Your Vioxx Lawyer - Celebrex, Bextra & Vioxx Side Effects 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.

FDA Presentation: COX-2 CV Safety: Celecoxib

DR. WITTER: Good afternoon. I am going to try and push along here to make up some time. I am a practicing rheumatologist. I have been with the FDA for almost ten years.

One of the first drugs that I was given at its 30-day IND stage safety review was celecoxib. So, although I could say a lot about it I am going to limit myself to the topic of interest to day and I will try to move along as expeditiously as I can.

Just to remind everyone, and we have been discussing this but it factored into my historical perspective in terms of why we did what we did, or what kind of discussions went on, to remind everybody that there are different reasons for drug exposure which have been talking about, acute and chronic pain for example. I will be talking later about some acute pain issues so, to some extent, I have two presentations that are tied together. But, you know, in this situation you are a patient; you have a reason to be taking it because of the pain. The issue of placebo control and how we might define placebo, and we can discuss that for quite a while, but in a short-term trial for example placebo control might generally be viewed as acceptable because there is rescue available. On the other hand, in a long-term chronic pain type study there are problems to deal with. It is not realistic; it is difficult, and that has impacted some of the ability for us and the sponsors to do the kinds of things we might want to have done.

On the other hand, if you are trying to prevent disease progression, such as the Alzheimer's and the polyps studies that we will be hearing about later today, one can classify them as subjects, not really patients, and so in this situation, again depending on the placebo, it may be more acceptable to conduct such studies.

So, having said that, let me just take this opportunity to thank the sponsors, past and present, be they from the industry or from the private sector or from government, for their efforts in this regard in this complex area and, more importantly, to thank the patients and the subjects for the topics that we have been discussing and will be discussing in the next few days. This is a very complex area of medicine but very important. So, we have the privilege of seeing some data today that we didn't see back when I started. And one of the points, if you take nothing else from my presentation, is that we have had a paradigm shift in this area. It has been a dramatic shift in terms of looking at safety events and the kinds of data that we have. So, one of the themes I am going to try to develop is exactly that.

So, this slide is to remind us all that there are available OTC, some of the medications we have been discussing, be they ibuprofen or naproxen, that have been available for a while and available for the studies for the most part that we have been discussing. Although we try, and I know the investigators try to limit that exposure, it is also a factor that I think has to be remembered, particularly here as we think through these data that we are looking at.

In preparation for the meeting then I also looked back--and not meaning to pick on any drug in particular but I went back to the diclofenac approval back in 1988 to try to give us all a sense of what were the databases available back then and how decisions were made. So just very quickly here, we had some pivotal trials in OA that involved 97 patients for 56 weeks. We had patients in pivotal RA trials that went on for anywhere from 6-12 weeks. We had Phase I/II trials, which are the PK kind of trials for example, with 950 patients or volunteers. Those were mostly 2 weeks. There were some supportive trials that had 11 patients for 12 weeks. We had some long-term open-label trials that involved 252 patients for about 38 weeks. So, I had one of the statisticians do this calculation for me and that comes out to be around 224 patient-years. So, keep that number in mind as we move forward.

I would also like to point out that as I was reviewing this I noted that there were two myocardial infarctions with diclofenac; none that I could see in the other comparators which, by the way, included aspirin and one of the adverse events that used to be looked at a lot was tinnitus and people would get evaluations for hearing loss. In any event, there were two MIs, one during double blind and one in the open-label trials. So, I just thought this might be of use as we think through where we are.

Part of my challenge here today is to present to you then a bit of a historical perspective and to try and merge some of the different approaches in terms how sponsors conducted the trials and how we subsequently analyzed the information.

So, I just want to step back just for a bit. I am presenting here the World Health Organization terminology that was used to define cardiovascular events in the celecoxib NDA base. These kinds of reporting systems have evolved over the years, as we all know, but just to give you a sense of what were some of the terms that were looked at in the original approval for celecoxib, just a few of them are listed here.

Then just to remind everybody that we, for the most part, will be describing and discussing today--at least I will--mostly serious adverse events. There is a regulatory definition for that. Deaths are obviously the hard endpoint which will be also discussed. The point I think has already bee made that in the celecoxib NDA these were spontaneous investigator reported events. They were not prespecified or not adjudicated. In my subsequent presentation what I will do is give you some information about the adjudication process and how sometimes that is problematic. Also, in discussing CLASS I would like to point out that the GI endpoints, because that is what the trial was intended to do, were prespecified and adjudicated but, once again, the cardiovascular events were not prespecified and not adjudicated. These were spontaneous investigator reports but we look at all this information.

