States. The so-called CLASS study randomized 8059 patients with OA or RA to receive a 6-month trial of celecoxib, ibuprofen, or diclofenac (Silverstein, Faich, Goldstein, et al., 2000). Patients were allowed to continue cardioprotective therapy during the study period. No difference in CV events was found among the three NSAIDs, and new-onset or aggravation of preexisting hypertension was lowest in the celecoxib-treated group. Other studies have shown similar results (Hudson, Rahme, Richard, et al., 2007; Solomon, Schneeweiss, Glynn, et al., 2004; Steinbach, Lynch, Phillips, et al., 2000; Whelton, White, Bello, et al., 2002). These data questioned whether the risk of CV toxicity identified during rofecoxib and valdecoxib therapy, which suggested a class effect (i.e., associated with the so-called COX-2 selective drugs), was indeed this, or rather a risk that attended some drugs and not others.

Numerous other studies followed the early efficacy trials and were intended to clarify the risk of CV events for individual drugs and the classes they occupy. Although some of the data are conflicting, a broad picture of the CV risk associated with NSAID therapy has emerged.

Studies of insurance claims data provide one source of evidence pertaining to this issue. An examination of data of Medicare beneficiaries, which consisted of new users of COX-2 selective NSAIDs (N = 76,082), new users of nonselective NSAIDs (N = 53,014), and nonusers of any NSAIDs (N = 46,558), suggested that most NSAIDs, when taken at standard doses, do not significantly increase the risk of CV events (Solomon, Glynn, Rothman, et al., 2008). These data showed that rofecoxib was associated with the highest CV event rates and that naproxen was associated with the lowest—even lower than nonusers. Certain patient characteristics (prior CV disease, hypertension, RA, chronic renal disease, chronic obstructive pulmonary disease, and age 80 years or older) were associated with elevated risk as well. Other large database studies (Ray, Stein, Hall, et al., 2002; Solomon, Schneeweiss, Glynn, et al., 2004; Graham, Campen, Hui, et al., 2005) suggested that the incidence of acute MI with celecoxib use was actually lower than that of other NSAIDs (Simon, 2007).

An extensive systematic review of observational studies that investigated CV events associated with a variety of NSAIDs found that celecoxib at doses of 200 mg/day or less was not associated with CV increased risk and that naproxen was not associated with any reduced CV risk (McGettigan, Henry, 2006). The highest risk among the nonselective NSAIDs was with commonly used doses of diclofenac (e.g., 150 mg daily). This finding may be notable in light of data indicating a relatively higher COX-2 selectivity of this drug than other nonselective NSAIDs (Graham, 2006; McGettigan, Henry, 2006; Vardeny, Solomon, 2008) (see Figure 6-1). Meloxicam, another long-acting nonselective NSAID that preferentially inhibits COX-2, at least in lower prescribed doses, has shown similar efficacy and a better GI profile than the comparable nonselective NSAIDs piroxicam and diclofenac and less peripheral edema and weight gain when compared with diclofenac (Smith, Baird, 2003). Studies are needed to evaluate this and other nonselective NSAIDs that possess relatively higher levels of COX-2 inhibitory activity.

Data from randomized controlled trials of celecoxib and other NSAIDs have yielded mixed results. A meta-analysis of randomized clinical trials evaluated the incidence of CV events in nearly 7500 patients exposed to celecoxib 200 to 800 mg daily; nearly 14,000 patients treated with diclofenac, ibuprofen, naproxen, ketoprofen (Orudis), or loxoprofen (available outside of the United States); and over 4000 patients who were given placebo (White, West, Borer, et al., 2007). The analysis failed to demonstrate a difference in the incidence of CV events associated with celecoxib compared with other NSAIDs or compared with placebo for up to 1 year of treatment exposure. Use of aspirin or presence of CV risk factors did not alter the findings.

