What a pleasure it is to join you today. I am Andrew Whelton from Baltimore, Maryland, from the Johns Hopkins University School of Medicine. What I propose to review today is a clinical issue that within the last several months has caused quite a considerable amount of concern on the part of many of our patients, and that is when they have the decision to make as to whether or not they will take the very important and new COX-2 specific inhibitor compounds, also known as COX-2s. (Please note that I have served as a consultant for several of the pharmaceutical corporations who have developed these drugs and are working on this topic. Some of the issues I discuss will reflect ongoing research and may have no relationship whatsoever to the package insert or label.)
Only within the last 12 to 18 months have we recognized that this very important category of analgesic, anti-inflammatory compounds does indeed, somewhat to our surprise, have an effect on renal function, and also, by extension and effect, on the cardiovascular system. It is those effects -- and the clinical implications -- which I want to unfold for you today.
From a scientific point of view, this story began just over 100 years ago. It is absolutely fascinating that the clinical issue that drove the development of the first drug in this category i.e., the first anti-inflammatory analgesic compound, aspirin, was clinical gastric intolerance. The man who synthesized or manipulated the molecule of salicylic acid to become acetylsalicylic acid or aspirin was a young synthetic chemist working for the Bayer Corporation. His father happened to be taking salicylic acid, which in its pure form is very erosive to epithelial tissues. So young Dr. Hoffman, as one of the early breed of synthetic chemists, set out to find a more tolerable form of salicylic acid, in essence trying several substitutions. When he put an acetyl side train on the compound, his father told him he tolerated it much better. It was the birth of this cascade of drugs.
Hoffman stopped at this point and came up with the brand name aspirin, and here we are, over 100 years later, and I have always thought it absolutely astonishing that aspirin remains the only proven drug to provide cardiovascular prophylactic benefit either in secondary prophylactic settings or indeed primary clinical settings.
The next point in the development of our story of discovery jumps to 1971. This was actually the first time we understood clinically and scientifically how these drugs work. It was the extraordinarily important work of Dr. John Vane, who told us about the existence of the cyclooxygenase enzyme system, which can damage cell products or arachidonate, and get processed and converted into prostanoids. Those prostanoids are thromboxanes, which come out of platelets or prostaglandins -- coming out in tissues such as the stomach, the kidney, and in sites of inflammation. For that work John Vane received the Nobel Prize in medicine.
The next part of the story was the unfolding of why there is a disconnect when we use anti-inflammatory drugs. As you will recall, we have the benefits of suppression of pain and inflammation on one hand but deleterious effects on platelet function, on gastric health and on kidney function on the other hand. So we have both clinical benefit and clinical mischief. The person who resolved that for us most notably was Dr. Phil Needleman and his team, although I acknowledge that there were at least three groups who contemporaneously pointed out the enzyme system -- as originally described through the work of John Vane and others-- exists as two isoenzymes. One of the enzymes, designated COX-2, is the target for the development of inflammation and pain, and the other enzyme, COX-1, is simply the one that allows our platelets to function, maintains the health of the gastroduodenal mucosa, and maintains health within the kidney.
So having recognized the existence of two COXs, then the pharmacologic rush was on to turn off COX-2 in order to turn off pain and inflammation, and this has led to an ever-expanding cascade of these very important coxib compounds.
We have an immense body of information about these drugs. Indeed, when the first two were developed, these drugs represented the largest pharmaceutical development program in the history of the whole industry. Really astonishing, and they have only been available to us for a couple of years in the clinical domain. Now the first one, which became available the very last day of 1998, was celecoxib, followed in the middle of the next year by rofecoxib and then, more recently, the third approved in the series here in the United States, valdecoxib.
We are currently working on a parenterally administrable form of valdecoxib. And, to complete the record, so to speak, of drug availability, another COX-2, etoricoxib, is available in limited parts of the world.
