May 16th, 2018
In his keynote at the 2018 Lown Conference, Dr. Vinay Prasad, practicing oncologist, researcher, and assistant professor of medicine at Oregon Health & Science University asked, “Could pharmaceutical companies make money by running clinical trials for spices to treat cancer?”
“Could pharmaceutical companies make money by running clinical trials for spices to treat cancer?”
Obviously spices don’t do anything to treat cancer, but with our permissive drug approval standards, insanely high prices for cancer drugs, and a little statistics magic, drug companies could theoretically profit from selling completely ineffective drugs, Prasad explained.
For those of you who missed the keynote, Prasad and his colleagues Dr. Christopher McCabe and Dr. Shaim Mailankody have now published a paper on this topic in Nature (Prasad also posted an accompanying “tweetorial“).
Here’s how drug companies could profit from selling spices:
The Food and Drug Administration does not require much evidence to approve a cancer drug – just one clinical trial based on a surrogate endpoint. Surrogate endpoints are not meaningful on their own, but they are meant to be indicators of “hard” outcomes like mortality. This means that a drug company can declare a drug successful if it shrinks a tumor by a certain amount or if patients can go a longer amount of time without recurrence of disease, even if the drug does not reduce all-cause mortality or increase quality of life.
“If you can fit a laser pointer between the curves, you can give the plenary speech at American Society of Clinical Oncology conference.”
The requirements for how much a drug needs to improve these surrogate endpoints are also low. For example, a cancer drug can be approved if it improves survival by just 2.5 months. Some outcomes of drug trials are so small that you can barely make out the difference between the placebo outcomes and drug outcomes on a graph. “If you can fit a laser pointer between the curves, you can give the plenary speech at American Society of Clinical Oncology conference,” joked Prasad.
Clinical trials are meant to determine whether a treatment is more effective than a placebo. Even if you run a trial and a some patients using the drug do better than patients taking the placebo, how do you know the difference is not just due to chance? Researchers use a statistical measurement called a “p-value” to determine the likelihood that an outcome was just due to random chance.
The conventional threshold for a p-value to determine that a study was significant is 0.05. Getting a p-value of 0.05 means that the outcomes of patients taking the drug were different enough from the placebo group that, if the drug were totally ineffective, there is only a 5% chance that these outcomes could occur. This is a very low probability – but not zero! By definition, a p-value of 0.05 means that, if you did 100 trials of an ineffective drug, 5 of them would be get that result, just by random chance.
Of course, doing clinical trials are expensive, and it wouldn’t make sense for a drug company to do 20 trials of a useless drug, in the hopes that one will be a false positive. Unless they could make an absurd amount of money selling the drug…
Prasad and his colleagues estimate that, at $22 million per clinical trial, it would cost a drug company $440 million to get a positive result just by random chance. So if a company can make more than $440 million per approval, they could theoretically make money by testing spices in your kitchen cabinet. Given the skyrocketing costs for new cancer drugs and the fact that the global cancer drug market is more than $100 billion, hitting the $440 million mark wouldn’t be that difficult.
Prasad and his coauthors aren’t saying that pharma companies currently test completely inert compounds, but “the drugs they are testing aren’t much better,” says Prasad. Our incentives for biomedical research have to change. We need to rely less on surrogate endpoints for cancer drugs, require more than one trial for approval, and rein in extreme prices for cancer drugs that don’t show a real benefit.