Comments

COMMENTS

LIMITATIONS OF SPONTANEOUS REPORTS SYSTEM

The post-marketing safety assessment of COX-2 inhibitors and market withdrawal of rofecoxib provided important lessons but left some unanswered questions as of the writing of this chapter. Perhaps the most important lesson is the limitation of spontaneous adverse drug reactions reporting system in the detec-tion of safety signals with a high background rate in a population using a drug of interest. In the US where millions of patients have used celecoxib or rofecoxib in 1999 and 2000, there has been no cardiovascular safety signal identified in the Adverse Event Report-ing System. The safety signal report from the Nether-lands was not widely publicized in the US. If there was no VIGOR or APPROVe, the cardiovascular risk of rofecoxib might not be recognized until much later. It is understandable that the traditional spontaneous reporting system did not detect the cardiovascular safety signal of rofecoxib. For rare outcomes that have been previously reported as drug-induced adverse events, such as Stevens–Johnson syndrome, liver fail-ure, or agranulocytosis, the prescribing physician’s level of suspicion may be high and the adverse event is more likely to be reported. For an adverse event like acute myocardial infarction in which the background rate is not rare and risk factors are well characterized, the prescribing physician may not readily attribute the myocardial infarction in a patient to the rofecoxib that the patient was using. For example, an overweight 59-year-old male smoker who had poorly controlled blood pressure and serum cholesterol started rofecoxib for his knee pain and then developed acute myocardial infarction in early 2000. Results from VIGOR were not yet available, the event could be explained by the patient’s existing cardiovascular risk factors (smok-ing, hypertension, and hypercholesterolemia), and the event would not be reported as an adverse drug reaction. This example illustrates the impor-tance of additional safety signal detection scheme to complement the existing spontaneous reporting system.

PRE-MARKETING AND POST-MARKETING TRIALS

According to current regulatory requirement, clini-cal trials of new NSAIDs like the COX-2 inhibitors only need to demonstrate short-term efficacy and safety. The study populations are usually relatively healthy and free from major comorbidity. However, once the drug is on the market, it is used in patients with a wide range of chronic diseases and concomi-tant medications and the new drug is used for peri-ods much longer than the study period of the pre-marketing trials. In all observational studies of COX-2 inhibitors that evaluated baseline comorbidity of study subjects, a substantial proportion had cardiovascular risk factors at baseline. Pooling data from multiple clinical trials (Konstam et al., 2001; Weir et al., 2003; White et al., 2003; White et al., 2004; Matchaba et al., 2005) to increase statistical power to evaluate risk of rare events is an important tool in safety assess-ment, but it does not address the issue of limited trial duration and non-generalizability to patients with cardiovascular comorbidity and concomitant medica-tions. Moreover, not all relevant safety information is included in published reports. Zhang and colleagues identified 502 reports involving COX-2 inhibitors and 331 had no event data on the occurrence of arrhyth-mia or renal complications (Zhang, Ding and song, 2006).

Sample sizes of the post-marketing trials of cele-coxib (CLASS), rofecoxib (VIGOR), and lumira-coxib (TARGET) were much larger than that of the pre-marketing trials and had larger statistical power to evaluate less common adverse events. Even so, they were not powered to precisely esti-mate relative risk associated with serious cardio-vascular outcomes. Moreover, low dose aspirin was allowed in only two of the three trials and provided limited information on potential interaction between COX-2 inhibitors and aspirin on gastrointestinal and

cardiovascular outcomes. Placebo-controlled trials of COX-2 inhibitors and non-selective NSAIDs would provide the most compelling evidence on the safety of these drugs, but these trials are ethically infeasible. For the active-control trials, long-term cardiovascu-lar safety of the commonly used comparator drugs, ibuprofen, naproxen, and diclofenac, has not been evaluated in clinical trials. Placebo-controlled results would have to come from study populations who did not require NSAID therapy, and APPROVe, APC, and PreSAP were such studies which demonstrated the increased cardiovascular risk among users of rofe-coxib and celecoxib. Not surprisingly, incidence of adverse cardiovascular events was much lower in these three trials than that observed among CLASS, VIGOR, and TARGET, raising questions about the generalizability of these results to patients who need NSAIDs. Trials need to be conducted among patients with coronary heart disease or cardiovascular risk factors to provide the most valid and generalizable answer to address the cardiovascular safety ques-tions of the COX-2 inhibitors. The Multinational Etoricoxib and Diclofenac Arthritis Long-Term program sponsored by the manufacturer of etori-coxib (Merck News Release, 2006) and the Prospec-tive Randomized Evaluation of Celecoxib Integrated Safety vs. Ibuprofen or Naproxen sponsored by the manufacturer of celecoxib (Cleveland Clinic Press Release, 2005) will provide more definitive answers.

