Zevra Therapeutics, Inc. (ZVRA) Earnings Call Transcript & Summary

Good morning, and welcome. [Operator Instructions] Slide 2, please. For the media and press, the FDA press contact is [ April Green ], her e-mail is currently displayed. Slide 3, please. My name is Dr. Robert Alexander, and I will be chairing this meeting. I will now call the August 2, 2024 Genetic and Metabolic Diseases Advisory Committee meeting to order. We'll start by going around the table and introduce ourselves by stating our names and affiliations. We will start with the FDA to my left and go around the table.

Good morning. I'm Patrizia Cavazzoni, I'm the Director for the Center for Drug and Evaluation & Research.

Good morning. I'm Dr. Peter Stein, I'm the Director of the Office of New Drugs in CDER FDA.

Good morning. I'm Dr. Maynard. I'm the Director of the Office of Rare Diseases, Pediatric, Urologic and Reproductive Medicine at CDER FDA.

Good morning. I'm Dr. Catherine Pilgrim-Grayson. I'm the Acting Director of the Division of Rare Diseases and Medical Genetics in FDA.

I'm Maura Ruzhnikov, I'm a clinical reviewer in the Division of Rare Diseases and Medical Genetics FDA.

Naomi Knoble, Associate Director, Division of Clinical Outcome Assessment, FDA.

Shawna Weis, I'm the acting lead pharmacologists in the Division of Medical Genetics, FDA.

Good morning. My name is Wonyul Lee. I'm the statistical reviewer in the Office of Biostatistics, FDA.

My name is Sydney Stern, and I'm a Pharmacokineticist with the Division of Translational and Precision Medicine, FDA.

Dr. Kishnani. Do you want to introduce yourself. Go ahead.

Yes. I'm Priya Kishnani, I'm clinical and biochemical geneticist, Duke University Medical Center.

I'm Kenneth (Kurt) Fischbeck. I recently retired from the NIH. So now a NIH distinguished investigator emeritus. Before that, I was Head of the neurogenetics branch in Intramural NINDS.

My name is Kiera Berggren. I'm a research speech language pathologist at Virginia Commonwealth in the Department of Neurology.

Good morning, everyone. I'm Gerard Berry. I'm a biochemical geneticist at Boston Children's Hospital and faculty member of the Harvard Medical School.

Moon Choi, Designated Federal Officer.

Robert Alexander, Chief Scientific Officer at the Banner Alzheimer's Institute in Phoenix and a research professor at the University of Arizona.

Wendy Chung, Medical Geneticist and Professor of Harvard Medical School and Chair of Pediatrics at Boston Children's Hospital.

Jon Mink, pediatric neurologist recently retired as well from the University of Rochester, currently a private consultant in the Rochester area.

Jean Le Pichon, I'm a child neurologist and a professor at Children's Mercy in Kansas City.

Good morning. I'm Sarah Chamberlin, I'm Founder and Executive Director of flok Health.

I am Elizabeth Heinze and I'm a patient representative.

I'm Walter Kraft. I'm a clinical pharmacologist and internist to Thomas Jefferson University in Philadelphia.

Good morning. I'm Andy Lieberman. I'm a neuropathologist at the University of Michigan.

Good morning. I'm Susan Ellenberg, Professor Emerita Biostatistics, Medical Ethics and Health Policy at the Perelman School of Medicine, University of Pennsylvania.

Good morning, it's Richard Kryscio. I'm a biostatistician at the University of Kentucky.

Hello, Cheryl Coon. I'm a psychometrician, I'm the Vice President of the Clinical Outcome Assessment Program at Critical Path Institute.

Good morning. I'm Carole Tucker. I'm Professor, Associate Dean of Research at the University of Texas Medical Branch.

Brad Glasscock. I'm the industry representative, I am the Head of Global Regulatory Affairs at BioMarin. And I just wanted to take a moment and congratulate the FDA for forming this committee. This is our inaugural meeting. It's a committee dedicated to focus on rare diseases, which is something that many of us in the rare disease community have been asking for. So thank you for establishing this committee and for -- it's my honor to represent the industry here today. So. Thank you.

Okay. Thanks, everyone. For topics such as those being discussed at this meeting, there are often a variety of opinions, some of which are very strongly held. Our goal is at this meeting be a fair and open forum for discussion of these issues and that individuals can express their views without interruption. Thus, as a gentle reminder, individuals will be allowed to speak into the record only if recognized by the chairperson. We look forward to a productive meeting. In the spirit of the Federal Advisory Committee Act and the Government in the Sunshine Act, we ask that the advisory committee members take care that their conversations about the topic at hand take place in the open forum of the meeting. We are aware that members of the media are anxious to speak with the FDA about these proceedings. However, FDA will refrain from discussing the details of this meeting with the media until conclusion. Also, the committee is reminded to please refrain from discussing the meeting topic during breaks or lunch. Thank you. Dr. Choi will now read the conflict of interest statement for the meeting.

The Food and Drug Administration is convening today's meeting of the Genetic Metabolic Diseases Advisory Committee under the authority of the Federal Advisory Committee Act of 1972. With the exception of the industry representatives, all members and temporary voting members of the committee are special government employees or regular federal employees from other agencies and are subject to federal conflict of interest laws and regulations. The following information on the status of this committee's compliance with federal ethics and conflict of interest laws, covered by but not limited to those found at 18 U.S.C. Section 208 is being provided to participants in today's meeting and to the public. FDA has determined that members and temporary voting members of this committee are in compliance with federal ethics and conflict of interest laws. Under 18 U.S.C. Section 208, Congress has offered FDA [indiscernible] to special government employees and regular federal employees who have potential financial conflicts when is determined that the agency's need for a special government employee services outweighs their potential financial conflicts of interest or when the interest of a regular federal employee is not so substantial as to be deemed likely to affect the integrity of the services, which the government may expect from the employee. Related to the discussion of today's meeting, members and temporary voting members of this committee have been screened for potential financial conflicts of interest of their own as well as those imputed to them, including those of their spouses or minor children and for purposes of 18 U.S.C. Section 208, their employers. These interests may include investments, consulting, expert witness testimony, contracts, grants, CRADAs, teaching, speaking, writing, patents and royalties and primary employment. Today's agenda involves a discussion of New Drug Application, NDA 214927 arimoclomol submitted by Zevra Denmark for the treatment of adults and pediatric patients, 2 years of age and older with Niemann-Pick Disease Type C. This is a particular matters meeting during which specific matters related to Zevra Denmark NDA will be discussed. Based on the agenda for today's meeting and all financial interests reported by the committee members and temporary voting members, no conflict of interest waivers have been issued in connection with this meeting. To ensure transparency, we encourage all standing committee members and temporary voting members to disclose any public statements that they have made concerning the topics, the part of the issue. With respect to FDA's invited industry representative, we would like to disclose that Dr. Bradley Glasscock is participating in this meeting as a nonvoting industry representative, acting on behalf of a regulated industry. Dr. Glasscock's role at this meeting is to represent industry in general and not any particular company. Dr. Glasscock is employed by BioMarin. We would like to remind members and temporary voting members that if the discussions involve any other products or firms not already on the agenda for which an FDA participant has a personal [indiscernible] financial interest, the participants need to exclude themselves from such involvement and their exclusion will be noted for the record. FDA encourages all the participants, advisory committee of any financial relationships that they may have with the firm at issue Thank You.

We will now proceed with FDA initial remarks starting from Dr. Cavazzoni.

Good morning. My name is Patrizia Cavazzoni, and I am the Director for the Center for Drug Evaluation and Research at FDA. I would like to welcome you to the first meeting of the Genetic Metabolic Diseases Advisory Committee. Also, I would like to thank all the participants in today's meeting, including patients, caregivers, advisory committee members and other groups. I will provide some initial remarks for this meeting. FDA convenes advisory committee meetings for a variety of different purposes. Generally speaking, the FDA selects products or topics for advisory committee meetings when there are complex issues that would benefit from broader expert input and discussion. Advisory committees allow the FDA to receive valuable input from many different groups such as clinicians, industry experts, academics, patients and caregivers when evaluating the potential benefits and risk of a new therapy. These are essential part of FDA's work. Many genetic metabolic diseases are very rare and impact a very small number of patients. For most, there are no available therapies. We recognize the challenges facing patients and families and the huge efforts to develop treatments for these rare diseases. Today, we're holding the inaugural meeting of the Genetic Metabolic Diseases Advisory Committee. This committee was specifically created to provide the FDA independent, knowledgeable advice and recommendations on technical, scientific and policy issues around treatment for genetic metabolic diseases. FDA is committed to facilitating the development of safe and effective drugs for rare diseases, including engagement with external experts and the rare disease communities to discuss its development programs. This new advisory committee will provide a dedicated forum for discussion of complex topics with specialized and diverse technical and scientific access in the field of metabolic genetics. I will now turn the meeting to Dr. Pilgrim-Grayson, the Acting Director of the Division of Rare Diseases and Medical Genetics to provide opening remarks for today's meeting.

We'll now proceed with FDA opening remarks starting from Dr. Pilgrim-Grayson.

