In the world of drug development, there has been a recent movement to speed clinical trials of experimental treatments to faster conclusions, with the goal of getting new medicines to patients sooner. That movement relies on what are known as “surrogate markers” (or proxies), which are possible predictors of a disease outcome that are supposed to forecast improvement of the associated disease months or years in advance.
Use of a surrogate marker—for example, lower blood glucose levels in a study of a drug for diabetes—means that many trials finish before it’s known if the disease actually improves or if patients are likely to live any longer than they would without the treatment.
Even though the FDA has indicated the level of evidence needed to support use of a surrogate and feel good about its reliability, we weren’t able to find that evidence.
Joseph Ross, MD
Now, a new study led by researchers at Yale School of Medicine and Emory University found a dearth of supporting evidence that these proxies coincide with better disease outcomes. Published online April 22 in the Journal of the American Medical Association, the study examined the strength of evidence for surrogate markers used in clinical trials for drugs that were approved by the Food and Drug Administration (FDA).
“There was a real paucity of evidence” that these surrogates are valid markers of a good disease outcome, said Joseph Ross, MD, professor of medicine and of public health at Yale and senior author of the study. “Even though the FDA has indicated the level of evidence needed to support use of a surrogate and feel good about its reliability, we weren’t able to find that evidence.”
That level of evidence is what’s known as a meta-analysis, a study that collects and analyzes findings from several different individual trials—a method that is widely viewed as the gold standard of clinical evidence. The FDA has previously said that the use of surrogate markers for drug approval should be based on meta-analyses demonstrating these markers’ associations with good patient outcomes.
Missing meta-analyses for many chronic diseases
Looking at 37 surrogate markers for 32 different chronic diseases such as hypertension or diabetes, Ross and his colleagues found that published meta-analyses did not exist for 60% of these markers. For most of the other 40%, where meta-analyses did exist, the strength of association between the markers and disease outcomes was weak. The researchers focused their study on diseases other than cancer, because comparable analyses have already been carried out for cancer treatments and reached similar conclusions.
The 37 proxies come from an FDA list of surrogate markers that formed the basis of prior drug approvals. Although the FDA does not approve the surrogate markers separately, once a marker is used in one successful drug approval application, it is considered fair game for any approval application targeting the same condition, Ross said.
It's not clear to Ross why so many of the markers on the published list lack this meta-analysis-based evidence. Ross and his colleagues now plan to conduct as many as possible of their own meta-analyses of surrogate markers that are insufficiently documented. It is a process that will likely take several years.
More transparency is needed for informed treatment choices
The FDA does not indicate when or on what evidence drugs were approved using these markers. In their study, the authors suggest that the FDA could increase transparency around the evidence supporting the list of surrogate markers so that physicians and their patients have more information about their treatment choices.
These findings do not imply that drugs whose approval was based on use of a surrogate marker don’t work. But these findings add a level of uncertainty to treatment choices that can be frustrating for patients, Ross said, especially in cases where medications are expensive or come with burdensome side effects. At present, the only solution is for doctors to be as open as possible with their patients about the existing evidence, or lack thereof, he said.
“I want us to go from a conversation between a physician and a patient that sounds like, ‘There’s this new drug; you should take it,’” Ross said, “to something more like, ‘There’s a new drug; it was just approved on a basis of 150 patients. We know that it will shrink your tumor, but we don’t know if you’ll live any longer, and here’s what it costs. Do you want to take it?’”