Indication: Analgesia, Dysmenorrhea Osteoarthritis, and Rheumatoid Arthritis
Reviewer: Kent Johnson, MD
Date: November 7, 2001
A. Approval for the indications of osteoarthritis and rheumatoid arthritis at a dose of 10mg/day and dysmenorrhea at a dose of 20-mg bid as needed.
B. Nonapproval of the acute pain, including opiod-sparing and prevention of operative pain. The only substantial multidose safety database is found in the Coronary Artery Bypass Graft (CABG) Surgery study 035. This study demonstrated an excess of serious adverse events including death in association with the use of paracoxib and valdecoxib 40 mg bid when added to ad lib parenteral narcotic analgesia. The allocation was 2:1 drug versus placebo and the population was highly enriched with patients at high risk for
cardiovascular thromboembolic events. Therefore, interpretation of these findings cannot be conclusive at this time. These finding warrants further investigation before valdecoxib can be considered safe and effective for the treatment of pain, particularly multidose therapy in the perioperative setting. The dose used in the CABG trial was eightfold higher than the dose proposed for approval for the treatment of osteoarthritis and rheumatoid arthritis and twice the dose proposed for the treatment of dysmenorrhea. In addition the proposed populations is distinctly different than the post-operative setting. The extensive safety database at 10-80mg daily in the arthritis safety database is adequate to support approval of the chronic therapy at 10 mg/day for arthritis and acute dose of 20 mg bid for short term use in dysmenorrhea.
2-SUMMARY OF CLINICAL FINDINGS
a) Adequate efficacy has been demonstrated in osteoarthritis and rheumatoid arthritis at 10mg/d with no additional efficacy at 20mg/d. The safety profile with chronic use in RA and OA is adequate at 10mg/d. At higher total daily doses, the findings of more hypertension and edema are frequently reproduced, and they are formally affirmed in a prospective manner in Trial 47, which directly tested the hypothesis of renal safety at 40 and 80 mg/day. In the analysis of older subpopulations over the age of 65 years edema and hypertension appear to be greater at 20 mg/day compared to 10 mg/ day.
b) Single-dose analgesia has been demonstrated at 20mg and 40mg in the dental, dysmenorrhea, with supportive data from other surgical models.
c) Two studies (024, 037) evaluating prevention or pre-emption of post-operative pain demonstrated the superiority of valdecoxib 20, 40 and 80 mg over placebo for the endpoints of time to rescue medication as well as proportion of patients taking rescue medication. There was no difference in pain intensity over the first 2 hours in study 024 and 4 hours in study 037. This finding raises concern over the value of preoperative management of post-operative pain, particularly in regards to the risk versus potential benefit.
Data from these two studies should not be considered for labeling until the overall clinical value of such treatment is further defined as well as the safety. Pre-operative dosing should be compared to post-operative dosing to adequately characterize the value of pre-operative treatment given the lack of differentiation in pain intensity between valdecoxib and placebo treated subjects.
d) Three studies of opioid sparing were submitted (Trials 35, 51, and 38). While mean opiod dose was lower in subjects treated with valdecoxib 40mg bid in the three studies, results were not replicated for 20-mg bid. Decreases in peak pain intensity were demonstrated in two of the three studies at 40 mg bid. This finding was not replicated for the 20-mg bid dose. Sparing of adverse events was not demonstrated in these studies.
In study 035 there was a statistically significant excess of serious adverse events associated with the use of valdecoxib 40-mg bid when added to ad lib narcotic therapy compared with narcotic analgesia alone. The value of "opioid sparing based on numeric differences in total opioid requirement is of unclear and unproven clinical benefit.
e) No efficacy advantage was demonstrated or suggested for valdecoxib compared to:
i. ibuprofen, naproxen and acetamenophen/oxycodone in anlagesic studies
ii. naproxen, ibuprofen or diclofenac in osteoarthritis studies
iii. naproxen in rheumatoid arthrits studies
3-OVERVIEW OF CLINICAL PROGRAM ANALGESIA:
This NDA consists of a program of analgesia trials to support a claim for acute pain, and a number of trials in osteoarthritis and rheumatoid arthritis to support a claim for chronic use in these conditions. The analgesia program tended to follow drug development programs for acute pain used in the past, relying heavily on single-dose demonstrations of efficacy compared to placebo and active controls, plus PK support demonstrating blood level stability over time and a satisfactory chronic risk/benefit from different indications (osteoarthritis and rheumatoid arthritis) to then extrapolate the safety for multiple-dose use in acute pain. The following is the sponsor’s request for claims:
An indication for the treatment of acute pain and dysmenorrhea at 40mg/d, with an
additional 40mg on day one if needed, and an indication for chronic treatment of the
signs and symptoms of osteoarthritis and rheumatoid arthritis at a dose of 10mg/day,
with the proviso that "some may receive additional benefit at 20mg/day."