Thank you, Dr. Cavazzoni. Good morning, everyone. I'm Dr. Catherine Pilgrim-Grayson, and I'm the Acting Director of the Division of Rare Diseases and Medical Genetics. I would like to welcome you to the first meeting of the Genetic Metabolic Diseases Advisory Committee. It's our privilege to host this inaugural convening today. Today, we will discuss the new drug application for arimoclomol proposed for the treatment of Niemann-Pick disease type C, which I'll sometimes refer to as NPC. I would like to extend my thanks to all who are attending either in person or virtually and participating in the meeting today. Thank you to the public attendees and especially the patients with NPC and their families. For those of you who have provided written comments or who will speak today, we are deeply appreciative, and we look forward to hearing your insights. We do recognize the significant unmet need for therapies for this disease. Thank you to all the members of the Advisory Committee. We appreciate that you have taken the time to review the materials and for joining us today to discuss the topics under consideration. Your perspectives and insights are also valuable to the agency. Before describing some of the complex issues we'll discuss today, I want to stress that we have not made any final decisions on the approvability of this application for arimoclomol. Our comments in the background package are preliminary and do not yet take into account today's proceedings. Our presentations should not be viewed as necessarily indicative of our final decision. In our review, we have faced many challenging considerations. And the reason we're here today is to gain your input into some of these scientific uncertainties so that we may incorporate them into our assessment. Next slide, please. Arimoclomol is a new molecular entity and is an orally available small molecule. The mechanism of action has not yet been fully elucidated, but the applicant proposes that it may increase transcription of several genes involved in lysosomal function and so facilitate the proper folding and maturation of certain NPC proteins. The proposed indication is the treatment of adults and pediatric patients 2 years of age and older with NPC. Next slide, please. NPC is a rare disease with an estimated prevalence of 1 to 3 cases per million in the United States. The condition is an autosomal recessive lysosomal storage disease resulting from biallelic mutations in NPC1 or NPC2. This function leads to progressive neurovisceral symptoms and the median age of death is 13 years. There's a clear unmet need. There are no approved treatments for NPC in the United States. The current standard of care is primarily supportive. Miglustat, which is approved in the United States for patients with other diseases is often prescribed off-label for patients with NPC and is considered the current standard of care by treating clinicians. Clearly, NPC is a devastating disease with substantial impact on patients and families. And addressing this need with effective therapies is a priority. In developing therapies, the voices of patients, families, their caregivers and health care providers have been invaluable. And I want to stress again my gratitude to the patients and families who have helped facilitate research and drug development. And thanks to those who participate in the meeting today. Next slide, please. Before I move to discuss the preview of the details of this application, I want to set the stage with a reminder about what constitutes substantial evidence of effectiveness for a given drug. And I would like to remind you that this applies to all drugs, including those that are for rare diseases and for common diseases. FDA has generally interpreted substantial evidence of effectiveness as a requirement for 2 adequate and well-controlled clinical investigations. FDA may consider data from one adequate and well-controlled clinical investigation and confirmatory evidence to constitute substantial evidence if FDA has determined that such data are sufficient to establish effectiveness. This approach is often used in development programs when it's not feasible or practical to conduct more than a single adequate and well-controlled trial. This does reflect agency openness to flexibility in the approaches, especially for conditions like NPC, where no satisfactory alternative therapies exist. Confirmatory evidence is generated from quality data from appropriate sources such as clinical data, mechanistic or pharmacodynamic evidence or animal data. You may hear me and my colleagues today refer to confirmatory evidence as additional evidence and the information provided as additional data. The quantity of this additional evidence needed to support effectiveness may vary and is impacted by the features and results of the adequate and well-controlled trial. It may be possible for a highly persuasive clinical investigation to be supported by a lesser quantity of confirmatory evidence, whereas a less persuasive trial may require a greater quantity of compelling confirmatory evidence to allow for a conclusion of substantial evidence of effectiveness. It's paramount to keep the current clinical context in mind that NPC is a rare and devastating illness with unmet need. We are aligned in our shared goal to have safe and effective treatments for NPC. We recognize that certain aspects for drug development that are feasible for common diseases may not be feasible for rare diseases and that development challenges are often greater with increasing rarity of the disease. We apply flexibility in the situation to address particular challenges posed by each disease while upholding our regulatory standard. Recognizing that is appropriate and necessary to tolerate some degree of uncertainty in the context of development programs for rare diseases. It is ultimately necessary that we have reasonable evidence to support our drug's effectiveness. Today's discussion will focus on whether the submitted data provide reasonable evidence of arimoclomol's effectiveness, which brings me to the questions that we'll be seeking your input on today. I just discussed the regulatory standards, substantial evidence of effectiveness. Today, we're asking for your scientific assessment. Your assessment of the information that was submitted by the applicant to support a conclusion that arimoclomol is effective. We are asking you whether the evidence currently in front of you from the single randomized controlled trial and the confirmatory evidence support a conclusion that the drug is effective in the treatment of NPC. Next slide, please. So just some brief words about the submission. The original submission for arimoclomol was received in July of 2020. The applicant conducted a single adequate and well-controlled clinical trial, which I will refer to as NPC-002 and proposed confirmatory evidence from in vitro, animal and clinical pharmacology data. The primary analysis compared arimoclomol to placebo on the main change in baseline to month 12 on the 5-domain Niemann-Pick disease type C Clinical Severity Scale, also known as 5DNPCCSS, the 5 domains being swallowing, speech, fine motor, ambulatory and cognitive functioning. The application received a complete response in June of 2021. This means that the FDA did not approve the application. This was due to the following deficiencies. There were concerns with the 5DNPCCSS, particularly with the swallow and cognition domain. There were concerns with the prespecified primary analysis for the 5DNPCCSS endpoint and uncertainty regarding the treatment effect. Weak and contradictory confirmatory evidence of effectiveness was also a concern. Next slide, please. In the current resubmission received December of 2023, the applicant modified the analysis of the primary endpoint by removing the cognition domain and rescoring the swallow domain. So now we have a rescored 4-domain Niemann-Pick disease type C clinical severity scale or R4DNPCCSS. The applicant also provided additional confirmatory evidence from clinical data and from nonclinical studies. Today, we will discuss whether the applicant has adequately addressed the deficiencies with this new information. There will be presentations from the applicant and from the FDA regarding these data. Next slide, please. I'll just provide a very brief preview of the post-hoc efficacy results of the R4DNPCCSS. The point estimate of treatment difference ranges from minus 1.5 to minus 1.2, depending on the analysis methods used and the results favor the arimoclomol arm. With regard to these results, however, the agency does note some remaining uncertainty with the R4DNPCCSS. Dr. Maura Ruzhnikov, Wonyul Lee, and Naomi Knoble will review for the study design results and uncertainties in detail. The applicant provided additional nonclinical, clinical pharmacology and clinical evidence. My FDA colleagues from the Division of Rare Diseases and Medical Genetics, the Office of Biostatistics, the Division of Clinical Outcomes Assessment, the Patient-Focused Statistical Support Team, the Office of Pharmacology and Toxicology and the Division of Translational and Precision Medicine have reviewed these data and we'll discuss them in detail. We will consider whether the potential confirmatory evidence, including the concerns that we have about them related to nonclinical clinical pharmacology and clinical data, we ask you to look at the strengths and limitations of them. Next slide, please. As you listen to the data presented today, consider the following: the existing uncertainty regarding the estimated treatment effect on both the 5-domain and the 4-domain NPCCSS, primarily focusing on the 4-domain rescored NPCCSS. The validity of the 5D-NPCCSS and R4DNPCCSS, the adequacy of the additional clinical and nonclinical data to serve as confirmatory evidence to support the efficacy of arimoclomol and the strength of the overall evidence to support the efficacy of arimoclomol. I will now conclude these opening remarks with a preview of the discussion and voting questions that we would like the committee to keep in mind as we hear the presentations today. The first question we'll ask you to discuss your assessment of the efficacy results of the NPC-002 trial. In your discussion, we would like you to comment on the 5-domain Niemann-Pick disease type C clinical severity scale and the rescored 4-domain clinical severity scale. And we'd also like you to discuss your assessment of whether the trial results demonstrate a treatment effect of arimoclomol on Niemann-Pick disease type C. We would also like you to discuss your assessment of the other data. particularly the additional clinical and nonclinical data with respect to support for the efficacy of arimoclomol, and we would like you to vote on whether the results of trial NPC-002 in concert with the other evidence support a conclusion that arimoclomol is effective in the treatment of patients with NPC. We would like you to provide a rationale for your vote, and if you vote no to provide recommendations for additional data that may support a conclusion that arimoclomol is effective. This concludes my FDA opening remarks. Thank you for your attention. I'll now hand the meeting back over to Dr. Alexander. We're looking forward to a productive meeting, and we look forward to your input in this important matter. Thank you.

Thank you, Dr. Pilgrim-Grayson. Both the Food and Drug Administration and the public believe in a transparent process for information gathering and decision-making. To ensure such transparency at the advisory meeting, FDA believes that it is important to understand the context of an individual's presentation. For this reason, FDA encourages all participants, including the applicants nonemployee presenters, to advise the committee of any financial relationships they may have with the applicant, such as consulting fees, travel expenses, honoraria and interest in the applicant, including equity interest and those based upon the outcome of the meeting. Likewise, FDA encourages you at the beginning of your presentation to advise the committee if you do not have any such financial relationships. If you choose not to address this issue of financial relationships at the beginning of your presentation, we will not preclude you from speaking. We will now proceed with the presentation from Zevra Therapeutics.

Good morning. I'm Louise Himmelstrup, Vice President for Regulatory Affairs at Zevra Therapeutics. I've been dedicated to the arimoclomol and the NPC program for more than 6 years. I would like to thank the FDA and the committee for their time to prepare for today's meeting on arimoclomol. Additionally, thank you to the NPC community, including patients and their families for their participation and support. Niemann-Pick disease type C, also known as NPC is an ultrarare neurodegenerative atypical lysosomal storage disorder. NPC affects between 600 to 900 patients in the U.S., which is far below the thresholds for both orphan diseases and ultra-rare diseases. NPC impairs the body's ability to transport cholesterol and other lipids inside cells. Over time, this accumulation causes progressive dysfunction of the nerves, brain and other organs. NPC is typically diagnosed in early childhood and causes premature mortality. The median age of death is 13 years. Despite decades of research and development, there are currently no approved therapies in the U.S. Arimoclomol is a small molecule, oral capsule that has been developed as a treatment for NPC. Results from our clinical program demonstrates that arimoclomol slows the NPC progression, which is a clinically meaningful benefit in a disease characterized by a relentless progression. Additionally, long-term data show that arimoclomol is well tolerated with an acceptable safety profile through 5 years. We are here today to review data supporting our proposed indication for the treatment of adult and pediatric patients aged 2 years and older. An important consideration for today's meeting is FDA draft guidance on demonstrating substantial evidence of effectiveness. For rare diseases like NPC with such a small patient population and a high unmet need, the guidance provides the regulatory pathway to demonstrate substantial effectiveness with a single, adequate and well-controlled clinical study that is supported by multiple sources of confirmatory evidence. Rather than providing 2 clinical trials like the standard approach for nonrare diseases, confirmatory evidence is used to substantiate the treatment effect observed in a single trial. As we will discuss today, Zevra's re-submission for arimoclomol aligns with this guidance and demonstrate substantial evidence of effectiveness that arimoclomol slows NPC progression. Our resubmission includes updates and new analysis that were all implemented based on recommendations the FDA included in its complete response letter to the original arimoclomol new drug application. First, our pivotal efficacy study is a positive, adequate and well-controlled randomized clinical trial. The trial showed a statistically significant result for arimoclomol on the prespecified primary efficacy end point. The treatment effect was also significant when we applied FDA's recommended [indiscernible]. And second, the confirmatory evidence for arimoclomol is consistently aligned, mutually reinforcing and taken together, confirms the benefit observed in the clinical trial. The confirmatory evidence comes from several sources, including clinical data, natural history data, data from our expanded access program, animal models of NPC and studies that revealed a better understanding of the arimoclomol mechanism of action. We are here today to answer the questions FDA raised in our complete response letter and posed to the committee today. First, we will address uncertainties about updates to the pivotal trial's primary endpoint, a composite endpoint known as the Niemann-Pick type C clinical severity scale. In agreement with the FDA, one revision included removing the cognition domain. Another update included revising the swallow domain scoring methodology based on FDA feedback and a qualitative study with NPC and swallow experts. We also will share information regarding the validation of this tool using objective functional measures. The NPCCSS is the only instrument validated for NPC. Second, for question about the estimated treatment effect of arimoclomol, we will show that the study met its prespecified primary endpoint with a prespecified analysis. The study also met the revised primary endpoint with FDA's recommended analysis. And the endpoint results were robust across various sensitivity analysis. And third, we will discuss the strength of the clinical and nonclinical confirmatory evidence. This includes evidence of benefit in our 4-year open-label extension of our pivotal trial and change in disease scores when transferring from untreated to arimoclomol treatment. Further, the favorable outcomes are supported by effect on the objective measure of weight and sustained stabilization over 3 years in the U.S. expanded access program. Our presentation will show how the pivotal study and confirmatory evidence together demonstrate the substantial evidence of effectiveness of arimoclomol. Here's the agenda for the presentation. We also have additional experts with us today. All outside experts have been compensated for their time and travel to today's meeting. Thank you. I'll now turn the lectern over to Dr. Patterson.

