In 5-Year Span of FDA Accelerated Approvals in Cancer, 43% Show Confirmatory Clinical Benefit

Edward R. S. Cliff, MBBS, MPH

More than half of agents granted accelerated approval for a cancer-related indication were unable to demonstrate clinical benefit in confirmatory studies, according to results presented during the 2024 AACR Annual Meeting and simultaneously published in JAMA.^1,2

Findings of a cohort-study showed that, between 2013 and 2023, the FDA granted accelerated approval to 129 agents for a cancer-related indication; 46 of these agents had more than 5 years of follow-up. Fifty-seven percent (n = 26) of the 46 drugs with accelerated approval did not show clinical benefit in confirmatory trials, 63% (n = 29) were converted to a regular approval, 22% (n = 10) were withdrawn, and 15% (n = 7) were ongoing after a median of 6.3 years.

“We found that the FDA is increasingly basing conversion decisions to regular approval on surrogate end points like response rate plus duration of response that don’t capture toxicity, and we found that it’s increasingly common practice for sponsors to run confirmatory trials in a different indication or population to the accelerated approval trial,” senior author Edward R. S. Cliff, MBBS (Hons), MPH, of the Program on Regulation, Therapeutics, and Law of Brigham and Women’s Hospital and Harvard Medical School, said in a press briefing during the meeting.

“Our recommendations include that the FDA should ensure that manufacturers run confirmatory trials that are powered to robustly assess clinically meaningful endpoints and clinicians should consider and communicate with their patients about any residual uncertainty of clinical benefit when they offer novel therapies to their patients,” he added.

The FDA’s accelerated approval pathway is based on surrogate end points that are deemed “reasonably likely” to predict clinical benefit. One-third of cancer drug approvals use the accelerated approval pathway, and more than 80% of accelerated approvals are in oncology. Previously reported studies have shown that, of the oncology drugs that received accelerated approvals between 2008 and 2012, 14% went on to show an overall survival (OS) benefit.

Therefore, investigators sought to address how the use of the accelerated approval pathway evolved since earlier analyses, how many drugs granted under accelerated approval show OS benefits as well as quality-of-life (QOL) improvements, and what evidence and what length of time is used to convert to regular approvals.

Investigators evaluated the data in 2 analyses. Fifty-nine drugs were included in the cohort, which totaled 129 drug-indication pairs because “many drugs are used across multiple indications,” Cliff added. The first analysis focused on the 46 agents that had more than 5 years of follow-up, and the second one looked at the 48 drugs that were then converted to a regular approval.

For the first analysis, investigators evaluated the efficacy outcomes used in pivotal studies that supported accelerated approvals in oncology between 2013 and 2017, which included response rate plus duration of response (DOR; 46%), response rate (41%), progression-free survival (PFS; 9%), OS (2%), and other (2%).

“For accelerated approval, looking at a preliminary measure such as response rate is a very reasonable way to get something accelerated approved,” Cliff said.

Furthermore, 20 of the 29 drugs converted to regular approval with clinical benefit, which comprised either improved OS and QOL (n = 7; 15%), improved OS with no QOL benefit (n = 7; 15%), improved QOL and no OS benefit (n = 6; 13%), or were converted without an OS or QOL benefit (n = 9; 20%).

In the second analysis, 40% (n = 19) of the 48 drugs that were converted to regular approval based on improved OS compared with 44% (n = 21) that were based on PFS, event-free survival, or disease-free survival, and 14% (n = 7) that were based on response rate alone or plus DOR. In the last group, 1 converted approval, which was pembrolizumab (Keytruda) in metastatic urothelial cancer, occurred despite a negative confirmatory trial.³

“This should prompt pause, so we looked in more detail at these 7 drug-indication pairs,” Cliff said.

The 7 indications were:

1. Bosutinib (Bosulif) in newly diagnosed chronic phase Philadelphia chromosome (Ph)–positive chronic myeloid leukemia (CML),
2. Asciminib (Scemblix) in phase Ph-positive CML in chronic phase that was previously treated with at least 2 tyrosine kinase inhibitors,
3. Selpercatinib (Retevmo) in RET-positive locally advanced or metastatic solid tumors that have progressed on or following prior systemic therapy or who have no satisfactory alternative treatment options, 
4. Dostarlimab-gxly (Jemperli) in mismatch repair–deficient (dMMR) recurrent or advanced endometrial cancer, as determined by an FDA-approved test, that has progressed on or following a prior platinum-containing regimen in any setting and are ineligible for curative surgery or radiation,
5. Pembrolizumab for unresectable or metastatic microsatellite instability–high or mismatch repair–deficient solid tumors, as determined by an FDA-approved test, that have progressed following prior treatment and who have no satisfactory alternative treatment options,
6. Capmatinib (Tabrecta) in MET exon 14 skipping–mutated positive metastatic non–small cell lung cancer, as detected by an FDA-approved test,
7. And cemiplimab-rwlc (Libtayo) in metastatic basal cell carcinoma that previously was treated with a hedgehog inhibitor or for whom a hedgehog inhibitor is not appropriate.

“The important thing to highlight here is that while response rate and DOR are very promising end points to get early ideas about efficacy or about disease activity, they really don’t give you a good sense because they don’t capture toxicity and they don’t capture the benefit to the whole population,” Cliff explained. “They don’t give you a good sense of the overall risk-benefit profile in totality.”

The accelerated and regular approval indications were also compared with one another as part of the analysis. Of the 48 drugs, 38% (n = 18) had the same indication, 38% were indicated for an earlier line of therapy, 17% had their regular approval broadened without moving to an earlier line of therapy, 6% had a narrowed indication in the regular approval, and 2% were changed in an alternate way.

“Romidepsin [Istoclax] in peripheral T-cell lymphoma, ibrutinib [Imbruvica] in mantle cell lymphoma, and venetoclax [Venclexta] in t(11;14) myeloma are examples where the indication has changed, and that has left some unanswered questions about the original accelerated approval indication,” Cliff noted.

The authors did note study limitations, which were that only confirmatory trial data were examined for clinical benefit evidence, and larger trials or those with more follow-up could have varying data. They also noted that 7 drugs in the first cohort had ongoing confirmatory trials so efficacy results were unavailable and published OS and QOL data were used, which means clinical benefit could have been overestimated.⁴

Editor’s Note: Dr Cliff did not cite any conflicts of interest and noted that the study was funded by Arnold Ventures.

References

1. Liu ITT, Kesselheim AS, Cliff ERS. Clinical benefit and regulatory outcomes of cancer drugs approved via accelerated approval. Presented at: 2024 AACR Annual Meeting; April 5-10, 2024; San Diego, CA. Abstract 918.
2. Liu ITT, Kesselheim AS, Cliff ERS. Clinical benefit and regulatory outcomes of cancer drugs receiving accelerated approval. JAMA. Published online April 7, 2024. doi:10.1001/jama.2024.2396
3. Fewer Than Half of Accelerated Approval Drugs Showed Clinical Benefit in Confirmatory Trials After Five Years. American Association for Cancer Research. Published April 7, 2024. Accessed April 7, 2024. https://aacr.ent.box.com/s/l668tbrh619q85r1ggjpjyz3mwngd8eu
4. Powles T, Csoszi T, Özgüroglu M, et al. Pembrolizumab alone or combined with chemotherapy versus chemotherapy as first-line therapy for advanced urothelial carcinoma (KEYNOTE-361): a randomised, open-label, phase 3 trial. Lancet Oncol. 2021;22(7):931-945. doi:10.1016/S1470-2045(21)00152-2