Your Vioxx Lawyer - Vioxx, Celebrex & Bextra FDA Transcript

Your Vioxx lawyer provides you the complete transcript of the February 16^th, 2005 joint meeting of the FDA's Arthritis Advisory Committee and the Drug Safety and Risk Management Advisory Committee. We have formatted the complete transcript of the three day conference for easy of navigation to provide you with the best possible vioxx information. To contact a Vioxx Lawyer, click here for a FREE case evaluation.

Cardiovascular Safety and the Risk/Benefit Assessment of Celecoxib

DR. VERBURG: Thank you, Dr. Feczko. Good afternoon, everyone. Again, my name is Ken Verburg. I lead the clinical research and development programs in arthritis and related conditions for Pfizer. In this respect, I have been studying celecoxib, valdecoxib and parecoxib for nearly eight years now.

My presentation over the next 40 minutes or so is focused to the cardiovascular safety of celecoxib and a risk/benefit assessment of this compound. I am joined here today by several of my Pfizer colleagues, as well as external experts in the field of cardiology, gastroenterology, rheumatology, epidemiology and other disciplines as listed on this slide. I will not spend the time to read each of the individually but they are here to contribute to the discussion afterwards.

So, what is Pfizer's position with respect to the cardiovascular safety of celecoxib and the risk/benefit of this compound? Our position is perhaps best summed on this slide in terms of conclusions. First, there are few therapeutic alternatives for patients with chronic arthritis pain. Patients who discontinue celecoxib then will likely turn to NSAIDs for treatment.

In our view, celecoxib is an effective and safe therapy for arthritis patients, and we base that on the following conclusions: First, celecoxib provides improved GI safety compared to NSAIDs. Secondly, all lines of evidence show that the cardiovascular safety of celecoxib is similar to NSAIDs for up to one year.

The caveat on this is that beyond one year little is known for any of these agents, and evidence for coxib versus NSAID class effect on cardiovascular safety is not established. Thirdly, rofecoxib appears to be distinct celecoxib and NSAIDs with respect to cardiovascular safety. Finally, only further study of NSAIDs and coxibs would define the longer-term cardiovascular risks against the known risks of GI ulcer complications.

I want to begin my discussion by going back and framing it in terms of the patients who require these therapies. In 2002, it was estimated that 1/3 adults suffer from arthritis or other related joint conditions, and that is an estimated 70 million individuals. Of these, about 7 million have significant impact on their daily activities.

Here are shown some data from the Centers for Disease Control, indicating that arthritis and other related conditions, joint conditions, results in significant functional impairment as compared to other diseases. About 39 million physician visits per year occur with arthritis patients and there are more than 500,000 hospitalizations due to arthritis each year.

NSAIDs are an important treatment option for arthritis patients. The American College of Rheumatology and other professional societies have indicated that first-line therapy is acetaminophen. That is perhaps an appropriate choice. But acetaminophen in many patients with moderate to severe forms of the disease does not provide adequate control of pain and other symptoms.

The data on this slide are from a double-blind, randomized, cross-over study in which one group of patients was randomized to receive first acetaminophen therapy for 6 weeks, followed by a washout period, and then to receive diclofenac in combination with a gastroprotective agent, misoprostol. In comparison, the second group was randomized first to receive a diclofenac-misoprostol combination, then following a washout period, was to receive acetaminophen for the subsequent 6 weeks.

Very quickly, the point that I want to make on this slide is that diclofenac offers significant improvements in terms of the Womack Target Joint Score, which is a composite score of pain, joint stiffness and physical function, as compared to acetaminophen at a total daily dose of 4000 mg, so a full dose of acetaminophen.

We have known for over two decades now that the efficacy of NSAIDs comes at a cost, and that cost is the risk of upper GI ulcer complications, that is, ulcers causing GI bleeding, perforation or leading to gastric outlet obstruction.

Largely, this risk was identified and characterized by pharmacoepidemiology studies. One such study is shown on this slide. Here we are showing the absolute incidence in terms of events per 1000 patient-years, the incidence of hospitalizations for GI bleeding or for perforations. Current users of NSAIDs are shown in the yellow line, subdivided by men and women, and they are compared to non-users of NSAIDs, shown in the white line.

What is readily apparent is that the absolute risk of hospitalization for GI bleeding or perforations increases substantially as a function of age. However, for each 5-year interval of age we see that NSAIDs increase the risk over non-users by approximately 4- to 6-fold. Of course, arthritis patients lie over to the far right-hand portion of this curve. This is the same population that is often characterized by cardiovascular risk factors or underlying cardiovascular disease.

Well, the discovery of the consistent enzyme and the characterization of the resulting biology around this enzyme led to the hypothesis, in 1992, that inhibition of COX-2 selectively would offer efficacy in the disease targets of arthritis and pain while obviating the side effects associated with the inhibition of COX-1.

Indeed, the discovery of celecoxib and the subsequent clinical development program supported that hypothesis. Here we show data from a trial of over 1000 rheumatoid arthritis patients in which efficacy was evidenced at 100 mg, 200 mg and 400 mg twice daily of celecoxib. What we see relative to placebo is that all of these doses provided significant efficacy in terms of the ACR-20 Responder Index. This efficacy was comparable to that observed with naproxen at a full therapeutic dose of 500 mg twice daily. The treatment period was 12 weeks in duration.

So, if we focus now on the right-hand panel of the slide, we are looking at the incidence over 12 weeks of endoscopic ulcers. What we see is that celecoxib at full therapeutic doses and a super-therapeutic dose was associated with similar incidences of endoscopic ulcers as compared to placebo treatment. This is in contrast to the naproxen treatment group which had an incidence rate of 25 percent and was significantly different than all other treatment groups. Thus, 1/4 patients treated in this trial over 12 weeks with naproxen was found to have an endoscopic ulcer.

This is data from a meta-analysis that will be published this month. This meta-analysis was based on 31 arthritis randomized controlled trials of Celebrex and included over 39,000 patients with osteoarthritis and rheumatoid arthritis, with a mean exposure of 7 months. The GI safety benefit is split into 3 different looks. The first is symptomatic ulcers and GI bleeding. The second is clinically significant blood loss, defined as reductions in hemoglobin of 2 gm/dL or more. Then also we focused on withdrawal due to GI intolerance.

As compared to NSAIDs, which are comprised basically here of naproxen, ibuprofen and diclofenac, we see that the relative risk for any of these events favors celecoxib, significantly so. This occurs at both the therapeutic doses of 200-400 mg or at any dose of celecoxib.

The data from the randomized, controlled trials has been further substantiated by observational epidemiology studies, and I will show data from two of these studies.

the first study that was published in 2002 evaluated the risk of hospitalization for upper GI bleeding with celecoxib, rofecoxib, the combination of diclofenac and misoprostol and NSAIDs. The point estimate of relative risk for the NSAID treatment group as compared to the non-users was 4. That relative risk agrees very well with a large body of literature evaluating NSAIDs in the epidemiology or observational setting. Celecoxib was similar to non-users in terms of the relative risk of hospitalization for upper GI bleeding.

The data from the first study was basically confirmed in the second study. In this case, the risk of hospitalization for GI bleeding was evaluated in patients with prior gastrointestinal disease who would be at high risk for subsequent GI ulceration.

Again focusing your attention over here, to the right, NSAIDs were associated with a relative risk of hospitalization for upper GI bleeding of a little over 3, significantly different from non-users. Celecoxib users had a similar risk of hospitalization as compared to non-users.

To sum then our conclusions regarding the safety benefit of celecoxib are stated as follows: The medical need for improved GI safety is fulfilled standard celecoxib. This is based on evidence from randomized, controlled trials in which celecoxib has a favorable GI safety profile versus NSAIDs, and also from emerging data from epidemiology studies which indicates that celecoxib is associated with a lower risk of hospitalization due to GI bleeding than non-selective NSAIDs.