This is a slide that Dr. Villalba had shown earlier. I have just added one column here, and the only point I want to make from this is that as we might look at events--and I am not going to talk about the various categories--these are different ways to look at cardiovascular events. We are all familiar with the APTC we are all familiar with. But I just wanted to make the point that as you look at the numbers and you make just a rough ratio comparison, they appear to be similar, leading one to make an assumption that the inferences that would be drawn by looking at any of these data sets, at least in a qualitative way, would be the same.

Turning specifically to Celebres, this is my reminder to you that this information is available on the web. It has been an effort that has evolved over the years. We have tried to put as much information as we can in our reviews so that all of you can have a chance to look at this information.

The original NDA was submitted on June 29 or 1998. It consisted of 51 studies. I have just listed them briefly here as to the types. There were 29 studies in Phase I. There were 14 studies that were arthritis patients either with OA or RA. There were 7 post-surgical analgesia studies. There was one long-term study which went out 2 years, study 024.

To remind everybody, although you probably weren't here, when we talked about the original approval of Celebrex at an advisory committee meeting, one of the things that I discussed in particular was this concept of dose creep, that patients tend to increase their dose if they are allowed to. I would also like to reemphasize the point that in any of these kind of long-term trials there is no controlling arm and that really makes it difficult to try and get a handle on how to interpret these events, particularly from a perspective of common events like cardiovascular events. So, in my own thinking anyway, you always want to have some kind of a controlling arm whenever you do long-term studies. Then, again, with this particular type of drug how OTC medications may impact some of these results.

That is just a summary of what I will be talking out and I will just point out what I will be talking about, which is the ADAPT trial and two other trials which will be discussed a bit.

The reviewing process for an NDA and particularly for this one when it came in--it was a very large database and so this was really a team effort and that is one of the things I want to stress here. This data is looked at by multiple people with multiple talents for long periods of time. So, there isn't just one person looking at the data; it is done as a team effort. In this case, for example, I was the primary medical officer to look overall efficacy and then to come to an overall conclusion about safety. To assist me was a renal/cardiovascular consultant. We also had another medical officer who reviewed the data and also paid attention to the cardiovascular results. We had a GI consultant who served as a secondary medical reviewer also looking at the safety data. Then we had people specifically looking at analgesic trials and the platelet safety trials. So, there really is a team of people who look at these results whenever they come in.

I am going to stick with just the OA and RA exposure because that is the most robust exposure that you have in here. What I am doing is displaying results from some of the consults that we had to the Division about these issues. I will be describing most of the results either in terms of patient-years or crude rates, and I will try and tie this together at the end to Kaplan-Meier approaches.

But just to give you a sense, in the original NDA in the controlled trials there was not a lot of information for exposure beyond 180 days, not surprisingly, but when you looked at the open-label trials we had a larger exposure. To the extent that these numbers make any sense, I am just pointing out a 16,208 patient-year exposure versus diclofenac, as I pointed out earlier, at 324.

Turning then to the cardiovascular mortality in the NDA database for Celebrex, the comparisons here in the information that we had are against placebo, Celebrex itself and the NSAID comparators in two different ways. They don't differ that much; there was a slightly different definition. Then also in the long-term open-label studies. You can see that there were not many events. When you do the math here and divide it using the patient-years to get an estimate of the crude mortality rate, you can see the highest number comes out here for the NSAID comparators in both situations. It also is higher than what was found when looking at the all known cardiac deaths in the long-term open-label arthritis experience. So, there didn't appear to be any large signals when looking at this particular outcome.

Turning then to serious adverse cardiac and renal events, i have again here the columns of placebo, differing doses of celecoxib and the NSAID comparators. When you look at these events overall there were no important differences. In fact, they looked worse for the NSAID comparators and the placebo looked roughly equivalent to celecoxib.

When you look at some of the individual events, and let me see if I can point to the particular events that have been discussed so far today, heart failure for example, there didn't appear to be any major differences between celecoxib and placebo; myocardial infarction, again there appeared to be no important differences between all the groups.

So, looking at this data in summary, there didn't appear to be any major clear signals that distinguished celecoxib as it appeared in the NDA database from NSAID comparators and, at least in some of these comparisons, from placebo as well.