It interesting to point out that when two separate development teams were working on celecoxib and rofecoxib, the exact three dimensional configuration of the enzyme upon which they would work, i.e., the COX-2 enzyme, simply was not known. Therefore, a very tedious and time consuming process within the development program had to be undertaken to screen. For example, 3,000 separate compounds were developed before making a decision to go with the celecoxib molecule as shown here, and a comparable approach before the decision was made to select the rofecoxib molecule.
Both of these drugs were developed in absolute commercial secrecy, and once we realized what the shapes looked like when the information came into the public domain, I've always thought it was quite extraordinary how similar they actually are -- C-shaped in configuration.
As they became available this is also the time framework in which we began to see the three dimensions of the channel in the enzymes which these drugs would turn off. And you will see on the right hand side of the slide the that the COX-2 enzyme, pain and inflammation, is substantially larger than the COX-1 enzyme. For example, we find in platelets it is indeed cylindroid in nature, so a drug that will block COX-1 in platelets -- out of necessity -- needs to be a fairly thin compound that will drop into the cylindroid channel. Whereas in the case of COX-2, the channel in the middle of the enzyme is C-shaped and the larger C-shaped drugs just fit right in beautifully in this channel, thus blocking it.
As a step in the development process, early on, since these COX-2 compounds would be used for the first time in humans and recognizing that they are metabolized within our liver, we were wise in saying "Gosh, maybe the drugs will fit the bill, maybe they will give all the efficacy we wish, maybe they'll give all the safety we assume they will give in the GI tract but since they are metabolized in our liver or the liver's of our patients, is it possible that, after extensive use, 100,000 or half a million patients, that we may be unfortunate enough to see severe liver toxicity?" This has certainly happened with other drugs, and you cannot predict these rare events with any degree of accuracy. The essence of it is that you need to be ready and therefore back-up compounds should also be ready. And that is why I show you on this slide the backup compound for celecoxib, our compound valdecoxib. The second, or back-up, compound for rofecoxib is etoricoxib. Now please note that at the bottom right hand portion of the molecules, these in essence are the anchors in the drug that hold them into the COX-2 channel. That is where we want them in our patients, because then they are preventing inflammation and the sensation of pain. These anchors, although quite similar looking when you first glance at them, in fact are different and produce an enormous impact on liver metabolism of these compounds.
Since this is relevant to the cardiovascular story, let me make one or two additional developmental comments here. At the bottom of celecoxib, one will see that the sulfur molecule is attached to an amide group or an amino group, so, by definition, you and I call this a sulfonamide. Valdecoxib has the same configuration. On the other hand, in the case of rofecoxib and indeed etoricoxib, a sulfur molecule in exactly the same position is now attached to a CH-3. By definition, we need to call that a methyl sulfide. So a subtle difference in the molecules yields tremendous impact on metabolism within the liver. And, as the cardiovascular story and particularly blood pressure issues unfold, these side chains may indeed have very important issues.
Now I am removing etoricoxib from the slide because, back at the end of the spring of last year, although this drug had gone through a human development program and the data had been submitted to the FDA for review, the drug was actually removed by the sponsor from the ongoing review process because of the development of cardiovascular complications, cardiorenal complications. So currently we have available three compounds, celecoxib, valdecoxib and rofecoxib.
An additional feature in the development program of valdecoxib is of interest. Once we knew what the enzyme channel looked like in the case of the pain and inflammation enzyme, it was possible to decide to put not just one anchor but a second anchor on the compound, as shown in the slide here. If you think about it in nautical terms, you anchor the boat, so to speak, at both ends. And of course, what that does for us is keep the drug in the enzyme, channel blocking the activity of the enzyme -- playing a role there for suppression of inflammation and suppression of the sensation of pain -- and that is precisely where we want the drug. That is just exactly where we want it, and doing a little manipulation like that is kind of straightforward. But nonetheless that generally leads to greater potency, and of course, a much greater selectivity for blocking the enzyme. I am a hard-nosed clinician, and one still has to do the prospective double blind trials to ensure that there is indeed a difference in efficacy between the compounds and such trials are not available to us as yet.
