SRP-9001 and Accelerated Approval: The Good, Bad, and Ugly
John Watkins, PharmD, MPH, BCPS
FDA's accelerated approval process frustrates payers. The advisory committee review of delandistrogene moxeparvovec (SRP-9001), Sarepta Therapeutics’ Duchenne muscular dystrophy gene therapy, highlights the strengths and weaknesses of this process.[1]
DMD, a progressive neuromuscular degenerative disorder affecting one in 3,300 males, is noticed when the child misses developmental milestones. Patients gain muscle strength until around age 6, when they plateau and begin to decline, losing ability to walk in adolescence, followed by a steady decline in upper body strength and eventual respiratory failure. DMD is caused by various mutations in the dystrophin gene. Four exon-skipping drugs have been approved by FDA in recent years; however, each of these addresses only the subset of patients that have mutations in the target exon, and their effect is modest at best.
Accelerated Approval
To gain FDA approval, a product must demonstrate “substantial evidence of effectiveness and sufficient evidence of safety…obtained from adequate and well-controlled studies.” The agency has some flexibility in applying these standards, especially when a drug receives accelerated approval. To qualify, it must be for “treatment of a serious or life-threatening condition.” The manufacturer must demonstrate “effect on a surrogate endpoint that is reasonably likely to predict clinical benefit,” or the endpoint can be “reasonably likely to predict an effect on morbidity or mortality, or other clinical benefit.” Severity, rarity, or prevalence of the condition and lack of effective alternatives also required.[2]
Sarepta’s presentation
At the meeting, Sarepta staff presented SRP-9001,[3] noting that the dystrophin gene is too large to fit the capsid of their viral vector. Sarepta scientists had studyied the dystrophins produced in patients with Becker muscular dystrophy, a less severe form with mutations of the dystrophin gene that produce shorter defective versions of the protein. With 3D structure analysis, they designed a gene that produces micro-dystrophin, a shortened version that retains the structural features that Sarepta’s scientists believed essential to function. According to Sarepta, this “evidence of nature informed rational design of SRP-9001.”
Sarepta presented three studies: a single-arm open label study 101 (N=4 patients, age 4-7), study 102, a randomized double-blind placebo-controlled trial (N=41, age 4-7), and ongoing open label study 103 (40 ambulatory and non-ambulatory patients for longer term follow up.) The primary clinical endpoint, the Northstar Ambulatory Assessment (NSAA), measures standardized performance of 17 specific activities, e.g., walking, rising from a chair, climbing a step. Each activity is scored as 2 (performed normally), 1 (performed with difficulty), or 0 (unable to perform), for a maximum score of 34. After 52 weeks, the difference in mean change from baseline NSAA score between treated and placebo groups in study 102 was a non-significant 0.8 (p=0.37).
The investigators conducted a post hoc subgroup analysis, age 4-5 and 6-7. The difference in the 4–5-year-old cohort was a statistically significant 2.4 favoring treatment, but the 6-7-year-olds showed no difference. The investigators hypothesized that this was due to heterogeneity in baseline NSAA. They showed a statistically significant difference between propensity score matched patients from other trials, and the treated 6–7-year-olds. However, propensity scoring cannot correct for unknown confounders, which are likely in a diverse population.
This presentation was followed by a public comment period, including anecdotal testimonies of major benefit to trial patients, described as life-changing, exceeding their expectations. Follow up was brief, but hope of a reasonably normal future life should not be minimized. On that high note, we adjourned for lunch.
FDA Staff report
In their afternoon presentation,FDA staff 4 raised a number of concerns:
During the trial program Sarepta changed the manufacturing process. These changes affected the purity of the product and essentially altered the dose received.
Staff questioned the validity of the mouse model used in animal preclinical trials.
A key argument for accelerated approval is the validity of a surrogate endpoint as “reasonably likely to predict clinical benefit.” Sarepta’s argument for biological plausibility of micro-dystrophin is based on historical data in Becker patients, whose dystrophin is different from Sarepta’s synthetic version. “Is Sarepta’s micro-dystrophin sufficient to retain the essential function of normal dystrophin?” Their answer was no.
NSAA scoring could be affected by motivation and effort, and the open label studies could be skewed by this. (An expert who trains NSAA users claimed that it is sufficiently standardized to preclude this if administered as she teaches.)
