TMT Review: Cardiovascular Safety of Celebrex
The review below was prepared by Taylor MicroTechnology, Inc. and was posted to www.masterdocs.com on February 5, 2005. More complete information was submitted to the FDA Division of Dockets Management on February 7, 2005, prior to the COX-2 Advisory Committee meeting (http://www.fda.gov/ohrms/dockets/dockets/04n0559/04n-0559-c000007-01-vol2.pdf, www.fda.gov/ohrms/dockets/ dockets/04n0559/04N-0559_emc-000002.txt and www.fda.gov/ohrms/dockets/dockets/ 04n0559/04N-0559_emc-000002-01.pdf).
As with most chronic-use drugs, most chronic pain drugs have not been subjected to the type of study (long-term, controlled evaluation of cardiovascular endpoints) necessary for adequate assessment of cardiovascular safety.
An exception is aspirin which has been conclusively shown to reduce cardiovascular events. However, aspirin also increases the risk of GI hemorrhage, so that for many patients aspirin therapy is not suitable for the treatment of chronic pain.
Since the introduction of selective COX-2 inhibitors, long-term controlled studies evaluating cardiovascular safety have been performed and there is now evidence that some COX-2 inhibitors as well as the conventional NSAID naproxen may increase cardiovascular risk (particularly heart attack, stroke and other thrombogenic events). Insufficient cardiovascular safety data are available for most NSAIDs. Note added 2/20/05: The above comment on naproxen was based on a December 2004 press release from the NIH on the ADAPT trial in Alzheimer's Disease which "indicated an apparent increase in cardiovascular and cerebrovascular events" among naproxen patients versus patients receiving a placebo. At the COX-2 Advisory Committee meeting in February 2005, this assertion was withdrawn and it was stated that the trial was discontinued because of the "practicalities" of continuing the trial and that "some post-hoc data composites barely reached [statistical] significance" for naproxen.
It is not yet clear whether increased cardiovascular risk is a class effect of COX-2 inhibitors, a class effect of non-aspirin drugs for chronic pain, or a drug-specific effect confined to certain drugs in the COX-2 or non-COX-2 NSAID drug classes.
A unifying hypothesis based on the thromboxane/prostacyclin pathways has been proposed as a basis for a class effect of COX-2 inhibitors. However, this hypothesis has not been fully supported by the actual data.
Significant differences in molecular structure and effects on blood pressure have been shown between different COX-2 inhibitors and there is some evidence that COX-2 inhibitors that increase blood pressure may be associated with greater cardiovascular risk.
The limited evidence suggests that increasing doses of some COX-2 inhibitors are associated with progressive increase in cardiovascular risk.
The following course of action seems prudent at this time:
Drugs for chronic pain should continue to be made available to patients with chronic arthritis or other causes of chronic pain.
Aspirin and acetaminophen may be excellent choices for many patients.
Dosage and duration of therapy of any drug for chronic pain should be limited to that necessary for adequate pain relief.
Particular caution in drug selection is necessary when treating patients who also have increased cardiovascular risk or an increased risk of a GI bleed.
Of the currently marketed COX-2 inhibitors in the United States, celecoxib (Celebrex) appears to have the most extensive safety evaluation and the most favorable safety profile. No evidence of increased cardiovascular risk has been shown at dosage of 200 mg/day or less.
The maximum Celebrex dosage should be restricted to 200 mg/day until additional safety data become available.
Direct marketing of Celebrex to the consumer should not be resumed, and should be discouraged for medications in general.
Celebrex should continue to be one of the options available for the treatment of chronic pain.
In early February, 2005, Pfizer released a Briefing Book including a meta-analysis of safety data from Celebrex controlled trials (Pfizer Briefing Book). This large database of controlled studies does not appear to raise concern about the cardiovascular safety of Celebrex. However, the average duration of therapy in those receiving >200 mg/day was only 2 months and Celebrex dosage may not have exceeded 200 mg/day in most patients. Additional analyses should be provided to clarify the dose-response and time-response relationships for cardiovascular events. Indices of cardiovascular safety should include myocardial infarction and stroke event rates as well as the modified APTC composite index used in most of the analysis. The basis for the cardiovascular event rate definitions used should be given further clarification. It is not stated whether p values in the meta-analysis were based on one-sided or two-sided testing. To the extent possible from the data available to Pfizer, tabulations of cardiovascular safety should include both Pfizer-sponsored and non-Pfizer-sponsored studies.
Over 44,000 patients were included (about 25,000 on Celebrex). Only studies for which Pfizer possessed the databases were included in the meta-analysis. Accordingly, the NCI-sponsored APC trial for prevention of sporadic colorectal adenomas was not included. This study is discussed separately.
The main index of cardiovascular safety was “Serious Cardiovascular Thromboembolic Adverse Events” and this was based on a modification of the APTC composite value.
This index is defined in Table 2 of the Pfizer Briefing Book and includes 19 event types, some of which are debatable (e.g., all five peripheral vascular terms dealing with venous rather than arterial thrombosis, and inclusion of "cerebrovascular disorder" and "cerebral hemorrhage").
