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

« Of Drug Research and Moneyball | Main | Regeneron Finally Makes It to the Market »

November 21, 2011

Avastin Coverage, Amended

Email This Entry

Posted by Derek

In response to the press coverage on the FDA's Avastin decision on Friday, a reader forwarded a revised and extended version of the New York TImes article that appeared soon afterwards. Here are some excerpts, which I think get across the thinking of many medicinal chemists and drug researchers. His contributions are bolded for emphasis, although it's not all that hard to see where the original ends and his revisions start.

"The commissioner of the Food and Drug Administration on Friday revoked the approval of the drug Avastin as a treatment for breast cancer, ruling in an emotional issue that pitted the hopes of some desperate patients against the statistics of clinical trials, two things that should never be compared, because that would be stupid.

The commissioner, Dr. Margaret A. Hamburg, said that the drug was not helping breast cancer patients to live longer or control their tumors, but did expose them to potentially serious side effects such as severe high blood pressure and hemorrhaging, making her decision very easy.

. . .The F.D.A. “recognizes how hard it is for patients and their families to cope with metastatic breast cancer and how great a need there is for more effective treatments. But patients must have confidence that the drugs they take are both safe and effective for their intended use.” Also, they shouldn’t take drugs that don’t work, so we thought that is was important that they stop eating 88 thousand dollar magic beans, and instead use drugs and medical procedures that work.

. . .Avastin will remain on the market as a treatment for other types of cancers, including forms of cancer that it actually treats, so doctors can use it off-label for breast cancer if they hate science. But some insurers might no longer pay for the drug, which would put it out of reach of many women because it costs about $88,000 a year.

Yet pressure came from the other direction as well the outcome was certain once the statistical analysis was done, so this could have been a much shorter article. The administration had pledged to make scientific decisions on the basis of science, which seems like a pretty good idea as well. That made it difficult for Dr. Hamburg to go against the pharmaceutical lobby, and easy to accept the conclusions of the F.D.A.’s own staff and the strong recommendations of the outside experts on its advisory committee.

. . .An initial clinical trial showed that Avastin, when combined with the drug (paclitaxel), delayed appeared to delay the progression of disease by about five and
a half months, compared to use of paclitaxel alone. However, the women who received
Avastin in the study did not live significantly longer and they suffered more side effects. As an example, high doses of sodium cyanide completely stops the progression of disease almost immediately and permanently, though women who receive this treatment don’t live as long and suffer more serious side effects from the control group.

. . .Many breast cancer specialists say that Avastin does appear to work very well for some patients, but that the effect gets drowned in a clinical trial that looks at overall results. Some doctors and patient advocates argued the drug should remain available for that reason. Representatives from large sugar companies also noted that their drug, placebo, works very well for some patients, but that effect is usually gets drowned in a clinical trial that looks at overall results. The FDA has yet to approve placebo for the treatment of breast cancer."

Comments (23) + TrackBacks (0) | Category: Cancer | Press Coverage | Regulatory Affairs


1. -=GiMP=- on November 21, 2011 2:49 PM writes...

Love it! Sign this man up.

Permalink to Comment

2. pete on November 21, 2011 2:52 PM writes...

Apt & amusing edits but then, ultimately, misses the target. Save the snide tone for a true boondoggle. Avastin ain't it.

Permalink to Comment

3. CMCguy on November 21, 2011 3:11 PM writes...

Are you actually endorsing these snide comments as I view as load of crap. If these represent the thinking of many medicinal chemists and drug researchers then perhaps they do need to get out of the lab and talk to the people who run clinical trails as and more so patients. That world is neither stupid, always has clear cut answers/data or favor euthanasia. Although still a very murky picture to trumpet this current decision only a objective science is misguided. To discount the observations of breast cancer specialist that works well for some patients would imply more science is need to address how to best use for this population. Even the NYT article mentions the follow-up studies were actually not against Taxol but different chemotherapy drugs and wouldn't that in itself suggest design issues in the follow-on studies. Still more questions that answers and I for one see no major problem with patients being consented with the high risk low probability of this option so will be between them and their doctor which is typically where this issue should end up.

Permalink to Comment

4. Sleepless in SSF on November 21, 2011 3:22 PM writes...

I think the FDA made the correct (only rational?) decision. However, I'm concerned that the snark in the last paragraph quoted is entirely off the mark. It doesn't seem impossible to me that there is a sub-population for whom Avastin is efficacious but that we don't have a marker to identify them.