So, the results that I just reviewed basically supported the fundamental hypothesis put forward in 1992 regarding selective COX-2 inhibitors. What emerged at the same time, however, was a concern over cardiovascular safety, and the first clinical evidence for an increased cardiovascular risk with a selective COX-2 inhibitor was observed with rofecoxib 50 mg once daily versus naproxen in the VIGOR trial.

At the same time, however, data emerged from the CLASS trial with celecoxib at 400 mg twice daily, 2-4 times the full therapeutic dose, demonstrating that the cardiovascular safety profile of celecoxib was no different than the NSAIDs diclofenac and ibuprofen combined introduction he CLASS trial.

For the remainder of my presentation this afternoon, what I would like to do is focus on cardiovascular safety using the following organization, and then conclude with some comments on risk/benefit.

So, let's begin with the longer-term studies evaluating celecoxib and its cardiovascular safety profile versus placebo treatment. Although we have been through this several times already, it stands to reason that we should spend a moment to define some of the fundamentals of the cardiovascular event definitions as we go through them.

So first, as we have heard already today, the APTC endpoint is a well-recognized endpoint with respect to the evaluation of cardiovascular therapeutics. It is comprised of non-fatal myocardial infarctions, non-fatal strokes or vascular deaths, as outlined on the slide.

The meta-analysis results that I will provide or review shortly have a similar construct to the APTC but they are based on investigator reports of serious adverse events to the company. In other words, there was not a process of adjudication and, of course, unlike cardiovascular endpoint trials, there were no definitions a priori about what a cardiovascular event would or would not be in order categorized appropriately.

Finally, we need to recognize that epidemiology studies rely on hospitalization for acute MI alone as their endpoint or in combination with coronary death predominantly.

So in collaboration with the National Cancer Institute, Pfizer and the NCI initiated two three-year placebo-controlled trials, beginning in 1999 or so, evaluating the effect of celecoxib on the prevention of sporadic adenomas. These trials are known as the APC and the PreSAP trials. The hypothesis being tested in these trials was that celecoxib would reduce polyp recurrence by greater than 35 percent in a high risk cohort, that is, patients who had a history of a prior adenoma.

Importantly, the setting allowed for the first longer-term comparison of celecoxib versus placebo. Trials of similar duration would be very difficult, if not impossible, to do in an arthritis population. Also, celecoxib was an obvious agent of choice here based on the emerging data demonstrating that it had superior GI safety to non-selective NSAIDs.

Dr. Ernie Hawk and Dr. Bernard Levin will review the results of these studies separately and go through them in some detail later this afternoon. To sum the results though, there was a significant cardiovascular risk associated with celecoxib in the APC trial and no such risk was observed with celecoxib in the PreSAP trial.

As we go through the studies and the data sets, it is useful and perhaps instructive to keep a score card of some of the study descriptions, as well as the patient populations because, as we have heard, none of these trials a priori were conducted to evaluate cardiovascular safety. That was not their primary objective. Thus, the types of patients, the durations entered into these trials can vary substantially.

So, beginning with the APC and PreSAP trial, you can see that over 1500 patients were enrolled in each of these trials. At the time study drug administration was discontinued there was about 2.5 years of exposure. And, the number of cardiovascular events, and these are APTC events, was 41 in the APC trial and 31 in the PreSAP trial. The mean age was about 60 and these patients were characterized by a fairly significant degree of underlying cardiovascular risk factors or cardiovascular disease. What is interesting is that there seems to be a little bit of a difference in the use of concomitant aspirin between the two trials, nearly twice as great in the APC trial as in the PreSAP trial.

Next, turning to a brief description of the Alzheimer's disease anti-inflammatory prevention trial, known as the ADAPT trial, this was a randomized, controlled trial. First of all, this trial was conducted and sponsored by the NIH. Pfizer's role in this study was to provide blinded study medication in the form of placebo and celecoxib only. This was a randomized, controlled trial evaluating celecoxib at 200 mg twice daily or naproxen at the over-the-counter dose of 220 mg twice daily versus placebo treatment. Elderly patients were enrolled in the trial, all greater than 70 years of age, who were at risk for Alzheimer's disease, that is, they had a first-degree relative with the disease.

Except for uncontrolled hypertension, there were not other restrictions for cardiovascular disease in order for patients to be eligible for the trial. The hypothesis being tested was that celecoxib would reduce the incidence of Alzheimer's disease by over 30 percent in a high risk cohort.

So, this represents the first longer-term placebo-controlled experience with a non-selective NSAID. Most of the results have not yet been disclosed into the public domain. What we do know though from the investigators is that naproxen was associated with a significant increase in all cardiovascular events, as well as all cerebrovascular events, versus placebo and no such effect was seen with celecoxib.

We also know from this trial, again in terms of preliminary information, that naproxen was associated with a significantly elevated increase of upper GI bleeding as compared to placebo treatment and, again, no significant difference was seen in celecoxib users.

Returning to our score card, in the ADAPT trial near 2500 patients were enrolled. The mean exposure at the time the trial was stopped was about 1.6 years per patient. The number of events reported and in the public domain appear to be about 70. Those are probably APTC endpoints-plus some. And, we know nothing about the underlying patient population at this point.

So, the conclusions that we can draw from the information so far, and as the week unfolds we will see if these conclusions hold, are as follows: In the APC trial celecoxib was associated with a cardiovascular risk as compared to placebo after approximately one year of continuous administration.

In the companion PreSAP trial no differences were observed with continuous treatment of celecoxib for up to 3 years of exposure.

In the ADAPT trial naproxen showed a cardiovascular risk compared to placebo over an exposure period of about 1.5 years, and this was in contrast to the findings with celecoxib.

When you distill this information down even at this point it is obvious that celecoxib requires further study to estimate the longer-term cardiovascular risk. An NSAID comparator in such a trial would be critical in our view.

Next let's turn to a comparison of celecoxib predominantly versus NSAIDs, and we will use a meta-analysis of the randomized controlled trials to do so. So, 41 completed randomized controlled trials and a total of over 44,000 treated patients were included in this meta-analysis. They were predominantly patients with osteoarthritis or rheumatoid arthritis. There was a small minority of patients with chronic low back pain, ankylosing spondylitis or Alzheimer's disease.

Of these patients, nearly 25,000 received celecoxib; 4000 receiving placebo; and over 15,000 patients were treated with an active comparator. We evaluated all doses of celecoxib, ranging from 50 mg to 800 mg daily. The primary NSAIDs in this comparison were naproxen, ibuprofen and diclofenac. The study durations ranged from 2 weeks to 1 year.

In terms of the comparisons to NSAIDs, the celecoxib exposure is shown down here, in the yellow box. Of the patients in the meta-analysis treated with celecoxib, 55 percent were treated for 3 months or longer; 12 percent were treated for 9 months or longer; and 4 percent, a total of 803 patients, were treated for 1 year or longer.

I too will review the results using the APTC-like construct, first reporting a composite endpoint of any cardiovascular death, non-fatal MI or non-fatal stroke, and then I will report the results of each of these components individually.

Back to our score card one more time, here now we are comparing and juxtaposing the study descriptions and the patient populations for the meta-analysis in comparison to the longer-term trials. So, we see that the number of patients is increased substantially in the comparisons of celecoxib to placebo or NSAIDs. We also take note of the fact that the mean exposure is much less, being on the order of only 6 weeks in the comparison of celecoxib to placebo, and on the order of about 3.5 months for celecoxib compared to the NSAIDs.

The number of events is fairly similar from placebo across to the other settings, about 2-3 time the number of events in the comparison of NSAID to celecoxib, meaning that each was approximately the same as the polyp prevention trials. And, there was a significant degree of underlying cardiovascular risk in this population but perhaps less so than in the polyp prevention trials.