Looking then at the data from the extension trial after the NDA in a bit different way, we configured the data to display the events of cardiovascular mortality based upon the last known dose that the patient had at the time of the event. So, that is what is displayed here. As you can see, you go from zero at 100 mg and up to 200 mg, 300 mg and 400 mg. When you do the math again using patient-year of exposure, there certainly appears to be a trend here. As you go up in the dose, the cardiovascular mortality goes up. These are small numbers and, again, it was difficult for us to place this in context with no controlling arm.

For example, if an event had been adjudicated away, and we don't do this, it would bring the rates down to what I have given here just for comparison's sake. So, we are aware of this; didn't know what to do with it; difficult to make comparisons without some kind of a controlling arm.

I would like to turn then to the SUCCESS-1, which stands for Successive Celecoxib Efficacy and Safety Study. In terms of what we are discussing today, this was a short 12-week study in patients with osteoarthritis. It had two comparisons with celecoxib, two different doses, naproxen and diclofenac. It was a large study involving 39 countries, lots of centers, and it was really intended to evaluate the homogeneity of efficacy and safety around the world. it was not intended as a cardiovascular outcome study. None of what I am discussing today was intended as a cardiovascular outcome study.

I have put up here a bit more of summary results to give you a sense--and these results were described previously at other meetings--between celecoxib, diclofenac and naproxen. What I have tried to do is highlight for you in yellow who has the most events. As you can see, for the most part with the exception of a small increase of cardiovascular events, there wasn't anything that in particular distinguished celecoxib from the other groups.

I have noted down here an update last month. There were, in fact, 8 myocardial infarctions in the 100 mg group; 2 in the 200 mg group; and 1 in the NSAID comparators. And, I have done the calculation for the rates to make some comparisons here. But as you can see, and I think the points are starting to emerge as we discuss more and more data, that when you look at the comparator NSAIDs they have their own sets of problems which we were certainly aware of as well.

Turning to the CLASS trial, in case you don't know, it stands for the Celecoxib Long-Term Arthritis Safety Study. I have highlighted the term arthritis here because, again, this is for the indication of arthritis and that is where this was studied. This is a unique trial. It was intended to mimic a real-world setting. We have been hearing criticisms that trials were not extrapolatable and generalizable so what we were trying to do, and the sponsor as well, was to come up with a trial that was in a more realistic setting.

I should point out that the only trial that was available, large outcome trial, was the MUCOSA trial published in 1995. So, this was a unique trial at the time. A lot of discussions went on about how to design this trial. One of the things that we had been discussing was aspirin use if indicated. Patients had either OA or AR. As we will be hearing more about, RA, we know, traditionally increases the risk of cardiac problems. In particular, there is something that just came out in Arthritis and Rheumatism this month which points out that RA doubles the risk of heart failure. This seems to be, according to the authors, an independent risk associated with the disease itself.

So, just to reiterate, this was designed as a GI safety study and it was intended to try and change the NSAID template regarding this particular outcome. This was not powered nor designed as a cardiovascular safety study.

This is a slide that back then, in 2001 when we discussed these particular CLASS and VIGOR trials and what we were bringing to the forefront at that point of time was this concept of 2X. So, let me just tell you a bit about the history of that. The X dose was intended to be the highest dose for the intended chronic indication. The idea of 2X was to give us some assessment of the robustness of the safety results. We have certainly heard, as somebody rolls in the door as an NSAID that, you know, we are safer. So, we wanted to see the data. We were also skeptical of the surrogacy for endoscopic results and how that might translate into rigorous outcomes. So, it was that kind of thinking that impacted upon the design of these kinds of trials.

Again just to remind you, at the time--and this is still the language in the GI portion of the FDA warning label, it describes in terms of looking at GI ulcers, gross bleeding of perforation, that there is one percent of patients if treated for 3-6 months who experience this event, but this occurs in about 2-4 percent of patients if they are treated for one year. So, this is data that we had previously known from other NSAIDs. And, I saw this on a slide earlier today--the kind of information that we had available from other databases, suggesting that there were lots of hospitalizations and lots of deaths associated with this particular adverse event.

Some of the baseline demographics then in terms of looking at the CLASS trial, the mean/median age was about 60 years; 11 percent of the patients were 75 years or older. These were mostly white females. Approximately 27 percent of the patients had RA; 10 percent had a history of GI bleeding or gastroduodenal ulcers; and about 21 percent were taking aspirin.

In terms of looking at the inclusion criteria and exclusion criteria, they were fairly open. Basically you needed to be able to give informed consent; that you required something like this kind of a medicine and that you were not pregnant. On the other hand, that you didn't have any active disease of any signals in terms of looking for hepatic events.