Good morning. My name is Marc Patterson. I'm a child neurologist at Mayo Clinic from where I am joining you remotely today. I've been involved in the care, management and research on patients with Niemann-Pick disease type C since my fellowship with the National Institute of Health, which began 34 years ago. I have the privilege of participating in the first controlled clinical trial that was executed at NIH and published in 1993 as well as in almost all of the subsequent trials of agents that have been studied in this disease. Perhaps most importantly, I have had the privilege of caring for several hundred families afflicted by this disease over this period of time. To give some more background on Niemann-Pick disease type C, as you've already heard, this is an atypical lysosomal storage disorder. Classic lysosomal storage diseases are those characterized by enzyme deficiency, which leads to the accumulation of substrate in the lysosomes. In contrast, the gene products of the 2 genes mutated in NPC produce proteins, which are involved in the trafficking of macro molecules. So this is a trafficking defect rather than an enzyme defect, the consequence of which is the storage of multiple substrates within the lysosomes. The disease is characterized by relentless neurological progression, which may first manifest in utero or early infancy or as late as maturity. The best data we have on incidents suggests that it is around 1 in 100,000 births, reinforcing the qualification of NPC as an ultra-rare disease. The disease is also highly heterogeneous and variable. Generally speaking, the core symptoms vary with age, and the common feature is a progression towards death. The rate of progression is determined by the age of neurological onset. Children with perinatal presentations often have severe liver involvement, which may be accompanied by pulmonary infiltrates. The mortality associated with this early onset is quite high within the first couple of years of life. The infantile presentation is dominated by neurologic progression. Children are typically hypotonic with delayed development initially and then show signs of neurologic regression, which is often rapid, but may occur more slowly in some cases. The most frequent presentation occurs in middle childhood where children have a more insidious onset. Some may appear to have difficulties with concentration and may be misdiagnosed as having attention deficit disorder. Others appear simply to be clumsy and in others uncontrollable epileptic seizures may occur. Many children suffer from sleep disorders. In the teenage years and early adulthood, the onset is dominated by cognitive impairment, presenting as an early onset progressive dementia. Many patients also have a dramatic psychiatric presentation, which may mimic schizophrenia, treatment-resistant depression or bipolar disorder. As the disease progresses, there is a tremendous burden on patients and caregivers, which only progresses over time. As you've heard, routine clinical care for patients with Niemann-Pick disease type C requires a multidisciplinary team. This is typically led by a neurologist or a geneticist, but requires the participation of multiple additional experts, including psychiatrists, gastroenterologists, physical medicine and rehabilitation specialists as well as occupational, physical therapists and speech therapists. Unfortunately, there are currently no FDA-approved therapies for Niemann-Pick disease type C. Routine care may include miglustat for appropriate patients. Miglustat is an imino sugar, which inhibits the synthesis of glucosylceramide and thus acts as a form of substrate reduction therapy. This is currently approved for certain patients with type 1 Gaucher's disease in the United States and is frequently used off-label in the U.S. to treat Niemann-Pick disease type C. However, because this molecule forms bonds with a broad range of ligands, including enzymes and the intestines, many patients may experience gastrointestinal adverse effects, which may lead to intolerance of the agent. The heterogeneity of Niemann-Pick disease type C makes it very difficult to measure this disease and its manifestations across a wide range of ages. There are no established surrogate endpoints or readily measurable biomarkers, which have been shown to address disease progression. Therefore, progression needs to be measured by using composite clinical scales, whose goal is to try to embrace the breadth of the disease and measure progression in a granular fashion. The Niemann-Pick disease type C severity scale was developed specifically to produce a valid and reliable tool that measures the severity of this heterogeneous disease, including its varying signs and symptoms across a range of ages in clinical trials. The development of the NPC clinical severity scale goes back more than 20 years. The scale was initially developed by colleagues in Spain. Subsequently, some of their data and concepts were used by our colleagues at the National Institute of Health, who developed the 17-domain NPC clinical severity scale. I had the privilege of being involved in the development of that scale and subsequently in the development of the simplified 5-domain version. These 5 domains, specifically ambulation, fine motor skills, speech, swallowing and cognition were selected to capture the key symptoms, which were regarded as the most important manifestations of the disease by patients, caregivers and clinicians. Clinicians score each domain from 0 to 5 using defined criteria. Higher scores indicate more severe clinical impairment. This scale is recommended by clinical experts for use as a primary outcome measure in clinical studies. The sponsors subsequently adapted the 4-domain NPC clinical severity scale based on recommendations from the FDA. The cognition domain was removed with agreement with the FDA to address concerns that a single item would be unable to fully evaluate a broad concept such as cognition in a 12-month trial. While cognition is clearly one of the most important manifestations of NPC, I agree that it makes sense to remove it from the assessment in the context of this clinical trial. Based on concerns from the FDA that the swallowing domain did not reflect a linear progression, a panel of NPC and swallowing experts reviewed the scoring methodology and made recommendations as to how it could be improved. Importantly, these experts only reviewed the methodology and did not make their recommendations based on study data. The revised methodology was applied to the original source data captured in the clinical trial. In my opinion, in the context of the arimoclomol clinical trial, the resulting 4-domain scale is appropriate for evaluating NPC disease progression. Since one of the topics for today relates to the adequacy of the revised swallow domain, let me walk through these changes next. The original scoring methodology for the swallow domain could yield incorrect equivalencies in disease severity. If the patient had no dysphagia, they were scored 0. If the patient coughed while eating, they were scored 1. Additional points were added depending on whether or not the patient experienced intermittent or consistent dysphagia with liquids or foods. The use of a feeding tube for supplemental feeding was automatically at 4 and if the patient only received nutrition through a feeding tube they were scored 5. In the complete response letter, the FDA identified situations where patients could reach the maximum score but have differing levels of swallow dysfunction. Let me illustrate with 2 examples. First, assume Patient A coughs while eating and has regular difficulties swallowing both liquids and solids, but does not need a feeding tube. The original scoring methodology would result in this patient receiving a score of 5. Then consider patient B, who requires a feeding tube for all of their nutritional needs. This patient would also end up with a maximum score of 5. I agree with the FDA's assessment that a patient who does not require a feeding tube at any time is not equivalent in swallow dysfunction to a patient who requires a feeding tube all of the time. The updated scoring methodology was designed to reflect linearity in disease progression. With this step-wise increase in the patient's level of swallow dysfunction is matched with a numeric point increasing score. Intermittent dysphagia is scored as a 2 and persistent dysphagia as a 3 regardless of whether dysphagia occurred with a liquid or a solid. The only way to receive a score of 4 or 5 is if the patient relied on a feeding tube. This updated scoring methodology was applied to the data that were collected during the clinical trial. Here is the full 4-domain NPC clinical severity scale, along with abbreviated descriptions for each scoring category. Higher scores reflect more severe clinical impairment with a maximum score of 5 in each domain and 20 overall. On each domain, each 1 unit change was designed to reflect a loss of complex function and increased disability due to NPC. Note that some domains do not take on values for all possible scores between 0 and 5. Jumps between these scores reflect greater degrees of worsening. Since we just reviewed the updated scoring for the swallow domain, I'll provide examples on other domains for how the scale is linked to patients reaching certain milestones that clinicians were trained to recognize and reliably score. For ambulation, the score of 1 means that the clinician observes the patient to be clumsy, striking objects and experiencing difficulty walking along a straight line as their age-related peers would be able to do. This individual would move to a score of 2 if they were observed to have a frankly ataxic gait with a broad base and difficulty making turns. Turning to speech. The score of 3 would mean that clinicians found the patients to be nonverbal, but still able to communicate their needs using hand gestures, such as the thumbs up or down pointing to pictures. The only worse thing that is possible in the speech domain is to reach a 5 for minimal communication, which would mean that the patient cannot communicate their needs in any meaningful way. As you can see, each step-wise change can be unambiguously assessed by a disease expert, and each step represents meaningful worsening. Deploying this tool was accompanied by specific training for investigators on how to collect the data based on the domains in the NPC clinical severity scale. And while we all try to be as consistent and precise as possible in administering instruments, we must bear in mind that we are dealing with neurologically impaired children across a broad range of ages. If you do not actually spend time sitting in an examination room with a child, particularly a neurologically impaired child, it may seem very reasonable to use a large number of scales to try to assess different aspects of the disease. But it's important to recognize that is not a realistic goal for most children with NPC. Those of us who are fortunate enough to have their own children or grandchildren or to care for them, recognize that children do not always wish to do what we ask them to do, whether they have a neurologic disease or not. And if they are neurologically impaired, it may be exceptionally difficult in some cases. Therefore, the well-defined scoring criteria must be applied using a flexible approach to accurately evaluate each domain in children. Additionally, the arimoclomol trial included several standardized procedures. Patients were scored according to a detailed scoring manual. There was also a requirement for the same NPC experts to evaluate patients at each visit. And all investigators, myself included, had to undergo rater training before enrolling patients in the trial. It's important to underscore that the NPC clinical severity scale is the only validated and appropriate tool to specifically measure Niemann-Pick disease type C severity and progression across multiple domains. The reliability and validity of the tool has been established at the domain level. With regard to construct validity, the correlations between NPC domains and relevant objective performance tests range from 0.45 to 0.99. The instrument has also shown high intra-rater and inter-rater reliability. Lastly, allow me to describe how we determine the minimum clinically important difference for the instrument. Using an anchor-based approach, we found that at one point worsening or greater on the 5-domain scale represented a clinically meaningful transition, reflecting a loss of complex function and increased disability. This finding was reinforced in qualitative interviews with patients, caregivers and NPC experts. The community consensus was the 1 point change in any domain score would be clinically meaningful. The majority of participants also reported that a slowing of progression in any one of the domains would be impactful. For example, this is the difference between a patient being able to eat by themselves or needing a caregiver to feed them or the difference between a patient having difficulty swallowing during eating some of the time or all of the time or between needing a wheelchair or not. Thank you very much for the opportunity to present and for your attention. I will now turn the presentation back to the sponsor.