It should be noted that there was the usual interaction with the sponsor regarding the scope and content of their development program. These interactions were more prescriptive in the case of OA and RA, as RA had been recently addressed in a Guidance Document, and the former had been the topic of a number of public meetings during which certain fundamentals such as trial duration, primary endpoints, and statistical methodology, were established. Thus, there was a priori agreement regarding data assessment in OA and RA, but the same cannot be said of analgesia. The agency, in collaboration with outside bodies, has been and remains in the process of formulating current analgesia guidelines, and, in particular, the nature of the evidence base needed to demonstrate efficacy in analgesia. A weakness in the approach used in the past is the extrapolation needed to assert multiple dose efficacy, rather than having data directly supporting this. In the past, this approach, although not ideal, was deemed acceptable given that agents were drugs which were
administered orally and usually showed identical dosing in both the analgesia and arthritis settings. Furthermore, pharmacokinetic parameters would suggest higher rather than lower levels on remedication in the acute multi-dose setting. In addition, in many to most acute pain settings, pain intensity typically diminishes rather than increase over time (suggesting that analgesia that is documented to be effective at the time of maximum pain would continue to be adequate as time passed.
An area where extrapolation cannot be made is in the assessment of dosing interval. Single dose efficacy data alone is less robust than comparative multi-dose data in assessing the optimal dosing interval. Although the division is exploring approaches which yields direct multiple-dose evidence and so depends less on extrapolation, the interactions for this NDA preceded this, so in this review the single-dose to multiple-dose extrapolation will be accepted. It is of note that supportive evidence of mulit-dose efficacy was submitted by the sponsor although studies were not designed to rigorously assess the multi-dose period.
The analgesia program consisted of nineteen trials -- seven dental, two dysmenorrrhea, and ten in various surgical settings. Only four were designed as multiple-dose trials. The other fifteen all were explicitly designed as single-dose, These trials could not provide multiple dose evidence of efficacy or dosing interval because: (1) The election of a very early time as the explicit, or de facto (as evidenced e.g. by its use in the powering of the trial) time-point for the primary analysis (ten trials using the 45 min PID for powering), and (2) the absence of re-dosing of short duration, active controls (ibuprofen or oxycodone/acetaminophen).
The dysmenorrhea trials were both 4-part crossover designs. Two surgery trials were designed to test the use of valdecoxib in a pre-emptive manner, given shortly before surgery. All trials were both placebo and active controlled except three which were designed to test a morphine-sparing hypothesis the pre-operative and the two pre-operative dosing studies. The three morphine-sparing trials allowed ad lib morphine use in both arms, so, in effect, they employed a "standard-of-care" as the control arm. The inclusion criteria varied widely across these designs, from patients undergoing the standard third molar extraction in the dental trials, to patients undergoing various modes of anaesthesia delivery (local, regional, spinal, general). This diversity has always been encouraged, as pertinent to any claim is a presumption of generalizability. In this NDA Trial 35 attempted to capture patients with substantial co-morbidity by enrolling patients who had undergone coronary artery bypass graft (CABG) surgery. This was an efficacy as well as a safety trial. The "COX-2 hypothesis" relates to organ specific safety; notably the upper gastrointestinal tract.
In discussions with the sponsor the division has emphasized the importance of rigorously testing the overall safety as well as upper gastrointestinal safety of valdecoxib. Given the evolving knowledge of selective COX-2 inhibition, this issue is of growing concern. This trial included a pre-defined basket of serious safety endpoints, called clinically relevant adverse events (CRAEs), which were to be formally adjudicated. In addition study 047 included renal safety endpoints in addition to asymptomatic endoscopically ascertained gastroduodenal ulcers as prespecified endpoints
The arthritis program consisted of early dose-ranging RCTs (Trials 15 and 16), followed by four standard efficacy trials (1 hip OA, 1 knee OA, and 2 RA), one active control, non-inferiority trial in OA (trial 63), and four formal safety trials – Trial 47 (OA/RA), 62 (RA), 48 (OA), and 53 (knee OA), all using a similar endoscopic ulcer primary endpoint, and one (47) also using a renal toxicity composite primary endpoint. These safety trials also collected validated efficacy endpoints, although not encompassing the full primary endpoint spectrum needed for formal efficacy evidence in OA or RA.
