Sekeres, Mikkael A. MD, MS
Two recent events reminded me of just how conflicted we, as oncologists, are about the role of Phase 3 studies in the cancer drug approval process.
Now, recall, a Phase 3 study is by definition a randomized trial in which patients are assigned to one of two treatment arms: an experimental arm, which includes a drug or drug combination (or treatment combination, if radiation therapy or surgery is being explored) which may be the latest and greatest thing in cancer therapeutics (let's call the drug “tumorkillimab”); and a control arm, which includes the standard therapy, or therapies, if no standard exists, or placebo/best supportive care, if ethical.
Phase 3 studies tend to require a lot of patients (often, hundreds), with the total sample size determined by the endpoint being studied, and the amount of difference in that endpoint thought to be clinically meaningful. Ideally, the endpoint would be overall survival—the ultimate measure of success in oncology, and one that is pretty cut-and-dry: you're either alive or dead. Trials looking for survival differences of a couple of weeks require a lot more patients than those measuring differences of a couple of years.
Phase 3 trials can take a long time to complete, particularly for slower-growing cancers. I remember being told, as a fellow, that if I were interested in studying whether tumorkillimab prolonged survival in early-stage chronic lymphocytic leukemia, I should design and start the trial immediately, so that I might have results by the time I retire. So, to save time and money, interim endpoints, such as response rates, or disease- or progression-free survival, often are used. Whether or not these routinely translate to overall survival is debatable.
Bevacizumab for Front-Line Treatment of Metastatic Breast Cancer
This issue came up with bevacizumab for front-line treatment of metastatic breast cancer, when the drug came before the Oncologic Drugs Advisory Committee (ODAC) at the Food and Drug Administration, in July of this year. I was a member of that panel. The findings from that meeting have been extensively discussed. The initial progression-free survival advantage of 5.5 months for bevacizumab + paclitaxel compared with paclitaxel alone dropped to <1.5 months in what were supposed to be confirmatory Phase 3 studies of bevacizumab combined with docetaxel or other drugs. It almost goes without saying that differences in survival were negligible, and not significant.
What to do? Part of ODAC deliberations include a public comment session, when anyone—and I mean anyone—can come up to a microphone and speak for or against recommending a drug's approval. This is our democracy in action. Woman after woman, all breast cancer survivors, testified that they were alive today, and able to participate in major family events, because they were treated with bevacizumab. These are heart-wrenching stories. But are individual stories enough to counter the science of well-performed trial results?
ODAC voted unanimously, or nearly so, that bevacizumab should not advance from accelerated to full approval in combination with paclitaxel or with docetaxel or other drugs, and that its breast cancer label should be withdrawn from the market.
2nd Event: New York Times Article about Two Cousins in PLX4032 Melanoma Trial
The second event was a story reported in the New York Times in September that profiled two cousins, both with advanced melanoma, who were enrolled on a Phase 3 study of PLX4032, a drug that demonstrated impressive responses in early-phase studies of melanoma patients with BRAF mutations, compared with dacarbazine.
Typical of such articles, this piece alternated paragraphs about the personal aspects of the cousins (they played together in swimming pools as kids; one received PLX4032, the other dacarbazine) with technical information and quotes from docs.
Although for the most part balanced about the advantages of Phase 3 studies (their scientific rigor, for example, and the importance of confirming initial promising results in what may have been more highly selected patient populations) and their disadvantages (such as the delay in providing disease-modifying therapy to desperate cancer patients), the story ends with the cousin receiving PLX4032 (read: tumorkillimab) attending the funeral of the cousin on the control arm. Apparently, the paper of record feels we should provide access to experimental drugs for all cancer patients.
What To Do?
Mikkael A. Sekeres, ...Image Tools
What to do? Let's be honest—we all care deeply about our patients, and want them to get better and to be cured, and we try like hell to get them the latest and greatest promising drugs. I take it as a point of pride that I have clinical trials of novel agents to offer patients who travel some distance to see me in Cleveland, and that I have treated patients with experimental drugs years before those drugs eventually got approved by the FDA. I was able to make a difference in the lives of cancer patients with drugs that were not yet widely available. Isn't that why we all went into oncology?
We could design trials that allow non-responding control arm patients to cross over to receive experimental drugs. This, however, compromises our ability to assess survival differences, as ultimately most patients would receive the drug being studied, diluting any potential advantage to the initial experimental study arm. Or, we could also insist that Phase 3 trials using interim markers, such as progression-free survival, also include quality-of-life endpoints. That way, if a survival advantage could not be demonstrated, at least we could ensure that treated patients were living better, if not necessarily longer.
Or, we could be patient, and wait for the results from the properly conducted study, the one that will assure us about the safety and efficacy of the latest and greatest cancer therapy, to guide our decision.
Because that, after all, is in the best interest of our patients. As Vladimir queried Estragon in Samuel Becket's absurdist play, “What are we doing here?” They do not move—which sometimes, for the moment, is the most prudent approach.
© 2010 Lippincott Williams & Wilkins, Inc.