The good news is that the FDA recently approved for use in ALS (amyotrophic lateral sclerosis) only the third drug meant to slow the progression of the disease – Relyvrio. However, the approval was highly controversial, because the data for the efficacy of this drug combination is questionable. The approval raises critical questions for what it means to take a science-based approach to medicine.
ALS is a neurodegenerative disease, similar to Alzheimer’s disease and Parkinson’s disease. These have proved to be a challenging class of diseases to understand and treat. In ALS motor neurons (both upper and lower motor neurons – the nervous system cells that allow for voluntary movement) die over time, causing weakness and the atrophy and death of muscle tissue. This is a fatal disease, with survival being between 2-5 years from time of first symptom. It is also relentlessly progressive, spreading throughout the body until eventually patients are not strong enough to breath on their own. At that point they are either placed permanently on a ventilator to help them breath, or they will die from the ALS.
More precisely, ALS is a type of disease, defined as degeneration of the motor neurons. But there are at least several, and perhaps many, specific pathophysiological diseases that cause ALS. It can be inherited, for example, or sporadic. It can be caused by a neurotoxin in some cases, and can occur with other neurodegenerative symptoms. But most patients with ALS are sporadic (no family history), idiopathic (no known cause), and follow a typical pattern of progression. It is a challenging disease to understand because many things happen at the cellular level on the way to these neurons dying, and it is difficult to know which processes are causing the progression and which are just a result of the cause. There also may be many causes working together, and when the cumulative stress gets beyond a certain point, the cells start to die.
There are currently three drugs approved to slow the progression of ALS, riluzole, edaravone, and now Relyvrio. The first two have a modest effect, prolonging survival by a few months. This can be significant for ALS patients, however, and the hope is that eventually we will build up a cocktail of drugs, each perhaps with modest effect but cumulatively with a significant effect. The ultimate goal, of course, is a “cure” which can be defined in this case as stopping progression (hopefully early enough to prevent functional decline). Reversing existing muscle loss is a different problem.
Relyvrio is a combination of two drugs, sodium phenylbutyrate and taurursodiol. The mechanism of action in ALS of the two drugs is currently unknown. However, phenylbutyrate enhances a pathway by which nitrogen is removed through the kidneys. It may also have effects on misfolded proteins, which can be a contributor to progression of ALS. Taurursodiol is a bile acid that may interfere with apoptosis – cellular death. These are speculative mechanisms, but not implausible.
Why is the FDA approval of this combination controversial? It’s because the data they relied on for approval is preliminary and not very convincing. It is, in fact, a phase II trial that focusses mainly on safety and only preliminarily at efficacy. Typically the FDA relies on two or more phase III trials, which are larger and more rigorous than a phase II. The study, called the CENTAUR trial, was published in the NEJM. The study enrolled 137 patient with ALS, with two-thirds in the treatment group and one-third in the placebo group. That is a relatively small trial for a clinical study, but not uncommon for ALS studies as it is a rare disease with short survival. Here are the results:
In a modified intention-to-treat analysis, the mean rate of change in the ALSFRS-R score was −1.24 points per month with the active drug and −1.66 points per month with placebo (difference, 0.42 points per month; 95% confidence interval, 0.03 to 0.81; P=0.03). Secondary outcomes did not differ significantly between the two groups. Adverse events with the active drug were mainly gastrointestinal.
The effect size is small and the statistical significance is slight. But actually, the results are much more suspect than that if you delve into the details. Here is a nice analysis published in a blog post on the Science website, which I will summarize. First, there were problems with randomization. The first 18 patients were all assigned to treatment, because the placebos had not yet shipped when the trial started. The next nine were then assigned to placebo, and then normal randomization took over from there. This is a small problem, but it partially unblinded the study and given the small numbers of subjects, small anomalies can be significant.
In my opinion the most significant issue is that 12 patients in the treatment group were on either riluzole or edaravone, the other approved drugs for ALS, while only two patients in the placebo group were. This is a critical asymmetry. Further, the primary outcome of the study is changes to a scale used to assess neurological function in ALS patients (the ALSFRS-R scale). The question is, do patients with ALS progress linearly along this scale as their disease worsens, or does progression follow another pattern? The study assumed a linear progression, but this may not be true.
Does all this matter? Absolutely. In fact, the FDA’s own analysis showed that if you make any changes to the choices made by the researchers in terms of how to analyze the data, the results become non-significant. If this sounds like p-hacking, that’s because it is. The company running the trial, Amylyx, chose the one analysis that showed a barely significant result.
Looked at from this perspective, the results don’t look good. If you ask – should SBM support the FDA approval of a drug combination with unknown mechanism of action based on a single preliminary study that is barely statistically significant and only as a result of clear p-hacking, the answer might seem obvious. But there is another layer to this case that makes it more complicated.
In a letter to the FDA [PDF], ALS clinicians asked them to approve Relyvrio for ALS. They write:
There is no cure for ALS, and most people with the disease die within 2-5 years of diagnosis. The unmet need is critical and pressing. That means that therapies that provide benefit, even if the effect is incremental, need to be available to ALS specialists and neurologists who are in the front lines providing care and treatment to people with ALS.
I actually know some of the people who signed this letter and worked with them. They are all serious and dedicated scientists and clinicians, and their opinion should be taken seriously. Because ALS is an almost universally fatal disease with few treatment options that are not very effective, we need to consider the risk-vs-benefit of using speculative treatments based on preliminary data. This drug combination is not magic – it has some reasonable plausibility. It is relatively safe. So the risk vs benefit, even if the possibility of benefit is small, is reasonable. What they are saying is – give us as clinicians treating these patients the option to decide if this is worth it for our patients.
I would consider Relyvrio to be an experimental treatment for ALS, despite the FDA approval, and its use to be compassionate care, not proven therapy. This is all well-established and ethically valid practice. The question is – how should the FDA have handled this? They have clearly moved recently to the position of approving treatments for neurodegenerative diseases at a lower standard than normal because of the seriousness of these diseases.
Also – there is a phase III trial in the works for Relyvrio, with results likely available in 2024. A larger and more rigorous trial should answer the question. I would not be surprised if the new trial shows no efficacy, although of course I hope that it confirms a clinical benefit. If it does prove with better data to be worthless, then the FDA should withdraw approval. In the end, that is a reasonable approach, as long as the more rigorous trials are done and the results affect approval and practice.
However, there is a final concern – whether it will be possible to recruit patients for a placebo-controlled trial when they can get prescribed the drug. Why take a chance on getting the placebo? If that turns out to be the case, then that would cast the FDA’s decision in a much more problematic light.