Elevidys in DMD: Approved Despite Trial Failed in Primary Efficacy Endpoint

The US Food and Drug Administration has given the green light for the first gene therapy that treats a rare form of muscular dystrophy (Duchenne muscular dystrophy, DMD) to be used in most people who have the disease and a certain genetic mutation. Last year, the drug – Elevidys, from the biotech company Sarepta Therapeutics – was approved (through accelerated approval pathway) to treat only children ages 4 and 5 with Duchenne muscular dystrophy, one of the most severe forms of inherited muscular dystrophies, who have a confirmed mutation in a gene called DMD that is associated with muscle strength. The FDA announced this past Thursday (June 20, 2024) that it had given traditional approval for Elevidys for ambulatory people 4 and older with a confirmed mutation in the DMD gene and accelerated approval for non-ambulatory people 4 and older with this mutation. There’s not enough data on safety to support its use in children under 4, the agency says.

This is another drug approved by the FDA although the pivotal study (so called EMBARK study) failed in the primary efficacy endpoint (the North Star Ambulatory Assessment (NSAA)). While secondary efficacy endpoints (such as Time to Rise From the Floor) were statistically significant in favor of the Elevidys treatment group, the secondary efficacy endpoints were supposed not to be tested if the primary efficacy endpoint was not statistically significant. 

FDA's action of giving Elevidys traditional approval is controversial. According to FDA's review documents, some of FDA internal reviewers (such as statistical reviewer) were against the approval citing there wasn't sufficient evidence, however, the head of FDA CBER division, Dr Peter Marks, overruled the functional group reviewers to approve this gene therapy.  

• Top FDA official overrules staff to approve gene therapy that failed trial
• Duchenne approval exposes FDA rift over Sarepta gene therapy
• Déjà Vu: CBER Director’s Intervention Saves Sarepta’s Elevidys Again
• How controversial was the decision by FDA’s Peter Marks to approve Sarepta’s gene therapy? Check its footnotes

Perhaps, Dr Peter Marks is right. We should not base the approval on the single p-value from the primary efficacy endpoint and we should not be a slave of the p-values. FDA's announcement says "In making this decision, the FDA considered the totality of the evidence, including the potential risks associated with the product, the life-threatening and debilitating nature of the disease and the urgent unmet medical need." In the rare disease setting, selecting a clinical scale as the primary efficacy endpoint can sometimes be a gambling decision. in DMD situation, FDA published a guidance for industry "Duchenne Muscular Dystrophy and Related Dystrophinopathies:  Developing Drugs for Treatment". The guidance did not recommend which endpoint should be used the primary efficacy endpoint. The guidance was indecisive in efficacy endpoints. 

"FDA has no defined set of required or recommended clinical outcome measures for studies in dystrophinopathies.  Although existing outcome measures developed for clinical trials and/or clinical care in dystrophinopathies or related conditions may be appropriate, FDA will also consider proposals for the use of novel outcome measures that are capable of measuring clinically meaningful effects in patients.  FDA encourages sponsors to propose and, if necessary, develop endpoints that can validly and reliably assess patients with a wide spectrum of symptoms and disease stages.  Sponsors should engage FDA early during the selection and/or development of efficacy endpoints.  The sponsor should include an assessment of multiple efficacy endpoints, when feasible, to characterize the breadth of effects on dystrophin-related pathologies, including skeletal, respiratory, and cardiac muscle function, even if the primary endpoint is only one of these measures. "

Even if the drug is effective, the statistical significance for the selected primary efficacy endpoint (NSAA total score in EMBARK study) may not be demonstrated. In the rare disease setting with the urgent unmet medical need, the strict statistical rules may need to be loosened, the sequential testing rule for controlling the overall alpha may need to be skipped, and the totality evidence from the trial needs to be considered in the decision making.

DMD, like the ALS amyotrophic lateral sclerosis, is a challenging disease for clinical trials. Majority of the late phase DMD trials failed. FDA's approval of Elevidys in DMD comes right after two recent failed trials in DMD by Pfizer and NS Pharma. Their clinical programs in DMD were stopped. 