A footnote to Table 2 states that "Stroke comprised the individual adverse events cerebrovascular accident, cerebrovascular disorder, and cerebral hemorrhage." Including "cerebrovascular disorder" as a "stroke" is debatable, particularly in patients with Alzheimer's disease. In addition, it would be useful to have event rates separately for cerebrovascular accident and cerebral hemorrhage, since the thromboxane/prostacyclin hypothesis would suggest that thrombotic cerebrovascular events might be increased with Celebrex whereas cerebral hemorrhage could be reduced.
Assurance is required that inclusion of many terms in the APTC and stroke definitions did not dilute a true drug effect on events related purely to arterial thrombosis.
Separate Kaplan-Meier plots for myocardial infarction + stroke, myocardial infarction alone, and stroke alone would be useful.
It should also be stated whether the decision to use the modified APTC composite index was made before or after the safety results had been examined.
Planned duration of therapy was at least 2 weeks and patients received Celebrex therapy “for up to 1 year”.
The average duration of therapy and the numbers of patients receiving at least one year of therapy were not clear from the initial review of the report. This information should be provided, together with the cardiovascular event rates by therapy duration.
It was stated that 7462 patients were “exposed to celecoxib >200 mg TDD for 1268 patient-years” which indicates an average duration of therapy for this subset of 2.0 months.
It should be noted that the greatest cardiovascular risk with Vioxx was reported as having been seen with duration of therapy of 18 months or more.
Celebrex total daily dosage varied from 50-800 mg/day and “the predominant exposure to celecoxib was in the range of 200 to 400 mg TDD…” (TDD = Total Daily Dose). The average total daily dose and the numbers of patients receiving >200 mg/day and >400 mg/day were not clear from the initial review of the report. Data on cardiovascular safety as a function of dosage should be provided. (It should be noted that in the NCI APC polyp study (IQ4-99-02-005), a study not included in the Pfizer meta-analysis, a preliminary report found a statistically significant increase versus placebo in relative risk of a composite cardiovascular endpoint of 2.5 at the 200mg bid Celebrex dose and of 3.4 at the 400 mg bid dose.).
Celebrex, in comparison with conventional NSAIDs, was associated with a significant reduction in the risk of "stroke" (p<.001; relative risk 0.31), and non-significant differences for cardiovascular risk: relative risks (and p-values) for “Any Cardiovascular Thromboembolic”, “Any Myocardial Thromboembolic” and “Myocardial Infarction” were 0.88 (0.40), 1.31 (0.213) and 1.58 (0.096) respectively. Thus, there was a trend towards an increased risk of myocardial infarction versus conventional NSAIDs.
Celebrex, conventional NSAIDs and placebo were comparable with regard to overall risk of serious cardiovascular/thromboembolic/cerebrovascular events.
Percentages of patients with "cardiorenal adverse events" (serious or non-serious) were comparable with Celebrex and conventional NSAIDs but higher than with placebo. Three individual categories (hypertension/aggravated hypertension, edema/edema generalized/edema peripheral, and cardiac failure/cardiac failure left/cardiac failure right) were associated with increased risk with Celebrex and conventional NSAID therapy compared with placebo. This is “consistent with published reports in the medical literature” indicating that NSAIDs (conventional and COX-2 inhibitors) “can be associated with cardiorenal effects”.
The Pfizer Briefing Book discusses three key Celebrex studies (APC trial, ADAPT trial, and Study IQ5-97-02-001) that relate to cardiovascular safety:
A recent analysis of the APC trial for prevention of sporadic colorectal adenomas identified a “statistically significant increase in cardiovascular events for patients treated with celecoxib 200 mg bid or 400 mg bid compared to patients treated with placebo.”
There was evidence of a dose-response relationship for this effect (relative risk 2.5 for 200 mg bid and 3.4 for 400 mg bid).
It is probable that most patients receiving Celebrex have effective pain relief at total daily dosage of 200 mg or less, and the US labeling for osteoarthritis restricts maximum dosage to 200 mg/day.
Preliminary results from the ADAPT Alzheimer’s prevention trial comparing celecoxib 200 mg bid, naproxen 220 mg bid (a fairly low dose of naproxen) and placebo “indicate significantly increased risk for gastrointestinal bleeding and for cardiovascular and cerebrovascular events in patients treated with low dose naproxen compared to patients treated with placebo at 18 months, but no increase in risk for these events in patients treated with celecoxib compared to patients treated with placebo…”.
It has been suggested that a recent reanalysis of Celebrex study IQ5-97-02-001 in Alzheimer’s disease patients raises concerns about cardiovascular safety. However, the evidence is not at all convincing.
Dr. Sidney Wolfe of Public Citizen wrote a January 31, 2005 letter to FDA in which he described this study as “an unpublished randomized placebo-controlled study by Pfizer, finished more than four years ago, that showed a significantly increased rate (3.6-fold) of serious cardiovascular adverse events and more than a doubling in the rate of cardiovascular deaths in people using celecoxib compared to those using a placebo in a study concerning Alzheimer’s disease.”
However, detailed review shows that Dr. Wolfe’s analysis is clearly flawed.
