The use of cyclo-oxygenase-2 (COX-2) inhibitors in OA and RA has been accompanied by an overall reduction in severe gastrointestinal events (i.e. upper gastrointestinal ulceration, bleeding, and perforation) and symptoms (reflux, gastritis) associated with non-selective NSAIDs. At the same time, their analgesic potential is at least as potent (but not superior) to non-selective NSAIDs. Despite these advantages, a theoretical, yet controversial, detrimental consequence of selective inhibition of cyclo-oxygenase-2 is the induction of a pro-thrombotic state with the potential to promote cardiovascular events (Myocardial infarction, stroke, etc). Clear-cut supporting evidence for this phenomenon is lacking. Several lines of evidence have implicated one COX-2 inhibitor in particular, rofecoxib (Vioxx), as being associated with an increased incidence of cardiovascular events, particularly when used in higher than standard daily doses. Yet still, uncertainties exist as to whether these findings represent true associations or are simply a result of biases within the various study designs. Here, Schnitzer & Farkouh et al analyze the GI (Lancet 364(9435):665-74, 2004) and cardiovascular events (Lancet 364(9435):675-84, 2004) encountered in the Therapeutic Arthritis Research and Gastrointestinal Event Trial (TARGET), a comparison of the pre-marketed COX-2 inhibitor lumiracoxib to naproxen and ibuprofen.
METHODS: Patients aged > 50 years with a diagnosis of primary OA (hip, knee, hands, cervical or lumbar spine)and no historical or clinical evidence of cardiovascular disease were randomized to receive lumiracoxib (400mg per day- a dose 2 – 4 times higher than the dose recommended for OA), naproxen (500mg twice a day), or ibuprofen (800mg three times a day). Patients with recent GI bleeding (within 1 year) or upper GI ulceration (within 30 days) and those receiving GI protective medications (i.e. proton pump inhibitors (PPIs), misoprostol, or full-dose H2-antagonists) were excluded. The primary outcome was the incidence of upper gastrointestinal complications (clinically significant bleeding, perforation, or obstruction from erosive or ulcer disease).
Secondary outcomes included a composite endpoint of cardiovascular morbidity and mortality. Patients deemed at high risk for cardiovascular disease were eligible for enrollment provided they were already receiving concomitant low-dose aspirin therapy (75-100mg per day) for at least 3 months prior to enrollment (randomization per group was stratified according to use of low-dose aspirin). Other anti-platelet agents were not permitted. Predefined clinical cardiovascular events were myocardial infarction, unstable angina, cardiac arrest, stroke, transient ischemic attack, deep venous thrombosis, pulmonary embolism, and congestive heart failure. Incident myocardial infarctions were defined based on clinical events for overt occurrences and by serial electrocardiograms for silent myocardial infarction. The study design was organized as two sub-studies, one comparing lumiracoxib to naproxen and the other comparing lumiracoxib to ibuprofen. Subjects were followed for 52 weeks with six visits at regular intervals.
RESULTS: Baseline characteristics were balanced across all groups and included approximately 75% women with a mean age (all patients) of 63 years. Approximately 25% of subjects were receiving low dose aspirin at entry. Despite the required exclusion of prevalent cardiovascular disease for enrollment, approximately 10% of subjects had a history of vascular disease. In fact, the proportion of patients with a history of vascular disease was higher in the sub-study comparing lumiracoxib to naproxen (12%) compared to the sub-study comparing lumiracoxib to ibuprofen (8%). In addition, more patients in the lumiracoxib vs. naproxen sub-study were H. pylor positive by serology. Subjects with diabetes, hypertension, and hyperlipidemia were balanced across groups. 17.5% of patients receiving non-selective NSAIDs discontinued the study based on adverse events (most commonly dyspepsia and other GI symptoms) compared to 15.4% of patients receiving lumiracoxib.
A total of 112 ulcer complications were included in the 52 week analysis:
| Lumiracoxib vs. Naproxen | |||
| Number of GI events/patients exposed | HR (95% CI) | ||
| Lumir | Naprox | ||
| Non-aspirin population |
9/3549 (0.25%) |
36/3537 (1.02%) |
0.24(0.12-0.50) |
| Aspirin population |
10/1192 (0.84%) |
14/1193 (1.17%) |
0.92(0.27-3.20) |
| Lumiracoxib vs. Ibuprofen | |||
| Number of GI events/patients exposed | HR (95% CI) | ||
| Lumir | Ibu | ||
| Non-aspirin population |
5/3401 (0.15%) |
28/3431 (0.82%) |
0.17(0.07-0.45) |
| Aspirin population |
5/975 (0.51%) |
5/966 (0.52%) |
0.92(0.27-3.20) |
109 confirmed or probable vascular events were recorded, 67 in the lumiracoxib vs. naproxen sub-study and 42 in the lumiracoxib vs. ibuprofen sub-study.