Importantly, when it comes to the issue of use of these drugs, for example, intraoperatively or immediately postoperatively, one would like to have an injectable format of the drug. We are working on that. We have an advanced program. I hope that in the next number of months we will have sufficient data to submit to the FDA to get this first parenterally administrable drug available.
I show you this slide only to make the point that valdecoxib and indeed all the other coxibs we have discussed so far are extraordinarily soluble and lipid. Of course, that is what allows them to penetrate into the brain and that is why they are so beneficial with respect to central suppression of pain sensation.
Because they are soluble and lipid, we cannot inject a lipid solution intravenously. So what one needs to do is to make the drug water soluble. An interesting twist to the story is that, in making valdecoxib water soluble in the form of parecoxib, it actually loses its clinical activity but, immediately upon injection, it is hydrolyzed back to the active moiety. So giving it by intravenous or intramuscular administration is a very neat way of packaging the drug.
We are particularly evaluating not only its efficacy but also its cardiovascular safety, in particular in the highest risk patients undergoing surgery such as for coronary artery bypass. Within a few months, I hope we will have it available.
Going back to the central theme of cardiovascular concerns, remember that we have had lots of experience over the years with the kidney side effects and he related cardiovascular effects with NSAIDs. We recognize that these side effects go in a cascade from the most common, such as abnormalities of salt and water metabolism, through to hypertension, congestive heart failure, problems of acute deterioration of renal function (often referred to as acute renal failure), and then much rarer issues, such as nephrotic range proteinuria with new onset of papillary necrosis, etc.
When it comes to salt and water, you cannot get more common than 100%. We all have a transient abnormality of salt and water balance when we take these drugs. When we first identified in the late 1990s that the new COX-2 compounds have an effect on normal kidney function as shown here by reduction of kidney production of prostaglandins, that was a big surprise. We hadn't anticipated that COX-2 would be in healthy tissues. After all, the algorithm proposed was that COX-2 is pain and inflammation and COX-1 is health.
Unfortunately, as with many things in clinical medicine, that theory turned out to be too simple. COX-2 is in many healthy tissues, not only in the kidney, but also in the brain, the female genital tract, in bone and, with great public health implications, on the surface of all epithelial-derived tumors. At the kidney level, we identified for the first time that COX-2 in healthy kidneys acts basically like an androgenous mild diuretic, so that when inhibited either with a COX-2 agent or a nonspecific COX-1/COX-2 inhibitor, we see what I show on the slide here: over the first day or two, there is a minor retention of salt and water, amounting to about 2 to 3 pounds in weight gain. But the kidney has so many compensatory limbs -- angiotensin, vasopressin, atrial natruretic peptide - that they compensate and, over the next couple of days, dump out in the urine the initially retained salt and water. So by about day 5 and 6, the average healthy person is absolutely in balance. We never even perceived that we would gain the 1 or 2 pounds as a result of the early inhibition of COX-2.
There's quite a different story in those with heart disease or vascular disease or kidney disease, because they undergo the same retention of salt and water over the first couple of days, cannot adequately compensate kidney function, over-shoot with retention and then, generally, at the end of one week of treatment, will manifest edema. Now we have never put that to the test, but let me add that in our recent trials we have found some 80% of all new onset edema associated with these compounds is visible or detectable clinically by the end of one week, which is quite extraordinary. It is important also to tell you that, when it comes to the issue of blood pressure, we as a clinical group didn't appreciate this until the mid- to early 1990s. This was also extraordinary because this is a very common problem and obviously was going on for decades, at least ever since the introduction of the modern antihypertensive drugs, because what we see is that if an NSAID is added during the treatment of hypertension, blood pressure goes up. So the whole blood pressure story here is disruption of blood pressure control. It is not new onset hypertension; in fact normotensives are only very mildly affected. So this now represents yet another excellent example of drug/drug interaction. On one hand, the antihypertensive drugs produce nice blood pressure control, but the addition of an NSAID or COX-2 disrupts that blood pressure control, and blood pressure goes back up.