Thank you, Dr. Patterson. My name is Dan Gallo, and I'm the Senior Vice President of Medical Affairs and Advocacy at Zevra. The arimoclomol clinical program included 3 studies in NPC, where disease severity and progression were evaluated using the NPCCSS endpoint. An observational study called Study 1, where we evaluated the rate of disease progression on the NPC-CSS in a heterogeneous population of NPC patients receiving routine clinical care. Our pivotal randomized placebo-controlled trial with Study 2, which involved a 12-month double-blind treatment phase where patients were randomized in a 2:1 ratio to arimoclomol or placebo. We also conducted an open-label extension of Study 2, where patients received arimoclomol for up to 4 additional years. Here is the design of pivotal Study 2. 50 patients were enrolled, including 27 who rolled over from Study 1. Patients were randomized 2:1 to receive arimoclomol or placebo for a 12-month double-blind phase. As in Study 1, all patients were permitted to remain on the routine clinical care and randomization was stratified by use of miglustat at baseline. After the double blind phase, patients were eligible to continue to the open-label extension, where all patients received open-label arimoclomol. 41 of the 50 patients in the double blind phase continued into the open-label extension. Here are the baseline demographics and disease characteristics. The average age of patients was approximately 11% and range from 2 to 19. The average time since first symptoms was about 8 years and most patients were on miglustat at baseline, which they were permitted to continue through the study. All three patients with double no mutations predictive of a more severe and rapidly progressive disease course were randomized by chance to the arimoclomol group. Study 2 met its prespecified primary endpoint, which was the change from baseline in the 5-domain version of the NPC-CSS at 12 months and assessed using a mixed model for repeated measures. The treatment effect was 1.4 points with an associated p-value of 0.0456. Now let's look at the results in the 4-domain endpoint. The results of both the 4-domain endpoint and the FDA recommended analysis method based on the while-on treatment estimand were congruent with our prespecified analysis. Arimoclomol demonstrated a clinically meaningful and statistically significant reduction in NPC disease progression compared to placebo from baseline to 12 months. The overall treatment difference was 1.5 points of less progression than placebo with a p-value of 0.041. Subgroup analyses for the 4-domain NPC-CSS show a treatment effect that consistently favors the benefit of arimoclomol by age, age at first neurological symptom, sex, baseline score and among patients with no double null mutations. Since all three patients with a double null mutation were in the arimoclomol group, a treatment effect relative to placebo could not be calculated for those with this mutation. For the subgroup of patients who did not take miglustat, the estimated treatment difference numerically favors placebo and has wide confidence intervals. As FDA outlined in their briefing document, this result is challenging to interpret and we agree for the reasons I'll review on the next slide. We saw a small number of patients in the study who did not receive miglustat as part of their routine care. In total, eight patients on arimoclomol and three on placebo did not take miglustat. This subgroup had substantial baseline imbalances that would predict a worse disease prognosis for the arimoclomol group. The average age at first neurological symptoms was 4 years with arimoclomol and 10 with placebo. The average baseline symptom severity score was nearly double in the arimoclomol group. And all three patients with double no mutations were randomized to arimoclomol. Therefore, we agree with the FDA's conclusion that these baseline imbalances that favor the placebo group make it difficult to interpret the subgroup who did not take miglustat. We also performed several sensitivity analyses, which showed that the primary endpoint findings were robust. The first line in the chart shows the updated primary analysis of the four domain scale based on FDA recommendations, which I showed previously. This analysis estimated the while-on-treatment estimand using an ANCOVA model. This is followed by additional sensitivity analyses that showed that the estimated treatment effect remained similar to the updated primary analysis. To ensure that the favorable results were not driven by outlying values we performed a nonparametric Wilcoxon rank sum test based on the ranks of the data and this too was statistically significant. The last row shows that the prespecified analysis based on the 5-domain scale using the MMRM model also achieved statistical significance. I want to acknowledge that the FDA provided many more sensitivity analyses in their briefing document. We agree with their conclusion that some nominal P values were below the 0.05 and others were above, but that the point estimates of the treatment effect consistently showed slower progression with arimoclomol compared to placebo. To further explore the effects of arimoclomol, next, we'll transition from statistical models to the actual patient level data on disease trajectory. On this plot, each point on the Y-axis will represent an arimoclomol patient. The X-axis will show their overall 4D-NPCCSS score from baseline to 12 months. The open dots represent the scored baseline. Green dots reflect 12-month scores that were improvements from baseline Purple dots reflect no change from baseline at 12 months, which is our goal, and pink dots reflect a worsening from baseline. After 1 year of treatment, 10 patients receiving arimoclomol had an improvement of at least 1 point and 12 had no change in disease severity. For placebo-treated patients, keeping in mind the 2:1 randomization, we see that the majority experienced a worsening of disease severity over the same time period. And no patients in the placebo group exhibited an improvement. Because of the 2:1 randomization, it is helpful to look at these data in terms of percentages. At 12 months, 29% of patients in the arimoclomol group had improved scores and 35% had no change. So 65% of patients were stable or improved. For placebo, no patients improved and 40% had no change. So, overall 65% of patients on arimoclomol had a favorable response compared to 40% of the placebo group with no change. In summary, across analyses, the clinical data from our pivotal trial consistently show that arimoclomol slowed the natural course of this devastating disease. We used the NPCCSS as our primary endpoint, which is a validated instrument that came out of the halls of the NIH. We updated the endpoint and statistical analyses based on FDA requests and recommendations. The trial showed a clinically meaningful and statistically significant effect at 12 months in both our prespecified analysis as well as using the FDA recommended while-on treatment estimand and statistical approach. And finally, the treatment effect was robust to multiple sensitivity analyses, all of which supported the benefit of arimoclomol. Thank you. I'll now turn the presentation over to Dr. Mickle to present the confirmatory evidence.

Thank you. I'm Travis Mickle, Co-Founder of Zevra and their senior adviser. I'll review the support of data from clinical and nonclinical sources, which provide confirmatory evidence of the effectiveness of arimoclomol. Let's start with the clinical evidence. The designs of our clinical studies enable pre-post analyses to further evaluate the benefits of arimoclomol. We compare patients who transition from routine care or placebo to arimoclomol, allowing patients to serve as their own control. We looked at the annual change in the 4D-NPCCSS scale. The first analysis includes 17 patients. In year 1, they received routine care as part of the observational study 1. In the second year, they were randomized to receive arimoclomol in the double-blind phase of Study 2. The second analysis includes 13 patients. In year 1, they were randomized to placebo in the double-blind phase of Study 2. In year 2, those same patients received arimoclomol in the open-label extension. For the second analysis, it's important to note that both patients and physicians were unaware of their randomized assignment during the double-blind for the first 2 years of the open-label phase. Let's start with the analysis of the annual change in the 4-domain scale. In the first part analysis, the 17 patients who received routine care and Study 1 had scores worsened on average by 1.6 points on the 4D-NPCCSS. But when those same patients received their arimoclomol, during the double-blind study, the following year, their annual disease progression was reduced to 0.9. The second analysis among the 13 patients who received placebo in the double blind and got arimoclomol in the open label showed similar results. Their annual worsening was 1.9% during the first year on placebo. But when they worsened, but they worsened by only 0.3 points after transitioning arimoclomol in the first year of the open-label extension. Both analyses show a slower rate of disease progression after initiation of arimoclomol. NPC can cause below average weight gain in children due to their swallowing issues and other complications. So weight is an important objective measure of development in these patients. Since children grow at different rates by age and sex, we evaluated annual changes using Z-scores. In the following analysis, we compared their weight to the CDC's childhood growth curve for their age and sex to determine the Z-scores. This standardized measure allowed us to appropriately assess weight changes in growing children. The annual change for a patient was calculated as the difference between Z-scores at baseline and the end of the 1-year period. Therefore, score of 0 indicates average growth, while negative values reflect below average growth in standard deviation units. In the placebo group, the average change in weight over that year was half of standard deviation less than what will be expected for children the same age. In the first year of the open-label placebo, patients have switched to arimoclomol had a dramatic increase in weight trend. For the patients on arimoclomol in the double-blind and the open label, we're closer to what you would be expected for their age compared to the healthy population. This trend mirrored what was seen in the previous example with a 4D-NPCCSS when patients initiate treatment with arimoclomol after transitioning from placebo in the double blind, but now with an objective measure. Next, let's review the natural history data. The FDA requested an analysis comparing patients in our open-label extension with external comparators from the NIH natural history cohort study, we had at least 4 years of follow-up data. To compare these nonrandomized groups, we used inverse probability of treatment weighting. Due to age differences between the groups, we focused on patients who are at least 4 years old. We waited the analysis based on baseline age, sex, miglustat use, age of onset and baseline score. In this analysis, we see that over 4 years, there is more than a 1 point difference in the rate of progression with arimoclomol treatment. These results mirror what we see in the double-blind and the previous examples of untreated patients switched to arimoclomol. We agree with the FDA that this analysis has limitations. Even with this in mind, we still see a consistent treatment effect favoring arimoclomol. Next, I'll share some data from the arimoclomol expanded access program. Outcomes in the U.S. EAP were collected over time across 14 sites under a clinical protocol. Physicians were trained on the administration of the NPCCSS and provided a scoring manual to ensure consistent evaluation. These data show changes in the 4-domain NPCCSS from baseline at every visit for 82 patients in the U.S. expanded access program. A slow rate of disease progression was observed in both children and adults. About half of the population in the EAP were children. But these data are difficult to interpret without a comparator group. To provide a comparator for the EAP patients in the United States, we utilize the NIH natural history cohort. The analysis was limited to patients aged 4 to 30 years to address the fact that the EAP data set included more adult patients and the NIH data set included more infants. No other inclusion or exclusion criteria were apply. This left 56 patients in the EAP dataset and 44 in the NIH. The average age, baseline, 4-domain, NPCCSS and miglustat use were higher in the EAP. This graph will show changes in the 4-domain NPCCSS from each patient's baseline at every visit. We see that the overall trend among arimoclomol-treated patients in the EAP was stable over the 36-month period. The trend in the NIH cohort was an increase relative to baseline. Here, each green triangle represents a change in score from baseline for a patient in the NIH data. As you can see, a few patients in the natural history study showed any improvement as indicated by the yellow shaded area. In fact, these red boxes reflect that three of the improvements in the NIH cohort were actually patients receiving arimoclomol. In contrast, when we look at the EAP patients treated with arimoclomol we see quite a few in patients improving. While these data are not a randomized comparison, the trend suggests a slower rate of disease progression with arimoclomol consistent with our other clinical data. Next, I'll review the nonclinical data supporting an effect from arimoclomol. Let's look at arimoclomol's mechanism of action. Lysosomes are responsible for the removal of waste products from a cell. NPC-1 proteins are primarily responsible for clearing free cholesterol from lysosomes, which is an important component of healthy cell and lysosomal function. As well as the overall process of autophagy, which is the recycling of cellular components. Let me describe the intracellular processes that lead to healthy lysosomal function. The process begins with transcription factors. When TFEB and TFE3 are localized to the nucleus of the cell, they activate the expression of the genes responsible or lysosomal protein synthesis. These genes are collectively known as the coordinated lysosomal expression and regulation network of genes or the clear gene network for short. Among the clear genes are those responses for NPC, which govern the expression of NPC1 protein. We will primarily focus our attention on the NPC1 gene and protein. In a healthy cell, this expression allows for functional protein to be formed and folded in the endoplasmic reticulum, matured in the Golgi and traffic to the growing lysosome where NPC1 can be embedded in the lysosomal membrane, once there, NPC1 with support from the other proteins expressed by the CLEAR network plays a critical role in removing cholesterol and other lipids from the lysosome. In a healthy cell, this leads to normal autophagy and helping neuronal function. However, in a patient with NPC, this process is disrupted by a mutated NPC gene. The process starts the same with the transcription factors located to the nucleus and activating the clear genes, except in this case, NPC1 is misformed. The mutated NPC1 gene has now expressed the wrong NPC1 protein, which in turn leads to mutated proteins being formed in the ER of which a few make it into the Golgi and even fewer mature and locate to the lysosome. Even if some do, they may be less effective at clearing cholesterol, leading to an accumulation of cholesterol, dysfunctional autophagy and neuronal death. In patients with NPC, arimoclomol improves lysosomal function by two unique pathways. Treatment with arimoclomol forces the transcription factors into the nucleus to activate the overexpression of the clear network genes. These genes in express more of the dysfunctional NPC-1 protein as well as the other proteins in the CLEAR network. With this overexpression, the ER now creates more NPC1 and other clear network proteins, of which more reach the Golgi, go through maturation and locate through lysosome. In most cases, the NPC1 protein is still not fully functional, although there is more of it in the lysosomal membrane, leading to more cholesterol clearance. CLEAR network proteins are also present in higher concentrations. Therefore, arimoclomol works through both an NPC1 dependent and independent mechanisms to improve autophagy reduce cholesterol accumulations and prevent cell death. Several lines of evidence supporting this mechanism were generated during the process of preparing the resubmission. To determine the mechanism of action for arimoclomol, we examined a number of in vitro models, but primarily focused on studies in fibroblast. These included wild-type and mutant fibroblast from patients with various genotypes. High concentrations of arimoclomol were required to measure a potential effect. This is common for fibroblast and generally not unexpected because they have a notoriously high cellular and potential target turnover rate when used for in vitro assays. In vitro doses were influenced by the model used, so the reserve ratio of in vitro, the human plasma concentrations in our studies can be difficult to interpret. Because NPC is an ultra-rare disease, our dosing in clinical trials was based on a safety margin from animal toxicology models. This approach to dosing was appropriate. As you can see in the table, patients in our clinical trials achieved arimoclomol exposures several times higher than those were effective in NPC mouse models. Arimoclomol upregulates the CLEAR network of genes, leading to an increase in NPC1 protein, which ultimately improves lysosomal function. Some of the data supporting these conclusions start with the in vitro models of gene expression, where we observe a significant upregulation of the NPC1 gene, resulting in an approximately 4.5 fold increase. This heightened gene expression subsequently leads to elevated NPC1 protein concentrations, where we see a similar magnitude of increase as with gene expression and as a consequence of increased NPC1 protein. There's a substantial reduction in the concentrations of cholesterol detected in NPC patient fibroblast following treatment with arimoclomol. This in vitro link from gene to protein to cellular function is further confirmed by the results of our in vivo studies. We studied two mouse models of NPC disease. NPC independent mice represent double null mutations with no functional NPC1 protein. NPC deepen at mice represent point mutations with some dysfunctional NPC1 protein. This type of mutation is more representative of the patients in our study and the NPC patient population overall. We evaluated several key endpoints in different studies. We looked at the ability of our arimoclomol to increase the levels of mature NPC1 protein in the brain. We examined myelin basic protein, a well-described marker of neuronal health and we evaluated survival. This graph shows mean levels of mature properly folded NPC1 protein in the brain of either wild-type mice in pink or an NPC dependent mice that were either treated with arimoclomol or were untreated. There was a comparable concentration of protein in arimoclomol-treated mice compared to that in measured in healthy wild-type mice. In addition, the concentration of NPC1 protein in treated animals was approximately 50% higher than that found in untreated animals. Examination of myelin basic protein gives a similar trend where concentrations of myelin basic protein were well above untreated animals. Given the wild-type control and the comparison to untreated NPC mice, this effect can only be attributable to arimoclomol. Improved brain concentrations on mature NPC1 and myelin basic protein also led to an increase in the survival of animals dosed with arimoclomol. Here, I show data from an in vivo models of NPC-dependent mice. More arimoclomol -treated animals survived through 16 weeks and no untreated animals survived past 18 weeks. With their arimoclomol survival was greatly improved. Mean survival times were increased from 16.7 weeks for untreated mice to 18.3 weeks in treated mice. Our nonclinical and clinical studies also reveal a consistent complementary mechanism with miglustat. And mechanistically, these complementary MOAs makes sense. As we already know, in the untreated lysosome proper function is impeded by an accumulation of cholesterol. As a result of poor lysosomal function, other lipid byproducts also accumulate. Glycosphingolipids lipids are one of these byproducts. With treatment of arimoclomol, we see enhanced clearance of free cholesterol. Since NPC1 is primarily responsible for the transport of cholesterol, glycosphingolipids are still present in high amounts, while improved, the lysosome is still impaired versus that of a healthy cell. We can see that miglustat through the inhibition of glucosylceramide creates less glycosphingolipids any processing in the lysosome. With the addition of miglustat, lysosomal function can improve the active removal of cholesterol by reducing the amount of glycosphingolipids reaching the lysosome. Evidence of these complementary mechanisms was seen in concentrations of cholesterol analyzed by filipin staining in NPC patient fibroblast. This slide shows the percent difference versus control in unesterified cholesterol detected for different combinations of arimoclomol and miglustat. With their arimoclomol alone, we see a dose-dependent reduction in cholesterol. At the highest concentration, this was a 42% reduction versus vehicle control. With miglustat alone, we see a similar dose-dependent effect with a 44% reduction in cholesterol. But we see even greater reductions in cholesterol when both are taken together at the maximum concentrations up to a maximum 78% reduction seen in the lower right compared to the control. This in vitro effect is just one part of the data collected on the combination of arimoclomol and miglustat. Here, we can see an overview of how these complementary MOAs have been explored in multiple models. The addition of miglustat had an effect with arimoclomol regardless of the model, the combination providing greater clearance of unesterified cholesterol in vitro than either drug alone. When these drugs were used in combination, we saw further enhancement of clear gene upregulation. And in vivo, we saw improved survival. These findings are in alignment with the clinical trial where we saw statistically significant treatment effect for arimoclomol on the primary endpoint among patients on both therapies. In summary, converging confirmatory evidence across multiple sources leads to one reasonable conclusion that the clinical outcome observed in the pivotal trial is both real and significant. Additional clinical data provided consistent evidence that arimoclomol slows disease progression and these effects are maintained long term. These benefits were also consistent with trends in weight. When arimoclomol patients were matched within NIH patients, arimoclomol saw slower disease progression over 4 years. In the EAP when compared to a similar set of patients in the NIH database, we also saw a slower rate of progression. arimoclomol targets the biochemical pathways most affected by NPC through the CLEAR gene network, which is critical for improved lysosomal function. This mechanism translates to tangible benefits in NPC mice with increased brain concentrations of mature NPC1 and myelin basic protein and increased survival. And the positive effects of arimoclomol shown in the in vitro and in vivo studies were enhanced with miglustat. But the overall most significant observation was the consistent benefit of arimoclomol observed across all studies and analysis. While each individual experiment has its limitations, the collection of mutually reinforcing data taken together confirm the results observed in our pivotal efficacy trial. I will now turn the presentation over to Dr. I. Dali.