So, first the results of the meta-analysis comparing celecoxib versus NSAIDs are shown on this slide. Here we are reporting the absolute number of events that occurred in each treatment group, and then the event rate in parentheses as events per 100 patient-years. What we have done here is we have taken all doses of celecoxib, 200 mg or greater, and combined them. So, we are looking at full therapeutic doses of celecoxib and super-therapeutic doses of celecoxib. In essence, this is 200 mg, 400 mg and 800 mg daily.

What we see here is that in terms of the composite event rate of cardiovascular death, non-fatal MI or non-fatal stroke there are no apparent differences between the two treatment groups, and that is generally true for cardiovascular death. There is an apparent increase in the event rate in terms of non-fatal MI in the celecoxib treatment group and a reduction in non-fatal stroke in the celecoxib treatment group when compared to the NSAID group.

If we evaluate the data in terms of the relative risk and 95 percent confidence intervals, the following results are evident: First, the relative risk for the composite endpoint is slightly below 1 favoring celecoxib, as was cardiovascular death and as was stroke. Non-fatal MI was slightly above 1 favoring NSAIDs. However, in all cases the 95 percent confidence interval for this comparison did not exclude 1 with the exception of non-fatal stroke in which the relative risk was nearly 3 times lower than in NSAID users.

If we now break this down a little bit further and evaluate the risk of celecoxib versus the NSAIDs that comprise the predominant exposure in the aggregate NSAID treatment group, i.e., naproxen, diclofenac and ibuprofen, we see no real pattern of difference in the comparison of celecoxib to these three drugs individually.

As a point of reference, I am also putting on this slide the comparison of celecoxib to placebo and you can see that the relative risk was 1.26 favoring placebo but not significantly different.

Next, if we subdivide the celecoxib 200 mg or greater treatment group into its component doses, again using the composite endpoint of cardiovascular death, non-fatal MI or stroke as the endpoint of interest here, we see no obvious dose-response relationship in the relative risk of celecoxib as compared to the NSAIDs.

Two of the trials in the meta-analysis included patients with substantial exposure to celecoxib. Those trials were the CAESAR trial and the CLASS trial. The CAESAR trial was a trial of a little over 800 patients with osteoarthritis who were treated with either celecoxib or diclofenac for a period of one year. The CLASS trial was a trial of nearly 8000 patients, with a median duration of exposure with respect to celecoxib of 9 months and 15 percent of patients treated with celecoxib were treated for one year or more.

Here we are showing the time to event analysis comparing celecoxib in these two trials and at doses ranging from 200 mg to 400 mg per day versus the NSAIDs used in the two trials, which were diclofenac and ibuprofen. We see no difference in the APTC composite endpoint between the two treatment groups through the exposure period.

So, our conclusions from the randomized, controlled trials are as follows: There is no association for increased cardiovascular risk detected with the use of celecoxib up to one year compared to the NSAIDs combined and also compared to naproxen, diclofenac or ibuprofen individually. And, a dose-related increase in cardiovascular risk with celecoxib was not apparent.

Next let's turn to a consideration of risk factors. We will stay within the construct of the meta-analysis. About 33,000 patients or so were available for this analysis with respect to cardiovascular risk factors and, again here we are going to be comparing celecoxib to the NSAIDs. The risk factors were based on either a history of hypertension, diabetes or hyperlipidemia or coronary heart disease as evidenced by a previous MI, a history of angina or other significant ischemia or revascularization procedure.

So, the patients with none of these risk factors are shown in the white bar; with one risk factor only are shown in the yellow bar; and with two or more risk factors are shown in the orange bar. Again, what we are showing is the absolute event rate in terms of events per 100 patient-years for celecoxib users over on the left, NSAID users over on the right, and here we show a breakdown by the composite endpoint and each of the components of the endpoint. What catches your eye is that as the patients are characterized with greater risk factors, the absolute event rates increase in both treatment groups.

Did they increase proportionally? This is the relative risk now comparing celecoxib versus NSAIDs by cardiovascular risk factors. So, no risk factors here; one risk factor; greater than two risk factors over to the far right. First a quick inspection across all three risk strata suggest that there were no significant differences in the relative risk between celecoxib and the NSAIDs, with the exception of non-fatal stroke in patients with no cardiovascular risk factors.

The second point I would like to make on this slide is that if you scan across to evaluate whether there were any patterns that were occurring that were altering the relative risk between the two treatment groups as a function of risk factors, that is not readily apparent when evaluating either the composite, cardiovascular death or MI. There is a trend for favorable comparison with stroke to become less favorable with additional risk factors, but I want to caution that this particular point estimate is based on a total of only 6 non-fatal strokes. So, as the confidence interval suggests, there is wide uncertainty around that point estimate.

If we use low dose aspirin as an indicator now for underlying cardiovascular disease, this slide shows the comparison of the relative risk for celecoxib versus NSAIDs again in the construct that we have shown before, and showing the relative risk in a non-aspirin cohort over here, on the left and the aspirin cohort, over here, on the right. Again, this is about 10-12 percent to the total population. Here we see that there are no significant differences in any of the endpoints of interest, with the exception of non-fatal stroke favoring celecoxib this time in the aspirin cohort. There is a noticeable change in the relative risk of cardiovascular death from the non-aspirin cohort to the aspirin cohort. But this again is shaped by very few events, as evidenced by the wide confidence intervals around the point estimate.

Next, if we return to the CLASS trial and evaluate the composite APTC endpoint by aspirin or non-aspirin use in terms of a time to event analysis, we see nonsignificant differences in the two time to event curves by log rank test in either the non-aspirin cohort or in the aspirin cohort. Approximately 22 percent of the patients in the CLASS trial were taking low dose aspirin.

So, our conclusions concerning risk factors are as follows: The cardiovascular safety profile of celecoxib remains comparable to NSAIDs regardless of cardiovascular risk factors as determined by medical history or use of low dose aspirin.

Next turning to epidemiology studies, 7 epidemiology studies have now been completed and reported as full-length publications in peer-reviewed journals, and have evaluated cardiovascular risk of celecoxib.

Returning to the concept of the score card but this time just evaluating the epidemiology studies in isolation, if you aggregate the studies together there was a total of over 30,000 myocardial infarctions in the studies for all treatment groups and over 1000 MIs in patients taking celecoxib. Both in terms of the number of events and in person-years, you can see that in terms of the total as well as for celecoxib users, these numbers are substantial in comparison to randomized, controlled trials.

I am going to review these studies individually very quickly, starting here with the study of Ray et al., published in 2002, in this study celecoxib at doses less than 300 mg daily or 300 mg or greater daily had a similar relative risk of hospitalization due to myocardial infarction or coronary death as compared to non-users. These results were very similar to the relative risk seen with ibuprofen and naproxen. In contrast, we see that the relative risk associated with doses of rofecoxib greater than 25 mg were significantly different than non-users.

The second study, published in 2003, showed that there was basically no difference between celecoxib, rofecoxib, naproxen or any of the other NSAIDs in terms of the relative risk of hospitalization for MI as compared to non-users.

In a third study, conducted by Solomon and co-workers at Harvard, found that celecoxib at all doses combined or subdivided into low and high doses was associated with a relative risk for hospitalization for myocardial infarction no different than non-users. In contrast, in this study as in the first study, rofecoxib at doses of 25 mg or greater was associated with a relative risk of 1,58, significantly different from non-users.

Next we turn to the study published by Kimmel and co-workers earlier this year. In this particular observational study celecoxib was associated with a significantly protective effect with respect to the relative risk of myocardial infarction, as were all NSAIDs combine. Rofecoxib was associated with no such effect when compared to non-users. Next we turn to perhaps the largest study conducted and published thus far. Here again we see, as in the previous trials that celecoxib was associated with a relative risk of myocardial infarction or coronary death, this time to remote use of NSAIDs, that is no different in terms of remote use. In contrast, here again we see that the relative risk associated with high doses of rofecoxib were significantly elevated. In terms of the NSAIDs, the point estimates of relative risk ranged from 1.06 for ibuprofen to 1.60 for diclofenac. None of those point estimates were different from remote users.