This is particularly important with ACE inhibitors and beta-blockers, because these lower blood pressure at least in part by producing a prostaglandin from blood vessel walls -- in essence prostacyclin production, and prostacyclin is a vasodilator. So peripheral resistance falls, and, as you know, blood pressure falls. If we then inhibit the prostaglandin production with an NSAID or to a variable degree with a COX-2 compound, then blood pressure goes back up because we have inhibited the vasodilator. And for those of our patients who develop edema, the blood pressure disruption is even worse.
In healthy blood vessels, we cannot find COX-2, but in diseased blood vessels, it is everywhere, as shown in this slide, where an incredible amount of deposition of the enzyme COX-2 can be seen in atherosclerosis. It turns out that atherosclerosis is a disease associated with a major degree of inflammation. So, separately, we are working on whether or not suppression of inflammation, i.e., COX-2 enzyme activity, in atherosclerosis will turn out to have major cardiovascular public health implications. It looks like that may be so and that suppression of inflammation in this disease process may be a key management issue that we have missed for decades.
The story clinically got a little bit more complex at the end of the 1990s. A hypothesis was proposed that these new coxib drugs might tilt the balance in individuals with vascular disease towards a pro-thrombotic state. Remember that coxibs have no effect on platelet function; they are simply too big to drop into the platelet channels. The platelets function normally and stick together or aggregate when they produce the thromboxane, which not only helps platelets to clump together - a desired effect in clinical bleeds, but also constricts blood vessels. And under normal circumstances, biologically, there is an oscillation in both directions for all of these activities. So the vasoconstriction is counter-balanced by vasodilatation, and this is done by the production of the prostaglandin, our friend prostacyclin, once again coming from blood vessel walls. However, it was shown by our colleagues Dr. Francesca Catella-Lawson and Dr. Garrett Fitzgerald, even in the setting of healthy adults, COX-2 specific inhibitor compounds (such as celecoxib and rofecoxib) could at least partially inhibit the vasodilatation. Thus, the proposal was made that maybe, if we look carefully, we might find a tendency towards increased thrombotic events such as strokes, myocardial infarction, etc.
At that point in time, we took a very careful analytical look at the thousands of patients included in what is generally designated as the NDA, the New Drug Application database. This covered multi-thousands of patients for celecoxib and for rofecoxib, and there wasn't any evidence of a prothrombic effect in those databases.
Next came the massive GI safety trials --CLASS and VIGOR. Indeed, the story of a cardiovascular signal evolved from the clinical trials.
The investigators working with celecoxib really didn't see any problem with blood pressure. These are data for adverse hypertension reports in the CLASS trial, shown on the right side. Remember the CLASS trial was at much higher dosing of the drug, in fact, supratherapeutic dosing, and extended for 9 to 10 months by comparison with the NDA, which was generally on the order of three months of exposure. Remarkably, in the CLASS trial, ibuprofen, the most commonly used drug in this category available over the counter, in fact, although the numbers are quite small, had more reported blood pressure problems associated with its use than seen with celecoxib or the other comparator diclofenac.
In contrast, with rofecoxib, it became quite apparent in the early development program that as one increased the dose, there was an increase in reported blood pressure problems. Not only was this seen in the NDA but, as you will see in the right hand portion of the slide, it was also seen in the VIGOR trial. It is important to appreciate that, demographically, the patients in these trials were quite different. In the VIGOR trial, it was rheumatoid arthritis and no aspirin used, whereas in the CLASS trial, it was osteoarthritis, rheumatoid arthritis and people with advanced cardiovascular disease such that they were using low-dose cardiovascular prophylactic aspirin. If we compare the time to onset of either myocardial infarctions or strokes or peripheral thromboses with or without a pulmonary embolus, we see that in the non-aspirin users in both the CLASS trial and the VIGOR trial, the time to onset of these thrombotic complications turn out exactly the same. In the VIGOR or the rofecoxib trial, at about 6 weeks, we began to see the emergence of a statistically significant higher incidence of the problems than was seen with the comparator drug, which in this trial was naproxen.
