The challenges that arise in the design of early-stage clinical trials for gene-based therapies was the topic of the workshop’s first session, which was moderated by Cindy Dunbar, a senior investigator at the National Heart, Lung, and Blood Institute (NHLBI). Richard Finkel, the neurology division chief in the Department of Pediatrics at Nemours Children’s Health System, spoke about natural history studies for Duchenne muscular dystrophy (DMD) and spinal muscular atrophy. Petra Kaufmann, the vice president for translational medicine at AveXis, reviewed the development of gene therapy for spinal muscular atrophy. Jonathan Kimmelman, the director of the biomedical ethics unit at McGill University, provided an overview of the ethical dimensions and recurrent challenges associated with early-phase research and first-in-human trials.
USING NATURAL HISTORY STUDIES IN CLINICAL DEVELOPMENT
When developing a clinical trial in a pediatric population, sponsors should carefully consider the study population and whether it includes newborns, infants, children, or adolescents, Finkel said. Key differences exist among these subgroups, such as differences in the volume of blood and cerebral spinal fluid in the body, which affect drug delivery and target engagement, as well as weight differences, which can affect dosing. Drug metabolism and excretion can differ as a child ages, which can affect drug exposure and a drug’s safety profile, and off-target effects may differ in children of different ages. A disease can also present differently in pediatric and adult populations, Finkel noted, so different outcome measures and study designs would be required for these two populations.
If a disease is present in both the adult and pediatric population, it is typical that clinical trials will take place first in the adult population, Finkel said, but he questioned whether this precaution is necessary. In certain situations, he said, there are persuasive arguments to start in children, especially if it is known that early treatment makes a difference, but an ethical challenge arises if a disease first appears in infants and the aim is to generate the most robust response to a drug.
DMD offers an example of why it is important to understand the natural history of a disease when thinking about gene therapy clinical trial designs, Finkel said. DMD is an X chromosome–linked genetic disorder that affects 1 in 3,500 boys, with an onset between ages 2 and 4 years old.1 Historically, boys with this disorder often lost their ability to walk by age 10, and almost none of those with the disorder were able to walk past age 13. Today, however, steroids can extend ambulation by about 3 years, Finkel said. Furthermore, the specific mutation that an individual carries plays a major role in functional outcomes, with some patients not losing ambulation until 20 years of age (Wang et al., 2018). It is critically important to understand the genotype–phenotype relationship when clinical trials are being designed, Finkel said.
As a second example, Finkel discussed spinal muscular atrophy (SMA), in which two genes—SMN1 and SMN2—play a role. Children missing a functional copy of SMN1 depend on the small amount of protein produced by SMN2, which prevents fetal lethality but is insufficient to prevent the progressive disorder from occurring. Because two genes are involved, there are two possible treatment strategies. One approach, taken by Biogen with its antisense oligonucleotide drug Spinraza®, modifies the SMN2 gene so that it produces more functional protein, Finkel said. Spinraza® was the first drug approved by FDA to treat SMA, and is delivered via intrathecal injection with four loading doses. Risdiplam is another SMA therapy that modifies the SMN2 gene in a similar way to Spinraza®, but the small molecule is orally active. Risdiplam is currently under clinical investigation by Roche, PTC Therapeutics, and the Spinal Muscular Atrophy Foundation.
The second approach, which AveXis took in developing the gene replacement therapy Zolgensma®, is a more traditional gene therapy approach where a single intravenous dose delivers a corrected copy of the SMN1 gene in order to replace the non-functional or missing copy of the gene, Finkel said. Zolgensma® received FDA approval in 2019 for use in children under 2 years of age.2 Because Spinraza® was available as an approved product, it was possible to conduct a randomized controlled trial for Zolgensma® with Spinraza® as the comparison control, Finkel said.
Transitioning to the topic of early-stage research and how those findings can help prepare for clinical trials, Finkel touched on the importance of animal models. Shortly after the SMN genes were identified by Judith
___________________
1 For a more thorough review of genetic changes in Duchenne muscular dystrophy and the implications for therapy, see Gao and McNally (2015).