In contrast, placebo-controlled trials designed to study a potential role for celecoxib in the prevention of colorectal adenoma revealed an increased CV risk in patients who received celecoxib. The Prevention of Colorectal Sporadic Adenomatous Polyps (PreSAP) trial demonstrated a reduced recurrence of colorectal adenomas in patients who received a single 400-mg dose of celecoxib daily; however, CV events occurred more often in patients who received celecoxib (2.5%) than in those who received placebo (1.9%) (Arber, Eagle, Spicak, et al., 2006). More concerning was the Adenoma Prevention with Celecoxib (APC) trial, which randomized patients to receive celecoxib 200 mg or 400 mg twice daily or placebo (Bertagnolli, Eagle, Zauber, et al., 2006). As in the PreSAP trial, a reduction in the recurrence of colorectal adenomas was noted; however, serious adverse CV events occurred in 18.8%, 20.4%, and 23% of those who received placebo, 200 mg of celecoxib twice daily, and 400 mg of celecoxib twice daily. This increase in CV events led the researchers to caution against the routine use of celecoxib for prevention of colorectal adenomas. Celecoxib administration in several long-term trials, including the APC and PreSAP trials, was halted as a result of CV events (Solomon, Wittes, Finn, et al., 2008). (See Chapter 7 for more on NSAIDs and prevention of cancer.)

The PreSAP and APC trials elucidated interesting findings related to the impact of both dose and dosing regimens on the level of CV risk associated with celecoxib (Solomon, Pfeffer, McMurray, et al., 2006; Solomon, Wittes, Finn, et al., 2008; Vardeny, Solomon, 2008). Vardeny and Solomon (2008) pointed out that the evidence for CV risk in the PreSAP trial, which administered a single 400-mg celecoxib dose daily, was less compelling than that of the APC trial, which showed significant risk associated with a 200-mg or 400-mg twice-daily dosing regimen. This was demonstrated in an independent review of combined data from the APC and PreSAP trials that found a hazard ratio (estimate of relative risk) for death from MI, stroke, or heart failure of 1.3 in patients taking 400 mg once daily (PreSAP), 2.6 in those taking 200 mg twice daily (APC), and 3.4 in those taking 400 mg twice daily (Solomon, McMurray, Pfeffer, et al., 2005). Similarly, an analysis of pooled data from nearly 8000 patients in six placebo-controlled trials, which included the APC and PreSAP data, assessed CV risk associated with three dosing regimens of celecoxib and found the lowest, intermediate, and highest risk associated with dosing regimens of 400 mg daily, 200 mg twice daily, and 400 mg twice daily, respectively (Solomon, Wittes, Finn, et al., 2008). Patients with highest baseline risk had an increased relative risk for CV events.

Notwithstanding the conflicting data, these studies suggest that there is some CV risk associated with celecoxib treatment, as there is with other NSAIDs. The finding of dose-dependency is relatively strong evidence of a pharmacologic effect. It is curious, however, that two studies observe that lower risk is associated with once-daily dosing compared with twice-daily dosing at the same overall daily dose (400 mg). This also may have a pharmacologic explanation in that the maximum plasma concentration of celecoxib is reached in 90 minutes and the half-life is just 1.5 hours (Paulson, Hribar, Liu, et al., 2000). Because prostacyclin levels require approximately 12 hours to recover following a single oral dose of celecoxib (McAdam, Catella-Lawson, Mardini, et al., 1999), it has been proposed that once-daily dosing (but not twice-daily dosing) of celecoxib may allow enough prostacyclin recovery and normalization to attenuate the prothrombotic effect (Grosser, Fries, FitzGerald, 2006; Vardeny, Solomon, 2008). Although further research is needed, the findings in other studies that utilized once-daily dosing support this theory (Sowers, White, Pitt, et al., 2005; Whelton, Fort, Puma, et al., 2001).