How does one explain this? The possibilities are that it was bad luck, it was just chance, or there were not enough patients in the trial to give what we call assurance of power for cardiovascular outcome issues. There was a lot of discussion as to whether naproxen might have given the cardiovascular protection but, in the massive epidemiology databases world wide, neither naproxen nor any other NSAID shows the protective role that aspirin at low doses contributes. And the FDA's current position is that aspirin remains the only proven drug for cardiovascular prophylaxis.
Clearly additional studies are needed, and some massive trial information is currently underway, so we will know this soon. Now, what about valdecoxib? In the current state of our understanding, we see no evidence of a cardiovascular outcome issue. It is interesting that the two sulfonamides do not have this problem, but we do begin to see it with the sulfone compound.
There are two additional epidemiology studies very recently reported. The one from Vanderbilt University, following up to the VIGOR trial, asks "Is it possible in a wide-spread group of individuals using these drugs, that at a high dose rofecoxib, the same sort of dose as used in VIGOR, that the signal may be there", i.e., an increase of myocardial infarctions or strokes? The data showed exactly that pattern; with rofecoxib doses exceeding 25 mg, there was a statistically significant increase in myocardial infarctions or other thrombovascular complications. This was very similar to VIGOR.
On the benefit part of the ledger, a recent epidemiology study coming from Canada, with a massive number of patients, asks a very simple question, "In day-to-day clinical practice operations, is that GI safety benefit clearly emerging?" They simply screened a massive number of people using the new coxibs and a massive number of people using the older traditional NSAIDs, and asked " Over a period of several months, is there a difference in hospitalization rate for GI bleeds?" And indeed, the results show that NSAIDs as a group were more likely to result in hospitalization. Further, with people taking nothing (controls) being one, rofecoxib came in at 1.9 and celecoxib came in at just the same as the background rate or control.
I point out that both rofecoxib and celecoxib were significantly lower with respect to hospitalizations then the NSAIDs in general. This is very reassuring, because it tells us yes, the main raison d'etre -- the initial one for the development of the coxibs -- is born out in a massive epidemiology database.
Recognizing that these drugs have an impact on blood pressure, I have undertaken a series of studies in the last year or two to try to sort out the blood pressure and edema part of the equation, cardiovascularly. We have initially done studies where the doses at which these two compounds have been used extensively. That would be 200 mg per day for celecoxib, 25 mg for rofecoxib. Then in individuals age 65 or older (who are more likely to develop edema) and only incorporating people who are being treated for blood pressure whose pressure nicely controlled, we asked the question "If they develop edema, can we clinicians see it at one week?" If they develop blood pressure problems, in other words, that drug/drug interaction I talked about earlier, can we see it at one week? If we are correct, we probably should. So we designed the trial to evaluate these issues at one week or two weeks and then to bring people back a month later and at 6 weeks to see if, in fact, some of these issues changed clinically. We have now done this in two massive trials representing over 2,000 patients studied in and we have often found that 70 to 80% of all new onset edema was indeed seen at the end of one week. The differences in edema rates were very reflective of what we had seen in new drug applications but now, for the first time ever, we had double blind side-by-side comparisons. And in each of these trials, the differences in edema were significant with less being seen in the case of celecoxib as is reflective of what we knew from the new drug applications.
