In 2008, a new drug then known as PLX4032 was under study for metastatic melanoma—the malignant mole. Melanoma, when widespread, was almost invariably fatal, and there had been no drug that really worked in this condition, although the standard treatment is a drug called dacarbazine, approved years ago by the FDA. Frankly, everyone who has cared for patients with melanoma and treated them with dacarbazine knows it really doesn’t work. 

PLX4032, on the other hand, was producing startling results. Part of a new wave of targeted treatments now being developed, PLX4032 was a drug that, for the first time in my memory as an oncologist who treated advanced melanoma patients for 40 years, produced good quality responses in most patients who got it. The majority of patients responded to it, and a substantial number went into complete remission. And it was safe—much safer than dacarbazine. No one had seen a drug do this before in melanoma. It was prolonging survival and providing a better quality of life with few of dacarbazine’s side effects. (You might ask why dacarbazine was approved in the first place. At the time, it showed promise, but not much. Yet because there was nothing else, it was approved. But unless someone figures out a new way to use dacarbazine that makes it effective, it has had enough years of testing to indicate that, by itself, it is nearly worthless.)

Despite the promising results with PLX4032, in 2009 the FDA required the company that owns it to do a trial in which some patients got the drug and others got dacarbazine. This was both absurd and unethical. In a New York Times article about the drug, some doctors expressed concern about doing this, but they were going along with it nonetheless. I can tell you one thing: no one working in the field who had this disease or had a family member with this disease would have allowed himself or his loved one to be randomized to the dacarbazine control arm. I sure wouldn’t have. In good conscience, you would have to say to the patient, “In my experience, dacarbazine almost never works and has bad side effects, but PLX4032 has been showing excellent and useful responses with minimal side effects. Is it okay with you if we give you dacarbazine?” I think I know what the answer would be.

Why was the FDA doing this? Under the Kefauver-Harris Amendment, or “Drug Efficacy Amendment,” of 1962, proof of efficacy was to be determined in “adequate and well controlled trials.” The act mentions only the use of historical controls— that is, data from previous studies. The amendment did not require new randomized controlled trials, as people often think. The requirement for these new trials—often an unnecessary impediment in early drug trials—was added by the FDA in its interpretation of the regulations.

Because of this rule, we needed to show that using PLX4032 increases survival compared with dacarbazine. That’s nonsense. What we really needed to figure out was how to make these quality responses from PLX4032 permanent. I can think of a number of ways to do this, but none of them involve comparing PLX4032 alone with dacarbazine. Patients should have refused to enter the control arm, and doctors should have refused to offer it. In my view, every patient with metastatic melanoma in the country who had exhausted all reasonable therapeutic options should have had access to PLX4032.

Today we seem to be mindlessly wedded to the use of randomized controlled trials, and they do have their place. But, the PLX4032 trials show, randomized clinical trials can be unethical. Doctors sometimes have strong beliefs about the effectiveness of treatments being compared in a randomized trial—often with good reason. And if they truly believe that the treatments are effective—while a placebo given to some patients is not— then it is their duty as physicians to tell patients so. Not many physicians take this position, but they should.

After two years, an international study involving almost a thousand patients and costing $100 million showed results that were so positive for the new drug, now called vemurafenib, that the study was stopped. The FDA approved vemurafenib for use in metastatic melanoma in August 2011. How many patients lost time, or their lives, in the meantime?

The FDA has also failed to recognize a critical point that is unique to developing new cancer drugs: the patients waiting for these drugs are dying. In this sense, cancer is different from diabetes or hypertension or arthritis, where patients live with their diseases for a normal or near-normal life span. New drugs for these diseases need to be safe enough to be given over the normal lifetime of the patient.

By contrast, cancer drugs are tested first in people who have, on the average, six to twelve months to live, like those patients with metastatic melanoma I mentioned before. The drugs need to either kill cancer cells or stop them from growing over a relatively short period of time measured in months, not years. There is urgency in the cancer field not present in other fields. And if a drug works to some degree, the same population of patients wants and needs access to it long before the creaky approval process allows. The FDA refuses to acknowledge this. The definition of effectiveness is in the eye of the beholder, and the beholder of record in the drug approval process is now an often inexperienced, overzealous, and overworked FDA reviewer.

This article is excerpted from The Death of Cancer by Vincent T. DeVita Jr., M.D., and Elizabeth DeVita-Raeburn, published in November 2015 by Sarah Crichton Books, an imprint of Farrar, Straus and Giroux, LLC. Copyright (c) 2015 by Vincent T. DeVita Jr., M.D., and Elizabeth DeVita-Raeburn. All rights reserved.