The evidence overall suggests that there is CV risk during all NSAID therapies, however the link with COX-2 inhibition suggests that drugs with COX-2 selectivity must be particularly scrutinized for this effect. This has been done with newer COX-2 selective NSAIDs, such as etoricoxib and lumiracoxib (Andersohn, Suissa, Garbe, 2006; Cannon, Curtis, FitzGerald, et al., 2006; Farkouh, Kirshner, Harrington, et al., 2004; Laine, Curtis, Cryer, et al., 2007; Schnitzer, Burmester, Mysler, et al., 2004). These drugs were not available in the United States at the time of publication (Medscape Medical News, 2005; Wood, 2007).

The TARGET study randomized over 18,000 patients with OA to receive lumiracoxib, naproxen, or ibuprofen (Farkouh, Kirshner, Harrington, et al., 2004). Patients were allowed to take low-dose aspirin during the study. The incidence of adverse CV events was low and did not differ among the NSAID groups, irrespective of aspirin use. Though not statistically significant, there were more nonfatal MIs in the lumiracoxib group than in the naproxen nonaspirin group. The lack of a placebo arm in this study makes it difficult to ascertain actual CV risk associated with lumiracoxib.

The MEDAL program evaluated pooled data from three trials involving 34,701 patients with OA and 9787 with RA who were given etoricoxib or diclofenac at recommended doses for a mean duration of 18 months (Cannon, Curtis, FitzGerald, et al., 2006). Patients in the two groups were reported to have almost identical rates of thrombotic CV events (320 patients in the etoricoxib group; 323 patients in the diclofenac group). This study has been criticized for the lack of a placebo arm and the use of diclofenac as the comparator NSAID (Graham, 2006; Hughes 2006; Rodriquez, Patrignani, 2006). Given the suggestion that diclofenac has a relatively high CV risk (Hernandez-Diaz, Varas-Lorenzo, Rodriguez, 2006; Kearney, Baigent, Godwin, et al., 2006; McGettigan, Henry, 2006), comparability between this drug and etoricoxib does not establish the safety of the latter drug (Graham, 2006). Although diclofenac is widely prescribed and does not appear to interfere with the antiplatelet effects of low-dose aspirin (Cannon, Curtis, FitzGerald, et al., 2006) (see later discussion regarding NSAID interference with cardioprotective aspirin), its use in this study complicates interpretation of the results.

NSAID Selection with Consideration of CV Risk

This conclusion obligates clinicians to be aware of evidence that supports different risk profiles for different NSAIDs and to consider relative risks when recommending a specific drug in clinical practice. The First International Working Party on GI and CV Effects of NSAIDs and Anti-platelet Agents suggested that either nonselective NSAIDs or COX-2 selective NSAIDs could be appropriate in patients with no evidence of elevated GI risk and average CV risk; naproxen was preferred in patients with no GI risk but high CV risk (Chan, Abraham, Scheiman, et al., 2008). The panel rated the use of COX-2 selective NSAIDs in patients with high CV risk as uncertain or inappropriate (see later discussion of NSAID selection in the presence of both GI and CV risk). As mentioned, an extensive review of Medicare beneficiary data confirmed the safety of naproxen from a CV perspective as well (Solomon, Glynn, Rothman, et al., 2008).

The AHA proposes a “stepped care” approach for the treatment of musculoskeletal pain in individuals with known CV disease or risk factors for ischemic heart disease (Antman, Bennett, Daugherty, et al., 2007). Others have incorporated similar recommendations into their guidelines for the management of high-risk cardiac patients (Anderson, Adams, Antman, et al., 2007). The first step of the AHA approach recommends acetaminophen, aspirin, tramadol, or short-term opioid analgesics. Nonacetylated salicylates, such as choline magnesium trisalicylate or salsalate (Disalcid), are a second-step option. If this does not control pain, a nonselective NSAID with low COX-2 selectivity such as naproxen or ibuprofen is recommended. The use of a COX-2 selective NSAID is suggested as a last choice. Of note is a review of the literature that suggested the use of a COX-2 selective NSAID with cardioprotective aspirin in patients with CV risk is preferential to nonselective NSAIDs (Strand, 2007). However, some experts have questioned the rationale of using a COX-2 selective NSAID with aspirin at all in patients with high CV risk citing a lack of research showing aspirin can sufficiently mitigate the absolute CV risk imposed by COX-2 selective NSAIDs (Hughes, 2007).