Thank you. My name is Christine I. Dali. I'm a child neurologists and Vice President of Clinical Science an Zevra. I will present the safety results supporting that arimoclomol is well tolerated with a safety profile that does not add to the patient's disease burden. The safety database for arimoclomol is considerable on an ultra-rare disease. We have treated a total of 668 individuals with arimoclomol across all clinical programs. Our Niemann-Pick, Type C, the primary safety data come from the 51 patients enrolled in Study 2. 28 of these have been treated 4 years or more. This slide shows the arimoclomol safety profile in Study 2. Most patients reported at least one adverse event during the 1-year double-blind phase. Nearly all adverse events in the study were mild to moderate in severity. The three most common AEs were vomiting, diarrhea and constipation. Which importantly occurred at similar rates across groups. Serious adverse events were more frequently reported by patients in the placebo group. And three patients in the arimoclomol group experienced an AE leading to discontinuation from the study. There was one death reported in the double-blind phase of Study 2. The patient was 8 years old and passed away following an event of cardiorespiratory is approximately 8 months into the study. The event was determined to be related to underlying disease and unrelated to study drug. Looking at the adverse events that led to discontinuation, there was one event of blood creatinine increase, and two events of urticaria angioedema. All events were moderate in severity and result without intervention following discontinuation of therapy. Arimoclomol is well tolerated and did not add to the high burden of NPC disease. We observed a similar incidence of adverse events between the arimoclomol and placebo groups in the 1-year double-blind phase. No new safety concerns emerge in the 4 years of the open-label study. This is also supported by data from an additional 206 patients with NPC that have been -- that have received arimoclomol for up to 3.5 years in the global expanded access program. In summary, and as stated in the FDA briefing document, no significant safety concerns or risks were identified with the use of arimoclomol in patients within NPC. Thank you. I will now turn the presentation over to Dr. Kristina Julich.

Thank you. I'm Kristina Julich. I'm the Chief of the Pediatric Neurogenetics Center at the University of Texas at Austin Dell Medical School and Dell Children's Medical Center. I've been involved in the Niemann-Pick Type C community for years, including actively managing patients on arimoclomol as principal investigator and the arimoclomol expanded access program. Now that we learned the pivotal study results and the additional supportive evidence of effectiveness, I'm glad to offer my perspective on the arimoclomol data. There are four key points that support the use of arimoclomol as a safe and effective treatment for patients with NPC. First is my assessment that the clinical results show a benefit for patients. The clinical trial used the appropriate endpoint, the NPCCSS to evaluate meaningful changes applicable across a heterogeneous population. Study 2 did show a slowing NPC disease progression. Second is that arimoclomol is safe. The clinical data show arimoclomol to be very well tolerated. This agrees with my own experience with patients who've been enrolled by me in the EAP. These data assure me of the safety profile and patients and their parents are comforted by the fact that arimoclomol doesn't add to the underlying anxiety caused by disease progression. Third is the additional clinical and nonclinical evidence confirming the benefits seen in the trial. Across the data, I see consistent evidence. This includes consistent biological evidence that arimoclomol has an effect on the disease pathology. And the results aligned with my personal experience with five patients who have enrolled in the EAP program. Some that have been enrolled for longer than 3 years. I wanted to my patients approximately every 6 months, and I've observed that several of these patients have remained stable, granted that this is anecdotal and uncontrolled but it aligns with the study and the other items. Finally, the mechanism of action makes sense and predict for a benefit. I'm very encouraged the sponsor has demonstrated a probable mechanism of action that is consistent with preservation of neurons and models of NPC and supports the reliability of the observed clinical benefits. Overall, the data show that arimoclomol will address an urgent need for a safe and effective treatment for NPC. I'll now turn the presentation over to Louise Himmelstrup.

Thank you. This concludes our presentation, and we are ready to take your questions.

Okay. Thank you, Dr. Himmelstrup. We will now take clarifying questions to the presenters. When acknowledged, please remember to state your name for the record before you speak and direct your question to a specific presenter, if you can. If you wish for a specific slide to be displayed, please let us know the slide number, if possible. Finally, it will be helpful to acknowledge the end of your question with a "Thank you" and the end of your follow-up question with "that is all for my questions", so we can move on to next panel member. So let me start. So it's Robert Alexander. And I have a question for Dr. Gallo I wonder if you have a display of the prespecified secondary endpoints in Study 2. And if you could -- if you could show us some comments on the results.

I'll ask Jason Connor to [indiscernible].

Jason Connor, statistical consultant at ConfluenceStat. Do we -- right, so I don't know that we have the slide with the secondary end points -- here's okay. There's some. And I would reiterate that the primary endpoint was achieved. And the primary endpoint is really the only endpoint be validated for the disease.

Sorry to interrupt, but these were the prespecified secondary endpoints in the study, no, I believe, right? These are ones which are derived from the NPC or worthy. I thought from the briefing document that there was that the secondary endpoints in the study were not statistically significantly different. Is that correct?

Yes, that...

You don't actually have a display with those prespecified lines from the study?

So there were trends in the secondary endpoint for the NPC-specific endpoints, but not much support from the other end point. So I think it's important to take a step back and the NPCCSS is the only validated tool for this disease and that's where we see a slowing of disease progression.

Just one last -- do you have -- the CGI was in the study though, right?

yes.

Do you have a display of the results?

I can tell you that there were no difference the CGI-I with a p value of 1.

Who had lined up? Dr. Fischbeck.

I actually have three questions. But I could do one now in the others later or if there's time.

You have extra time, so you can probably do all three.

Well, just one question about the post-hoc analysis. I'm not a statistician. But my understanding from statisticians is that post-hoc analysis is not statistically valid, I guess, because you can pick and choose from the data to get the result that you want. Here, it's a little -- it's interesting because the FDA encouraged to post hoc analysis and you followed the FDA a recommendation. So it might be good to hear from the FDA about that later or why they recommended a post-hoc analysis. But I wonder if you have any justification using the post hoc analysis here. In other words, going back and reanalyzing the data from the study rather than doing -- gathering new clinical data or doing a new clinical study.

I'd like to turn it over to Jason Connor, our statistical expert to reply.