Next, turning to the results of Shaya at al., these investigators used an APTC-like endpoint in their observational trial. They found that celecoxib alone, rofecoxib alone or the coxibs combined were similar in terms of adjusted odds ratio to non-naproxen NSAIDs in their study.

Finally, the publication of Levesque et al. reports the following, they show that celecoxib, subdivided into low and high doses, was no similar to non-users in terms of the relative risk of MI. Here again we find the observation that rofecoxib at doses of 25 mg or greater was associated with a significantly elevated risk, and the NSAIDs were generally also similar to non-users.

These investigators also subdivided the evaluations of the compounds into non-aspirin users and aspirin users. We see that either in the non-aspirin cohort or the aspirin cohort the relative risk of celecoxib is similar to non-users. We find a significantly elevated risk in patients taking rofecoxib at doses of 25 mg or greater.

Trying now to sum all the previous observations into one trial, the results are shown here. They are subdivided by low doses and high doses of celecoxib and rofecoxib. We see that celecoxib at either low doses or high doses is similar to non-use with respect to risk of MI. In contrast, rofecoxib at high doses is systematically associated with an elevated relative risk of MI as compared to non-users.

So, our conclusions from that rapid review of epidemiology studies are that the risk of myocardial infarction with celecoxib as used in the real-world population, that is at the doses that patients actually take and the duration for which they actually take them, is consistently similar to non-selective NSAIDs in non- or remote use of NSAIDs. These findings are in contrast to the increased risk associated with rofecoxib. The available data suggests that the risk of MI is similar for low and high doses of celecoxib.

So, turning now to a consideration of the topic of mechanism, a unifying hypothesis that would attribute cardiovascular risk to the coxib class only conceivably could explain the VIGOR results, the APPROVe results with rofecoxib and the APC results with celecoxib. But it couldn't explain the consistent comparability between celecoxib and the NSAIDs as viewed in the meta-analysis or versus non-use in the epidemiology studies, and could not explain the lack of effect of celecoxib in the PreSAP trial or other results in the ADAPT trial where the non-selective naproxen was associated with increased cardiovascular risk, unlike celecoxib.

So, if we try to sum up all of the clinical observations to date, it would appear that the absolute cardiovascular risk associated with coxibs may be small in terms of the order of magnitude, but the risk may be different between compounds within the class, and that non-selective NSAIDs may carry the same risk. Therefore, that draws into question the clinical significance of the prostacyclin-thromboxane imbalance and its importance in leading to a prothrombotic state.

In support of the hypothesis, NSAIDs may not provide effective blockade of platelets even though thromboxane production is reduced throughout their entire dosing interval. This would be more or less a unifying hypothesis across both classes. It would unify the coxibs and the NSAIDs together if this was the case, to some degree.

Alternatively, what all these compounds have in common, which was discussed this morning, is that they inhibit COX-2. But by doing so, they not only inhibit the formation of prostacyclin but they also inhibit the formation of other prostaglandins that are formed by COX-2 activity and subsequent enzymatic activity.

So, the data on this slide go back to point one on the previous slide. Here we are showing the effect of a single dose of ibuprofen 800 mg on platelet aggregation responses, over on the left. What you can see here is that following ibuprofen administration in normal health volunteers there is a significant reduction in the platelet aggregation response to arachidate, but that this effect is largely eliminated by 8 hours and the platelet aggregation responses return to essential control levels. This occurs despite significant inhibition of thromboxane-A2.

Over on the right-hand panel is the time course of the urinary excretion of the major prostacyclin metabolite, and here we can see that ibuprofen comparably inhibited the urinary excretion of this metabolite to a degree comparable to that seen with celecoxib, and that the inhibition was significant even at the 6-12 hour time period.

As was mentioned this morning, this effect will vary from NSAID to NSAID, but using the example of ibuprofen, it is conceivable that, as again was mentioned this morning, mixed inhibitory, non-selective COX-1 and COX-2 inhibitors could act in terms of platelet function, in terms of vascular effects as selective COX-2 inhibitors during a portion of their dosing cycle.

Alternatively, COX-2, as I mentioned previously, in the vasculature has been linked to several cardiovascular disease states, and the up-regulation of COX-2 expression not only results in the production of prostacyclin, which would then lead to downstream beneficial effects on endothelial function and prevention of platelet aggregation, but has also been shown to increase the production of prostaglandin E2 which, again through a cascade, could result in reduction in plaque stability ultimately. Also, COX-2 could lead to the formation of thromboxane-A2 which would obviously have effects opposite of those to prostacyclin.

That particular scheme would suggest that the results of COX-2 inhibition might be more complicated than just focusing on prostacyclin, and also that the clinical outcomes of such effects might be more difficult to predict than we would envision.

If we move this consideration of mechanism a step further and ask the question, well, if that may be the case in the coxibs and the NSAIDs are all alike, then what may account for the differences that are seen with rofecoxib as compared to the other agents in terms of cardiovascular risk?

We should not forget in this conversation, and this was brought up this morning, that each of these compounds has unique pharmacology, pharmacologic activity, unique pharmacokinetics and other properties that could mitigate or worsen a cardiovascular risk profile that is embedded in a mechanism-based effect.

What we are showing on this slide is some of the recent work and so all these compounds may be characterized by that. Rofecoxib has been heavily studied in this respect, as has celecoxib. What has emerged from some of these studies is that rofecoxib and/or some of its metabolites may have pro-oxidant effect which could ultimately lead to increase in blood pressure or thrombosis. Do we know for sure that this is an overall effect of rofecoxib? No, but it is clear from the clinical literature that rofecoxib has been associated with elevations in blood pressure to a degree that are not seen with other agents, whether they be non-selective NSAIDs or celecoxib.

The most recent example of this is shown on this slide. This is the third of three studies now that basically confirm the same observations. So, at equal efficacious doses for osteoarthritis, that is, 200 mg of celecoxib once daily, 25 mg of rofecoxib once daily, or 500 mg twice daily of naproxen, we see that over both 6 and 13 weeks of therapy with these agents in patients with osteoarthritis and treated for hypertension, as determined by 24-hour ambulatory blood pressure monitoring, that rofecoxib was associated with a significant elevation in systolic blood pressure at both the 6- and the 12-week time point, and was significantly elevated as compared to the celecoxib and naproxen treatment groups.

So, we know from outcome studies that reductions in this order of magnitude in terms of systolic blood pressure can have a significant impact on cardiovascular mortality and morbidity.

In summary then, it is not established that the prostacyclin-thromboxane imbalance contributes tot he effects observed for coxibs or NSAIDs clinically. Furthermore, the underlying pharmacology is more complex, involving other prostaglandins and pathways and raises the potential even for benefit for COX-2 blockade. And, there is emerging evidence for molecule specific mechanisms.

Finally some brief concluding remarks on

333 risk/benefit of celecoxib as it currently stands, just by way of recap, celecoxib in comparison to the NSAIDs in terms of GI safety within the randomized, controlled trial setting has a lower incidence of clinically significant GI outcomes, and in epidemiology studies has similar risk of hospitalization for GI bleeding versus non-users.

Celecoxib versus NSAIDs in terms of cardiovascular safety--the randomized, controlled trials indicate that there is a comparable cardiovascular safety profile versus the alternative therapies. The epidemiology studies indicate that there is a similar cardiovascular risk in celecoxib users as compared to non-users.

In conclusion, the overall risk/benefit assessment of celecoxib is as follows: In the currently approved arthritis indications the risk/benefit of celecoxib remains positive relative to NSAIDs. There is comparable efficacy that demonstrates a GI safety benefit, and it has comparable cardiovascular risk based on the data that we have currently.

