About this Author
DBL%20Hendrix%20small.png College chemistry, 1983

Derek Lowe The 2002 Model

Dbl%20new%20portrait%20B%26W.png After 10 years of blogging. . .

Derek Lowe, an Arkansan by birth, got his BA from Hendrix College and his PhD in organic chemistry from Duke before spending time in Germany on a Humboldt Fellowship on his post-doc. He's worked for several major pharmaceutical companies since 1989 on drug discovery projects against schizophrenia, Alzheimer's, diabetes, osteoporosis and other diseases. To contact Derek email him directly: Twitter: Dereklowe

Chemistry and Drug Data: Drugbank
Chempedia Lab
Synthetic Pages
Organic Chemistry Portal
Not Voodoo

Chemistry and Pharma Blogs:
Org Prep Daily
The Haystack
A New Merck, Reviewed
Liberal Arts Chemistry
Electron Pusher
All Things Metathesis
C&E News Blogs
Chemiotics II
Chemical Space
Noel O'Blog
In Vivo Blog
Terra Sigilatta
BBSRC/Douglas Kell
Realizations in Biostatistics
ChemSpider Blog
Organic Chem - Education & Industry
Pharma Strategy Blog
No Name No Slogan
Practical Fragments
The Curious Wavefunction
Natural Product Man
Fragment Literature
Chemistry World Blog
Synthetic Nature
Chemistry Blog
Synthesizing Ideas
Eye on FDA
Chemical Forums
Symyx Blog
Sceptical Chymist
Lamentations on Chemistry
Computational Organic Chemistry
Mining Drugs
Henry Rzepa

Science Blogs and News:
Bad Science
The Loom
Uncertain Principles
Fierce Biotech
Blogs for Industry
Omics! Omics!
Young Female Scientist
Notional Slurry
Nobel Intent
SciTech Daily
Science Blog
Gene Expression (I)
Gene Expression (II)
Adventures in Ethics and Science
Transterrestrial Musings
Slashdot Science
Cosmic Variance
Biology News Net

Medical Blogs
DB's Medical Rants
Science-Based Medicine
Respectful Insolence
Diabetes Mine

Economics and Business
Marginal Revolution
The Volokh Conspiracy
Knowledge Problem

Politics / Current Events
Virginia Postrel
Belmont Club
Mickey Kaus

Belles Lettres
Uncouth Reflections
Arts and Letters Daily
In the Pipeline: Don't miss Derek Lowe's excellent commentary on drug discovery and the pharma industry in general at In the Pipeline

In the Pipeline

« Complaints From a Fossil | Main | The Tar Pit Beckons »

October 20, 2005

This Had Better Be Good

Email This Entry

Posted by Derek

I wrote a brief wrap-up on the FDA's concerns about the new Bristol-Meyers Squibb / Merck diabetes drug Pargluva (muraglitazar). It's officially "approvable", but the FDA wants more cardiovascular safety data before it can be sold. But just this morning the JAMA web site has rushed out an article from a team at the Cleveland Clinic on the drug's clinical trial data. (Accompanying editorial here). It's very disturbing, in more ways than one.

At the time, I said that "By my reading, the cardiovascular event profile of the drug subjects looks slightly but noticeably worse than that of the placebo group. There are plenty of possible extenuating factors, and the number of patients involved is small, but I think that this is going to be a problem for the companies during the FDA hearing. Here's the list of questions the FDA has proposed for discussion (PDF again), and you can see that edema and cardiovascular safety loom large. . ." That's fine, as far as it goes, but I didn't dig far enough into the data, and I wonder if the advisory panel did, either.

What the authors of this new paper have noticed is the number of patients taking a low dose of muraglitazar - lower than the companies ended up seeking approval for. They didn't show enough beneficial effects for that dose to be worthwhile, but since muraglitazar's cardiovascular problems appear to be strongly correlated with dose, these patients also had no cardiovascular events at all. The problem is that these patients were included in the risk calculations, and that makes the drug look safer than it would be under real-world conditions.

