The news of Amylyx Pharmaceuticals’ Phase III failure for its amyotrophic lateral sclerosis (ALS) drug, Relyvrio, and its subsequent market withdrawal has sparked deliberations throughout the healthcare community about the ongoing struggles in neuroscience drug development.
Although this particular announcement is recent, the failure of late-stage clinical trials in neurology is an all-too-familiar tale. Over 10 years, only 53.1 percent of investigational drugs in neurology transitioned from Phase III to New Drug Application (NDA) or Biologics License Application (BLA)—lower than in any other disease area except ophthalmology and oncology. This is significant, as Phase III trials are often the longest and most expensive trials to conduct.
But the problem goes deeper. The likelihood of approval from Phase I was only 5.9 percent in neurology—also much lower than in most other disease areas.
We keep repeating the same mistakes, hoping for different outcomes
This is a key moment in time to assess the limits of the current methodologies, especially for central nervous system (CNS) disorders. There is a compelling case to change course from an over-reliance on genomics and subjective measures. In the case of ALS, for example, genetic assays have yielded very little predictive power in identifying those who will develop the disease. Yet, we include almost no exposomic markers in clinical trials. These markers are critical, as they measure the life course of environmental factors. We continue to focus solely on “nature” when the “nurture” aspect might unlock the answer. We also detect and characterize ALS and so many other neurological conditions almost entirely on symptoms, rather than a true molecular profile.
To grasp the magnitude of this issue, we need to ask ourselves: what is the molecular definition of ALS? If you are scratching your head, you are not alone. The reality is that we do not have a clear understanding of the disease at the molecular level. This is true for neurological conditions at every life stage; autism spectrum disorder is a prime example of a childhood disorder defined by observed symptoms.
This is not the case for other areas of medicine. Consider type 2 diabetes. The definition of the disease, its severity, and its response to treatment are based on a biochemical profile. If we are to make progress in finding effective treatments for CNS disorders, it is essential to cast a broader net beyond a gene-centric approach and reliance on symptom-based diagnostics.
We have looked under the lamppost for long enough. Now, we just have to accept that the car keys are somewhere else.
A research setback does not signify a lack of progress; rather, it offers a chance to identify what works and what doesn’t. However, it is the responsibility of life science leaders to ensure that their organizations learn from these failures and pivot to better approaches so that new medicines become available for the patients who need them.
It’s time we started thinking about time
By now, many of my colleagues would point out—and rightfully so—that numerous non-genomic tests are indeed done during CNS drug trials. The problem is not what we measure in blood, but when. We, as a profession, have become reliant on “snapshot” testing—testing that only captures one or a few disconnected moments in time. Contrast the two or three occasions when we do blood tests during a clinical trial with the simple fact that the CNS is remarkably dynamic. Its many intricate parts are constantly humming in perfect harmony when healthy. It’s like a beautiful symphony. Yet, how do we truly measure the intricate complexity of this symphony if we only listen to two or three moments of music?
It’s time to move away from measuring a large number of things to measuring a large number of times. We need to measure the CNS at a frequency that is aligned with how it functions. We need a technology that can measure our physiology at hundreds of time points. Today, we have the tools to do so. Assays using just a single strand of hair can provide between 500 and 1,000 time points of molecular data across hundreds, if not thousands, of molecules.
Some temporal biomarkers have already been adopted. One prominent example of this effort can be seen in the case of multiple sclerosis (MS), where magnetic resonance imaging (MRI) has proven to be a powerful tool for diagnosing the condition.
The time dimension is where the secrets of the CNS are hidden.
A new frontier in CNS trial biomarkers
Despite the incorporation of biological insights in some CNS trials today, they often provide a partial and limited understanding of disease pathways. It’s important to introduce innovative perspectives that prioritize the discovery of novel biomarkers through objective testing methods that go beyond genetics. Seeking out novel biomarkers that rely on objective testing can offer invaluable insights.
One area that biopharma has only scratched the surface on is the utilization of the exposome—a measure of all the environmental exposures an individual experiences throughout their lifetime, even prenatally, and how their biological response to those exposures relates to health. It is time to make these types of measurements more mainstream when it comes to clinical research.
Incorporating such novel biological molecular measures into drug development has the potential to significantly impact both drug development and patient care. These biomarker-driven trials open up avenues for identifying new treatment targets, accelerating the development of therapies that alter the course of diseases, and refining treatment plans customized to each patient’s needs. Moreover, biomarkers offer the advantage of early disease detection, enabling prompt interventions and ultimately leading to better outcomes for patients.
The industry must recognize the constraints of genetics and phenotyping testing and embrace innovative approaches that go beyond traditional methods. This calls for a concerted effort to identify new biomarkers based on objective testing methodologies that consider both environment and genetics, while also prioritizing temporal dynamics over snapshot testing.