There is shared uncertainty of the cardiovascular safety beyond one year of continuous treatment for all of these therapies. Thus, further studies are planned by Pfizer to evaluate the longer-term GI and cardiovascular safety of celecoxib versus NSAIDs in arthritis patients. Thank you.

DR. WOOD: Thanks very much. Questions from the committee? Yes?

DR. DWORKIN Could you go back to the blood pressure slide, and do you have any data on what it would look like if you had 400 mg daily of celecoxib? I think you just showed 200 mg. I am sort of interested in 200 BID.

DR. VERBURG: We do not have a direct comparison of 200 mg BID celecoxib versus rofecoxib. We have done a 24-hour ambulatory blood pressure trial evaluating 200 mg BID of celecoxib relative to placebo, and we have found very minor differences in the blood pressure profile of celecoxib at that dose as compared to placebo. That is as close as I can come to that.

DR. WOOD: Curt?

DR. FURBERG: I think the focus of your presentation troubles me a bit. You really spent most of the time on comparative trials, and if you are really interested in safety comparing two active drugs is not the best way to go. You get much better information by looking at the placebo-controlled trials.

I think we are here to answer two questions, is Celebrex safe? And I think what you talked about is not going to help us answer that question. We need to look at the placebo-controlled trials.

You answered the second question, is the safety of Celebrex different from the NSAIDs? Let's come back to the placebo-controlled trials. There is information in the briefing document from Pfizer that you did not bring out, and I would like to refer people to table 4 which presents a summary on the Celebrex experience in placebo-controlled trials, and it is showing risk ratios of 1.7, 1.8 versus placebo for thromboembolic events--trends that are not too dissimilar to what we see in other placebo-controlled trials of the other COX-2s. I think in addition to that, you did not address at all the issue of heart failure that we talked about earlier. We were informed that in the APPROVe study there was a 4-fold increase in heart failure in that placebo-controlled trial. For Celebrex, if anything, it is worse. If you look at your table 6, although there are small numbers, there is a 6-fold increase in heart failure, statistically significant, and that is not mentioned.

So, if you are going to talk about safety, my plea is that let's look at all aspects of safety, including the thromboembolic events and heart failure, and let's pay a little bit more attention to the placebo-controlled trials because, as has been said over and over again, we really don't know the safety profile of the various non-selective NSAIDs, and to compare to those drugs is not very informative. Thanks.

DR. WOOD: Do you want to respond to that?

DR. VERBURG: I think the only way to respond to that is actually review some of the data. Why don't we take a look at a couple of the slides? So, why don't we go to slides C-112?

Going back to the meta-analysis, with the caveats that it is 11,000 patients and it is 6 weeks of exposure and it is roughly 31 events. So, we are shaping conclusions based on a very small data set over very small durations. The composite endpoint for placebo was 1.4 in terms of events per 100 patient-years as compared to 1.8 for celecoxib. In terms of cardiovascular death and MI, you can see that the results were lower with placebo and there was no difference in non-fatal stroke.

Indeed, if you plot these out in terms of relative risk, you find that the point estimate of relative risk for three of these endpoints favors placebo but the confidence intervals are fairly substantial, indicating very low precision around those points.

DR. WOOD: What was the exposure for that?

DR. VERBURG: Six weeks.

DR. WOOD: I think we should emphasize that. You missed that out. Just go back to the slide.

DR. VERBURG: This is 1.7 months of exposure.

DR. WOOD: As long as we have that on the record.

DR. FLEMING: But in essence, if you are doing a non-inferiority, if you want to show you are not worse, there is a major issue of you are not giving very long exposure here as to whether you might be really underestimating excess risk.

DR. VERBURG: In our view, that is why we did not focus on these data in the presentation. We felt it was really non-informative and we would really leave the discussion of placebo comparisons over longer term to Dr. Hawk and Dr. Levin when they present.

DR. WOOD: Well, why don't we put up the Kaplan-Meier curve from the trial, the APC trial?

DR. VERBURG: Again, I don't have those data.

DR. FURBERG: For any myocardial thromboembolic events the relative risk is 1.77. So, I don't know why you have that discrepancy. You have much lower relative risks in your slide than in the briefing document that was sent to the committee members.

DR. VERBURG: I probably should step back, it is a little bit different construct in my presentation than in the briefing book, and it was really based on our desire to get to what was a more meaningful endpoint and that was the APTC. I don't think that the differences between the analyses in any way change the overall conclusions.

DR. FURBERG: Well, we may disagree on that point. How about heart failure then?

DR. VERBURG: Sure. Let me just check my notes here. Can you bring up for me C-248? These are data that were provided in the briefing book I believe--

DR FURBERG: Correct.

DR. VERBURG: --comparing celecoxib to placebo in terms of reports of adverse events from investigators, not adjudicate, hypertension and peripheral edema and cardiac failure, and celecoxib is associated with a significant increased incidence of all these events, as are all non-selective NSAIDs.

Let's go to the next slide.

DR. WOOD: What duration?

DR. VERBURG: Same duration.

DR. WOOD: Six weeks treatment?

DR. VERBURG: A mean of six weeks of treatment.

DR. FURBERG: So, the patients didn't even have a chance to develop heart failure. You raised their blood pressure and caused fluid retention and you followed them for a few weeks. They didn't have a chance to get into heart failure.

DR. VERBURG: So, let's step back. What we are doing is we are trying to determine some cardiovascular safety parameters from trials that were designed to test fundamentally the efficacy of arthritis.

DR. FURBERG: Sure.

DR. VERBURG: So, again, we have recognized all of the faults in what we are doing. There is no getting around that. But if we want to see what the data look like in order to form some conclusions, this is what it looks like. We hear the criticism but, again, these are from NDA trial databases of 12-week, placebo-controlled trials to evaluate efficacy in arthritis. So, we are limited by the purpose of those trials.

DR. FURBERG: Yes, but these are trials that you designed and set up, and you are not providing the answers that we need to evaluate the efficacy and safety.

DR. WOOD: I don't understand the answer to the last question. You are telling us you don't have the data that you published in The New England Journal two days ago with you in this presentation of a placebo-controlled trial?

DR. VERBURG: I do not. That trial was conducted by the National Cancer Institute.

DR. WOOD: You are welcome to download a slide from The New England Journal. They have a web site that let's you do that.

DR. VERBURG: And we will cover that topic later. I just don't have a slide with that in my presentation.

DR. WOOD: Any other questions? Byron?

DR. CRYER: Yes, throughout your presentation you suggested that there may be cardiovascular risk, specifically thrombotic risk associated with non-selective NSAIDs. You suggested this mechanistically with ibuprofen and with naproxen based upon the ADAPT trials from observations.

My sense and my understanding of the literature is that there are no good data with non-selective NSAIDs to suggest an increased cardiovascular risk when one looks at meta-analyses, specifically a meta-analysis published by Garcia Rodriguez as recently as 2004. The relative risk of ibuprofen was right at 1 and there was a relative risk for an overall reduction of events, albeit modest, associated with naproxen.

My specific question to you is that in the ADAPT trial you stated that the increase in events with naproxen was significant. My question is do we, in fact, know whether that increase was statistically significant because my assessment of the math from the ADAPT trial, given the limited data that we have, is that it is mathematically unlikely that the increase in events with naproxen would be statistically significantly increased.

DR. VERBURG: We have not seen the data so I think it is speculation. My interpretation of what was put into the public domain is that there were significant differences, but without having the data to review I can't answer that.

DR. CRYER: But I think your wording is very important and somewhat misleading because you specifically say "significant" and many of us, when we hear the word significant, we are led to a conclusion that that is a statistically significant increase. And without having the data, as you just said, I think it is just a little misleading. All we can say for now is that there was a numerical increase which, if not statistically significant with naproxen, could have been entirely due to chance.

DR. VERBURG: Point taken. Thank you.

DR. WOOD: Ralph?