The Cleveland group's recalculations now put the risk of cardiovascular events with clinically relevant doses of muraglitazar at 20% higher than the placebo group, and at 67% higher than the combined placebo-standard of care group. (That includes patients treated with pioglitazone, a PPAR-gamma compound that's been approved for some years now). Put that way, this sounds like a huge increase, but it's important to remember that both of these figures, though real, are pretty small. The placebo group had about 34 events per 1000 patient years, and the drug treatment group, in the new analysis, had around 40 events. So, back-of-the-envelope, for every thousand patients on muraglitazar, you might expect an extra 6 cardiovascular incidents per year. The similarities to the Vioxx data are not hard to spot, and in fact the authors of this paper have been very much involved in that controversy as well.

But I'm not going to push that comparison. This is a different case than Vioxx, a drug that (for many patients) really does seem to do more than existing compounds can. The problem here is that muraglitazar (and all the PPAR alpha-gamma compounds that have gone into development) was supposed to be better for cardiovascular outcomes than the plain PPAR-gamma compounds that are already out there. Needless to say, it was also supposed to be better than a damned placebo, which it isn't. The entire dual-PPAR-agonist idea is in trouble. The whole point of adding PPAR-alpha activity was to improve blood lipid profiles, and pretty much the whole point of doing that is to improve cardiovascular health. The first part is working, but the second part, the important part, just doesn't seem to be happening. Looking at the data, I find it hard to imagine why anyone would take muraglitazar over the exisiting therapies, when there's no evidence for what is supposed to be its main advantage.

As if that weren't bad enough, there's also a background worry about cancer rates with PPAR compounds. The muraglitazar data aren't totally reassuring on this front, either. Other compounds in this class died because of carcinogenicity in long-term rodent studies, and muraglitazar is the first compound to actually make it past such studies. But the data submitted to the FDA show that rats given the compound at high doses do indeed show bladder cancer - it just seems to be less of a problem than it was for the earlier compounds from Merck, Kyorin, Novo, Dr. Reddy's, et al. For a marginal compound, though, this is a real issue.

I don't necessarily think that the people at BMS (and Merck, a latecomer to this compound) were sitting around wondering about just how to snow the FDA. But it would certainly cheer me up if I could rule that out, wouldn't it, now? At the very least, the companies weren't being as critical of themselves as scientists have to be, and they've committed a mistake that would flunk a PhD candidate or get a paper tossed back from a well-refereed journal. Something has gone seriously wrong here. We're supposed to be better than this.

What on Earth were they thinking, submitting data in a way that makes it look like they were trying to pull a fast one with the cardiovascular risk factors? Now, of all times? Who knows, maybe people at BMS had just convinced themselves that things were fine, somehow - the capacity for human self-deception is limitless. But didn't anyone at Merck turn pale and have to sit down when they saw these numbers? I didn't realize how bad the situation was back in September, but even then I wondered about this, saying: "I can't predict which way this one is going to go, and neither can anyone else. But post-COX-2 is a bad time to be coming to the FDA with possible low-level cardiac risks in your clinical data. . ." Now that the risks look even worse, I'm baffled. You people want the sky to come down on your heads?

Comments (6) + TrackBacks (0) | Category: Clinical Trials | Diabetes and Obesity | The Dark Side


1. John Johnson on October 21, 2005 11:42 AM writes...

Absolutely nuts. For better or worse, the FDA has had a rather itchy trigger finger lately when it comes to clinical holds if cardiovascular safety is involved. I'll be very interested to see the followup on this, as will just about everybody else in the industry.

Permalink to Comment

2. Anonymous on October 21, 2005 12:38 PM writes...

all you have to do is check the fda website. . both the company presentation is there as well as the fda review. there is no "cover up" here. the data is clearly presented and the death data is there for everyone to see: 18 deaths in the pargluva arm, 2 in actos arm, 1 with placebo, 9 cv deaths in pargluva, 0 with actos, 1 with placebo, 7 cancer deaths with pargluva, 1 with actos, 0 with placebo.