2 See FDA News Release at https://www.fda.gov/news-events/press-announcements/fda-approves-innovative-gene-therapy-treat-pediatric-patients-spinal-muscular-atrophy-rare-disease (accessed January 26, 2020).
Melki and colleagues in 1995, researchers recognized the need for animal models to study the disease, he said. Arthur Burghes and his colleagues generated a knockout mouse model containing a human SMN2 transgene that recapitulates the severe type 1 form of SMA and predicts response to drugs (Monani et al., 2000). In the years that followed, Finkel said, additional models of SMA were developed and characterized in zebrafish (McWhorter et al., 2003), flies (Chan et al., 2003), and, more recently, pigs (Duque et al., 2015).
On the clinical side, researchers found that disease severity was reduced in children who have additional copies of the SMN2 gene (Feldkötter et al., 2002). Finkel and his colleagues went on to show that infants with the most severe form of the disease (Type 1, or Werdnig-Hoffmann disease) who had two copies of SMN2 exhibited greater morbidity and mortality than infants who carried three copies of the SMN2 gene (Finkel et al., 2014). Furthermore, Finkel and his collaborators developed outcome measures that assessed motor function, strength, and weakness (Finkel et al., 2014) as well as electrophysiological markers (Swoboda et al., 2005) and a neurofilament marker (Darras et al., 2019) in order to predict the disease course in untreated and treated patients.
Natural history studies of SMA patients were valuable because they showed that age is an important variable with regard to the change in motor function, Finkel said, and there was significant functional variability among the ambulant patient population (Mercuri et al., 2016). There is a need to understand the characteristics of the patient population and to notice that certain subgroups are more amenable to change without drugs, he said. When one drug candidate, valproic acid, was tested in young children, the improvement seen in some of the children was apparently not a result of the drug but instead was due to variations in the natural history of the disease. The standard of care (which can involve non-invasive ventilation support, nutritional support, physical therapy, and other interventions) produced a marked improvement in survival from the 1980s into the 1990s without any drug therapy, Finkel said, and is another factor to consider when designing a clinical trial (De Sanctis et al., 2016; Oskoui et al., 2007).
Currently, Finkel said, a Phase 3 open-label trial using a historical control is under way with Zolgensma®. Zolgensma® (also known as AVXS101) is being tested in symptomatic infants with SMA type 1, the severe form of the disease. Data so far have shown that this therapy produces a marked improvement in event-free survival, Finkel said, with those children who are treated as soon as symptoms appear responding best to the gene replacement. Treatment shortly after birth in the pre-symptomatic period appears to generate the most robust response not only with Zolgensma®, but also in the case of Spinraza® and Risdiplam.
In the gene therapy development process, it is helpful to talk to FDA early and often, Finkel said, adding that a paper written by a group of FDA staff (Xu et al., 2017) was helpful in summarizing important lessons about the development of gene therapies, specifically from the perspective of developing Spinraza®. In summary, Finkel said that pediatric studies have particular challenges and regulatory requirements and that understanding the nuances of genotype and phenotype associations can help in the design of an efficient clinical trial. It is necessary to provide standard of care to minimize patient variation, he said, but this adds a second treatment variable to any clinical trial.
DEVELOPMENT OF GENE THERAPY FOR SPINAL MUSCULAR ATROPHY
Continuing on the topic of SMA, Kaufmann said that this disorder affects approximately 1 in 10,000 live births worldwide and that around 1 in 54 people carry the genetic defect responsible for causing this disease (Mendell et al., 2017). The SMN2 modifying gene plays a role in determining the phenotypic severity of the disease, with the most severe form affecting infants and with milder forms that may even present as late as adulthood. Approximately 60 percent of individuals with SMA have type 1 disease, the most severe form, and these individuals show symptoms before age 6 months and are never are able to sit up, and most never see their second birthday. In type 2 disease, which affects approximately 30 percent of patients and first appears between ages 6 and 18 months, individuals can sit, but are never able to walk, and more than 30 percent of these individuals will die by the time they are 25. In type 3 disease, accounting for about 10 percent of patients, individuals can walk but may lose that ability over time (Lorson et al., 2010; Verhaart et al., 2017). Type 1 disease, because of its severity and lethality, was the strongest candidate for the first gene therapy trial, Kaufmann said.