• Pfizer’s Phase III DMD gene therapy trial misses primary endpoint
• NS Pharma's Duchenne therapy Viltepso fails confirmatory trial

Sarepta's Elevidys in DMD (EMBARK trial) and Amylyx's RELYVRIO in ALS (PHOENIX trial) faced similar situations: they received preliminary approval from the FDA but were required to conduct confirmatory studies to demonstrate clinical benefit. However, both trials failed to achieve statistical significance in their primary efficacy endpoints. The key difference lies in the EMBARK study, which showed significant differences in secondary efficacy endpoints, whereas the PHOENIX study failed to demonstrate statistical significance in any endpoints - therefore, the fate for Elevidys (traditional approval is obtained) and Relyvrio (the product was withdrawn from the market) is totally different now.  

Functional Unblinding in Double-blind, Randomized, Controlled Trials (RCTs)

Earlier this month, FDA advisory committee declined to endorse Lykos Therapeutics’ application to market its psychedelic drug MDNA, also known as ecstasy, as a treatment for post-traumatic stress disorder (PTSD). The panel voted 9-to-2 against the treatment when asked if data showed MDMA’s effectiveness, and 10-to-1 against when asked if the benefits of MDMA outweighed its risks. Both of Lykos Therapeutics’ pivotal studies (MAPP1 study and MAPP2 study) are positive. With relatively small sample sizes, both studies showed highly statistically significant results that MDNA-assisted therapy is highly efficacious in individuals with severe PTSD. The failure in earning the endorsement from FDA advisory committee was not due to the study results, but due to the concerns about the study design and the study conduct, specifically, potential abuse, functional unblinding, and expectation biases. Functional Unblinding was one of the sticky issues discussed during the advisory committee meeting. 

Functional unblinding refers to the situation in a clinical trial or research study where individuals involved (such as participants, investigators, or assessors) gain access to information that reveals the identity of the treatment or intervention being administered. Functional unblinding can happen inadvertently due to various reasons, such as unintended disclosure of treatment details (accidental unblinding), observation of side effects specific to a treatment, or recognition of differences between treatment groups. Clinical trial sponsors usually implement strict procedures to prevent accidental unblinding and manufacture the matched control treatments/using double-dummy technique to prevent recognition of differences between treatment groups. 

However, functional unblinding due to side effects (or treatment emergent adverse events (TEAEs)) may occur with any investigational drug. If the treatment group (or investigational drug) can cause significantly more side effects, study participants, investigators, or assessors can guestimate which treatment group the study participants are assigned to. For example, IGIV causes headache events; niacin causes red or flushed skin, sotatercept causes telangiectasia, bleeding events, and increased hemoglobin levels,... participants or investigators may be able to guesstimate if the participants are receiving the investigational drug based on these unique side effects (adverse events). 

Functional unblinding is an especially important issue in psychedelic drug (such as Lykos' therapeutics's MDNA) clinical trials. In FDA's guidance for industry (June 2023) "Psychedelic Drugs: Considerations for Clinical Investigations", 'functional unblinding' was discussed as the following. It raised the functional unblinding issue and provided some solutions for preventing/handling the functional unblinding: 

Functional unblinding is a critical concern in clinical trials where the investigational drugs have unique and distinctive side effects. Functional unblinding can introduce bias (consciously or unconsciously), affect participant behavior, influence how outcomes are measured and interpreted, and compromise the objectivity of the study's outcome.

If functional unblinding is suspected, it is difficult for sponsors to demonstrate that the functional unblinding does not occur. One approach to investigate/assess the functional unblinding is to employ questionnaires before the study unblinding to ask the participants and investigators which treatment group they think the participants are receiving. However, I see no or very few sponsors doing this. I described this in an earlier post "Assessing potential unblinding due to imbalance in side effects through exit questionnaires".