A Closer Look
Design Dilemma: The Debate over Using Placebos in Cancer Clinical Trials
Ask a clinical researcher in almost any field except cancer for the “gold standard” of assessing the effectiveness of a new therapeutic drug and you will probably get the same answer: a randomized double-blinded placebo-controlled trial, in which neither patients nor their doctors know who is receiving the active treatment and who is receiving the inert substance.
By contrast, cancer clinical trials—for practical and ethical reasons—rarely are blinded or include placebo control groups. Most conventional chemotherapy agents have distinctive, often severe side effects that make blinded trials impractical. Moreover, many patients and researchers assert that in trials for a life-threatening condition such as cancer, placebos are inappropriate and that all participants should receive active treatment.
Indeed, placebos are rarely used in most early-phase trials, which are not usually randomized studies and are often first-in-human studies of new agents. Mary Schwartz of Newport News, VA, was diagnosed with stage IV adrenal cancer in 2004 at the age of 35. She has now been cancer free for 6 years after being treated with an investigational chemotherapy regimen in a phase II clinical trial at the NIH Clinical Center. Would she have considered enrolling in a randomized placebo-controlled trial had one been available and offered to her? Absolutely not, she said.
“Because of the rarity of my cancer and the almost certainty of fatality, I would not have chosen to participate if there had been a possibility of getting a placebo,” she said. “I would have wanted to know I was getting the investigational treatment.”
When Placebos Make Sense
Now, however, with the emergence of novel molecularly targeted anticancer agents, some cancer researchers have begun to challenge tradition and are conducting placebo-controlled trials of new anticancer agents. They cite two major reasons why they believe such trials are now feasible and, in some cases, necessary.
First, unlike conventional chemotherapy drugs—which are frequently given intravenously and produce obvious side effects such as nausea, vomiting, and hair loss—many molecularly targeted agents are taken in pill form and cause side effects, like fatigue or headaches, that can be hard to distinguish from effects of the cancer itself or from other medical problems unrelated to the cancer.
And second, because many molecularly targeted agents halt or slow tumor growth without causing tumor shrinkage (the conventional measure of an anticancer drug’s effectiveness), placebo-controlled trials are needed to show that tumor stability is a treatment effect and not the tumor’s natural behavior.
“When a tumor shrinks, that’s presumed to be a direct treatment effect. It’s exceedingly rare for cancer to shrink spontaneously,” said Dr. Richard L. Schilsky of the University of Chicago, past chair of the Cancer and Leukemia Group B, an NCI-supported Clinical Trials Cooperative Group, and senior author of a Journal of Clinical Oncology paper arguing that placebo-controlled cancer clinical trials are scientifically feasible, are ethically justifiable in certain circumstances, and may be necessary or desirable to meet regulatory standards for drug approval.
But “if lack of cancer growth is your endpoint, it’s more complicated,” Dr. Schilsky continued. “I have patients I’ve been observing for years, with no treatment, who have had no tumor growth.”
Dr. Schilsky and his co-authors stress that the use of placebos in cancer clinical trials is ethically acceptable only when there is no effective therapy for the patients’ stage of disease. The authors also stress that patients randomly assigned to a placebo must receive best supportive care; that is, treatment of pain and other symptoms but no anticancer treatment.
Randomized trials comparing active treatment to placebo plus best supportive care led to the Food and Drug Administration’s (FDA) approval for the molecularly targeted anticancer agents sorafenib for advanced kidney cancer and sunitinib for gastrointestinal stromal tumors. Placebo use is also ethically acceptable in trials with an “add-on” design, in which patients randomly assigned to one arm receive standard therapy plus the investigational drug while those in the other arm receive standard therapy plus a placebo. Trials of this design led the FDA to approve erlotinib for advanced pancreatic cancer and for advanced non-small cell lung cancer.
A novel type of trial design minimizes placebo use but can still reveal whether stable disease is a treatment effect or natural tumor behavior. All patients in the trial initially receive the investigational drug. Those whose tumors shrink remain on the drug, whereas those who experience severe side effects or whose tumors grow are withdrawn from the trial. Patients with stable tumors are then assigned in a randomized double-blinded fashion to either continue active treatment or receive a placebo. Those whose cancers progress are unblinded and, if they were receiving the placebo, are put back on active treatment.
Trials designed in this way demonstrated the activity of sorafenib in advanced kidney cancer and also showed that the drug carboxyaminoimidazole was no more effective than a placebo in treating that disease.
A Check against Bias
The use of a placebo control group can also serve as a check against the subtle bias that may be introduced when physicians know that one group of patients in a trial is not receiving active treatment, noted Dr. Jeffrey Abrams, associate director of NCI’s Cancer Therapy Evaluation Program.
“You may unintentionally get more scans or x-rays of patients in the observation arm of the study, and so you may find disease progression sooner than you might find it among patients in the investigational arm,” Dr. Abrams said. “When there is no known effective treatment for the patient’s stage of disease, the use of a placebo as an alternative to observation only in the control arm can help control for this type of bias because both investigator and patient are blinded to treatment assignment.”
The use of a placebo can also help researchers better understand a new drug’s side effects, continued Dr. Abrams. “Without a placebo, there’s a tendency to attribute every reported side effect to the study drug. But in a placebo-controlled study, you may find that patients on the placebo arm report some of the same side effects. So you get a more accurate picture of the true side effects of the investigational drug.”
Dr. Richard Pazdur, director of FDA’s Office of Oncology Drug Products, said he sympathizes with the view that patients who are facing a life-threatening condition like cancer want to have active treatment. “There is almost always an alternative to using a placebo in a cancer clinical trial,” he said. “What you have to ask is: What will the use of a placebo achieve that could not be achieved with another trial design?”