There are different types of natural history data available on SMA patients, Kaufmann said. She and her colleagues at AveXis used patient registries and associated medical charts to get an initial understanding of the course of the disease, she said, noting that these data sources do have significant limitations, particularly missing data or limited time course data. To get a fuller picture of the natural history of type 1 disease, Kaufmann collaborated with Finkel and his colleagues and a group at Columbia University on a cross-sectional and prospective natural history study (Finkel et al., 2014).
Researchers were grateful for the patients and families who participated in this study because they made a significant time commitment to the study and recognized that there might not be direct benefit for their families,
Kaufmann said. She added that it would have been better to have a fully prospective study with regular, more frequent visits, but that would have placed an even greater burden on patients and their families. Later, the National Institutes of Health did fund a more complete, prospective natural history study of type 1 SMA by the Network for Excellence in Neuroscience Clinical Trials (NeuroNEXT),3 which confirmed what she and her colleagues had found. The NeuroNEXT natural history study also provided more data on changes in motor function and electrophysiological measures that were fit for use in clinical trials (Kolb et al., 2017).
One of the important breakthroughs that enabled gene therapy trials for diseases that affect the central nervous system, Kaufmann said, was the discovery that AAV vectors can cross the blood–brain barrier and robustly express transported genes in cells throughout the brain and spinal cord (Foust et al., 2009). This vector was used to design a construct that would enable immediate and sustained expression of the SMN1 protein and produce a rapid onset and durable therapeutic effect, the latter of which was possible because the vector targets non-dividing neurons. The vector was also designed not to integrate into the human genome (Naso et al., 2017; Thomas et al., 2003).
In the Phase 1 clinical trial, three patients were given what was thought to be the minimally effective dose based on animal studies, and another 12 patients were given the proposed therapeutic dose (Mendell et al., 2017). There was an initial 2-year safety follow-up period, and the patients will be followed for another 15 years. “In a new field it is critical that we have a good understanding of what happens to the patients in the long term,” Kaufmann said. The patients who received the gene therapy construct had improved survival, motor function, and motor milestone achievements, she said, and the patients treated with the proposed therapeutic dose had early and rapid motor function improvements.
At 24 months following the gene therapy administration, all patients were alive and did not need permanent ventilation, and 11 of 12 patients in the therapeutic dose group could sit without assistance for longer than 5 seconds. Furthermore, two patients in that cohort were able to stand and walk independently, and so far, patients who received the therapeutic dose continue to achieve and maintain motor function milestones, Kaufmann said. As Finkel had noted, Zolgensma® was approved by FDA in May 2019 for the treatment of pediatric patients under age 2 years who have mutations in the SMN1 gene. Subsequent studies have shown that it is best to treat patients at a very young age, even before they display symptoms.
___________________
3 For more information on NeuroNEXT, see https://neuronext.org (accessed January 26, 2020).
In conclusion, Kaufmann said, robust natural history data with frequent visits are important for developing treatments for patients with rare disease who could benefit from gene therapy. Directly engaging with patient groups that can help with recruitment and represent the patient perspective in regulatory and drug development contexts was incredibly important, she added. Finally, she reiterated how important the altruistic participation of patients and families in natural history studies is for the development of efficient gene therapy clinical trials.
ETHICAL DIMENSIONS OF FIRST-INHUMAN GENE TRANSFER TRIALS
Kimmelman provided a high-level overview of research ethics and discussed those concepts within the context of gene therapy clinical trials. Research that is conducted in humans is ethically challenging because human beings can feel pain, suffer, and have their own interests, Kimmelman said. One purpose of research on humans is to generate information that will be useful for other people in the future. As a result, an important aspect of research ethics involves ensuring that trial-generated information does not get adulterated, either during the study or when health care systems take up that information. Another important aspect of research ethics focuses on the highly trained and specialized individuals who conduct research on human beings, Kimmelman said. This highly trained workforce should be concentrating on the most productive lines of research in order to generate highly reliable information about the safety and efficacy of a particular treatment. Doing so may help to ensure that health care dollars are not wasted on expensive and ineffective therapies, he said.