DMD progression is heterogeneous, the drug effect is moderate. Therefore, blinding is essential.
Results from the primary endpoint of trial 102 were not significant, and results of the subgroup analysis are inconsistent. Possibly there is benefit in 4-5-year-olds, but this small sample is underpowered.
The older group did not benefit, except in comparison with external controls. The reviewers did not find this convincing.
The safety profile is comparable to that of other AAV vector gene therapies. Without adequate evidence to identify the patients likely to benefit from SRP-9001, non-responders won’t have access to later products that may be more effective. Hepatotoxicity, cardiotoxicity, and myositis in study 102 were of special concern.
Discussion and Vote
The committee voted on this question:
Do the overall considerations of benefit and risk, taking into account the existing uncertainties, support Accelerated Approval of SRP-9001 —using as a surrogate endpoint, expression of Sarepta’s micro-dystrophin at Week 12 after administration of SRP-9001 —for the treatment of ambulatory patients with Duchenne muscular dystrophy with a confirmed mutation in the DMD gene?
The committee consisted of 14 voting members, including an articulate and well-informed adult patient, whose comments during the discussion were thoughtful and helped the group understand the patient perspective. Members wrestled with the decision as they cast their votes. This treatment benefits to at least some patients. Risks are modest, and the main harm is disqualification for possible future treatment options. Some worried that approval would cause patients to drop out of the ongoing trial, precluding the long-termconfirmatory evidence they wanted to see.
In the end, 8 members voted YES and 6 voted NO.
I left the meeting frustrated. Sarepta applied for accelerated approval after their trial failed to produce the evidence required for a traditional approval, instead of conducting more studies to get further evidence. With its less stringent requirements, accelerated approval allowed Sarepta to bypass this work and bring the product directly to market. That is not the purpose for which accelerated approval is intended. FDA’s leniency in requiring confirmatory evidence is reason for skepticism.
Given the risk/benefit profile of SRP-9001, the possibility of a large benefit (but we can’t say how probable), the modest risk of adverse effects, and not being eligible for later treatment if this one fails (unknown probability), one would be likely to agree with the members that voted YES. Let patients and their parents decide which risk to take. But the FDA is not permitted to address the elephant in the room: this treatment will cost several million dollars per patient. Payers are stewards of limited health care dollars. When you add opportunity cost to the equation, the balance tips quickly to NO.
ICER has suggested mandating a much lower price for accelerated approval, giving patients access and providing some revenue for the manufacturer to fund further research and reap some profit. The full price would be charged only upon final approval after the full evidentiary standard is met. This provides an incentive to complete the research expeditiously.
So, how would I have voted? My inner left-brain evidence reviewer says, “No. Absolutely not! Bad precedent.” My heart yearns to say “yes” and is secretly glad the majority agreed. Clearly, some of the trial patients have experienced transformative results, and kids waiting to be treated suffer more damage with every year that passes. I know hopeful parents were watching. I share their joy. The bioethicist in me agrees. The balance of evidence suggests likely meaningful benefit for many (if not most) patients with relatively modest downside, other than cost.
The managed care pharmacist says, “I hope Sarepta will be held to their promise to produce the confirmatory data. We’ve seen too many past cases where that evidence was never delivered.”
References:
[1]. FDA Cellular, Tissue, and Gene Therapies Advisory Committee, May 12, 2023. Materials are available at https://www.fda.gov/advisory-committees/advisory-committee-calendar/cellular-tissue-and-gene-therapies-advisory-committee-may-12-2023-meeting-announcement-05122023#event-materials.
[2]. Witten C. Accelerated Approval. Briefing document for May 12, 2023 meeting. Available at https://www.fda.gov/media/168091/download.
[3]. O’Malley P, Sarepta Therapeutics. SRP-9001 (delandistrogene moxeparvovec) for Treatment of Duchenne Muscular Dystrophy. Sponsor’s meeting presentation. Available at https://www.fda.gov/media/168095/download.
[4]. Adu-Gyamfi E. Surrogate Endpoint Concerns, BLA 125781 Application for Accelerated Approval of delandistrogene moxeparvovec (SRP-9001). Analysis by the Advisory Committee Planning Working Group Office of Therapeutic Products Center for Biologics Evaluation and Research, FDA. May 12, 2023. Available at https://www.fda.gov/media/168092/download.