Study details and key findings

For the study, researchers examined data from 129 cancer drugs that were granted accelerated approval between 2013 and 2023.

Among 46 drugs that had been granted accelerated approval between 2013 and 2017, 29 drugs (63%) were ultimately converted to full FDA approval, while 10 (22%) were withdrawn. An additional seven drugs are still in ongoing trials, with a median follow-up of 6.3 years.

However, only 20 (43%) of the drugs granted accelerated approval demonstrated a clinical benefit in confirmatory trials. Nineteen of these drugs (41%) failed to show improvement in quality of life or overall survival for patients in confirmatory trials after five years of follow-up. 

The time cancer drugs spend in the accelerated approval process has also decreased. In 2013, it took 9.9 years on average for drugs with accelerated approval to be pulled from the market if confirmatory trials did not show a benefit for patients. By 2017, this had decreased to 3.6 years.

The researchers also found that the evidence used to convert accelerated approvals to full approvals changed over time. Between 2013 and 2020, all conversion decisions were based on either progression-free or overall survival. However, between 2021 and 2023, 37% of conversions were based on patient response rates.

According to the authors, patient response rates do not capture information about a drug's toxicity or whether it has helped a patient live longer.

"It doesn't tell you about the net clinical benefit. It tells you that there's a benefit on a scan," said Edward Cliff, a hematologist affiliated with the Program on Regulation, Therapeutics, and Law (PORTAL) within Brigham and Women's Hospital and Harvard Medical School and the study's senior author. "Someone can have a shrunken tumor and have terrible neuropathy, and actually they're worse off. Or people die of infection, which we see a lot with [hematological] malignancies."

Commentary

According to John LaMattina, who led Pfizer's research division between 1999 and 2007, the findings are not surprising — or concerning. "Anybody involved in R&D would know that not every one of these [drugs] is going to come back with a confirmatory trial works," he said. "Am I shocked that I see this result? No. In fact, in some ways, I take heart."

However, Reshma Ramachandran, a family medicine physician and health services researcher at Yale University, said the findings were worrisome.

"Accelerated approval comes with this promise that you're going to confirm clinical benefit. So if more than half of the drugs that you have undergoing accelerated approval aren't doing that, that's concerning, just from a big-picture standpoint," Ramachandran said.

According to Ramachandran, doctors need to pay closer attention not only to whether FDA has approved a drug, but also to the strength of the evidence used to support the decision.

"When we prescribe things, we talk about the benefits of a product and the risks of the products that are very clear cut. When things are more gray, uncertain, it's difficult for us to kind of walk through with patients what we know and we don't know," she said. "A lot of clinicians don't pay attention to the level of the trials that support FDA approval, we're usually looking at the outcome — did FDA approve this?"

For their part, the study's authors say the findings aren't meant to be an argument against using drugs that have been granted accelerated approval. Instead, they emphasize the importance of communication between doctors and patients about the potential benefits and uncertainty of these treatments.

The authors also encouraged drugmakers to collect quality-of-life data more regularly during confirmatory trials and FDA to push companies for stronger evidence of clinical benefit to support full approval for a treatment.

"Plenty of these drugs end up being good drugs, and we'd love to know that sooner, and we'd love to know that in a more robust fashion," Cliff said. "It's not necessarily about good drugs versus bad drugs. It's about how can we achieve the most certainty with the most clinical information to guide patients and clinicians." (Reed, Axios, 4/8; Wosen, STAT+ [subscription required], 4/7; Liu et al., JAMA, 4/7)

Artificial intelligence (AI) is creating opportunities for cancer care providers to personalize treatment plans and reduce spending on ineffective treatments. Read on to discover three ways AI can help providers make better decisions in oncology.