All potential drugs, including gene therapies, need to undergo an arduous vetting process of clinical development and be proven safe and effective, Kimmelman said. The first stage in the development process is considered a learning phase in which researchers try to determine how best to use the therapy. For example, developing a cell-based therapy for amyotrophic lateral sclerosis requires knowing the appropriate dose of cells to deliver, the type of immunosuppression regimen required, and the timing of the drug administration that achieves optimal benefit.
The next task, Kimmelman said, is to use this information to design and conduct rigorous clinical evaluations, typically randomized controlled trials. Deciding when to initiate early-phase clinical trials requires a basic understanding of the potential risks and benefits of the proposed treatment, and that in turn involves demonstrating potential efficacy and safety in high-quality preclinical studies. In the case of gene and cell therapies, FDA pays closer attention to the quality and design of preclinical studies (e.g., blinding and the selection of appropriate controls) than it does for small
molecule drugs, Kimmelman said. Ultimately though, broader decisions about risk versus benefit are ethical judgments made by IRBs and ethics committees, he said, and FDA is clear about delegating those types of ethical judgments. One problem with this approach, he said, is that some IRBs may not be well equipped to carefully vet all of the preclinical evidence and to make judgments about the prospect of direct and societal benefits.
Another recurrent challenge in the conduct of early-phase research is deciding whether to consider exposure to the new intervention to be therapeutic. There are two main reasons why it does not make sense to call access to interventions and early-phase studies therapeutic, Kimmelman said. The first is that during the early stages of clinical research it is not yet clear if an intervention will be effective because the goal during this phase is to identify the materials, practices, and beliefs needed to combine with an intervention to elicit its activity. Claims of therapeutic benefit in first-in-human clinical trials should be met with skepticism, Kimmelman said. A second reason it does not make sense to call early-phase clinical studies therapeutic, he said, is that there is not strong evidence to suggest that participation in an early-phase clinical trial will provide any therapeutic benefit, given that most early-phase research studies fail to find evidence of efficacy and safety. In his own research, Kimmelman said, he has observed that the fraction of patients who enter a Phase 1 clinical trial and receive a drug at a dose that will ultimately receive FDA approval for their indication is about 1 in 70, while 10 to 15 percent of participants in a Phase 1 trial will experience a grade 3 or grade 4 adverse event (unpublished finding).
Looking beyond an assessment of safety, Kimmelman said that a major challenge of early research studies is to define the lowest effective dose and the optimal timing of delivery, particularly for interventions that are expected to be expensive, such as gene-based therapies. Another challenge is to avoid conducting an uninformative trial—that is, one that does not provide results that are of value to patients, researchers, clinicians, or policy makers (Zarin et al., 2019). Whether a trial is considered to be informative is based on its relevance, design, feasibility, integrity of analysis, and reporting (Zarin et al., 2019). In particular, reporting the results of early-phase research is critically important so that subsequent researchers or health care systems can use the data in patient decision making (Fung et al., 2017). “It is critical that we recognize that if patients have participated in clinical research, we honor their sacrifice and contribution by making sure that we are promptly and completely reporting the results of their participation,” Kimmelman said.
DISCUSSION
Following the three presentations in this session, there was a panel discussion moderated by Dunbar, and workshop participants had an opportunity to ask questions of the speakers. Topics during the panel discussion included early-phase trials for pediatric patients, patient stratification, and ways to improve clinical trial readiness.