Lastly, while the analgesic and anti-inflammatory effects of the non-selective NSAIDs may be similar at optimal doses, their cardiovascular safety profiles may not be the same. The meta-analysis of clinical trials involving NSAIDs by Kearney and colleagues and the meta-analysis by McGettigan and colleagues clearly indicated that the NSAIDs are not the same with regards to adverse cardiovascular effects (Kearney et al., 2006; McGettigan and Henry, 2006). This heterogeneity of cardiovascular effects has major implications in the selection of comparison groups in large safety trials.

THE ROLE OF OBSERVATIONAL STUDIES

As large-scale clinical trials are costly and time-consuming, evaluation of cardiovascular safety of COX-2 inhibitors with existing automated data has been an efficient way to provide important safety information on a timely basis. Thirteen of the fifteen studies summarized in Table 47.2 are based on auto-mated data sources, further demonstrating the util-ity of these data systems in rapid response to drug safety signals. The observational studies on the COX-2 inhibitors did suggest increased risk of serious cardiovascular thrombotic events among rofecoxib users, especially at dosages higher than 25 mg per day. However, the study designs for these reports were not the same, the comparator drugs were differ-ent, and important confounders, including smoking, body mass index, and use of non-prescription low dose aspirin, were not accounted for in the analysis in several studies.

RISK–BENEFIT ASSESSMENT OF THE COX-2 INHIBITORS AND NSAIDS

The therapeutic role of COX-2 inhibitors needs to be interpreted in the context of the risk–benefit profiles of the agents. Difficulties experienced by regula-tors are discussed by a senior FDA officer (Kweder, 2004) and by the director of the Uppsala Monitor-ing Center (Edwards, 2005). For both the COX-2 inhibitors and non-selective NSAIDs, their princi-pal anticipated beneficial effects are the analgesic and anti-inflammatory effects, and no single agent or class of agent has been shown to have superior efficacy than others. The risks may involve multiple organ systems and are not restricted to the gastroin-testinal and cardiovascular systems. Liver, renal, cutaneous, and hematologic toxicities are impor-tant issues to consider in the risk–benefit calculus. While the COX-2 inhibitors are associated with less gastrointestinal complications than selected NSAIDs for patients not taking aspirin, how they compare against the combination of NSAID and a proton pump inhibitor or an H₂ blocker or misoprostol is not known.

Another factor that may affect the risk–benefit profile of NSAIDs and COX-2 inhibitors is the devel-opment of new indication. Celecoxib has already been shown to decrease the development of rectal polyp among patients with familial polyposis and it has been shown to decrease the recurrence of colorectal adenoma among those who had a history of adenoma removal (Arber et al., 2006; Bertag-nolli et al., 2006). The efficacy results of APPROVe will provide more information on the potential use of COX-2 inhibitors in the prevention of colorectal adenoma.

In addition to overall risk–benefit assessment, regu-lators and clinicians need to carry out the assessment among subgroups of patients defined by specific risk factors, including those for gastrointestinal compli-cations and cardiovascular disease. For example, the risk–benefit calculus for a 70-year-old over-weight man who has osteoarthritis, coronary heart disease, and prior history of gastric perforation is very different from that for a 35-year-old woman who has no history of heart disease or gastroin-testinal complications and needs pain medication for rheumatoid arthritis. Another issue that has not been adequately addressed in the large COX-2 inhibitor trials is the effect of duration of treatment, which has major clinical implications. More systematic synthe-sis of data and quantitative risk–benefit assessment for the non-selective NSAIDs and COX-2 inhibitors are needed.