The Pfizer study report is provided at http://www.clinicalstudyresults.org/documents/company-study_76_0.pdf as a PDF file apparently prepared on January 24, 2005 by a Manhattan company called Global Document Solutions. It is not clear who performed the statistical analysis and who wrote the report.
Dr. Wolfe apparently derived his conclusions of increased cardiovascular risk by selecting an arbitrary set of serious cardiac event types and stroke and adding up the numbers of occurrences on Celebrex and placebo. This subset of the data gave a count of 3 (2.1%) in the 140 placebo patients and 22 (7.7%) in the 285 Celebrex patients from which he concluded that “there was a statistically significant increase in the composite of all serious cardiovascular events in patients getting Celebrex compared to patients getting placebo”.
Dr Wolfe appears to have made the elementary error of summing the number of episodes in this subset, rather than the number of patients who had one or more of these types of event. As an example, there were 5 Celebrex patients with “Cardiac Failure” and 2 Celebrex patients with “Pulmonary Edema”. Pulmonary Edema is a type of Cardiac Failure and it is likely that this resulted in double counting of 2 patients who had Cardiac Failure as manifested by Pulmonary Edema. The limited data provided in the Pfizer report do not allow correction for this potential problem. Thus, Dr. Wolfe’s statistical analysis is not valid.
Dr. Wolfe also reported “There were two deaths (out of 140 patients) in which cardiovascular diagnoses were mentioned in the placebo group and nine deaths (out of 285 patients) in which cardiovascular diagnoses were mentioned in the group getting celecoxib. This also represents a statistically significant (p=.04, more than 2-fold) increase in the rate of cardiovascular deaths in people getting celecoxib compared to those getting a placebo (from table on page 7 of the Pfizer results).”
However, examination of the table on page 7 shows that Dr. Wolfe “cherry picked” the deaths in the table to make his point. There were 4 deaths on placebo (2.9%) and 13 deaths (4.6%) on Celebrex (with all deaths occurring between December 1997 and January 1999). Since the current concern about Cox-2 drugs is an increased risk of heart attack or stroke, this is the most relevant subset of deaths to examine – 3 deaths on Celebrex (1.1%) and 1 death on placebo (0.7%). The other “cardiovascular deaths” included by Dr. Wolfe were “cerebrovascular disorder” (rather than cerebrovascular accident, i.e., stroke), ruptured aortic aneurysm, pulmonary embolism, atrial fibrillation, subdural hematoma, and one patient in whom 5 causes of death were listed, the first two being “emphysema” and ”respiratory insufficiency” (with none of the other causes including heart attack or stroke). One could make an argument for including pulmonary embolism in this analysis (since it is normally a thrombus-induced event) but none of the other conditions have been linked to Cox-2 drugs and they should not have been included in Dr. Wolfe’s analysis.
The Pfizer report states “A statistically significant difference favoring placebo in adverse events was observed for certain CV-related body system terms (Cardiovascular Disorders, General; Heart Rate and Rhythm Disorders; Myo, Endo, Pericardial & Valve Disorders). These differences were primarily driven by the individual terms cardiac failure, fibrillation atrial, and angina pectoris.”
This statement is puzzling.
In any case, it is traditional to specify in the protocol the hypotheses for which statistical significance testing will be performed; if not, the results should be described as hypothesis generation rather than hypothesis testing, with appropriately conservative conclusions.
The Pfizer report also states that baseline imbalances existed between treatment groups for certain cardiovascular risk factors. This is an important observation but additional information on this is required and the analysis should be adjusted for the effects of these baseline imbalances.
It is also worth noting that two interim analyses were performed during this study, so that the statistical significance testing should also be adjusted for this multiple testing.
A safety monitoring board independent of Pfizer was responsible for assessing Celebrex safety in this study and apparently did not express concerns during or after the study.
It is TMT's view that there will be continuing concern about the safety of Celebrex until additional long-term controlled data are available. In the meantime, responsible physicians should not cherry pick from the available data so as to generate conclusions for or against the drug.
A TMT tabulation of key results from Pfizer Study IQ5-97-02-001 is shown below:
|
Pfizer Study IQ5-97-02-001 |
|
|
|
|
|
|
Numbers of Patients |
% of Patients |
||
|
|
Celebrex |
Placebo |
Celebrex |
Placebo |
|
Dose (mg/day) |
400 |
0 |
- |
- |
|
Number of Patients |
285 |
140 |
- |
- |
|
Any AE (Adverse Event) |
229 |
105 |
80.4 |
75.0 |
|
Discontinued because of AE |
34 |
14 |
11.9 |
10.0 |
|
Serious AE |
73 |
32 |
25.6 |
22.9 |
|
Death |
13 |
4 |
4.6 |
2.9 |
|
Death from Heart Attack or Stroke |
3 |
1 |
1.1 |
0.7 |
This document was prepared by Taylor MicroTechnology, Inc., 196 E 75th Street, New York, NY 10021. Questions or comments may be sent to info@masterdocs.com.
TMT Home Page Cardiovascular Safety of COX-2 Inhibitors Go to Top of Page