Permalink to Comment

3. Derek Lowe on October 21, 2005 1:44 PM writes...

Anon, that's true, and those are the numbers I was looking at in September when I thought that there might be trouble. But even those don't match up with the JAMA paper's re-analysis. I quote, with emphasis added:

Table 3 illustrates the adverse events occurring in the 5 reviewed clinical trials conducted in diabetic patients with muraglitazar dosages of 5 mg/d or less. Table 4 illustrates the event rates, RRs, and P values for muraglitazar compared with pioglitazone or placebo. The primary outcome measure (all-cause mortality, nonfatal MI, or nonfatal stroke) occurred in 35 of 2374 muraglitazar-treated patients (1.47%) vs 9 of 1351 control patients (0.67%) (RR, 2.23; 95% CI, 1.07- 4.66; P = .03). A more specific outcome measure, substituting cardiovascular death for all-cause mortality, occurred in 27 of 2374 muraglitazar-treated patients (1.14%) vs 7 of 1351 control patients (0.52%) (RR, 2.21; 95% CI, 0.96-5.08; P = .06). A more comprehensive outcome measure adding CHF and TIA events to the composite yielded an incidence of 50 of 2374 for muraglitazar-treated patients (2.11%) vs 11 of 1351 control patients (0.81%) (RR, 2.62; 95% CI, 1.36-5.05; P = .004).

Permalink to Comment

4. jeep on October 26, 2005 3:18 PM writes...

The CV issue may not be real. I havent heard of anyone coming up with a good mechanism why it should cause more CV events/deaths, and the numbers are pretty small. the JAMA paper was data culled from the FDA website, so they really dont have all the relevant data, timelines, etc. I think more data will tell the real story.

Permalink to Comment

5. AG on October 27, 2005 1:14 PM writes...


This comment acknowledges the huge shortcoming a diabetic agent would have if only afforded a monotherapy or severely restricted combination therapy indication(s). None the less, I found Dr. S. Nissen's commentary lacking in missing the "obvious" that he is want to see as he was in casual observation of the muraglitizar CV AE data.
>>>>JAMA excerpt form discussion with my comments in ( ) >>>
The possibility of an interaction between muraglitazar and other antidiabetic therapies must also be considered. Most adverse cardiovascular events occurred in studies in which muraglitazar was combined with glyburide or metformin. .....(No, essentially ALL the adverse events occurred with a metformin or glyburide addition. Very interesting, how Dr. Nissen seems to minimize this as he magnifies surrogate markers, ala IVUS when it serves him yet minimizes them in conventional diabetes study design here. There appear with this PARTIAL DATA SET to be only 2 TIAs - one at 1.5mg (~250 pts.)and one with a 5mg dose (~250 pts) and 1 Stroke in the muraglitizar monotherapy arms 0f ~ 500 pts total. The subtherapeutic 0.5mg dose had zero events. Interestingly, there were 1 MI and 1 Stroke in the pioglitazone 15m arm of ~ 25o pts total. Review the Cardiovascular events table.)However, the number of events in any single study is too few to draw definitive conclusions about the RRs of muraglitazar with or without concomitant therapy with other agents. (This statement appears to be INCORRECT with regard to WITHOUT concomitant therapy. HOW could someone like Dr. Nissen miss this? I don't believe he could have overlooked anything. Do you?) Other dual-PPARs in development may or may not exhibit similar hazards. (I don't think he has DEFINITIVELY identified a HAZARD. Has he? And given the genetic – efficacy complexities known with PPARs/PARs, isn’t it reasonable to look closely at monotherapy events, alone and in comparison to combination? They have simply applied what statistical tools were suitable to a partial data set...and choose the words of "exhibiting...hazard." The most obvious statement in the world for a researcher to say is ...more study is needed. But VIGOROUS STATISTICAL ANALYSIS OF ALL DATA, including data since NDA submission to present is what is needed. If that analysis is not enough to guide the FDA with a safety first mindset, THEN I can buy Dr. Nissen’s “saving” grace. I see many sides to the players involved, including political complexity being applied to this ‘situation’ from CCF to the FDA. I wish it were pure science, a ‘high road’, on ALL sides. Time may not tell.

Permalink to Comment


Remember Me?


Email this entry to:

Your email address:

Message (optional):

XKCD on Protein Folding
The 2014 Chemistry Nobel: Beating the Diffraction Limit
German Pharma, Or What's Left of It
Sunesis Fails with Vosaroxin
A New Way to Estimate a Compound's Chances?
Meinwald Honored
Molecular Biology Turns Into Chemistry
Speaking at Northeastern