The analgesia trials were assessed by (1) the improvement in pain over time, (2) the time to the onset of analgesia, and (3) the time to need for re-dosing or rescue medication. All three of these should be substantially inter-correlated, so all were tested at the p<0.05 level, and no adjusting for multiplicity was done. However, this threefold endpoint approach was not appropriate for the morphine-sparing trials as they did not collect time to onset of analgesia, nor did they allow rescue medication. In these trials, two measures were used: (1) pain relief, and (2) morphine spared.
By the endpoints noted above, the following number of trials demonstrated efficacy (by either all three endpoints showing statistical significance at a p<0.05 or two endpoints, in the case of the morphine-sparing trials): 10mg/d – 4 trials (#5, 14, 35, 24), 20mg/d - 8 trials (#5, 14, 35, 58, 59, 11, 24, 37), 40mg/d – 6 trials (35, 58, 59, 72, 24, 37). So few comparisons to active controls were statistically significantly different, either superior or inferior, that this evidence base is not further considered. Using the criteria of replicated success in two of the three pain models – dental pain, dysmenorrhea, and post-surgical pain, and excluding
dosing at 80mg/d or 40mgbid – given evidence to suggest an unacceptable risk-benefit at these levels, the data support clear single-dose efficacy of 20mg, and 40mg. There was no replication of the efficacy of 10 mg based on pain intensity differences.
The clinical relevance of opioid-sparing was not adequately demonstrated. Pre-emptive administration of valdecoxib 20, 40 and 80 mg was associated with longer time to rescue medication compared with placebo in Trials 24 and 37 and number of patients who took rescue medication. Pre-emptive versus post operative dosing efficacy was not tested and peak pain intensity over 2 and 4 hours respectively in the two studies did not differ between placebo and active treatment arms. Therefore the benefit of pre-emptive treatment is not
clear, especially in view of the safety concerns in the post-operative setting.
The trials performed for the demonstration of efficacy in RA and OA were conventional and adequate in design. They included three formal efficacy trials in OA (two placebo control trials and one non-inferiority trial using only an active control, and two in RA, both placebo controlled. There were also safety RCTs with safety parameters as primary endpoints that also measured efficacy. These studies employed less standardized athritis efficacy endpoints such as patient and investigator global assessments and time to
dropout due to inefficacy.
The analysis of the efficacy results for RA and OA in this NDA were relatively straightforward. Valdecoxib did demonstrate efficacy at the 10mg and 20mg/d dosages in replicated data by usual comparisons with placebo arms, and there were no obvious threats (e.g. a differential dropout pattern) to the validity of these conclusions. Although no formal active control, non-inferiority evidence was pre-specified and pre-agreed upon in this NDA, this NDA, like others in the past, included numerous comparisons with active controls – and these were within the range of what has been seen with prior NDAs. There was no added benefit at 20mg/d, compared to 10mg/d.
Note: The review proper contains numerous adverse event tables which are supplied for reference, as the global safety experience of valdecoxib will likely bear critically on approval and labeling. Review comments are made in each section of these databases, but all relevant safety considerations are captured in the discussions of safety and risk / benefit here in the Executive Summary.
With two notable exceptions – edema and hypertension, valdecoxib was comparable to the standard non-steroidal agents used as active controls in the trials, except for some evidence supporting fewer GI adverse events, and some lessening of opiate side effects (e.g. constipation, dizziness, etc.) in trials with those as active controls. These findings will be reflected in the AE tables in the label. The finding of a greater incidence of edema and hypertension at doses above 20 mg/day, almost uniformly in the databases and clearly when prospectively addressed in formal safety Trials 47 and 62, is of concern, The relationship
between these events and the signal of more vascular events at 40mgbid dosing in the predisposed population of Trial 35 (CABG) is unclear. The excess of serious cardiovascular thromboembolic events in the valdecoxib arm of the CABG trial (see analgesic safety table #12) is of note as the entire study population received prophylactic low dose aspirin as part of the standard of care in this setting to minimize just such events. Given the emerging concern over a possible pro-thrombotic action of certain agents in the COX2 class, these data are of concern. These findings were seen at high dose in the peri-operative setting, not in the chronic safety studies of similar high doses.
