March 20, 2024 | When it comes to research on amyotrophic lateral sclerosis (ALS), every stakeholder—most especially patients—knows time is of the essence. From diagnosis to death, life expectancy from the debilitating neuromuscular disease can be two years or less. “The time people give us is currency,” says Indu Navar, founder of the nonprofit Peter Cohen Foundation operating as EverythingALS.
Through that lens it becomes clear the only way forward is by eliminating unnecessary repetition with open science and precompetitive collaboration. For ALS researchers, the focal point of these efforts is the creation of digital biomarkers that can shorten clinical trials and the odds of their success with fewer required participants, she says.
For tech entrepreneur Navar, this is personal. Her husband Peter Cohen died of ALS without the benefit of a speedy diagnosis, let alone access to the kind of large-scale research efforts required for making life-changing therapeutic breakthroughs.
In 2020, a year after his death, EverythingALS was born to bring patient-centricity, technology innovation, and data science to bear on the complex and dehumanizing disease. Creation of an open innovation platform for knowledge-sharing and citizen-driven research is the key enabler.
One of the big needs in ALS research is better outcome measures for tracking disease progression in clinical trials, according to James Berry, M.D., a neurologist at Massachusetts General Hospital who serves on the scientific advisory board of EverythingALS. That would allow for more rapid and effective studies not just for ALS but neurodegenerative diseases in general.
Berry is co-principal investigator for the Asymptomatic Gene Study for early diagnosis of ALS and scientific advisor to the Austen Speech Study and Radcliff multi-modal study, the three trials developed by EverythingALS that aim to advance digital biomarker research for the early detection of ALS disease progression and facilitate remote patient monitoring of changes in participants’ speech and other modalities over time. Audio and video speech data collected for the Austen Speech Study is shared currently with researchers interested in using machine learning to analyze the speech of neurologically impaired individuals.
The Pooled Resource Open-Access ALS Clinical Trials (PRO-ACT) database, the largest publicly available repository of longitudinal ALS clinical trials data, has helped sponsor companies and academic investigators design more effective studies, says Berry. “We think we can do the same thing with speech... to improve understanding of quantitative motor speech analysis.”
The 12-item questionnaire known as the ALS Functional Rating Scale-Revised (ALSFRS-R) is the current mainstay in trials and does its job of capturing how people function in the aggregate, Berry says. But it leaves a lot of areas of function under-investigated—for example, by having patients rate the quality of their speech on a scale of 0 to 4, omitting potentially important details related to speaking rate, articulation, and intonation.
It’s important to measure small changes that can happen during a neurology trial where patients might be followed for only six to nine months, points out Navar. Over time, they might translate into big changes that impact patients in a positive way.
Navar’s husband was 49 years old when he experienced his first symptoms, including difficulty in lifting the front part of his foot. At the time, the couple was investing in and advising startups. He had worked at Amazon for 18 years and helped build the Amazon Cloud while she was on the founding team of what is now WebMD and for many years ran Serus Corporation, a software as a service company for the manufacturing space that was acquired by a public company.
It took two years for Peter to be diagnosed with ALS, she says. As with many neurological diseases, a wait-and-see approach is often taken no matter how early patients see their doctor. Ruling out other causes of the symptoms can be expensive and time-consuming.
Speech analysis algorithms may be of some assistance in the diagnostic process, says Berry, but are likely to augment rather than replace existing tests in the arsenal. A small group in the ALS community is working on quantitative motor speech analysis but more as a potential outcome measure for trials to speed drug development than to bring a new diagnostic to market.
So many speech analysis algorithms now exist that they are commodities at this point, and useful in the field, Berry says. But they measure only one aspect of speech, which is comprised of multiple subsystems and distinguishing characteristics, and therefore capture “a very small portion of what is going wrong in a complex disease like ALS.”
Advanced algorithms might capture more information by “listening” to people talk and extracting quantitative information that distinguishes normal from disordered speech, using the same natural language processing technology behind voice assistants, he adds. “We’re making strides in that direction very rapidly and... my guess is that we’ll end up with many very good algorithms.”
EverythingALS has made citizen-driven research possible by building a community of people who come together for online events where they can routinely hear presentations from experts about the research they’re doing and why it’s important, says Navar. Patients can also use the EverythingALS platform as a place to talk to one another during what they call “happy hour.”
About 7,000 people are subscribers who have attended some of these events, and another 300,000 people from 50 countries consume the content on social media channels such as YouTube, she adds. A subgroup of these individuals who have either been diagnosed with ALS or think they might have it come to the EverythingALS website where they learn about enrolling trials such as the ongoing Austen Speech Study.
The caveat in this case is that they must be able to speak, an ability that patients ultimately lose and forces them to drop out the study. Roughly 1,500 people have signed up for the trial since its launch in 2020 but only about one-third of them have contributed speech data, which was recently released, Navar reports.
It’s an agile, results-oriented research process that incorporates learning and re-implementation, she continues. “Our purpose is to learn and go deeper into what we learn and for that we need a constant feedback loop.”
EverythingALS has a long list of industry partners—the consortium includes companies such as AbbVie, Johnson & Johnson Innovative Medicine, Eli Lilly, Bristol Myers Squibb, and Mitsubishi Tanabe Pharma America, among many others that are providing financial support—who work together in a precompetitive way on developing better endpoints for ALS trials.
In terms of communication with ALS researchers like Berry as well as pharma companies in the consortium, it’s a true collaboration and not just a quarterly presentation, says Navar. Meetings happen weekly with the data science team at EverythingALS.
Besides quantitative motor speech analysis, EverythingALS is incorporating other types of digital biomarkers into ongoing and future studies, “to try to bring the same successes we’ve had in speech to things like actigraphy and movement, which are also critical to ALS,” notes Berry. “I see a very bright future there.”