Jason Connor again. I agree. Usually, we think of post-hoc analyses as something goes wrong in the clinical trial, and we throughout these patients, think about other endpoints. Here, the trial hit its primary outcome and the analyses that were done post-hoc were done at the invitation of FDA. Even the worst patients in these trials, we saw all three were randomized arimoclomol and never less the primary endpoint was achieved, both with the prespecified endpoint and with the updated one. The post-hoc analyses that were done removing cognition it still hit statistical significance, the post-hoc analysis that was done, changing the swallowing domain, as you heard Dr. Patterson say that swallowing domain change we've done by experts in the swallowing field without knowledge of the data. So it's not like oftentimes as a statistician, I hear people ask me to, well, what if we throw by these patients? This home is tricky, let's not include him. None of that happened here. So absolutely, they're post talk, but they were all done changing the scale without knowledge of individual patient-level data. the original analysis was achieved and the updated analysis, even the updated post-hoc analysis, yes, it still achieve statistical significance, both changing the endpoint and updating to their kind of preferred demand. So right, it's post talk, but it really just supports hitting all those prespecified endpoints.

[indiscernible] then let's go to Dr. Kuhn.

Cheryl Kuhn. My first question is for Dr. Patterson. I assume he is still online. Yes. Yes. Okay. I'd like to understand the process for signing values to each of the response categories in the scale. There is a lot of information about the Swallow Domain and it that made a lot of sense. But the ones where it's jumping from two categories to four, and there's nothing in between. Was there a clinical reasoning for that? Or was there a psychometric reasoning? And what sort of evidence is there to support those score assignments.

Yes. So I'll turn it over to Dr. Patterson.

Yes. Thank you. It's Marc Patterson here. Thank you for the question. As I say, this score was a modification of the 17 domains for originally developed at NIH -- main process was simplified the domains was selected as we said, independently by clinicians and by patients and their caregivers. I just wanted to emphasize those points first. And the -- what we wish to have was a scale where there were very clear distinctions between each level. And as we said in this case, I think you're probably alluding to the each domain where there's a jump from three to five between a non-verbal patient who has functional communication skills [indiscernible] no communication. This is an easy distinction to make and one which is clear and these test by the clinical radars and is -- I think it's meaningful this is apparent. Similarly, in terms of the fine me manipulation, you can see a jump between mild and moderate dysmetria or dystonia, which again is a fairly clear distinction. I should emphasize that, again, that the rates were all trained in advance. In fact, when the scale was developed, it was based on video recordings of patients who are actually examined by me. The raters who develop the scale were all NPC experts who independently rated these patients, and you've already seen the results of the reliability and validity of the scale determined in that fashion. I hope that answers your question, but I'm happy to expand further if it would be helpful.

That was helpful. And I do have a second question. Do we have time?

Yes. Go ahead.

Okay. In the briefing materials as well as the presentation, it was stated that the primary endpoint met statistical and clinical -- clinically meaningful levels of difference. What was your definition for clinically meaningful? What were you using to make that statement? And that's my last question.

Yes. So the clinical meaningful was determined based on two methods. So we did an anchor-based method. And then we have done interviews qualitative studies with the patients, caregivers and clinicians to determine what is a meaningful difference to them. And the results of those studies are that, that any preventing worsening is clinically meaningful.

Dr. Ming did you have a question?

Yes, I have a couple of questions about the rating scale to follow up on the previous question about rating scale. I understand the explanation about the 2-point change in ambulation, fine motor skills and speech. But I wonder in particularly in the clinical trial setting, if you enter the trial at, say a, with a two on ambulation and you get a bit worse that requires a 2-point increase. And so particularly in a study like this, I'm wondering if the data have been evaluated for potential bias over representation of those who are more likely to have a 2-point change on their next pump rather than one. And then the second question is about ceiling effects. It looks like from the briefing materials, at least one individual enrolled at 20 on the 4-domain scale, which is at the ceiling and therefore, could not possibly get worse. Can you address this, please?

So we will address it in your two questions. So for the first one, I'd like Dr. Patterson to explain what happens between, for example, toxic gate to the next level of four. For your second question, I'll ask Jason Connor [indiscernible]. Dr. Patterson.

Dr. Marc Patterson here. Yes. Thank you. Yes, I want to make sure I understand the question Dr. Ming is asking. One of them, I think, probably requires statistical response, and I don't have the broad data to answer that because I think he was concerned that they might be biased towards certain types of [indiscernible] progression, if I understand this question correctly. And all I could say is that we attempted the rate of patients in a consistent fashion. I don't know from my personal experience that I use any bias. But again, I think to answer that question would require looking at detailed data and analysis, and I don't have those -- that level of granular data is trying to be at the moment.

Right. So this is Jason Connor again. Right. So I agree, there are those 2-point changes. We can show you. So you saw this plot on the whole for the whole NPCCSS score. This is showing you the ambulation only change, and apologies that pink is sort of harder to show up here. But you can see, for example, there is one placebo patient there at the bottom that went from one to four. So they, in fact, moved from one to two and then that 2-point jump. In other words, I see the arimoclomol patient there in the middle. But I think one key to is this was a 12-month trial, so many trials we run are relatively short, but 12 months here did allow for longer times to see that progression over the disease. But I agree if there were minor changes worsening within the one step, that may not be captured, but as Dr. Patterson has said, part of the scale was the reproducibility is high because the difference in these stages are pretty evident and reproducible to raters.

Do you want to answer the question about being a ceiling on the one subject?

Can I see that slide back up? Right. And when I think there are ceiling effects, for instance, in ambulation 5 is wheelchair bound. And so when someone has wheelchair bound, they can't go up higher in the score, but I apologize if there was a more subtle part that question, if you could repeat it.

So not just on this specific item, but on the whole 4-domain or prior to that, the 5-domain. Looks like at least one individual was at the ceiling for the total score at the time of enrollment.

That's right.

Do you know was it more than one? All I can say is the min and max.

Yes. So can we see -- I think it's 33, 33. Here we go. Right. So there were two placebo patients and two arimoclomol patients who both had achieved the ceiling by the end of the trial. So -- and you can see the one placebo patient, in fact, didn't have much worse to get there and went to 20. But there were two in each group. And just as a reminder, there's 2:1 randomization. So that's effectively twice as many in the placebo group. But we did see some of those ceiling effects.

Dr. Ellen Berg?

This is Ellen Berg. So there were, I think, eight who, eight patients who dropped out, seven on the drug arm and one on the placebo arm. Can you tell us when they dropped out at what time and what the reasons were. We know that there was one death and three who dropped out because of adverse events, but I'd like to know the reasons why the other dropped out. And then I would also like to know on I think it's Slide 33, whether any of the patients who showed improvement were people who dropped out early, and that improvement was based on an imputed score.

Yes. I'll hand over to Christine i Dali for the reason for discontinuation and then continue on to Jason Connor for the second part of the.....

And the timing of the discontinuation at what point do they...

Yes.

Thank you, Christine i Dali, VP, Clinical Science, Zevra. So we had -- as I said, 3 patients that discontinued due to safety, that was fairly early on. Then we had 2 patients who met the early escape criteria that was prespecified in the protocol that if patients met these criteria early on in the study, they were unblind and switched to open-label treatment. Then we had 1 patient who withdrawn due to consent, which was also during this study, and then we had this 1 fatal event.

And at what point did they drop out?

Yes. I need to get that slide up, hopefully, but you can maybe discuss....

Right. Sorry, I think we can show this. So Jason Connor again. Professor Ellenberg, so this shows the -- all the patients you're referring to. So the 2 urticaria patients, their last follow-up is -- it was at 3 months. And you can see the answer to the other part of your question, both of those patients were actually illustrating improvement at the time of that drop out. And then the urea/creatinine increased patient #6. We track all the way to 8 months. That patient was also seeing a difference. So in respect to your question back to CO-33, there were a total of 10 arimoclomol patients who experienced improvement, 3 of those were patients that you see here who didn't go all the way out to 12 months. So these values were used in FDA's then recommended while on treatment as demand. But even then when we were doing arm imputing process and imputing them to the placebo median of 1, still saw statistical significance. So treating for instance, these 3 patients, all who actually observed improvement out to 3, out to 8 months when we'd use the placebo median for imputation and treat them as 1 worse, still achieved statistical significance in the ANCOVA model.

All right. I'm not sure I really follow that. So...

Well, I'm happy to put that back up and go slower or to answer specific questions.

Why don't you show Slide 33?

Okay. Sure.

And which ones of those did not actually have a 12-month value?

Okay. So -- and can I get SB4 on my screen. So I don't know which specific dots, but there are 3 patients, a minus 1, a minus 1 and a minus 2. So like I said, and I apologize, I don't know exactly which of the 10 people who improved those are. But 3 patients. One, who improved by 2, who we -- and improved by 2 in just 3 months of follow-up and then 2 patients improved by 1, 1 at 3 months and 1 at 8 months when they've left the trial due to AEs. And then so 7 of the other blue dots who were improved were patients who completed the whole 12-month trial.

All right. Just let me remind everyone to state your name before speaking. Let's go to Dr. Kishnani online.

This is Priya Kishnani, and I had a question around the measures. So I understand that cognition, we removed that from the 5-point -- the 4-domain. So was cognition assessed in any other way like using the lighter or anything else to know how these patients were doing? And another question is that -- as I read through the briefing package, to me, it appears that the benefit really is in the patients who have received both miglustat and the arimoclomol, is that a fair statement? I needed some clarification on that.

Yes. So to begin with the cognition domain, we did not add other measures of cognition in the trial. And so -- and the reason why we removed the cognition domain is that it is -- it's a very broad concept in the NPCCSS scale, and we did not have an objective measure to validate it up again. For your second part of the question around miglustat. So what we know is that the miglustat and arimoclomol, they have different complementary effects or mechanism of action. And the nonclinical data suggest that the arimoclomol has a benefit on its own. It also show benefit that is enhanced when used together with miglustat. This study design for Study 2, which is in a very small population. It was designed to evaluate arimoclomol benefit in a representative population of NPC treated with routine clinical care. So it makes it difficult to assess the contribution of miglustat in that question to fully answer the question whether there's an additive effect due to -- also due to the imbalances that we show in the group that did not receive miglustat. But -- so overall, the benefit is -- seen in this trial is on top of routine clinical care. So there's no one side fits all recommendation for -- since patients might not be good candidates for miglustat. So I -- maybe you'd appreciate a comment from Dr. Marc Patterson on -- his perspective if there are any specific groups of people we should not treat.

I just want a clarification what you're calling routine standard of care? Is that with miglustat? Because to us, as clinicians, that's a very important aspect since it's not approved here.

Routine clinical care is considered -- or miglustat is considered part of routine clinical care. 80% received the miglustat as routine clinical care in the trial, which is in line with the general use in the U.S.

Okay. Dr. Lieberman?

Yes. Andy Lieberman. I just wanted to gain some clarification on the proposed mechanism of action to make sure, which individual and the group is best at -- yes. So...

Speak in to the microphone.

Was it necessary to have this induction of the CLEAR network for any of the phenotypic rescues that you showed in the briefing package, Philippine or effect on EndoH sensitivity? And was it -- was this CLEAR network induction seen in any of the animal models? And was it -- if you look at all the gene expression changes induced by arimoclomol, is it by like a hypergeometric test or some sort of statistical analysis, does the CLEAR network pop out because you showed us maybe half a dozen genes that go up? Just trying to get a sense of the strength of this data set for the mechanism of action.

I'll hand over to Dr. Travis Mickle.

Yes. Travis Mickle from Zevra. Dr. Lieberman, I think those are all very good points. When we first looked at the research project that was brought forward with arimoclomol, we're looking across all the different in vitro and vivo models that were possible. I think to answer the first part of your question, there is no good phenotype/genotype comparison that we could make from the clinical data to the in vitro data. So we had access to various -- sorry, I'm just bringing up a slide if I need it. We had access to various patient fibroblasts that we could study the effects in. There was no systematic approach. Look, there were 10 of the CLEAR genes that were actually investigated. Not all of those 10 showed upregulation in the same degree. And then we focused on the ones that gave us the particular effect that we were looking for to look at the various genes. So let me bring up the actual -- if we could look at the CLEAR gene network upregulation. So I'm looking for actually just the different genotypes that we measured in the patient fibroblast. So we may not be able to get you that particular slide. If the Chairman would be so generous to allow....