DR. D'AGOSTINO: I just want to get clarification from you. Given the discussion we had previously with the APPROVe trial and waiting 18 months before you started seeing a separation of serious events, and so forth, how do you respond? I mean, your presentation was talking about six weeks, a year at most. So, how do I interpret your presentation? And I was going to ask about the placebo trials also.

DR. VERBURG: So, the purpose of my presentation really was to go back and review what we know about the cardiovascular safety of celecoxib in the approved indications for this drug, which are osteoarthritis and rheumatoid arthritis. We reviewed all of the data that is available to review the safety of that drug versus placebo or versus alternative therapies. Subsequent speakers I think will expand into other indications that are currently being explored.

DR. D'AGOSTINO: So, your presentation would leave it that we really don't know what to make out of any long-term use? DR. WOOD: Wait a minute. It is one thing to say you presented the data for placebo-controlled trials in the approved indications, but it is not reasonable to say you presented all the data in placebo-controlled trials. The largest placebo-controlled trial presented in The New England Journal you haven't presented and you say you don't have the data here. That just doesn't pass the laugh test. Here it is, do you want it?

DR. VERBURG: I have seen it.

DR. FECZKO: Just for clarification of this, the APC trial will be presented I think later on this afternoon by Dr. Hawk. It is sponsored by the National Cancer Institute. We were not part of that trial. We are not privileged to the data. We were just given some top-line commentary about the data. The same holds for the ADAPT trial. We were not part of that data safety monitoring board or the results of that trial. I believe that is planned to be presented on Friday.

DR. D'AGOSTINO: My concern is the conclusions which we heard. I mean, you know something is coming down the line and why were these conclusions given as opposed to saying here is what we have at this point in time and walking away from it? It is a very positive presentation.

DR. WOOD: Dr. Manzi?

DR. MANZI: I think probably my questions can wait until they review the APC trial because it really has to do with the long-term issues.

DR. WOOD: All right, thanks. Dr. Shafer? DR. SHAFER: One might think I am fixated on low dose aspirin here, and perhaps I am. But once again we have three bits of information on low dose aspirin. We have table 4 in the handout that Pfizer prepared or the document that Pfizer prepared which again shows that actually the risk factors that existed, in fact, got worse on low dose aspirin.

We have in the APC trial, which will be coming up, table 4 from those data, again showing that the risk factors maybe were ameliorated a little bit but still with low dose aspirin the risks persisted. So, we don't have a protection, if you will, from low dose aspirin.

Then in your own slide 48, now in 48 it is not a placebo-controlled result and it is not blinded, but we can use the relative risks in the ASA versus non-ASA used for the other drugs to see that in the case of the high-dose rofecoxib group low dose aspirin conferred no protection.

Do these data, this sort of persistent signal that low dose aspirin provides no protection--are those data that actually pretty strongly support your contention that there are other mechanisms besides the COX-1 and COX-2 balance at play here?

DR. VERBURG: I am not sure that I follow where you are taking the question.

DR. SHAFER: You had suggested that perhaps there is something else besides the COX-1-COX-2 balance.

DR. VERBURG: Right.

DR. SHAFER: If it is the COX-1-COX-2 balance low dose aspirin ought to make these COX-2 drugs look like non-selective drugs.

DR. VERBURG: Correct.

DR. SHAFER: The fact that low dose aspirin doesn't do that repeatedly would look to me to support your contention that there is something else going on, and that is what I am asking. Is this something that Pfizer has considered? Have you had more thoughts on that?

DR. VERBURG: Only to reiterate some of the thoughts that I think were brought up this morning, and that is that this would not necessarily obviate or alter any changes in blood pressure that might occur with these drugs. It might but it might not. Also, it sort of lends itself to is there other molecule-based pharmacology that could moderate or modulate the effects that one sees from one compound to another? But that is about the extent of it. DR. WOOD: Garret, this keeps coming up. Do you want to address this?

DR. FITZGERALD: It is always difficult to address a straw-man when the construct is laid out and the arguments are assembled. I find the aspirin story really straws in the wind as opposed to anything definitive. For example, a comment was tossed out about blood pressure. We have absolutely no information as to whether low dose aspirin impacts on the blood pressure elevation by COX-2 inhibitors by controlled experience.

I think as far as the mechanistic issues that we talked about this morning, we would only expect aspirin to have a diminishable effect as opposed to an abolitional effect on that type of hazard because, as I mentioned this morning, it isn't a prostacyclin-thromboxane yin-hang balance. Prostacyclin acts as a more general constraint on factors that transmit cardiovascular risk. So, I find the arguments unpersuasive.

As far as molecule specific effects are concerned, it is quite true that almost every drug has multiple mechanisms of action that relate to dose-response relationships. But, in contrast to the mechanism we discussed this morning, the in vivo basis for the molecule specific effects are tenuous to non-existent and that includes the pro-oxidant effect of rofecoxib which is based on one paper in the literature using quantitative estimates of oxidative stress that those of us in the community view as highly questionable. Thank you.

DR. WOOD: I think your job, Garret, is to take Dr. Shafer for a drink and make sure that the two of you have sorted this out tonight! Dr. Domanski?

DR. DOMANSKI: I was just waiting for discussion of APC and I can wait a bit longer.

DR. WOOD: All right. Dr. Dworkin?

DR. DWORKIN: Yes, given the results that you allude to for the APC trial, suggesting that you don't really begin to see a difference until after a year, do you think it is going to be ethically possible, going forward, to do long-term placebo-controlled trials of celecoxib? You were suggesting that we need to do that, but I am not sure how given the results that we have in The New England Journal.

DR. VERBURG: I don't but I really want to address the question of ethics. I think I will step back and answer the question as follows, the APC was not the only trial which you will hear today. There is also another trial that shows that there was no risk associated with celecoxib. What does that inform us about the true risk of celecoxib over the long run? Relative to placebo, the drug may carry a cardiovascular risk. That I don't think is something that is known entirely. If the risk is there it seems to be small because it is not seen on a consistent basis. You could throw in the ADAPT trial. The results there are shown to be the same.

So, our sense is that you know something about the long-term cardiovascular profile of celecoxib. You know nothing about the long-term cardiovascular effects perhaps of non-steroidals. Yet, many patients would take them continuously. So, I don't know that it necessarily would be unethical. In fact, you might suggest that it would be mandatory for us to go and evaluate that. Patients have a need and a desire to know what risks they will be taking with their drug, not just in comparison to alternative therapies but what is the true risk if they decided not to select any therapy at all.

DR. WOOD: Allan?

DR. GIBOFSKY: Just a comment. I think it is important when we consider the safety issue to bifurcate the safety issue because there may be a dichotomy between how we are approaching it. I think some are approaching it with is the drug safe, while others are approaching it with is the drug safe for the intended use as prescribed in the label? I think those are two very different issues.

The test of whether a drug is safe or not, to test it across all indications is one thing. To test it across all other indications is something else. So, I really think we need to discuss safety in the context of intended uses. Many drugs, when tested for unapproved uses, will turn out not to be safe, whereas they may very well be for the indications for which they are approved, and that is why I think we have to be a bit relative in our discussion as well as being absolute.

DR. WOOD: Dr. Friedman?

DR. FRIEDMAN: Two points, first, you touched briefly on the issue of blood pressure. Surely, there must be ways of getting some good data on what celecoxib really does to blood pressure. The data you have shown are from relatively small numbers of people, followed for a very short time, and we don't know anything at all about what other medications or how they were otherwise protected. Do you have any plans for getting better, longer-term information in a more consistent way?

DR. VERBURG: Well, I think what I would like to do is turn the discussion over to Dr. Welton who has been studying the blood pressure effects of celecoxib and NSAIDs for many years. He can at least recapsulize for you what we have and perhaps also indicate what the future directions might be.

DR. WELTON: Thank you so much. Andrew Welton, from Baltimore. I have, I have to tell you quite frankly, been itching to get up here to the microphone to clarify at least the clinical aspects of the evolution of the blood pressure story because I do not think it has come across entirely clearly either this morning or this afternoon, specifically, the human component thereof.