What is particularly interesting biologically is that systolic blood pressure would be the dominant one of interest, since these were older individuals. Lo and behold, most of the change is there at one week. With one of the compounds, it is a little bit more in terms of blood pressure disruption at two weeks and it doesn't go away. It is still there at 6 weeks, so clinicians need to be aware of this. We see it reproducibly in both of these trials. A curious observation is that with celecoxib, pressure does not go up; it goes down. It is only a very slight reduction and as one would look at it you might say "well it's trivial," but it is telling us something and it is a message and we need to sort out.
A sizable number of the patients in the overall cohorts of these two trials did develop very impressive elevations of blood pressure, i.e., going up greater than 20 mm gets one's attention clinically. Also curious, it looks like something continues to happen over the duration of the study i.e., week 1 to 2 to 6 so that, at least in one of the coxibs, there is what looks a trend for an increase in the blood pressure problem. We don't know what causes that trend as yet. We are working on it, but when it comes to specific antihypertensive drugs, as we would predict based on the mechanism by which they works, ACE inhibitors and beta-blockers require the most attention, while there is no interaction with calcium channel blockers, which of course are not prostaglandin dependent with respect to the mechanism by which they work. Then why is it that we see such dramatic differences clinically, in edema and in blood pressure, between celecoxib and rofecoxib? We need to sort that out.
One of the nagging issues in these trials was that we were measuring blood pressure when patients came back to the clinic. We had not done it over 24 hours. So we have very recently completed an enormous blood pressure trial using 24-hour blood pressure monitoring, all done in Type 2 diabetics or selecting those at vascular risk, and all of these patients were treated with an ACE inhibitor. The bottom line for this trial is that the differences between these two coxibs are indeed maintained throughout the 24-hour cycle, and the differences are reflective of what we are seeing in the earlier trials. In this trial, the findings at 6 weeks and 12 weeks were comparable.
We are currently studying that issue. The coxib story is that gastrointestinal safety is the same for them all that the efficacy (once you make a little adjustment for dosing) is the same for all of these coxibs, and, of course, none of them has an effect on platelet function. That is the COX-2 story. But, cardiorenal and cardiovascular outcomes are different - with differences in edema and blood pressure. There must be something in the molecules or the metabolism or some biological explanation. So we are working currently on a series of different hypotheses in these issues. Clearly, these remain hypotheses until we prove them or disprove them in clinical circumstances.
For example, the COX-2 compounds celecoxib and valdecoxib are both sulfonamide compounds. So we have the question whether or not the enzyme carbonic anhydrase, which is in our kidneys, eyes and brain, to name just a few places, is of great physiologic importance. If deep in the kidney, carbonic anhydrase was inhibited by celecoxib, i.e., acting just like acetazolamide, which is the most potent inhibitor, then maybe this might influence retention of salt and water. This might also be an important influence on blood pressure. Lo and behold, the side chain to the immediate left of acetazolamide is the sulfonamide side chain. This is identical to that seen on celecoxib or valdecoxib, and you remember that is the side chain that anchors these drugs into the COX-2 enzyme. It may end up having completely different biologic actions elsewhere, such as in the kidney or in blood vessels that influence blood pressure and edema.
So we have relevant studies underway in the kidney and studies to see if we can perhaps reduce intraocular pressure, which has importance in glaucoma. There's a very important recent study showing an effect of some coxibs on endothelial function in blood vessels and that it is a beneficial effect. And there are some interesting studies underway evaluating the interaction of these drugs, coxibs of different variety interacting with cell membranes as they penetrate through the cell membrane to get access into the cells and then block the activity of COX-2.
So an exciting cascade of studies is underway. I am convinced that, within the next year or two, we will have a much better idea of how these drugs produce the variable cardiovascular impacts that I have just reviewed for you. In the meantime, in patients who are aged, who have treated hypertension or heart disease, just look for early edema formation, early pressure destabilization, and early onset of congestive heart failure, which is linked to the edema story. They are all easy to treat, and by doing so we may reduce cardiovascular outcome issues, but we still need to be hard-nosed clinicians and do the appropriate cardiovascular outcome trials.
Thank you. It has been a pleasure.