If that is the case, are women who are benefiting from Avastin treatment going to be cut off by their insurance companies? I haven't seen anything addressing continuing treatment; does anyone know if there's anything to prevent that from happening? The balance of evidence dictates that we not treat new metastatic breast cancer patients with Avastin, but grandfathering (grandmothering?) current patients could prevent pointless agony for some women.

Permalink to Comment

5. MadScientist on November 21, 2011 4:22 PM writes...

The FDA definitely came to the only correct decision, which is to remove the breast cancer "umbrella" approval, preventing many patients from being treated with a very expensive drug that may not work or possibly cause more harm. It's also a clear illustration of the "personalized" aspect of disease therapy. Avastin obviously works in a small subset of patients, and withdrawing the approval does not prevent oncologists from using it off-label for those cases that benefit. Medical science has learned that the efficacy of a drug isn't just a statistical occurence, but that a well-defined (though yet unknown) factor is responsible for these variations. The FDA's decision simply means that labeling changes will be instituted, and it will not be considered a first-line treatment for every newly-diagnosed breast cancer patient.

Avastin needs to be studied more intensively to determine which populations might benefit. Cancer of any sort, besides being a nightmarish diagnosis to receive is an equally miserable disease to treat effectively. What works for Patient X at the moment may not work anymore a few days from now, as the surviving cells mutate and elicit new defenses.

The practice of medicine should always be considered an assortment of large, ongoing clinical trials. The outcome from this one clearly came up short.

Permalink to Comment

6. Algirdas on November 21, 2011 5:16 PM writes...

CMCguy "works well for some patients"

Sleepless in SSF "sub-population for whom Avastin is efficacious"

MadScientist "Avastin obviously works in a small subset of patients"

Does any of you have a reference for this? Point me to a survival data plot which has a tail of outstandingly long-lived survivors in Avastin treated vs control group.

If not, are all three of you oncologists working on breast cancer, who know of a systematic, statistically meaningful study with such survival data that is about to be published or is in preparation?

If not, does any of you work at Genentech and have inside knowledge of trial data?

I suspect none of the above options apply. Far, far more likely, you are just spouting nonsense, essentially claiming that Genentech and FDA are idiots, for not presenting sub-population data (data so well known that even random commenters on the internet know all about it) and for ignoring it, respectively.

Permalink to Comment

7. Vader on November 21, 2011 5:29 PM writes...

"The FDA has yet to approve placebo for the treatment of breast cancer."

On the evidence, they have, in fact, done so at least once already.

Permalink to Comment

8. qetzal on November 21, 2011 8:15 PM writes...

I second Algirdas' comments. Where's the evidence that Avastin "obviously works" in some patients?

Permalink to Comment

9. alaric on November 21, 2011 9:54 PM writes...

At ESMO there was some information out in thje personalized medicine session about a soluble VEGFA assay Roche is working on that separates out the PFS curves pretty good, for recent enough studies where the samples haven't been compromised for sVEGFA by older preservatives. (OS is never going to happen. There are like eight lines of therapy in MBC.) There's an AVADO sVEGFA retrospective from SABCS 2010. It'll take forever though to run a proper regulatory-acceptable prospective study, even if they can accrue to it after all this.

Permalink to Comment

10. Sleepless in SSF on November 21, 2011 11:13 PM writes...

@Algirdas: Where is all the hostility coming from? I said I didn't think it was impossible that a sub-population existed but my main question was what happens to such a group when approval for an indication is revoked.

We should be very careful about demonstrating that no sub-populations exist if we're going to use that as a basis for withdrawing treatment. I'm not sure I'd react all that kindly if you insisted that treatment had to be withdrawn from my non-progressing wife or daughter, especially if your justification was that they couldn't possibly be benefiting since they didn't constitute a "tail of outstandingly long-lived survivors."

There's a big difference between withdrawing approval for future treatment and withdrawing treatment from current patients.

Permalink to Comment

11. Jose on November 21, 2011 11:34 PM writes...

Check out something called "evidence-based medicine"- it's the newest fad sweeping the rational universe. The elves, unicorns and Orcs might not like it, though...

