FDA顾问委员会以3:9否定Capricor的DMD细胞疗法疗效证据

A panel of experts today voted against Capricor’s cell therapy for Duchenne muscular dystrophy (DMD), capping off an advisory committee meeting marked by statistical confusion.
Capricor’s therapy, deramiocel, is meant to restore heart and skeletal muscle function in DMD patients. Capricor’s approval application is specifically for DMD-associated cardiomyopathy, a common complication of the rare genetic disease, and the panelists were asked to vote only on deramiocel’s effectiveness in cardiomyopathy.
Speaking to Fierce the day before the committee met, Capricor co-founder and CEO Linda Marbán, Ph.D., cautioned that she doesn’t have a crystal ball but “cannot imagine why somebody would vote against this.” Nine experts ultimately did, with only three voting in support of deramiocel’s efficacy.
While advisory committees provide recommendations, the FDA is still the ultimate arbiter of approvals and can choose to go against the experts’ consensus. A decision on deramiocel is expected from the agency by August 22.
Patient representatives Buddy Cassidy, Ph.D., and Debra Dunn voted in favor of the therapy’s efficacy, and were joined by the adcomm’s acting chair Evan Snyder, M.D., Ph.D., a neurologist at Sanford Burnham Prebys Medical Discovery Institute in San Diego.
“As a scientist I would have voted no,” Snyder said when explaining his decision. As a clinician, however, “I voted a highly qualified yes, because I don't want methodologic issues to close the door on something that could have something there.”
The adcomm centered on Capricor’s resubmitted biologics license application (BLA) for deramiocel, which was based on new data from the phase 3 HOPE-3 trial. Capricor’s original rejected submission used data from the earlier Hope-2 trial and its open-label extension, the results of which the company and regulator firmly disagree on.
Capricor put forth its revised BLA following the announced departure of Vinay Prasad, M.D., the former director of the FDA's Center for Biologics Evaluation and Research, who oversaw the previous rejection of deramiocel and numerous other rare disease therapies during his turbulent tenure.
Deramiocel consists of specialized heart cells, called cardiosphere-derived cells, sourced from donor cadavers. The treatment is given via IV infusion every three months.
Overall, experts considered the results of Capricor’s Hope-3 trial too “fragile,” meaning that they were overly dependent on the type of analysis performed and on the exclusion of certain patients.
Statistical snafu
Like the Hope-2 trial before it, the FDA and Capricor fundamentally disagree on whether or not the Hope-3 study of deramiocel hit its primary endpoints. Capricor announced the trial succeeded last December, but the FDA maintains that the trial in fact failed, and that Capricor’s claims of efficacy are due to numerous changes the company made to its statistical analysis plan.
In Capricor’s presentation at the adcomm, Marbán condemned the FDA’s use of statistical analysis plan (SAP) 1.1, “which was never formalized to evaluate this data," and which she said "is simply not appropriate and conceals a true treatment effect.”
“For reasons we do not understand why, the FDA has seized upon that unfinished, unsigned draft SAP and performed numerous analyses based on this incomplete draft,” Marbán said. “This would be like your professor grading your term paper on an early draft you have never even submitted.”
Results from Hope-3, using Capricor’s preferred SAP version 3.0, were published in The Lancet during the adcomm.
During its own presentation later in the day, the agency defended this decision.
"The models specified in SAP 1.1 are consistent with those specified in SAP 1.2 through 1.5,” Tingting Zhou, Ph.D., an FDA statistician, said during the agency’s presentation. "Notably, the analytical models and missing data handling are similar across all SAP versions and protocol versions that preceded the interim futility analysis.”
Through the meeting, committee members largely sought to clarify details of the different analysis plans, with several expressing concerns over the extent of Capricor’s changes and others taking issue with the FDA’s failure to provide feedback on SAP 2.0, which was ultimately supplanted by SAP 3.0.
