Thursday, December 29, 2011

My name is FDA, and I approve of this message

As the new year begins and election season accelerates, we will be hearing phrases like that more and more.  In the context of a political commercial, it is often pretty clear what is being approved and the basis for the approval.

But what about new drugs?  How does the FDA decide whether or not to approve a new cancer drug, and what exactly is being approved?  In the era of targeted therapies, which are incredibly expensive to develop, and therefore incredibly expensive for patients and their insurance companies, these are crucial questions.



I think the case of bevacizumab (Avastin) is a perfect example.  Avastin was the first of a new class of drugs that treats cancer not by directly killing tumor cells, but instead by attacking the blood vessels that feed a growing tumor, essentially attempting to starve the tumor of oxygen and nutrients.  Avastin was approved by the FDA for the treatment of renal cell carcinoma (kidney cancer) based on its ability to prolong the life of patients with this disease.   Avastin was given a provisional FDA approval to treat metastatic breast cancer, based on its ability to delay progression of the tumor.  It was big news last month, however, when further studies failed to show an improvement in "overall survival" (meaning how long the patient lives from the time she begins treatment) in women with metastatic breast cancer, and the FDA withdrew its approval.

Avastin is in the news again today.  In today's issue of the New England Journal of Medicine there are two reports (links here and here) showing that Avastin prolongs Progression-free Survival (PFS; the time from beginning treatment until the tumor gets worse) but not Overall Survival (OS) in women with newly diagnosed ovarian cancer.  Based on these results, Genentech, the manufacturer of Avastin, declared that they will not seek approval from the FDA to treat women with ovarian cancer using Avastin, because they know approval will not be granted without an effect on OS.

Overall survival, an extension of life expectancy, is clearly the ideal for a cancer drug.  I certainly treat all of my patients with drugs that I expect will allow them to live longer (hopefully to cure, but even in patients who are not likely to be cured, I would like them to live longer).  But is that the only goal of cancer treatment?  Perhaps there is a benefit to an increase in PFS.  If I told you that your child was going to die in 10 months no matter what I did, but that there is a treatment that will keep his tumor from growing for 7 of those months, and that during those 7 months he would develop no new tumor-related symptoms, would you want him treated with it?  Probably that would depend on the side effects the drug causes, right?  If the drug was very toxic, you might decide that treatment isn't worth it, but if all that happened to your child was high blood pressure that was easily controlled by medication, you might say yes.

The primary result the FDA wants to see before approving a new cancer drug is an improvement in OS.  Although FDA guidelines do allow for approval based on an improvement in PFS, the degree of improvement required for PFS-based approval is much more strict than what is required for OS-based approval.  I believe that there may be cases where even a modest change in PFS would be a more appropriate standard.  Demonstrating a change in OS requires large studies enrolling many patients.  For rare diseases, studies like this may not be feasible.  In that circumstance, FDA approval based on a change in OS may be an unattainable standard, and PFS may be a reasonable alternative.

Some may ask whether lowering the standard will change the entire drug approval process, because PFS is an easier standard to reach.  Concerns could be raised that no pharmaceutical company will ever try to reach the OS standard (which is more expensive and more time-consuming) if PFS is sufficient to be granted FDA approval, and thus we will never know which, if any, drugs prolong patient survival. 

These are valid concerns, but I believe they can be addressed.  Perhaps a two-tiered approval process, with a lower tier for drugs that affect PFS and a higher tier for drugs that affect OS, with financial incentives for reaching the OS standard, would be a solution.  I'm sure there are others.  But the status quo gives the impression that if a drug doesn't change total survival time, it has no benefit.  And I'm quite sure that's not true.

Related Posts:
Access to experimental drugs for dying patients
Medicine from the Sea
I Can Buy it Over the Counter: FDA Followup

7 comments:

Lorin Decker Buck said...

Thanks for writing about this topic, David. I read the story in The Post this morning and had questions (what's new, right?). First question: If Avastin keeps the tumor(s) and therefore the disease from progressing, how is it that the endpoint (i.e., death) remains the same? To a layperson, that seems counterintuitive.

Will's Dad said...
This comment has been removed by the author.
Doctor David said...

Hi Lori,
PFS would translate into an overall survival benefit if tumors grew at a steady pace. Unfortunately, they don't. So, what we sometimes see is a tumor stay stable for a while (longer PFS), but once the tumor starts to grow, it grows more aggressively than previously, and so the overall survival time is unchanged.

You said "First question" which implies more are coming??

Hi Will's Dad,
Yes, the FDA is supposed to protect the safety of patients. As you (unfortunately, all to well) know, cancer treatments (all of them, not just chemo) cause side effects. So what the FDA is really doing is deciding whether the effectiveness of the drug is good enough to justify the inevitable side effects (safety). The issue is how to judge effectiveness. Is the most important measure of effectiveness how long the patient lives? Or is a delay in tumor growth sufficient?

The answer to your second question is variable... but in the FDA's published guidelines, for a drug to be approved it has to extend survival (or whatever endpoint is being used) more than some comparison. So if there is no treatment, any extension compared with how fast the disease "runs its course" is enough. If there is already a standard treatment, the new one gets compared to the standard.

emedicalpoint said...

Hi,

I have added your blog name in my blog list.

http://doctors-emedicalpoint.blogspot.com/

Erin Glankler said...

Thank you for another thought-provoking post. Here are my questions: does the FDA have to approve a cancer drug for each type of cancer individually? For example, could a drug be approved to treat AML but need a whole new set of trials to be approved for treatment of ALL, even if it is supposed to affect all leukemia cells the same way? And about how long does the process take to get a drug approved? During the time when it hasn't been approved or rejected, is it only available to people in specific clinical trials? And where do the Phase I, II, and III clinical trials come in? Thanks in advance.

Doctor David said...

Hi Erin,
Thanks for the questions. Answering them is a bit complicated... so I think that will be a topic for an upcoming posting. Sort of a follow-up to this one.

Olivier Sanar said...
This comment has been removed by the author.