Q: Musella and Tenenbaum recently proposed a new way, called conditional approval, for the American FDA to move potentially useful drugs to a patient market. They wrote that safety would be covered and efficacy assessed by a registry. What do you think of that idea?
A: Imagine if there were a way to speed up the discovery and testing of drugs for cancer. Al Musella and Marty Tenenbaum, founders of two cancer patient advocacy organizations, think they have just such a plan.
They have proposed giving new cancer drugs conditional approval, allowing them on the market after Phase I (safety) trials in as few as 50 patients. Patients and their doctors would be free to use the new, and unproven products, provided all patients enroll in a registry. The patient’s (de-identified) clinical information (pathology reports and biomarkers, for example) would be included in the registry. This would allow the FDA and researchers to determine, by the end of a pre-set conditional period, whether or not the drug is safe and effective when used on a larger population.
Sounds like a great idea, doesn’t it? It eliminates Phase III efficacy trials, which are slow to complete, expensive to conduct, and drive up drug prices. Musella and Tenenbaum’s plan would give patients ready access to potentially life-saving treatments, speed up approval, reduce the burden on the FDA, bring down the cost of drug development, and create a registry for cancer patients that would open up research avenues. What’s more, Japan, Canada, and South Korea have conditional pathways for new cancer treatments. Why not us?
Because it’s a bad idea that will waste a lot of money and hurt patients. Have we all forgotten the recent history of ineffective and harmful cancer treatments? Autologous bone marrow transplant, or ABMT, and Avastin (bevacizumab) are two that come to mind.
ABMT was developed in the 1980s and was used on at least 41,000 women with metastatic breast cancer. Oncologists, transplanters, and the press embraced the treatment enthusiastically on the basis of a single study comparing it to historical controls. It took more than a decade to accumulate enough patients in randomized controlled trials (RCTs), which showed in 1999 that ABMT was worse than standard chemotherapy—it killed about 10 percent of patients. Insurers paid more than $3.4 billion over the course of the decade for a treatment that actively harmed patients.
A decade after that, Avastin was approved for breast cancer on a fast track in 2008, based on “time to progression.” While there wasn’t any proof that time that progression directly affects how long patients live or their quality of life, Genentech, the manufacturer, argued that it was a good marker of improved patient outcomes. By 2010, worldwide sales of Avastin had hit $6.8 billion.
But once again, it took an RCT to show that the drug caused significant side effects and offered no survival benefit. A few patients may actually have been helped by the drug, but there is no way to predict ahead of time whether patients would be helped or harmed. The FDA withdrew Avastin’s approval for breast cancer treatment in 2011.
Imagine Avastin under a conditional approval plan. Women on the drug would have enrolled in a registry, and that registry might have turned up the side effects that emerged in the actual post-marketing trial. But it could not have shown lack of efficacy.
There are several reasons for this, the main one being that outcomes for people on a registry have to be compared to something. That something would be historical controls–similar patients who have had other treatments. The benefits of most cancer treatments are modest at best, and it’s very easy to be fooled by historical controls unless a new drug has extraordinary power to extend life or cure the disease.
There are other ways to solve the problems of high drug prices and slow development of truly innovative drugs for rare cancers, which Musella and Tenenbaum cite as their motivation for recommending conditional approval for cancer drugs. Congress could let Medicare negotiate drug prices. It could increase the FDA’s budget to speed up approvals. (FYI, the FDA already approves drugs faster than any regulator in the world.) Insurance companies could be required to cover treatment for patients in randomized controlled trials.
Even though Musella and Tenebaum’s registry is a bad replacement for RCTs, it’s still a great idea. Manufacturers can track defects in cars and toasters better than the US tracks outcomes and safety of new medical products. Registries should be required for every patient who receives an implantable device or a new drug for any condition. Let’s use registries for what they’re good at, which is spotting harms once a product is used in a large population. Don’t erode the approval process for the sake of speedy access to drugs of uncertain benefit.