SAFETY REREPORT ON TRIAL 35
This analgesia study comparing paracoxib/valdecoxib and placebo in patients undergoing CABG surgery was designed to test opioid-sparing, as reported in the Analgesia Efficacy section of this review. This trial was also specifically designed to test a safety hypothesis, using a pre-defined basket of safety endpoints, called "clinically relevant adverse events" (CRAEs), which included many serious vascular endpoints. The trial was powered using both a morphine sparing and a CRAE event rate calculation. In the trial analysis, there were 80 such events (25.7%) in the 311 paracoxib/valdecoxib patients, compared with 23 (15.2%) in the 151 placebo patients (p=0.012, by Fisher’s Exact). The patient numbers for the particular events are shown in Table 1 below.
Note: The reader is also referred to an in depth analysis of this important trial in the Parecoxib Medical Review by James Witter M.D. PhD. It is attached in toto in the appendix.
Table 1: Clinically Relevant Adverse Events (CRAEs): Prespecified Endpoint event placebo
deaths 0 4
myocardial infarction 1 1
cerebrovascular accident 1 9
deep venous thrombosis 0 3
pulmonary embolism 0 2
congestive heart failure 1 4
renal dysfunction / failure 7 29
infection 11 29
pulmonary complication 4 19
pericarditis 1 4
GI event 0 4
major non-GI bleed 2 0
Discussion: These data, along with the other analyses in Dr. Witter's review (appendix) are manifestations of an increase in vascular events rates, which coupled with the signals seen elsewhere in this database (for example, Trial 47 and the adverse event tables shown later in this review) all contributes to the concern that there may be a component of increased thrombogenicity associated with this agent.
A total of 22 deaths have been reported in the NDA and the 120-Day Update. Fifteen occurred during a blinded trial, five in open extensions, and two in an ongoing cancer pain trial (Trial 40) which remains blinded.
The total double-blind exposure for all doses of valdecoxib is 1283 patient-years (107, 323,397,316, 142 patient-years for 1-5mg, 10mg, 20mg, 40mg, and 80mg valdecoxib total daily dose, respectively) , compared to 291 patient-years for naproxen, 248 patient-years for diclofenac, 40 patient-years for ibuprofen, and 161 patient-years for placebo. Thus, the crude death rate in the unblinded controlled studies for valdecoxib is 0.9% (12/1283) compared to 0.52% (3/579) in comparator NSAIDs (p=NS, by Fisher’s Exact). Given the 2:1 (parecoxib/valdecoxib: placebo) randomization in the CABG trial the 4 deaths in that study may bias the rates. This study was in an enriched population for serious cardiovascular adverse events and used a dose not proposed for chronic use. The rates excluding this trial are 0.6% for valdecoxib compared to 0.5% for the NSAID comparators.
The rate of cardiovascular thromboembolic deaths ( including arrhythmia, MI and PVD) in the controlled database was 0.5% (6/1283) for valdecoxib and 0.3% (2/579) for the NSAID groups combined. Excluding the CABG study the rates for such events was 0.3% (4/1283) for valdecoxib and 0.3% for NSAID comparators. The number of events was small and there was no pattern seen based on dose or duration of therapy. Excluding the CABG trial there was no clear signal for differences in event rates between valdecoxib and comparator NSAIDs. A large outcome study employing chronic dose therapy would be needed to address this issue further. Such as study would include overall safety including cardiovascular, renal and GI endpoints as well as overall deaths and serious adverse events.
Requested Cardiovascular Safety Analysis in High Risk Patients
Concerns have been raised regarding COX-2 selective agents and cardiovascular safety following outcome study (VIGOR) of one such agent. Based on these concerns a subanalysis of cardiovascular events in high risk patients was requested by the reviewer. The following tables do not suggest a higher risk of cardiovascular events in an enriched population of "high risk" or "at risk" patients for valdecoxib compared to the NSAID comparators. The small number of patients exposed precludes robust comparisons.
HIGH RISK PATIENTS*: Rates of Serious Thromboembolic Cardiovascular