Other general AHA recommendations when NSAIDs are used are to select patients at low risk for thrombotic events, prescribe the lowest dose required to control symptoms, add cardioprotective aspirin (with a PPI) to patients at increased risk of thrombotic events, and regularly monitor for sustained hypertension, edema, worsening renal function, and GI bleeding. The AHA cautions that effective pain relief with the use of NSAIDs may come at the cost of an increase in risk for CV or cerebrovascular complications (Antman, Bennett, Daugherty, et al., 2007).

NSAID Selection in Presence of Both CV and GI Risk

Patients who are at risk for both GI and CV adverse effects present special challenges in determining how to proceed with NSAID therapy. Although the use of naproxen or ibuprofen combined with nonaspirin antiplatelet therapy (e.g., clopidogrel [Plavix]) has been suggested (Hughes, 2006), there are few data that support this approach.

The expert panel of the First International Working Party on GI and CV Effects of NSAIDs and Anti-platelet Agents suggests that the initial choice of an NSAID should depend on the patient’s CV risk, whereas the addition of gastroprotective therapy depends on the number and severity of GI risk factors (Chan, Abraham, Scheiman, et al., 2008). The panel rated naproxen plus a PPI or misoprostol as appropriate, regardless of patient age, in patients with both CV and GI risks. Non-naproxen NSAIDs and COX-2 selective NSAIDs, with or without gastroprotective therapy, were rated as uncertain or inappropriate in such patients. Some researchers recommend avoiding full doses of NSAIDs with a long half-life, such as naproxen (Fick, Cooper, Wade, et al., 2003).

A retrospective cohort study using government databases including patients 65 years and older concluded that celecoxib may be safer from a GI perspective than other NSAIDs in older adults receiving cardioprotective aspirin to minimize CV risk (Rahme, Bardou, Dasgupta, et al., 2007). Others have suggested a low-dose COX-2 selective NSAID plus low-dose aspirin plus a PPI if NSAID therapy is used in patients with major GI and CV risk factors (Laine, White, Rostom, et al., 2008).

It is important to remember that avoiding NSAIDs altogether may be the best approach for high-risk patients (Laine, White, Rostom, et al., 2008). In fact, the expert panel of the First International Working Party on GI and CV Effects of NSAIDs and Anti-platelet Agents recommends avoidance of NSAID therapy entirely in patients with high CV risk and multiple GI risk factors (e.g., patients older than age 70 who are receiving concomitant corticosteroids, low-dose aspirin, and other antiplatelet agents or anticoagulants) (Chan, Abraham, Scheiman, et al., 2008). Opioid analgesics are among the safest analgesics from a GI and CV perspective and should be considered in patients with pain who present with significant risk for NSAID-induced adverse effects.

NSAIDs with Minimal Effect on Platelet Aggregation

Although NSAIDs do not block platelet function sufficiently to be useful as antithrombotic agents, most of the nonselective NSAIDs do interfere enough to increase bleeding time. For this reason, these drugs may carry relatively higher risk for patients with bleeding disorders, such as hemophilia. In such populations, the least risky drugs classified as nonselective NSAIDs are presumably those that are least likely to exert a COX-1-mediated inhibition of platelet function, such as nabumetone, meloxicam, choline magnesium trisalicylate, magnesium salicylate (Arthriten, Backache, Doan’s), and salsalate (Disalcid). The COX-2 selective NSAIDs (e.g., celecoxib) have no effect on bleeding time and should be considered as well if not contraindicated (Visser, Goucke, 2008) (see Chapter 8 for perioperative NSAID use).