Here we are again. Back in February (see the Feb. 27th post), the FDA asked for more data for Imclone's cancer therapy, Erbitux, saying the existing studies were not sufficient to approve the drug. The Wall Street Journal's editorial page threw a memorable fit about what they saw as the FDA's intransigence (to which piece I responded on June 18th.)
Now comes AstraZeneca to the FDA, similarly looking for fast-track approval of their similarly targeted (see the August 20th post) small molecule therapy, Iressa. The FDA's advisory panel met today, and ended up appoving the drug. The reviewer comments in the briefing documents suggested on Monday, though, that AZN was in for a hard time - and because of similarly unconvincing data.
And once again, the Journal weighed in today with a table-pounding editorial. Allow me to comment on their worldview:
Earlier this year, Astra-Zeneca reported very encouraging results from a couple of small trials totaling about 400 patients. The drug shrank tumors in 10 to 19% of lung cancer patients who had not responded to chemotherapy, and improved symptoms in about 40%. Most importantly, it appeared to add to the length and quality of life.
Sounds impressive when you put it that way! Let's look at the trial that AstraZeneca is using as their showpiece: 216 patients with non-small cell lung cancer who had failed standard chemotherapy. Unfortunately, the FDA contends that only 139 of those patients had truly not responded to earlier treatments. (The presence of these patients muddies the statistical evidence quite a bit, and it's just this sort of thing that helped to get Imclone's application in trouble.)
How many of these 139 responded to Iressa? Ten percent. Is a 10% response rate good enough to provide real-world clinical benefits? Is it enough reason to approve a drug on an accelerated schedule? Both those questions are very much open to argument, and even if the answer to the first one is "yes," reasonable people can believe that the answer to the second one is "no"
How about those improved symptoms? We have AstraZeneca's word that 40% showed some improvement in coughing and shortness of breath. But these results are not compared to any sort of control group, making them very hard to interpret - actually, it flat out makes it hard to determine that they're not an illusion. Other medications were also administered during the trial, and the study design makes it difficult (perhaps impossible) to say if those were responsible.
Most importantly, it appeared to add to the length and quality of life. Studies of Iressa in other cancers, such as head-and-neck, and yielding similar results.
That last statement is, unfortunately, true. The overall response rate in the head and neck trial (reported in May at the ASCO meeting) was 11%. As for quality of life, measuring that is hard enough under any circumstances, and measuring it without a control group is, I believe, basically impossible.
How about the other studies? The ones that this impassioned editorial doesn't mention? The two lung cancer studies that dosed Iressa plus standard chemotherapies - you know, the studies that actually had control groups? Iressa didn't have a 10% response rate in those. It had a zero per cent response rate - it didn't add to the effects of either one of the standard agents at all. The FDA found itself in the position of being asked to approve a drug that has completely failed to work in two studies, and shown marginal effects in two more.
. . .it is actually a repudiation of the speedier approval process the FDA has come to accept in drugs for terminal disease. . .that means approval base don smaller, so-called Phase 2 trials, along with compassionate use data to help establish safety. . .compassionate use programs can make a nit-picking bureaucrat's life difficult, by getting good drugs into the hands of many doctors and patients, who thereafter become a constituency urging formal FDA approval.
Nice use of the adjective. Compassionate use can also make a drug company's life difficult, as an avalanche of requests comes pouring in. It can make a scientist's life difficult, because the data obtained are usually of poor quality - heterogeneous and not well controlled.
But let's get emotional, since the Journal does: compassionate use can also make things damned difficult for terminally ill patients and their families. Remember, patients know they're getting an experimental drug, their last chance for survival. The FDA heard from some of those patients today, and I'm glad that they're still here to testify. And let's talk about those "good drugs": remember, in the lung cancer trials, 90% of the patients who took Iressa as monotherapy did not respond. And 100% of the patients who took it in combination therapy saw no added benefit. We didn't hear from these patients today. Many of them aren't around to talk about how hopeful they were that this new drug might save their lives.
The point here isn't to quibble with the FDA's suggestion yesterday that the data could be better. Studies can always be larger.
Yep. That helps. And drugs could always show some convincing efficacy; that would help, too. That's one good definition of the data being better.
The point is that this isn't a good reason, and certainly not an ethical one, for delaying approval. Particularly in cases of terminal disease, any safe drug with even a hint of effectiveness should be brought to market as quickly as possible.
An uncontrolled trial of powdered milk could end up showing "a hint of effectiveness," guys. Iressa's better than that, of course, but where do we draw the line? If we don't insist on solid statistics from well-run trials, we might as well just throw open the floodgates. The FDA is trying to get companies focused on proving that their therapies actually do something, while (because of the state of the industry) many companies are focused on trying to get their drugs on the market by the quickest route possible, cutting the clinical trials as thin as they can. Imclone's a much more spectacular example of that, but AstraZeneca should still count itself lucky to have gotten Iressa through.
Reading the editorials that the Journal has pumped out over the last few months, you get the mental picture of mustache-twirling FDA baddies snickering as cancer patients expire all around them. It's a caricature of the truth. It feels odd for me to stick up for a regulatory agency, but I'll stick up for this one. Speaking as a researcher in the industry, I can say that the FDA drives us all nuts, but we need them. We need people to poke holes in the studies, to question the data, to give us a hard time. It's called science. It's how we've made it as far as we have.