Importance of randomized, controlled trials and need for a thoughtful, considered approach to development of rational combinations are highlighted by the increased mortality risk seen with pembrolizumab treatment in two halted multiple myeloma studies, FDA officials say.
The troubling mortality results from two multiple myeloma trials of Merck & Co. Inc.’s PD-1 inhibitor Keytruda (pembrolizumab) suggest important lessons for future immunotherapy studies of combination regimens, US FDA officials say.
The higher rate of death in the pembrolizumab-containing arms in both the KEYNOTE-183 and KEYNOTE-185 studies underscores the importance of randomized, controlled trials and the need for sponsors to consider possibly incorporating control arms in earlier-phase studies to assess toxicity.
In addition, the pembrolizumab experience highlights the need for a thoughtful and considered approach to development, including ensuring that there is a good rationale underlying the combination regimen proposed for testing, agency staff said.
At two recent meetings, Richard Pazdur, director of FDA’s Oncology Center of Excellence, and Nicole Gormley, a clinical team leader in the Office of Hematology and Oncology Products, weighed in on their views of lessons learned from the two failed Merck trials.
In July, FDA imposed a clinical hold on the two Phase III studies (as well as a partial clinical hold on a cohort in KEYNOTE-023) due to an increased risk of death for patients receiving the immune-checkpoint inhibitor with immunomodulatory agents compared to the control group. Merck had halted new patient enrollment in the two studies a month earlier due to the mortality imbalance. (See sidebar for coverage from our sister publication Scrip).
KEYNOTE-183 was a randomized, controlled trial of Celgene Corp.’s thalidomide analogue Pomalyst (pomalidomide) and low-dose dexamethasone with or without pembrolizumab in patients with relapsed and refractory multiple myeloma who had received at least two prior lines of therapy. Patients on the pembrolizumab arm had a 61% increased risk of death compared to the control group.
In KEYNOTE-185, newly diagnosed multiple myeloma patients who were ineligible for autologous stem cell transplant were randomized to Celgene’s Revlimid (lenalidomide) and low-dose dexamethasone with or without pembrolizumab. Patients in the pembrolizumab arm saw a doubling in their relative risk of death compared to the control group.
In an Aug. 31 statement, FDA released detailed results from the two KEYNOTE studies and said the agency was working with sponsors of other PD-1/L1 inhibitors to examine other trials in which the checkpoint inhibitors were being studied in combination with other agents for treatment of hematologic malignancies. (Also see “FDA Eyeing Other PD-1/L1 Drugs With Clinical Hold On Keytruda Myeloma Trials” – Pink Sheet, 31 Aug, 2017.)
The agency’s action led to partial clinical holds on other studies, including some multiple myeloma combination trials in the CHECKMATE development program of Bristol-Myers Squibb Co.’s PD-1 inhibitor Opdivo (nivolumab). (Also see “Deciphering US FDA’s Keytruda Safety Announcement” – Pink Sheet, 25 Sep, 2017.)
Mortality Data An ‘Aha’ Moment
The mortality results from the two KEYNOTE studies surprised and dismayed industry, clinicians and FDA, resulting in soul searching for the reason behind the adverse findings and how these results may inform future combination studies.
Speaking from the audience during a panel discussion on combination therapies at the recent Friends of Cancer Research (FOCR) annual meeting, Pazdur said one of his biggest disappointments in 2017 was seeing the survival curves from the two pembrolizumab trials in multiple myeloma – results which were not a chance finding. “We saw in two trials almost a doubling in the deaths,” Pazdur said, describing it as “one of those ‘aha’ moments.”
“Our paradigm of … let’s just add something onto existing therapy may not be the correct.” – OCE Director Pazdur
The OCE director asked what has been learned from the pembrolizumab experience in multiple myeloma and proceeded to partially answer his own question.
“Our paradigm of … let’s just add something onto existing therapy may not be the correct,” Pazdur said. “That’s probably more for cytotoxic drugs; it probably doesn’t work for immunotherapies.” He noted that in the pembrolizumab studies three drugs with immunological properties were being used together.
“That is a sobering awakening that has not been well discussed in the oncology community,” Pazdur said, adding that he hoped it would be discussed at the American Society of Hematology’s annual meeting in Atlanta, which starts Dec. 7.