Though not the trial’s primary endpoint, deramiocel’s approval in DMD-associated cardiomyopathy hinges on left ventricular ejection fraction (LVEF) data, a measure of the heart’s ability to pump blood. Capricor’s original plan was to analyze the change in LVEF in patients after treatment, but the biotech pivoted the day before data unblinding to instead analyze ranked change in LVEF using a different type of statistical test.
Even with Capricor’s last-minute analysis change, the final p-value for the LVEF data presented to the committee still was not statistically significant, sitting at 0.09.
"In some ways, this discussion is sort of moot, in as much as the sponsors' analysis, as presented in The Lancet and presented to us, show that they did not meet this endpoint statistically,” committee member John Teerlink, M.D., a cardiologist at the San Francisco Veterans Affairs Medical Center,” said during the discussion.
Teerlink was also concerned by data suggesting patients given deramiocel experienced increases in left ventricular volume.
“Whenever I see increases in volumes, that's a no-go area,” he said. “In other areas, drugs that increase ventricular volumes actually end up hurting people.”
Though deramiocel is under consideration only for approval in DMD-associated cardiomyopathy, Marbán urged the experts at the adcomm to consider deramiocel’s full potential in the disease.
“We respectfully ask that your vote reflect the potential of deramiocel to address the disease as a whole, and whether these boys deserve a chance to have deramiocel in their arsenal of opportunities to treat DMD,” Marbán said.
During the adcomm’s public comment section, multiple patients, patient advocates and physicians urged the FDA to add a voting question for the skeletal muscle data. Many spoke of deramiocel’s positive effects on preserving muscle function and helping patients maintain independence.
The adcomm’s acting chairperson Snyder ultimately said FDA guidance does not allow for adding a voting question but encouraged the committee to share their full views on the PUL 2.0 data.
Like the secondary ejection fraction endpoint, Capricor and the FDA also disagree on whether the primary skeletal muscle endpoint—improvement in a measure of upper limb function called PUL 2.0—was met.
While initially meant to be analyzed as an absolute change in PUL 2.0 score, Capricor updated the analysis while the trial was ongoing—but again before unblinding—to assess percent change from baseline instead. Capricor contends this change was scientifically warranted, based on a 2025 paper from experts not affiliated with the company, and that the PUL 2.0 endpoint was met however one looks at the data.
“Hope-3 met the primary endpoint,” Craig McDonald, M.D., a pediatrician and neuromuscular specialist at the University of California, Davis and national principal investigator for the Hope-2 and Hope-3 trials, said during Capricor’s presentation. “Primary analysis gives a mean difference of 4.55% in favor of deramiocel, with a p-value of 0.029, which corresponds to a 1.2-point absolute change in total PUL 2.0.”
A change in PUL 2.0 score of one point or more is generally considered clinically meaningful. A pre-specified sensitivity analysis of the absolute change in PUL 2.0 produced a p-value of 0.05, McDonald added, right on the cutoff for statistical significance.
“The conclusion regarding the primary endpoint does not depend on the choice of scale,” McDonald said.
Per the FDA, however, the opposite is true.
"The results are unambiguous,” said Prateek Shukla, M.D., an acting branch chief in CBER’s general medicine division. “The primary endpoint, mean change from baseline in PUL 2.0 total score at month 12, showed a mean difference of 0.66 points favoring deramiocel with a p-value of 0.24. This is not a near miss. The study did not meet its primary endpoint.”
While the experts saw more encouraging signs in the PUL 2.0 data and felt compelled by the patient testimony asserting deramiocel’s effectiveness in upper limb function, they still found the data unconvincing and too dependent on the kind of analysis used. But for future research, the panel highlighted upper limb function as an area for Capricor to pursue further.
“There may be something in the upper limbs, but it's really not proven here, either quantitatively or qualitatively,” committee member Steven Pavlakis, M.D., a pediatric neurologist at SUNY Downstate Health Sciences University in Brooklyn, said. “But if there was anything to look at going forward, that would be it.”