Issues with Conducting Early-Phase Trials in Pediatric Patients
There is an inherent conflict between a first-in-human trial not being considered therapeutic and the requirement to offer a pediatric population a chance of therapeutic benefit, Dunbar said. How can that be dealt with, she asked the speakers, especially with regard to consent forms? There are exceptions to the general rule against designing Phase 1 studies for therapeutic outcomes, Kimmelman said, especially when there is exceptionally detailed mechanistic data. Another potential exception would be when similar interventions have already been tried in other genetic diseases using the same (or very similar) vectors, and a safety profile is established.
When there is no treatment for a disease and there is the prospect of a gene therapy, parents will still often sign their children up for a trial because they are desperate for any improvement, Finkel said. “The real obligation, I think, is on the physician serving as the investigator to frame the discussion carefully and look at the risk and benefit,” he said, adding that his institution has a policy where a patient advocate is present in his discussions with parents to make sure that he and other investigators are presenting information in an unbiased manner and that parents are truly understanding the potential benefits and risks. Kaufmann agreed that it is critical to be transparent with parents and to have good information available for them. “I think the more we partner with patients, the more we have strong patient groups who can provide that kind of information to parents and patients, the better off we are,” she said.
Data Collection and Patient Stratification
Speakers were asked whether they believed clinical trial data collection was robust enough to be able to understand which patients are most likely to benefit (or not benefit) from the therapy. The more information that researchers can collect, Kaufmann said, the better chance they have of being able to stratify patients in the future. The therapies being discussed in this session apply to a very small group of the broader population, Finkel said, and it is not clear how generalizable the results are, even when the therapy
is highly effective in the study population. Once a drug is commercially available, real-world data from registries can be collected, he added, which can be very valuable for understanding patient stratification.
Tools and Approaches Useful for Clinical Trial Readiness
Speakers were asked by a workshop participant if they could identify tools or models that they have found useful in their work, specifically for determining the minimal effective dose of a gene therapy. High answered that large animal models are very useful in predicting therapeutic doses, efficacy, and safety. Kaufmann agreed, adding that forming partnerships with patient groups and potential investigators early, while preclinical work is ongoing, is helpful for trial readiness. Finkel said that his group has been finding it useful to develop informative biomarkers for early readouts regarding safety and efficacy.
The topic of continuous monitoring devices was brought up by a workshop participant, who asked the speakers if they were considering those types of approaches as a way to collect frequent data from patients. This is an area that needs to be explored, Kaufmann said, especially for use with patients who may need to travel for treatment and assessment. Such technology would also provide data from patients in their natural environment, as opposed to the clinical environment. The challenge, she said, is dealing with the flood of data that such devices would generate. Patient privacy might also be an issue with such devices. Another challenge, High said, will be to correlate the data from wearable devices with more standard measures.
Exploring Future Opportunities and Challenges
A workshop participant asked if there were procedural and ethical differences between gene-based therapies and cell-based therapies. High answered that viral vectors used to deliver gene therapies, such as AAV, are similar to other specialty pharmaceuticals in that they are manufactured and shipped to a pharmacy, whereas cell therapies require a far more complicated infrastructure that resembles that used for bone marrow transplants. There are concerns, she said, that if a cell-based therapy for a disease such as sickle cell disease works, creating the infrastructure to treat as many as 100,000 people could be difficult.
Panelists were asked if they see gene therapies being used in the future for more common, chronic diseases such as osteoarthritis. It will take a great deal of experience and safety data from rare disease indications before the field will start thinking about more common chronic diseases, Kaufmann said. There will also be the problem of scalability that will
require innovation in manufacturing processes to address. Dunbar noted, however, that gene therapy is already being used in the cancer field.
Speakers were asked if it might be possible to use the results from one trial in a rare disease to shorten the development time for a similar disease, say all retinal diseases, rather than starting from zero for each disease. High responded that timelines for clinical development have shortened over the 30 years that gene therapy clinical trials have been run, and she said that FDA’s draft guidance documents issued in 20184 should help further shorten timelines.
___________________
4 See Cellular and Gene Therapy Guidances at https://www.fda.gov/vaccines-blood-biologics/biologics-guidances/cellular-gene-therapy-guidances (accessed January 26, 2020).
This page intentionally left blank.