Although the ASH program does not list any presentations specific to the KEYNOTE-183 and -185 findings, the safety profile of immune-checkpoint inhibitors in blood cancers seems certain to be a focus of discussion given the adverse mortality seen in the KEYNOTE myeloma trials.
In addition, the ASH program lists abstracts from other pembrolizumab studies in different multiple myeloma settings. These include a pilot study in smoldering multiple myeloma, and a Phase II trial of pembrolizumab plus lenalidomide and dexamethasone as post-autologous stem cell transplant consolidation in patients with high-risk multiple myeloma.
Take-Home Points For FDA
Gormley summarized her take-home points from the pembrolizumab combination therapy trials in multiple myeloma during a Dec. 1 workshop on cardiovascular toxicities in immuno-oncology.
“We’ve not learned all the lessons yet from this experience and that this is something that is still undergoing active investigation within our agency and with some of the other sponsors, but I think there are several things that we’ve learned from this,” she said.
At the top of Gormley’s list was the importance of having randomized, controlled data.
“Oftentimes there’s desire to accelerate drug development … with reliance on Phase II trials and things like that, and I think this example really underscores the importance of having randomized trials,” she said. “We would not have observed or been able to pick up on the safety signal if the trial had been conducted in a different setting other than a randomized, controlled setting.”
“We would not have observed or been able to pick up on the safety signal if the trial had been conducted in a different setting other than a randomized, controlled setting.” – FDA’s Gormley
Gormley’s comments echo remarks by Roger Dansey, Merck’s senior vice president of oncology clinical research, at the FOCR meeting.
The myeloma study results were “extremely sobering,” Dansey said. “We clearly did not expect or ever want such an outcome. “
“You look back at what we did, it seemed reasonable at the time just based on the clinical evidence,” he said. “And it’s clear evidence of why [you] do randomized trials … because without that randomization, in the context of an uncontrolled trial you would not have picked up a signal.”
Gormley suggested there could be an opportunity for sponsors to learn about unexpected toxicities sooner in development by incorporating control arms in earlier-phase trials “not necessarily for strong efficacy assessments, but even just evaluating safety differences that would allow you to have more comfort in the safety findings that are being observed.”
It may be prudent for sponsors to follow a more measured approach to all of the phases of clinical development, Gormley said, also citing the importance of having a firm understanding and “rational approach to the combinations that are being pursued.”
In the multiple myeloma setting, there were very few preclinical models to support the combination of immunomodulators and PD-1 inhibitors, she said. “I think having more information in that setting to really be informed about the mechanism [is] really imperative.”
At the FOCR meeting, Kenneth Anderson, a myeloma specialist at Harvard Medical School, said although there were preclinical combination studies of immunomodulatory drugs and checkpoint inhibitors that showed at least additive activity, “I’m not sure in my mind that the preclinical models are adequately reflecting at least toxicities.”
The pembrolizumab experience shows that “as we combine therapies, we need to be more vigilant, not less vigilant, for an unexpected toxicity,” Anderson said.
More attention also needs to be paid to the how differences in patient population and disease setting may impact an immunotherapy combination regimen’s toxicity profile, Gormley said.
“The multiple myeloma patient population in particular generally represents an older patient population that has significant co-morbidities already, and so I think having better attention paid to that aspect,” she said.
Moving Away From Use In Early Disease
The negative results from the two KEYNOTE studies led FDA to restrict the multiple myeloma treatment settings in which immune checkpoint inhibitors could be tested.
Gormley discussed the clinical holds that FDA issue for other PD-1/PD-L1 combination regimens in multiple myeloma after the pembrolizumab mortality results became available.
“Many of the holds …. required that the trials be conducted in a later line of setting, such that we really felt that especially if these trials are going forward they were not appropriate for evaluation in disease settings such as smoldering multiple myeloma or newly diagnosed patients that have available therapeutic options with known clinical benefit,” she said, adding that FDA also required informed consent be updated.
FDA also has taken steps to update drug labeling to reflect the adverse mortality results seen in the pembrolizumab studies. On Nov. 30, the agency approved labeling supplements for Revlimid and Pomalyst to add warnings about the increased mortality seen when pembrolizumab was added to standard multiple myeloma treatment with a thalidomide analogue and dexamethasone.
“The addition of a PD-1 or PD-L1 blocking antibody to a thalidomide analogue is not recommended for the treatment of patients with multiple myeloma outside of controlled clinical trials,” labeling for the Celgene products states. Keytruda labeling contains similar language.