scorecardresearch Skip to main content
STAT | Pharmalot

A case for taking risks on drugs for the dying

Steve Jobs died of pancreatic cancer.Justin Sullivan/Getty Images/Getty Images

This story was produced by Stat, a national publication from Boston Globe Media Partners that will launch online this fall with coverage of health, medicine, and life sciences. Learn more and sign up for Stat's morning newsletter at

Imagine you are diagnosed tomorrow with pancreatic cancer. The latest medical odds would give you a 6 percent chance of surviving more than five years. But if your doctor said you had prostate cancer — a disease with many more treatment options — then the likelihood of being alive in five years would skyrocket to 99 percent.

Clearly, the medical need is greatest for pancreatic cancer. So, you’d think that our regulatory system would adjust the drug approval process to raise the possibility of bringing life-changing drugs to patients.

You’d be wrong.

Right now, the Food and Drug Administration will not approve a drug if there is more than a 2.5 percent chance that it doesn’t work — no matter the disease or how poor the prognosis.


There is a cold logic to this approach. And in a recently published paper, Massachusetts Institute of Technology finance professor Andrew Lo took the FDA to task for not being more flexible.

Lo believes the FDA should be willing to alter the odds based on the severity of the disease and availability of treatments. His argument is rooted in the dry language of statistics but is, nonetheless, compelling because it appeals to common sense.

“For terminal patients with no existing treatments, it seems to make sense to be more lenient and approve drugs that, in the end, may not be effective,” said Lo, director of the laboratory for financial engineering at the MIT Sloan School of Management. “There are desperate patients out there who view this differently than the regulators.”

Of course, no one wants faulty drugs on the market. But some degree of error is inevitable when weighing the risks and benefits of new medicines.

The challenge for the FDA is to get the balance right. Set the bar too high, and good drugs could be delayed or rejected.


Lower the standards, and a drug that proves ineffective — or worse, harmful — could be unleashed.

It’s a difficult needle to thread, with inherent uncertainty and high political stakes.

The FDA has generally taken a cautious stance. But with dying patients clamoring for better treatment options, Lo’s study proposes a less risk-averse approach.

For pancreatic cancer, for instance, Lo calculates that the acceptable chance of a “bad” drug hitting the market should be 28 percent.

While that may sound like a startlingly high chance for mistakes, it also means that many more useful drugs could get into the hands of patients much faster. Conversely, he believes the math dictates that only 1 percent of prostate cancer treatments should be ineffective, since there are already several beneficial medications available.

Many academics agree that it’s time to toss uniform risk thresholds. “Why should we use a one-size-fits-all threshold for all diseases?” said Jagpreet Chhatwal, a health economist at Massachusetts General Hospital’s Institute for Technology Assessment, who was not involved in Lo’s study. “Why not look at more individualized risk?”

Lo uses an analogy about speed limits. Society is willing to accept a higher degree of risk by allowing drivers to reach 65 miles per hour on the Mass. Pike — even though there is a likelihood of crashes and mishaps at that speed. But the risk tolerance drops in a school zone, where the speed limit is a fraction of what it is on interstate highways.


Lo’s paper arrives as Congress is looking to speed drug approvals. Under the 21st Century Cures Act proposed this year, the FDA would be empowered to approve added uses for drugs without having to rely on randomized controlled clinical trials, the gold standard for gauging benefits and risks. Such legislation would effectively make it more likely for approval errors to occur.

In this light, officials are paying close attention to proposals that may alter existing touchstones. But they’re not enamored of Lo’s ideas.

“It’s an overly simplistic analysis,” said one FDA official who asked not to be named. While conceding that new ways to assess risk are under consideration, the official said the FDA already has ways to help patients with life-threatening diseases, including a program that speeds the process without sacrificing standards for evidence.

That may be true. But while the FDA has been approving more drugs in recent years, the question remains: How many others are being tossed to the wayside, given the FDA’s inflexible approach to risk?

FDA officials have been driving slowly for a long time. They may now want to hit the gas and take a risk on more potentially life-saving drugs.

Ed Silverman can be reached at Follow him on Twitter @Pharmalot. Follow Stat on Twitter @statnews.