So if you will bear with me for a moment, I will tell you, if I might have slide C2-42, that sequence, please? I would point out that this is a fascinating story that first came to our attention with NSAIDs in 1993. These were the observations of Janet Pope, who was then a first-year rheumatology fellow, who pointed out in this meta-analysis, published in The Archives of Internal Medicine, that, indeed, all NSAIDs, when compared with placebo, do distort blood pressure and elevate blood pressure.

If I might have the next slide, the following year we learned something else in an additional meta-analysis. That was, once again, that NSAIDs disrupt blood pressure, the mean increase being 55 mm, but particularly learned that this dominantly emerges during the treatment of hypertension, which then set up the issue that maybe we are looking at an issue of drug-drug interaction.

If I might have the next slide, this was about the time frame with respect to the start of the first two coxib development programs and, therefore, we were very mindful of the importance of blood pressure as these drugs went into a human evaluation.

I show you here the data for the osteoarthritis studies as they were incorporated into the new drug application. You can see, scanning from left to right, that there really isn't much in the way of hypertension adverse events reported, and here we are at the mercy of the investigators doing the trials. In the CLASS trial, as Dr. Verburg already pointed out, additionally not much in the way of blood pressure.

If I might have the next slide, taking exactly the same approach, using NDA osteoarthritis trials for the second of the coxibs, this then gave us the emergence of a very obvious dose-correlated increase in hypertension events but, again, at the mercy of the investigators doing the trials. This wasn't correlated with specific elevations in blood pressure.

Next one, please. It was at this point that I and my colleagues thought the only way to resolve this correctly is to do head-to-head, prospective, double-blind, randomized trials. And, the logical subset in which to do these studies is, in fact, patients who are being treated for hypertension because this was emerging now more as a story of disruption of blood pressure control rather than the genesis of new onset hypertension.

In brief, our first trial was powered to look for a 3 mm or greater difference in blood pressure effects between the two coxibs using that because it is a guidance rule from our colleagues in the Cardiorenal Division of the FDA. The essence of it is it showed in treated hypertensives early disruption of blood pressure with rofecoxib, as seen on your left; continued for 6 weeks of observation; and not seen with celecoxib.

This was reasonably curious. Standard rule of thumb, make sure your observations are correct. So, on the right-hand side of the panel it shows repeating these studies in over 1000 people.

Next one please. The additional issue that emerged--

DR. WOOD: Try and get to the point quickly because you are answering a single question and we are running really short of time.

DR. WELTON: I understand. Mr. Chairman, I beg your pardon. Bear with me for a moment.

DR. WOOD: One moment.

DR. WELTON: Over 24 hours pressures are sustained. Next one. There are differences in the antihypertensive drugs. There are differences seen in the responses of the drugs also at the doses that cause comparable efficacy.

Next one, please. Let me simply wind up. If I might have the next one, please. As you will see at the top right-hand side, what it shows is that if you shift in the population blood pressure by as little as 2 mm, on the right-hand side at the bottom, you can see the reduction and mortality. So, these small changes in blood pressure in large numbers of patients are very, very important.

I would end to answer the question of Dr. Nissen that he asked earlier on, if I might have C-28-3, and that, Mr. Chairman, is my final point.

DR. WOOD: It really is.

DR. WELTON: Here we are showing elevations of greater than 20 mm Hg and it does show between these two coxibs there are important differences in these big swings in blood pressure. I regret I cannot show you placebo results in this trial because we didn't incorporate them but that speaks to your earlier question, Dr. Nissen.

DR. WOOD: Thanks a lot. Curt?

DR. FURBERG: I just wanted to say for the record that we have some missing information. There is a fairly large number of studies sponsored by the NIH that have information on cardiovascular outcomes. An effort was initiated to get that information together but no real follow-up. So, it looks to me like the NIH has dropped the ball and not provided the information that we need from those other trials.

DR. WOOD: Cardiovascular outcomes in what? In celecoxib?

DR. FURBERG: Yes, with Celebres, yes.

DR. WOOD: I see.

DR. FURBERG: So, I think we should request that information and, if necessary, even go to the director.

DR. WOOD: Tom?

DR. FLEMING: I commented earlier about how one struggles to try to interpret the data when there are such short-term interventions, the 41 trial meta-analysis that if you focus on the placebo control you only have 6 weeks of treatment. It certainly tempts me to focus much more on the half a dozen studies that have longer-term follow-up.

You mention in slide 36, the CAESAR trial and the CLASS trial, although diclofenac is the control and, as Dr. FitzGerald said, is that Celebrex with hepatic side effects? What does it mean if there is not a difference? Interestingly though, when you look at the CLASS trial and the non-aspirin users there is also an ibuprofen arm and the summary that is given here is in atrial SAEs, anginal SAEs, MI and thrombophlebitis. There are four times as many events on Celebrex than ibuprofen in the non-ASA users.

If we go to the placebo-controlled trials, we have seen that in the APC trial there is a three-fold increase in the rate of CV death, MI and stroke. Another placebo-controlled trial that you didn't mention is the Alzheimer's trial, the 9702001 trial, that being placebo-controlled is of interest, and it had I think a doubling in the rate of targeted events.

Then, the last issue related to this is the PreSAP and the ADAPT trials will also be very informative, and I am very confused in exactly what you do know. I think someone has already alluded to. On slide 821 it is written as though you know that these results will be neutral or favorable.

So, I have a multi-component question here, am I interpreting this--can you tell us more about the Alzheimer's 9702001 trial? And, what exactly do you know today about the PreSAP and ADAPT trials?

DR. VERBURG: Let me start with the second one first. So, the PreSAP results will be reviewed by Dr. Bernard Levin later this afternoon in full detail. So, those results will be disclosed to the committee. For the ADAPT trial I know no more than what has been published, what has appeared in the newspapers, nothing more. DR. FLEMING: And what about the 9702001 trial?

DR. VERBURG: Right. So, let's go to slide C-214. Let's talk about this for a minute. So, the Alzheimer's trial, study 001, was a small randomized trial comparing celecoxib 200 mg twice daily to placebo over one year of treatment. Notice that the randomization was 2:1 and that the mean patient exposure was on the order of about 10 months or so.

This goes back now to the concept that we used in the briefing book and we will update this in a minute, but for any cardiovascular event you can see that there were 3 events versus 11 events. There were 4 myocardial events in total. Two of those I believe were angina and 2 were MI. Cerebrovascular events are listed here and then further down.

Of course, these are based on investigator reports to us. Also, if we go back--

DR. FLEMING: Well, before you leave this slide, which I guess you have just done--is there data that you have on heart failure as well?

DR. VERBURG: I am sure we do. I just don't have that right at hand but we can certainly get that for you. I just don't have that in my presentation.

I am looking for the background medical history in this trial. Do I have the wrong slide number? So, what concerned us a little bit about the results of the trial you can see here, again coming back to my comment earlier, when the purposes of the trial are not cardiovascular in nature, they can be heavily confounded because you are not controlling for distribution of patients by risk factor. So, what you see here is a trend for a higher degree of underlying risk in this patient population.

Also, I want to add one comment--

DR. FLEMING: Although somewhat modest I would say when you are looking a relative risks of 2 in the outcomes. A valid point, small numbers, but it doesn't explain a large part.

DR. VERBURG: So, we didn't entirely dismiss it there either so we took it one step further and, in fact, at about the time of the CLASS and the VIGOR results we did employ a blinded adjudication process of all cardiovascular events, serious cardiovascular events that Dr. William White, who is with us today, conducted along with some of his colleagues. That trial was published several years ago.

Could I have slide C-217? That article that Dr. White wrote was targeted to arthritis patients. At the time, he and his co-workers also adjudicated the events from the Alzheimer's trial. Dr. White, if you would care to make a comment? I think you are most informed on these results.