Yes. Why don't we come back to that so -- because I know we've a number of other members who have questions. Dr. Chung?

Wendy Chung. Most of the data you've shown us today is for the 4-domain severity scale. And I'm just curious because we're asked to comment on the 5-domain scale in general. I understand cognition was difficult to assess. But was this just noisy and there was back and forth, did it add nothing and everyone was not changing, but can you either give us an impression or show us any of the data on the 5-domain scale for those individuals and how they were doing over time?

Particularly on the cognition domain?

It could be the cognition sub portion of the scale or the 5-domain scale overall. But just to give a sense, and that was the original prespecified analysis, it would be helpful to understand that.

Yes. So the prespecified 5-domain NPCCSS analysis is what you see here at the bottom. So that's on the hypothetical estimand MMRM analysis.

So for instance, there's a beautiful graph you've got by participants, individual participants and then moving back and forth. You probably didn't prepare that in advance, but do you have that for the 5-domain?

No. You're thinking about this graph. I don't have it for the 5-domain. I don't have it for the 5-domain. I can see if I can get it after the break.

Impressionisticlly, was there just qualitatively any comments on how people change?

I think Jason Connor can provide a bit more flavor to it.

It's Jason Connor again. Yes, I think your first observation was right that it's just kind of noisy. I mean, for reasons to describe the cognition, especially when a lot of these kids are already neurologically undeveloped at 2, 3 years old, it's hard to precisely identify changes at that level and even in some of the older kids too. But it was basically noisy and didn't add anything either way. Yes.

Okay. Thank You. Dr. Kryscio? I got to turn your microphone on.

Okay. Its Richard Kryscio again. I've a question, just a clarification from Dr. Connor. Could you put up Slide 29. And -- we've heard a lot about dropouts. How were the -- when people were dropped out, how did you get 34 measurements at month 12?

Dr. Connor?

Jason Connor again. Right, so this plot -- the final analysis is just an ANCOVA. So it's not like the MMRM, it's including things at every time. So this used everyone's last known value graphically at each time, which is why you're seeing kind of ends not changing. But the -- and this sort of implies an MMRM, but we showed this graphically, but the final analysis is an ANCOVA that just uses a patient's last observed value before treatment or some of the patients that had rescue, but actually then stayed on arimoclomol, we use theirs even after the unblinding. That way, in the progressive disease, we thought that was conservative as the patients could continue to progress.

Can you clarify what assumptions you make when you're doing an ANCOVA and there's missing data?

So right. So if we can go back maybe to SD-4, they already have it set up for me here. Right. So in the primary analysis -- I should say, the primary updated analysis, we use patient's last known value, not necessarily just the 12 months. So for instance, the 3 patients here that we saw the urticarias, the urea/creatinine, we did use their improvement. The patients who had early escape, you can see the 1 patient, in fact, went to 12 months. So we had their whole 12-month data. Patient Level 5 with the early escape was already at the maximum part of that scale. And then the patient who died, the value was used for their last known measurement, which was 18 out of that maximum of 20. And even when we do sensitivity analyses and supporting FDA's sensitivity analyses, for example, here's even FDA's sensitivity analyses. And then these in the bottom, I'm happy to speak to the ones in the bottom, which we were able to replicate. These are even imputing patients randomized to arimoclomol, who were unable to complete the trial as if they had received placebo. And in all of those cases, you can see the treatment effect holds up even if the right tail of the confidence interval is just above 0, we see these treatment effects still on the order of 1.2 in the kind of the worst case scenarios.

But in all those analyses, what assumptions are you making?

So I think in the ANCOVA, we're not making any assumptions other than using the last known value that, that would project forward. Understandably, that's maybe anticonservative in a progressive disease like this, which is why I brought up this slide which is showing kind of worst-case scenarios and assuming, in fact, patients never even got arimoclomol. So there was the 1 patient that saw an improvement up to 8 months. This is assuming sort of the patient didn't have arimoclomol at all and, in fact, it was imputed to a 1 point change or in the multi-mutation average, I think that patient in FDA's case averaged to 1.5 worsening. So the bottom rows here kind of these worst-case scenarios that again show a 1.2 point improvement.

Yes. I'm less concerned about the value and the assumptions you're making when you're doing slight imputation.

So again, in the ANCOVA, we didn't impute anything. We used the last known value prior to treatment -- or prior to going up treatment.

Ms. Berggren?

Kiera Berggren. I was looking for a little more clarification on how you define dysphagia in this when you revised the scale?

Yes, I'd like to invite our experts, Dr. Lisa LaGorio to reply.

Good morning. I'm Dr. Lisa LaGorio, and I'm an Assistant Professor and a Speech Language Pathologist at RUSH University Medical Center in Chicago. And it's been my pleasure to have been working with Niemann-Pick patients and their families for the last 10 years. And I've also was one of the swallowing experts that was part of developing the rescoring of the -- the revised scoring of the swallow domain. So you asked about how are we defining dysphagia? So let's bring that slide up and show you. Okay. Let's take a look at on the right-hand side, that's the revised scoring. So normal with no dysphagia. So that meant these people were eating and drinking regular foods and liquids as appropriate to their age. And since all patients were over the age of 2, they were eating a regular diet. Coughing while eating was those people that might have a random cough come up once in a while. Intermittent dysphagia was defined as patients who might have more trouble if they were tired or if they were -- it's time of day. But if they really focused on eating, focused on drinking, maybe used a special sippy cup or a straw then they -- and so then they didn't -- were considered to not have dysphagia, whereas the people that had true dysphagia, the Level 3, they were people that were on modified diets, modified liquids, but were still orally eating. Then Level 4 were those people who were still orally eating, but weren't able to maintain enough caloric intake or enough hydration. So 50% or more orally. If it was less than 50% orally, they needed a supplement. So they got their supplementals through the G-tube. So they were Level 4. And then Level 5 were all those people that were G-tube only. And all of these levels were written out on the training manual. So let me just bring that slide up. And the training manual people were -- lets bring that one up. They were trained. They saw videos, they scored it and they were really good inter and intra-rater reliability with all of the scoring.

And that was going to be my next question. What was that inter and intra-rater reliability? Do you have the numbers on that by any chance?

I don't have the numbers, but they were like 0.9 or something. Yes.

Yes. Anyway. Would you mind stay up there for a minute? I just had a question about who actually -- how this -- the swallowing data from the study was rescored. And I assume the NIH study had to be rescored too, to create comparable the 4-item version of the scale, like -- first off, is that correct that the actual was -- the data was rescored for the analysis for the 4-item scale?

Yes. The -- after the revision of the swallowing scale, they did go back and rescore.

Yes. And so also the NIH data had to be rescored too, I assume, or the natural history study data had to be rescored?

With regards to validation, we reran the analysis on content validity comparing the [ actual norms ]...

Yes. Not just really talked about the mechanics of the rescoring. So like that was done by the sponsor. And was it done blinded to treatment group or visit or they just rescored the data? Without knowing?

Yes. So the qualitative study was performed based on an interview -- a protocol, a psychometric analysis and a recruitment plan that was reviewed by the agency before doing the study. And so that was not based on data that was based on an interview on the scale itself.

Okay. Dr. Kraft?

Walter Kraft. Some of the exposure response or dose response in the preclinical package was variable. And with that in mind or optimization, the mechanism of action putative was location or translocation of the transcription factors to the nucleus. Is there any thought as to exactly how that small molecule would facilitate that translocation as far as a mechanism?

I'd like to turn over to Dr. Travis Mickle.

Yes. Travis Mickle, Zevra. We don't have a particular molecular target that's been identified at this point. We do know the various steps that are all involved and have seen whether that's protein concentration increases in NPC1 or gene upregulation. And then, of course, the downstream consequences with the in vivo studies of increased survival NPC1 protein in the brain and so forth.

And if I could just follow up with was there a dose dependency for adverse events in the preclinical animal models? And that's all my questions.

So by adverse events, I mean these were disease models that we had, so fairly adverse as far as all the animals themselves. The main one that we measured more objectively, of course, was survival. So if we could bring up that survival from the presentation. And I'll just show that again. So in this particular case, this is just 1 of the 2 models. This one is actually the double now, which we didn't show in the main presentation. We could also have the point mutation. But you can see here in survival, we do see an increase versus that of the disease model. And then in the point mutation that I described before, there's a more substantial increase with arimoclomol. Now the dose dependence here was variable. So there seemed to be a point in which the lower doses did not achieve any difference. And at the higher doses due to the fact that they were administered through their drinking water, there seemed to be a taste diversion or some other effect where you had this inverse kind of U-shaped curve. But in every one of our studies, we did observe at least a minimal survival benefit, if not substantial like in this particular instance, where we can distinguish the effect that we see in each one of those studies with survival.

Okay. Let's -- if we -- let's go to Dr. Tucker.

Carole Tucker. Two quick questions. The first one is about the measure. There's been -- when I look at this measure -- sorry, I just lost my train of thought doing that. Did we look at the measurement characteristics across domain? So for instance, the fine motor skill in the very last category, it says a growth motor limitation, but it's in the fine motor skill. So I'm just curious about any factor analysis or domain or anything where we've looked at dependency across domains within those. That's a question probably for Dr. Patterson, and I'm looking at Slide 20. And then I've one other quick question.

Dr. Patterson, can I ask you to comment on the severe step of the fine motor skills?

Yes. I don't have the data on factor analysis, which is being asked about and perhaps the statistical team can comment on that. If I understand the thrust of the question, it is that each domain might not be purely measuring the function of domain. And I think that's a reasonable question to pose. The disease, of course, affects multiple levels of the nervous system and separating one from another to some extent, is artificial. So one would typically expect that someone who has severe impairment of fine motor skills would typically have quite pronounced gross motor impairment as well as language and swallowing impairment. They do tend to travel [indiscernible] a clinical observation, which has been consistent over the years. Perhaps you can bring up the data on one of our statistical consultants who could respond because I think you're asking more about a statistical analysis of these data.

Actually, I think what you stated kind of covers it a little bit. Being familiar with other similar scales for other severe diseases, we do see within the more mild cases not as big a relationship, say, between fine motor or speech or growth. It's really at the high level. so it seems like this measure combined, there may be some overlap or additional measurement. And then my other fairly quick question is actually back to the measurement. And in this case, I'm looking at Slide 29 and Slide 30. So what I'm noticing is in the placebo group, you only had 4 people that at a baseline had greater than 8. And so there seems to be -- you can also go to the next slide if you want. These are changed scores. And when I add that 2.1 and I know the group differences, but the placebo group mean started out at 6.7. This is probably for the person -- well, maybe it is you. And the -- I cannot say it whatever, the study drug started out at 9.2. And so -- just out of curiosity, does that tend towards the ceiling effect or a difference if more people in the placebo group were milder and more people in the treatment group were already maybe near of worsening? So you may not see a significant change in the group that's worse because you're seeing kind of this coalescence across response categories and domains. Sorry, I'm not stating that clearly, but Dr. Patterson you're shaking your head, do you think you...

I think it's more a question for our statistical expert, Jason Connor, so I'd like to turn it over to him.