Comparison of patients achieving CV endpoint according to treatment and use of low-dose aspirin
| Lumiracoxib vs. Naproxen + Ibuprofen | ||||
| Number achieving CV endpoint (%) | HR (95% CI) | p value | ||
| Lumir | NSAID | |||
| Non-aspirin population |
Total = 35 (0.50%) MI = 14 Stroke = 12 Unstable angina = 5 TIA = 3 |
Total = 27 (0.39%) MI = 9 Stroke = 8 Unstable angina = 5 TIA = 5 |
1.22(0.74-2.02) | 0.43 |
| Aspirin population |
Total = 24 (1.11%) MI = 9 Stroke = 11 Unstable angina = 5 TIA = 4 |
Total = 23 (1.07%) MI = 8 Stroke = 9 Unstable angina = 6 TIA = 1 |
1.04(0.59-1.84) | 0.89 |
Subjects receiving low-dose aspirin had a higher number of composite vascular events across all groups. Incident stroke, cardiovascular death, DVT, PE, and congestive heart failure did not differ significantly between lumiracoxib treated subjects compared to those treated with non-selective NSAIDs, either collectively, separated by sub-study, or separated by treatment with low-dose aspirin. For myocardial infarction in particular, no significant difference was detected between lumiracoxib and the combined non-selective NSAID groups.
Non-selective NSAIDs were associated with significantly higher mean changes in systolic and diastolic blood pressure compared to lumiracoxib (systolic: +2.1 mm Hg vs. +0.4 mm Hg respectively (p<0.0001), diastolic: +0.5 mm Hg vs. -0.1 mm Hg respectively (p<0.0001)).
CONCLUSIONS: Lumiracoxib is associated with fewer serious GI complications than those encountered with either ibuprofen or naprosyn. This effect was reduced if concomitant low-dose aspirin was used. The proportion of cardiovascular events encountered in subjects treated with lumiracoxib was statistically similar to those encountered in subjects receiving naproxen or ibuprofen and did not differ according to concomitant therapy with low-dose aspirin.
EDITORIAL COMMENTS: This is the first published study of COX-2 specific inhibitors in arthritis patients to prospectively evaluate cardiovascular adverse events (with prespecified definitions) along with gastrointestinal adverse events. Because the frequency of these adverse events is small, a huge study (18,325 patients randomized!) was necessary.
On the surface, lumiracoxib appears to significantly reduce GI ulcer complications while not increasing risk for cardiovascular events, compared to conventional NSAIDs. However, it should be noted that patients at highest risk for GI events were excluded from the study (prior GI bleeds, recent peptic ulcers, and those already taking GI protective medications such as PPIs). Therefore, the relative risk of serious GI complications with all three medications is likely to be underestimated with regard to the general population. Furthermore, as in prior studies of COX-2 specific inhibitors, the addition of low dose aspirin negated the GI benefit of lumiricoxib compared to conventional NSAIDs. Therefore, patients requiring low dose aspirin are not likely to receive any GI benefit from lumiracoxib compared to conventional NSAIDs.
Regarding the CV adverse event data, there was a slightly higher frequency of CV events in the lumiracoxib, compared to conventional NSAID, group but this was not statistically significant. It should be pointed out that, in the recent Vioxx APPROVe study, significant cardiovascular risk was not identified until after 18 months of therapy. In light of this, a final interpretation of cardiac risk associated with lumiracoxib cannot be made until studies with treatment periods of 2 or 3 years are available. It is also important to note that patients treated with low dose aspirin had a > 2-fold higher incidence of CV events than non-aspirin treated patients, whether they were being treated with lumiracoxib or conventional NSAIDs. This increase may reflect inability of aspirin to convey cardioprotection against both COX-2 specific and nonselective NSAIDs. Alternatively, low dose aspirin may be acting as a surrogate marker for patients with higher CV risk who would have developed CV events regardless of COX-2/NSAID treatment. Only a controlled trial with randomization to aspirin or placebo in patients without CV risk factors could distinguish between these two possibilities.