DR. WHITE: Thank you. William White, University of Connecticut Cardiology Center. So, these were done in accordance with the other clinical trials that you have heard, using strict criteria between two blinded adjudicators. As you can see, there was a 2.9 percent incident rate in the placebo group and a 3.5 percent rate in the celecoxib group, which was not statistically different.

To answer the heart failure question, there were just too few cases of adjudicated heart failure, not different between the two treatment groups.

DR. WOOD: So, these were adjudicated events that had already been reported? Or, where these prospectively defined? DR. WHITE: Yes.

DR. WOOD: So, tell us what you did.

DR. WHITE: I am not sure what you are asking, were the cases prospectively defined when the study started?

DR. WOOD: Right. DR. WHITE: No.

DR. WOOD: So, maybe somebody should comment on that. Richard, to you want to comment? Okay, well, we will get to that.

DR. FLEMING: For heart failure you said there were two adjudicated cases. They broke out in what manner?

DR. WHITE: i believe it was equal in each group. It was a very small number. There was either one and one or two and two. I can't recall, to tell you the truth.

DR. FLEMING: Why don't we check?

DR. WHITE: We will check.

DR. WOOD: Any other questions? Dr. Shafer? DR. SHAFER: This does not involve aspirin.

DR. WOOD: Thank goodness!

DR. SHAFER: One of the things we are looking at is overall safety, and you brought up the subject about alternatives, NSAIDs being the alternative. What data are there about celecoxib GI tolerability versus NSAIDs when combined with a proton pump inhibitor?

DR. VERBURG: I am not aware of any data that evaluate GI tolerability issues--

DR. WOOD: There is lots of data on that.

DR. VERBURG: There are data with respect to complicated ulcers, but with respect to whether patients stay on therapy longer with celecoxib alone versus the combination of an NSAID and, say, a proton pump inhibitor, I am not aware of any such data.

DR. WOOD: Do you want to take that, Steve? No? Actually, the last sponsor presented some of that data in their presentation.

DR. NISSEN: I want to explore with you for a moment the issue--you have several times used the term "equally effective doses" and this is important. In several of the trials we see a relationship between dose and the amount of cardiovascular toxicity. It is particularly important because you have done a lot of blood pressure comparisons between rofecoxib and celecoxib and one of the arguments I have certainly heard is that the equivalent dose of celecoxib to 25 mg of rofecoxib is 200 mg BID, not once a day. So, I would be very interested in understanding that, particularly when you consider that there is a much shorter half-life and, you know, particularly if you do an ambulatory blood pressure study the effect of the drug may be gone toward the end of the dosing interval, which would tend to bias the study in favor of celecoxib. So, could you address any data that you have that indicates that 200 mg once a day has the same effectiveness as 25 mg of rofecoxib? DR. VERBURG: 200 mg of celecoxib in terms of 25 mg of rofecoxib in terms of effectiveness?

DR. NISSEN: Yes, I want to know about efficacy, and then I would also like to know about any blood pressure comparisons of 200 BID to 25. I am trying to understand. You have made a case that the drugs have a very different effect on blood pressure and I am testing that a little bit with you to make sure that we got that right.

DR. WHITE: Do you want me to answer that?

DR. VERBURG: Yes.

DR. WHITE: Well, I have conducted two controlled clinical trials in this regard. The first one was done about three or four years ago in 178 patients treated with celecoxib at 200 mg twice daily versus placebo twice daily, specifically in hypertensives treated with an ACE inhibitor to bring out the worst-case scenario with regard to interference with the drug. The 24-hour systolic blood pressure difference was 1.3 mm Hg between celecoxib at 400 daily and placebo, which was not statistically different. That was giving it BID.

Now, in the other realm, not placebo controlled but published two weeks ago in The Archives of Internal Medicine, was a 500-patient study in which patients with osteoarthritis of the hip or knee, plus hypertension, plus type II diabetes, also treated with a angiotensin blocker were then randomized to celecoxib 200 daily, rofecoxib 25 daily and naproxen 500 BID. At 6 and 12 weeks into the double-blind period a very formal cluster of arthritis efficacy assessments were made using the same standards for any arthritis drug, and they were, in fact quite equivalent using Womack and Visual Analog Pain Score and so forth.

So, from the patient perspective at 6 and 12 weeks, they were therapeutically equivalent. At those same endpoints as you already saw, there was a significant pharmacodynamic interaction between rofecoxib and perhaps the underlying treatment because there was very little salt and water retention, evident based on edema and weight gain, with about a 4.2 mm increase in 24-hour systolic pressure. That was a sustained increase during the daytime.

With regards to naproxen and celecoxib, there was no such increase seen, yet, there was clinical equivalence with the regards to anti-inflammatory responses. That is pretty much what there is. There are no other studies like those.

DR. VERBURG: Dr. Simon, do you have a comment?

DR. SIMON: Yes, I was one of the authors of the hypertension study in the Archives. As a hypertension study and as a rheumatologist, why would I be involved in such a study? In fact, I was involved to ensure that the outcome measures for osteoarthritis, as measured by Patient Global and Womack, which is a functional outcome scale, and the VS scale for pain would then be the appropriate way to look at equivalence of benefit.

The data sets that suggest that there isn't equivalence in this kind of analysis of 200 mg versus 25 are really based on different ways to look at the evidence, such as night pain and other aspects of components of some of these outcomes. This was really a very robust way that is, in fact, typically used for approval at the FDA in determining efficacy of a particular therapeutic. And, we were able to demonstrate both at 6 weeks and at 12 weeks that there were equivalent benefits. But you are absolutely correct, differences in half-life, if you ask different questions will give you different responses.

DR. WOOD: Do you really want to say something because I really want to get to the next--all right.

DR. BRAUNSTEIN: Yes, I just want to show--

DR. WOOD: Be quick.

DR. BRAUNSTEIN: Well, I will show you actually the pharmacologic responses for COX-2 inhibition.

DR. WOOD: All right, go ahead.

DR. BRAUNSTEIN: This shows you the average 24-hour inhibition of COX-2 for different doses of rofecoxib and celecoxib. This is a standard ex vivo PGE-2 inhibition assay. What you can see is that there is a dose response, as we know, for all NSAIDs to inhibit COX-2, and over 24 hours celecoxib 200 mg twice a day has the equivalent COX-2 inhibition of approximately rofecoxib 25 and celecoxib 200 once a day is roughly the same as rofecoxib 12.5.

May I have 233? These are the results of a clinical study looking at Patient Global Assessment in response to therapy, acetaminophen 4000, celecoxib, rofecoxib 12.5 and then rofecoxib 25. Without getting into an argument--although statistically rofecoxib 25 had the greatest effective, you can see that rofecoxib 12.5 is the dose that has the most similar efficacy to celecoxib 200 once a day, you know, similar to what you would expect based upon the pharmacologic and the pharmacodynamics.

DR. WOOD: All right, thanks. Let's move on to the next presentation, which is from the FDA and is by Dr. Witter.

If you or a loved one has been injured by a cox-2 specific inhibitor such as Bextra, Celebrex or Vioxx, please fill out our FREE case evaluation form. A Client Relations Representative will contact you shortly to discuss your potential case. Feel free as well to call us at (800) 476-6070 or email us at clientrelations@weitzlux.com

Weitz & Luxenberg is no longer accepting Vioxx cases.

see also:

Welcome Your Vioxx Lawyer - Vioxx, Celebrex & Bextra FDA Transcript - Welcome
A transcript of the Bextra FDA meeting.

Concluding Comments Your Vioxx Lawyer - Vioxx, Celebrex & Bextra FDA Transcript - Concluding Comments
Your Vioxx lawyer provides the complete transcript of the FDA meeting.

FDA Transcript - Day 1 Conference for Safety - Vioxx, Celebrex & Bextra FDA Transcript
Read here for the complete transcript of the FDA meeting for Vioxx