The aspirin use in CLASS deserves some comment. It was at less than or equal to 325 mg daily. Again, this was if patients needed for cardiovascular events. But the use was not stratified in the CLASS trial. The dose and the duration of use also varied. It wasn't a constant. So, I think this is one of the things that we had discussed back in 2001, that it was probably not a good idea to try and draw any firm conclusions from the aspirin use from this trial, and that only observations and possible directions for future studies might be the most value for this particular study.

To give you a sense then of the exposure in the CLASS trial, it was again a large trial. In terms of making some comparisons here, this one trial to the extent that we believe, or you believe, patient-years of exposure and how adequately that assesses risk, there was three times more information in this one trial on diclofenac than we had in other trials, the point being that, you know, we had been very comfortable with much larger databases in this regard which is a good thing.

The exposure, in terms of looking at the durability and long-term, is listed here for celecoxib, diclofenac and ibuprofen. The patients who were exposed from 12-15 months, there were not many patients in the diclofenac group compared to the other two arms. This was a confounding observation and when we were trying to understand some of the benefits in this trial we got into discussions of informative censoring, which I won't get into today, but this was a factor in terms of trying to understand and put some of these results in context.

Turning to deaths, I have listed here--and this is the same information I have talked about at other advisory committee meetings--there were 36 deaths overall, 19 in the celecoxib group, 9 i diclofenac and 8 in ibuprofen. I have calculated roughly the patient-years here for comparison. No important differences, at least to my eye. Most of these patients were 65 years or older. Most of these deaths were from cardiovascular events. There were 11/19, or 58 percent, in celecoxib and roughly the same in diclofenac, a bit more in the ibuprofen group.

In terms of looking at this data, and in spite of my own caution earlier about looking at aspirin versus non-aspirin, it is exactly what I am going to be doing to give us a sense of what the data look like. Again, here are the three treatment groups. This displays all deaths and this displays the cardiac deaths, broken down this time into all patients, those that use aspirin and those that were not using aspirin. When you look at this data in terms of all-cause mortality, again, there do not appear to be any point differences. When you look at aspirin users there were more events in the ibuprofen group. When you look at non-aspirin users there were more in the diclofenac group. This pattern basically held through when we looked at the entire study for cardiovascular deaths. There was the same trend.

Turning then to serious cardiovascular events in the CLASS trial, here again is displayed a comparison between aspirin users and non-aspirin users. Looking at the groups, you can see then that there were not as many patients that did take aspirin so the numbers are smaller; the patient-years of exposure are smaller. But, nonetheless, here are the results. When you look in the aspirin users and at the issue of myocardial infarction you can see that there were more of those in the ibuprofen group. When you look at the combined atrial endpoint, which was a combination of atrial fibrillation, bradycardia, tachycardia--I am not remembering one of them, anyway, it was a composite endpoint that we had come up with to get a handle on this. There were more events in the ibuprofen group. For combined anginal disorders, which was a combination of unstable angina and coronary-artery disorder, there were more in the diclofenac group.

Looking at the non-aspirin users and looking at the same types of endpoints, in this situation it looks different in that there are more events in the celecoxib group than in the other two comparators--small differences but differences nonetheless.

What I have tried to do in this slide is to put together some of this information in terms of looking at APTC-like events, recalling again that these were not adjudicated. I don't want to diminish the importance of APTC so I am calling it "like" events. So, I have just simply added up cardiovascular deaths, MI and strokes to give us a sense of what this endpoint might look like if it had been done, and you can see in this comparison that there are more of these events in the ibuprofen group versus the other two.

This is Kaplan-Meier analysis that I took from one of the publications that I have listed up here, by Dr. FitzGerald, in Nature/Drugs Discovery, in 2003. There also was something by Dr. Strand and Hochberg in 2002 in Arthritis and Rheumatism. I put this slide up here to try and make some comparisons for us. This displays the Kaplan-Meier analysis for serious thromboembolic cardiovascular events, arguably in the most important population to look at, in the non-aspirin users, and as you can see from this particular analysis celecoxib appears to be between the comparators here. It might not be showing up well in the back. This is diclofenac; here is ibuprofen.

What I have displayed over here then is to give us some comparisons, if one looks at true rate comparisons with these number of events or patient-years to tie back to earlier looks of the data, and again it is probably hard to see, this is 0.97, 0.7, 7.45 versus 1.78, 1.33 and 0.8. the point being is that there do not appear to be any important differences in the conclusions or inferences that are made no matter how you look at this data.