Permalink to Comment

12. NoDrugsNoJobs on November 22, 2011 10:17 AM writes...

There is no question that avastin was active against the tumors, even in the trials showing less of an effect on progression free survival, the progression free survival was very statistically significant. With regard to actual survival and the question of whether some women did better on the drug, I believe there was a trend in the combined studies towards increased survival but only a trend. But here's the thing...the drug has a lot of serious side effects, some of them fatal which were more than placebo. Given the fact that the drug showed a statistically significant increase in progression free survival, an overall trend towards increased survival despite an increase in fatal side effects, seems logical to assume that the drug is helping some women just as it is hurting some women (through side effects). The net result is an increase in progression free survival but no significant effect on overall surrvival. The fact that the drug was approved and its removal controverial is not surprising because it is a tough question - this drug is still approved by the EMEA and many organizations of practicing oncologists. It does work, its just a question of net benefit and that is as much a social policy issue as anything else. Trying to over simplify the issue does not do justice to the importance of the issue nor to the intelligence of those taking that route.

Permalink to Comment

13. CMCguy on November 22, 2011 1:52 PM writes...

#6 the "works (very) well for some patients" is taken directly from the post as a quote repeated from the NYT article. This theme/statement has been in most the accounts I have read about Avastin use in MBC so while I can't point to specific studies when such is attributed the "breast cancer specialist" (read oncologist, which I am not but spent most career around such) I do suspect has some relevant value. I think there may have been info in the recent FDA meeting that presented this as suggested trend but was not "strong or conclusive" enough. Sure evidence is more anecdotal than statistically validated however obtaining "systematic, statistically meaningful study with such survival data" is much easier to talk about than implement. I further presume Genentech (whom I do not work for) and others have attempted to identify sub-population bio-markers to foretell best chance for success and have failed find anything (yet I trust).

Permalink to Comment

14. Cellbio on November 22, 2011 2:34 PM writes...

@ Nodrugs

Overall, excellent, excellent post. I do think, however, that the logic you put forward deserves a bit more fleshing out, in my humble opinion, as someone who has spoken to patients, and doesn't want to inflame the discussion, just pursue the concepts more deeply.

Does Avastin really help "some women just as it is hurting some women"? Or, is it possible that the net benefit, as you nicely put it, would always be as measured so far because there is not a subset of better responders, just a population that shows the mix of pharmacological impacts which include a very modest benefit across much of group, with a low rate of adverse events pulling against that group benefit?

I know personalized medicine has great promise, and some real examples, but this core issue of responder subset vs. modest benfit to a larger group is one I have faced personally in trying to practice the craft. I saw first hand how rooted the concept of "responder subset" was in the minds of sponsors and our small company's founders. So much so, that when I presented clear and obvious evidence that there was no responder subset, just a small, measurable but clinically insignificant pharmacological impact, no one would agree and we all went forward to find the biomarkers associated with this non-existant group. And believe me, it is easy to find candidates in post-hoc analysis, as most any cut of the small trial population is not associated with a uniform distribution of efficacy or adverse events.

It is from this perspective that I appreciated Algirdas' robust request for evidence of such a group, as I fear it has become common language to express small clinical benefit as evidence itself for responder subsets. I didn't hear hostility, nor disrespect for patients or the clinical trials, just frustration.

Permalink to Comment

15. soon2bdesigner on November 22, 2011 5:01 PM writes...

A very important post, thanks. Another point to consider is that NONE of the trials so far have evaluated Avastin as a stand alone therapy, all have been combination therapies. If you consider that there will always be a heterogeneous response in a population, even for an approved drug, then it's easy to see how some patients could be seen as "super responders." But the fact of the matter is, Genentech has been unable to show that there are more "super responders" in the study arm [drug + Avastin] than there are in the control arm [drug only]. The studies were not designed to show this effect in the first place - it's an afterthought, as far as I can tell.

Finally, let's remember what "progression free survival" is NOT. It is NOT an extension of life, and it is NOT an improvement in the quality of life. In fact, it IS (generally) a measure of the size of a tumor, nothing more, nothing less. Surveys administered to patients in all studies showed that quality of life was NOT improved by addition of Avastin. It's worth debating whether ANY magnitude of progression free survival, even if highly significant, in the absence of improved quality of life or overall survival, is reason to approve a drug, but that's a debate for another day.

Permalink to Comment

16. metaphysician on November 22, 2011 5:45 PM writes...