Right. And the -- this is Jason Connor again. Yes, ceiling effects and baseline effects are always important considerations in a scale that does have borders like this. But the ANCOVA that we fit models, the changed score based upon both baseline miglustat use and baseline NPCCS score. So -- If we -- thinking back to the original models from grad school, it's like saying comparing 2 people with the same baseline, what difference do we expect? And that's everything you see in these plots. So this is all controlling for that baseline score.

Okay. Let's go back to Dr. Kishnani.

Yes. I actually wanted us to look at Slide #34. Actually, let me just look one moment here. Slide #32 actually. And looking at -- I'm sorry, just give me a moment. It's like 30. If you could move there, the double functional null mutation at the bottom, where it says it's not available. Is that because there was no placebo group? And if that is the case, then I'd really like to know. I take it that those were the patients who were under 2 years of age when they had their first neurological symptoms, but then is there a way for us to look at them in Slide #33, like who are those patients?

Yes. I'd like Jason Connor to respond. First, I want to clarify that the 3 double null patients were randomized by chance to the arimoclomol group, and we did not stratify for double null.

Jason Connor again. Right. To reiterate, we couldn't find a treatment effect or estimated treatment effect in that group because they were not in the placebo. So we didn't have a comparator there, just the way the randomization worked out. And those 3 patients are 2-, 3-year olds, who went from 14 to 20. So in the arimoclomol group, the 2 patients who went up to 20, they're at the top of the plot. Those are both double null patients who started the trial at 3 years old. And then the third is a 2-year-old patient at the bottom left of the plot who went from a 2 to a 4. But as you see here, the 2 worst patients on arimoclomol were both those young double null patients who we'd expect to have rapidly progressive disease.

I see. So they did do worse than others on -- is that?

Right. Yes, the 2 worst arimoclomol patients up in the top here, where -- were the double null patients. And we did see 1 double null patient randomized arimoclomol, who only progressed from 2 to 4.

Got it. And as you look at Slide 34, where it says 35% of the patients worsened, do we have any deeper phenotyping on who these patients are, where they started, what their characteristics are so that there can be a better understanding?

Yes, I'll ask Jason Connor to comment.

Jason Connor again. So we did look and try to predict, for example, who is most likely to improve or worsen. We try that in all trials, and it's challenging in all trials, it's particularly difficult in a trial of just 50 patients like this. So we looked at that, but it was difficult. As you said, the double null is predictably getting the worst. But otherwise, we were unable to predict who might be most or at least likely to benefit from arimoclomol.

And in terms of safety, for the urticaria and the angioedema that these patients experienced, these were different patients, correct? I believe it was in 3.

Yes. So it was 2 patients experiencing urticaria and angioedema.

And the way they got better was they discontinued study drug or...

They discontinued and then it resolved. Yes.

And so there was no further investigation to understand what the cause for this was? Were they sick? Was there an infection, COVID, something that altered their immune system?

I'd like to ask Christine i Dali to comment.

Christine Dali, VP Clinical Science. Yes, these 2 patients with urticaria happened at the same month in the same country. They developed during 1 day, this urticaria and angioedema. And due to the risk of unblinding the whole site, we decided just together with the PI to discontinue the patients and all symptoms resolved. We did not do any challenges later on for the same reasons.

Okay. Dr. Fischbeck?

Yes. Maybe related to that. One second I lost my place. Yes, on -- or separately, the -- there were no NPC2 patients enrolled in this study. And I wonder if they'd be -- you'd want to have them included in the label even though they weren't studied. It's much less common than NPC1. And maybe that would apply to the double nulls as well.

So we did include NPC2 in the inclusion criteria, but no patients enrolled in the study with mutation in NPC2 gene. So we do not have clinical data to support that effect. But if we go back to the mechanism of action, we've both the NPC1 dependent pathway and the independent pathway. So the -- in theory, the NPC2 patients should benefit from the pathway with the upregulation of the genes and improved autophagy and -- yes, and overall cell biogenesis improvements.

Can I ask the last question I've?

If it's quick because we only have.....

Yes. It's quick. Well, I think it's quick. The nonclinical data, which there is some new data included in the analysis was based on mice that were -- or the treatment was started at 3 weeks of age, well before they became symptomatic or manifesting the disease at 9 weeks. Also, it was administered in the drinking water rather than given by gavage, which is -- which we've dealt with these issues as these are more difficult, but it's, I think, more accurate in terms of how much drug the animals are getting. And I wonder if you've any justification for that. Just not treating after disease onset as the patients are presenting and using drinking water rather than gavage.

I'll turn it over to Dr. Travis Mickle to respond.

Yes, Travis Mickle with Zevra. The rationale to treat the animals at a younger age was based on research suggesting that there could be an advantage to do so. Again, our intent even in the patients is to provide that treatment as soon as possible. If there's going to be a clinical benefit, you want to slow that progression and see what that effect would be in mice. Now the issue with gavage versus drinking water, I certainly understand very well. And ideally, you'd want to use gavage. Dr. Lieberman is one site in the world that does these studies very often as well as the other in Oxford, which, again, as part of our academic research partner here, that was the limitations of the laboratory, they could only perform this in a drinking water scenario.

Thanks. Dr. Ellenberg?

Susan Ellenberg. I've a few remaining questions if there's time. But I want to go back to Dr. Kryscio's question about the assumptions under the analysis, and this is really for Dr. Connor. So he said there's no assumptions made in the ANCOVA, but if you're using the last observation carried forward, then there's certainly an assumption. The assumption is that they're not going to change after the last observation. And since there were more dropouts on the treatment arm, it seems like there'd be more -- it was more patients without the opportunity to decline further after they dropped out, and I wanted to see if I was misunderstanding something.

Yes. I'll turn to Dr. Jason Connor.

Jason Connor again. Yes, thanks for letting me clarify. I agree completely on those assumptions of not having time to progress if we're using those last values. And that's why I tried to show this slide with Dr. Kryscio. So yes, if we're just using the ANCOVA and it's subtle, I think, with that plot implies we're estimating a 12-month treatment effect. And that's what I was trying to say we weren't with the while on treatment as demand that FDA recommended. We were looking at the effect at the last time the study drug was taken, which was 12 months or may have been before. This -- and we've our own, but I'll show up here because I think this is more conservative, but it really still supports the point, as this is looking at 12-month effect sizes. And the 3 methods at the bottom, the worst case, the multiple imputations, the placebo median. These are all looking at cases, really assuming the patient wouldn't even get the drug at all that are using placebo estimates, meaning the patient didn't even stand to benefit from arimoclomol. So we had the patient who was still actually improved by 8 months. And as a reminder, we saw no one improved in the placebo group, but we had a patient leap for an adverse event after showing an improvement up to 8 months. And then the bottom estimates here show -- in fact, they hadn't received drug and are just placebo patients and do get worse. So even in these circumstances, when we allow those few patients who had adverse events who dropped out to be considered like placebo patients, where they didn't have an opportunity to benefit even out to 12 months in their progressive disease, we're still seeing improvements of 1.2 to 1.3 points.

Okay. So when they dropped out -- after they dropped out, they did not remain in the study at all? They did not have any further measurements?

I think they -- I don't know if the patients who dropped out for safety, if we kept getting -- okay. So I'm hearing no for that, that we didn't have those further measurements.

Okay. Can I ask anything else?

Sure.

And -- so for the patients with escape, we did use their further measurements. So the patients who were on arimoclomol had progressive disease and escaped, they actually changed over to arimoclomol because we didn't know if they were placebo or not. So we actually used their measurement for ongoing. So we did go up to 12 months for their measurement.

Okay. I wanted to know if the patterns of adverse events differed by study arm such that there might have been some unblinding because of the high types of adverse events people had?

Okay. So I'm seeing no, but I'd invite Dr. i Dali to answer that question.

Christine i Dali, Clinical Science, Zevra.

Can you adjust the mic. I can barely hear you.

Okay. Sorry. I think I could pull this up. So this is an overview of the adverse events during the double blind. And as you can appreciate, the same amount of events in the arimoclomol group and in the placebo group. And we really did not show any differences between the 2 groups account.

I understand that there was no difference in terms of the total. But what I'm asking is, is there a difference in the pattern? So for example, somebody dropped out on the treatment arm for urticaria. Now if a lot of people got sort of mild itching on the treatment arm, but didn't on the placebo arm, that could have caused some unblinding, because that's what I'm asking?

Yes. But we didn't see anything on that at all. No differences in the pattern.

And I've....

And let's go to Dr. Coon first because we're running out of time.

Cheryl Coon, I think you showed the patient level analysis plot for the overall 4-domain as well as the ambulation. And in the FDA documents, the revised swallow is also shown. Do you have it for the other 2 -- other 3 domains, that individual plot please?

Yes. I'm showing speech here, which is the first domain here. So you see again, the same pattern with improvements in the arimoclomol group and no improvements in the placebo group. Let me show the fine motor skills as well. And here you'll recognize the pattern. So that was the 2 remaining domains of the 4-domain.

Okay. And I just want to go back to this because maybe just me, I'm having trouble understanding. So the difference between the 5-item -- the 5-domain and the 4-domain versions, one difference is the cognition domain was eliminated. But the second difference is that the swallowing value were recalculated based on the new scoring rules. Is that correct?

Yes, that's correct.

Okay. And then when you do these comparisons versus natural history like the NIH, you went through that same rescoring exercise with the NIH data? Or is that the original scoring for the swallowing domain?

For -- particularly for the NIH data set, it's the original 4-domain scoring without revising the swallow domain score.

So it's a little bit apples and oranges, perhaps. I mean, so it was a different scoring rule for the NIH, when you're seeing those curves with NIH versus your data, the swallowing domain was scored differently in the NIH versus what you're showing us for....

No. So we've used the same scoring methodology on both our data and the NIH data. So you see there the 4-domain without revising the swallow domain. So we're comparing the same domain -- the same scoring for both for comparison.

I think that people were nodding that you didn't rescore that. Did you recalculate the individual swallowing values in the NIH data is my question?

No, we did not recalculate for the NIH data.

Okay. All right. Thank you. Andy -- did you want to ask one more question, Dr. Ellenberg before we go to lunch?

The difference between the use of miglustat was intriguing. You showed results of comparing the 2 treatment groups and those who did and did not take that. But I'm interested in whether you compared within the treatment arm, those who did and did not -- who were and who were not taking miglustat at the time? Is there seem to be any interaction there?

Yes, I'll ask Jason Connor to comment on that.

Jason Connor again. I don't think we formally compared or did any statistical test given that there were just 8 patients randomized to arimoclomol who were not on miglustat. And I think all those patients had a reason not to be. Maybe Dr. Patterson can speak to that, but it's hard to compare because there's usually a reason patients weren't randomized to miglustat, for example, I think younger patients couldn't be -- none of the double nulls could be progressed faster. So there were fundamental differences and who did and didn't get miglustat in there.....

I think it was 28 versus 5.

28 versus 8, I believe.

26 versus 5 on Slide...

Yes. 26 versus 5 and then plus 3 double null. So there were 8 arimoclomol patients not on miglustat, 5 were sort of the standard mutation and then 3 really double nulls.

Right. And given that the numbers are quite small, still was there -- do they -- there seem to be any qualitative differences in how they responded to?

So the off miglustat patients responded less well than the patients who were upon miglustat. Yes.

Okay. Thanks. We'll now break for lunch. We'll reconvene again in this room at 12:20 Eastern Time. Please take any personal belongings you may want with you at this time. Panel members, please remember that there should be no chatting or discussion during the lunch break -- no chatting or discussion about this topic. Additionally, you should plan to reconvene around 12:15 p.m. to ensure you're seated before we reconvene at 12:20. Thanks.