I would like to turn then to the Alzheimer's study, 001, which has been discussed a bit today. This was under an IND in a different division, Neuropharmacologic Drug Products. We were aware of this study. This information had been discussed previously.

This was a study that was started in 1997. It was a double-blind, placebo-controlled trial that lasted for a year. It was a comparative study of celecoxib for the inhibition of Alzheimer's disease. One of the results in terms of efficacy conclusions was that celecoxib did not limit progression in this situation.

There were other studies that were ongoing at the time, 004. This was an open-label study looking at long-term safety. This study was terminated when the results of 001 were made available.

There was another study under this IND, 002, which was a placebo-controlled, long-term study. It had vitamin E co-use in it as well. This was intended to look at brain size by MRI and to look at Alzheimer's disease-associated proteins and inflammatory mediators to get a sense of mechanisms. This study was also terminated due to the results of the 001 study. So, the IND was inactivated in July of 2001. As we have been preparing for this meeting it came to our attention, the following letter which I just want to bring to your attention regarding this particular study. I am just highlighting a few things here rather than showing the whole letter. But this was basically a letter from the DSMB that was involved in this particular study. What the letter points out is that the trial was conducted between 1997 and 1999; that there were, according to this letter, no adverse events to support stopping the trial while it was ongoing, however, at final review there was an excess of cardiovascular-related and other risks but it was difficult to interpret, according to this letter, because of the small sample size which made relative risk and odds ratios unreliable. This was conducted in a frail and fragile population that had substantial co-morbidities and concomitant medications, making it difficult to know how to generalize these results. It was commented that there were indications of failure in randomization in baseline cardiovascular disease and cardiovascular medications, meaning in particular that there were more in the celecoxib group than in the placebo group.

The letter went on to state that the members were concerned that this data had not previously been made available, other than i an abstract form, and they were concerned about this because this may be the only information available in medically ill elderly populations with placebo control. Looking then at cardiovascular events, I have summarized them briefly here comparing the Celebrex 200 mg versus placebo. I just summarized the events. With the one exception of cerebrovascular disorder, there don't appear to be any differences in all the adverse events--deaths overall, cardiac deaths, serious adverse events, cardiovascular, and no matter how you look at it--congestive heart failure, atrial fibrillation and then I made another APTC-like calculation here, they all wind up on the celecoxib side of the ledger here.

This is also information that we had available to us in preparing for this meeting in terms of addressing the issue or randomization. These are results from the sponsor that you saw already. When you look at the Celebrex group there were imbalances in terms of hypertension, diabetes, those that had bypass surgery, those that had history of ischemia and those that had history of coronary-artery disease. Whether or not this, in a small trial, is enough to explain the results is to be determined.

So, that is what I have to say today. Thank you.

DR. WOOD: Thanks a lot. You also have not covered the APC trial. Right? That is sort of surprising. Does the committee want to go on to the next two presentations and wait for questions to Dr. Witter at that point? Let's do that. Let's go on to the next two presentations.

DR. FLEMING: Could I have jut one?

DR. WOOD: Sorry.

DR. FLEMING: Just on slide 35 as you were presenting those 001 results, it is certainly noteworthy that there is a pretty consistent excess across all of these key categories for Celebrex. We talked, for example, about heart failure adjudication. It is kind of hard to adjudicate something in a blinded way when all the events are in the one arm. I don't know if the adjudication committee was aware of how the results broke out before they did their adjudication. In any event, we were told those broke out at 1/1 after adjudication. They were 5/0 before. So, that seems difficult to justify as well. So, I look at this as one of a small number of placebo-controlled trials with a fairly long period or treatment exposures. So, this is of some relevance.

DR. NISSEN: Tom, did you attempt to do a p value there from those numbers?

DR. FLEMING: For which aspect of this?

DR. NISSEN: Well, say, APTC-like or just the serious AEs? Is that going to be significant? DR. FLEMING: Probably borderline.

DR. WOOD: You know, the elephant in the room is the next trial so let's move on and see if we can get some of these questions dealt with afterwards. Let's go on to Dr. Hawk's presentation.

DR. WHITE: Do you mind if I make one comment, jut for cleaning the air? The adjudication committee was not aware of the results when they looked at the data at all.

DR. WOOD: Right. Let's come back to that point later because there are lots of problems with that adjudication. Let's go on to the next two talks.

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:

Answers What all Vioxx users must read - Questions and Answers on Vioxx safety
Important Questions and Answers on Vioxx by Committee Members

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