That's what I was thinking, actually: "How do you know the 'super responders' are actually benefiting from Avastin? How do you know they wouldn't be 'responding' better than average on any drug, or none at all?"

Permalink to Comment

17. WB on November 22, 2011 8:23 PM writes...

Dripping with sarcasm, but spot on in its assessment. Great re-write, I love it!

Permalink to Comment

18. qetzal on November 22, 2011 9:28 PM writes...


From a quick skim of FDAs decision (linked in Derek's previous Avastin post), I don't see a trend toward increased overall survival. Some trials showed nonsignificant increases of a few months in the Avastin group. But other trials showed the opposite: nonsignificant decreases in overall survival for the Avastin group.

Permalink to Comment

19. lgg on November 22, 2011 9:46 PM writes...

I was offered a slot in an Avastin trial (TNBC) in 2010. I said no, no, no - the data were not convincing that the chance of being a super responder was better than being a super get-worser...or at least, of having awful side effects and not living any longer. This post hit the nail on the head. I linked it to my FB so all my friends who think this was a travesty for me know that I have a brain in my head. Resources could go to therapies that might really work, which is not avastin.

Permalink to Comment

20. Northwest AJ on November 24, 2011 3:58 AM writes...

All the while I've just been curious as to whether anyone's been trying to, oh, examine the genes or receptors in in tumors that seem to be responding and see whether they can find a particular difference in them from the ones that definitely aren't. Because that would be the difference between anecdote and data, wouldn't it?

Permalink to Comment

21. late to the party on November 26, 2011 9:39 AM writes...

@20 agreed. Have any patients been genotyped? Does the small group of patients who respond well (if truely existant) have some key similarities in their cancer genome?

Permalink to Comment

22. Cellbio on November 26, 2011 11:03 AM writes...

@20 and 21,

Making lots of measurements does not really change the nature of the observation, so it remains observational, and maybe not just an anecdote, as it could yield a hypothesis drawn from a group, but not really solid data from a properly designed study.

One challenge, if we assume there is a responding group, is that these true responders are embedded in a placebo responder group. There is no way to sort the "true" responders to therapy from the ones who responded by chance but happened to be in the treatment arm. Though one can mathematically subtract placebo rates from treatment arm rates and come up with a delta, often used to suggest responder subset, one can't subtract them from the genomic analysis. With clinical results that have small deltas over placebo, the types of studies that fuel this sort of analysis, a repsonder definition may yield more patients in the placebo group than would be found in the delta over placebo (e.g. placebo rates is 20%, treatment effect is 30%, placebo out numbers true responder 2:1). I think you'd be lucky to have a delta that is larger than placebo. Trying to run genetic analysis in tiny groups polluted with as many or more chance responders than true responders makes it tough to find anything robust.

Permalink to Comment

23. Northwest AJ on November 27, 2011 6:09 AM writes...

On the contrary, making lots of measurements changes what observations you're able to analyze statistically, which might make the picture totally different. Or it might not. But it takes trying to find out.

One way to test whether a particular gene/gene product is having an effect on the efficacy of a drug, I would think, would be to analyze what percent of non-responders have that gene in their cancers. (There aren't generally placebo arms in cancer drug trials for ethical reasons.)

If, say, 10% of people respond to a particular drug - a low rate and not worth keeping the drug around, by itself - but when checked for $gene, it's found that, say, 90% of responders have it and only 1% of non-responders do, then you've just found that something which previously appeared useless could actually be really useful in context of genotyping.

Of course, all of this is totally speculative; they would need to go needling around in a haystack to analyze the data in this fashion (although since it's common for breast cancer to have some genetic testing done, they might be able to run it through a few analyses based on what is known, if they haven't already.) I just feel that it's worthy and responsible to at least try to investigate differential genetics in light of mixed field reports. I'm put in mind of this because I was just reading Mukherjee's "The Emperor of Maladies" and these kinds of things have been a hugely significant factor in previous cancer discoveries.

Permalink to Comment


Remember Me?


Email this entry to:

Your email address:

Message (optional):

The Last Post
The GSK Layoffs Continue, By Proxy
The Move is Nigh
Another Alzheimer's IPO
Cutbacks at C&E News
Sanofi Pays to Get Back Into Oncology
An Irresponsible Statement About Curing Cancer
Oliver Sacks on Turning Back to Chemistry