Randomization and placebo groups are giving way to biomarkers, real-world data from wearables, and multi-center and adaptive trials.
The European Medicines Agency (EMA) defines a disease as rare if it afflicts fewer than 5 people in 10,000. The list includes well-known diseases, such as sickle-cell disease, and more-obscure diseases, such as the neurodegenerative disease aceruloplasminemia, the autoimmune disorder Schnitzler syndrome, and Waldenström macroglobulinemia, a type of lymphoma.
Even within one rare disease, many subtypes can exist that require personalized approaches to therapy. As one example, EMA’s chief medical officer Steffen Thirstrup mentions cystic fibrosis, which affects around 1 in 2,500 babies born. “When I left medical school 30 years ago, we knew that cystic fibrosis was caused by a genetic defect, but now we know that different mutations in the genome can cause this disease,” he explains. “So now, you actually start to subsegment this rare disease, which adds additional complexity to finding patients and screening them to find out whether they’re eligible for a trial.”
Some of the treatments for rare diseases, such as gene editing, are relatively new and therefore require different approaches to demonstrating safety and efficacy. As Peter Marks, director of the FDA’s Center for Biologics Evaluation and Research, explains, difficulties with testing treatments for rare diseases start at the pre-clinical stage. “As we move into the era of genome editing, some of our paradigms, such as using animal models to do toxicology testing and pre-clinical testing, only take us so far,” he says. For example, a drug might work as desired in an animal model but create undesirable effects in humans.
As one such example, Annemieke Aartsma-Rus, professor of translational genetics at the Leiden University Medical Center in the Netherlands, mentions the muscle disease myotubular myopathy, which affects 1 in 50,000 male newborns worldwide. There is “a dog model where gene therapy worked really well, preventing muscle pathology, but some patients experienced severe liver problems,” when the drug was tested in humans, she says. As it turned out, those patients already had liver complications, which the treatment exacerbated. As Aartsma-Rus notes, “A treatment can have different effects in a pathological tissue than in a healthy tissue,” and that might not be identified in animal models.
Aartsma-Rus develops treatments based on antisense oligonucleotides for patients with very rare mutations or even individual cases, so-called n-of-1 trials. Side effects of personalized therapies will be seen only after they are tested in humans, she says: “So far, there’s not really a solution for predicting this, and the only real proof is your actual patient.”
Academic–industry collaborations might enhance pre-clinical safety studies. “Pharmaceutical companies probably have a lot of data on what is safe and what is not, but [academics] don’t have access to that,” Aartsma-Rus says. She works on Duchenne muscular dystrophy, which affects 1 in 3,500 male neonates. For this disease, “companies are sharing the outcome measures of placebo groups, but not the treatment groups,” she says. Aartsma-Rus believes that sharing data on rare diseases should be mandatory, “especially for things that don’t work and in regard to safety — things that are toxic,” she says.