28 Jun 2022

Part 1 of 2

Kristin Holloway opened the inaugural Holloway Summit, held May 18-20 in Miami, by recounting her family’s experience with frontotemporal dementia (FTD). The events leading up to the diagnosis of her husband, Lee, may seem all too familiar to families grappling with the disease. As Kristin realized in hindsight, her husband’s behavior had started shifting long before, indeed, soon after their marriage and before the birth of their child. An ambitious tech entrepreneur who had co-founded the internet security company Cloudflare, Lee started sleeping in, skipping work, and lost interest in seeing friends. Like many spouses of people with FTD, Kristin initially chalked up her husband’s behavior to various stressors in his life. It took a visit from Kristin’s mother to make it painfully clear that something was wrong. After months of agonizing attempts to drag her reluctant husband to psychiatrists, and ultimately to neurologists, the family received the devastating diagnosis: Lee Holloway, at the age of 36, was suffering from behavioral variant FTD.

Kristin decided to do more about FTD than take care of her husband. She set up a fund with the Association for Frontotemporal Degeneration (AFTD). Part of her gift goes to funding an annual meeting around a specific goal, and for the initial one, that goal was accelerating the development of digital assessment tools for this disease.

Organized by AFTD, the summit brought together researchers in academia and industry, as well as business and nonprofit foundation leaders. The two days of talks and discussion brought experienced FTD investigators who are steeped in the confounding complexities of these diseases face-to-face with young entrepreneurs who have no patience for the slow pace of progress in this field.

While the first symptoms in FTD vary from person to person, one constant is that they only become clear to loved ones in hindsight, at a time when the affected person is far down the path of disease. Merely obtaining a diagnosis can be a challenge in this highly variable constellation of disorders. No disease-specific treatments exist. Clinical trials struggle to enroll sufficient numbers of participants, and to develop meaningful endpoints that capture the diverse manifestations of the disease. All this is why digital health technologies—which could reach anyone with a smartphone and keep tabs on the fluctuating symptoms of FTD—are so sorely needed. Having such markers would speed the path to treatments for the disorder.

Relative to its big cousin, Alzheimer’s disease, FTD is relatively rare, occurring in approximately 11 in 100,000 people. It is also marked by staggering heterogeneity at every turn, from its genetic causes to the type of pathology that spreads in the brain to the behavior, speech, and motor disturbances that arise. Behavioral variant FTD is the most common form, and some 40 percent of it is caused by an underlying pathogenic mutation. Primary progressive aphasias, which are most often sporadic, come in two forms marked by halting speech, or problems finding words or recognizing common objects. Certain movement disorders, including the atypical parkinsonian disorders corticobasal degeneration and progressive supranuclear palsy, and amyotrophic lateral sclerosis, also lie along the spectrum of frontotemporal lobar degeneration (FTLD), the pathological descriptor for this constellation of disorders.

Confusingly, many people develop symptoms of more than one type of clinical variant—for example, some people develop both bvFTD and ALS, while others suffer from a blend of behavioral and language symptoms. Vanishingly few specialists exist who can diagnose the disorder, let alone direct participants to clinical studies or trials. For the few who have found their way to a diagnosis, participation in clinical studies can become prohibitively burdensome as symptoms worsen, especially for those who live far from major medical centers.

And herein lies the potential of digital health technologies, said AFTD’s Penny Dacks. By developing digital tools to monitor symptoms from a participant’s home, scientists aim to dramatically expand access to clinical studies. Candidate digital biomarkers and other digital measures are currently being tested in international cohorts of pathogenic mutation carriers, and researchers hope that these will help pave the way to earlier detection and diagnosis among people with sporadic forms of the disease. Though improved diagnosis was not the main focus of this meeting, Dacks believes that objective digital measures that detect the subtlest of early symptoms may prove useful on that front.

Trouble is, although a plethora of researchers in academia and industry are at work to develop these tools, none have moved beyond the exploratory stage. “Folks working on innovation in technology have not always connected with folks working in the regulatory space,” Dacks said. “We need to have more cross-stakeholder discussion so we can move these tools beyond the idea stage.”

Rhoda Au of Boston University set the tone of the summit by calling for fundamental change in clinical research. For one, Au noted, the infrastructure needed to conduct clinical studies in the traditional way—in which patients visit a clinic and undergo a series of in-person evaluations with specialists—doesn’t exist in much of the world. China, for example, has roughly 300 neurologists for a population of 1.3 billion people, Au said. Large swaths of the world are excluded from clinical research.

As opposed to the scarce neurologist—let alone the elusive FTD specialist—the smartphone is ubiquitous even in developing countries. Within the next two years, 7 billion people are projected to hold these computers in the palms of their hands, Au said, making the smartphone the most penetrating technology in the world. With a further push from the COVID-19 pandemic, the use of smartphones and tablets has grown substantially even among the oldest ranks of the population. Capable of tracking myriad health and lifestyle factors that fluctuate from day to day, such devices have the capability to pick out emerging patterns from the chaos of daily life.

Some researchers translate traditional in-clinic tests, such as the neuropsychological inventory, into a digital format. Au takes a broader view of what should be done. Equipped with multiple sensors that would passively and continuously monitor a person’s daily habits, from movements within and outside of the house to sleep patterns to time spent with friends, smartphones can discern subtle changes in behavior that would go undetected by quarterly or biannual rounds of traditional in-clinic tests.

Frequent monitoring is well-suited to capturing changes associated with FTD, Au believes. At the earliest stages of neurodegenerative disease, symptoms crop up inconsistently, and different family members might report different sets of behaviors and symptoms. “That’s what digital biomarkers are going to have to mimic,” Au said. “This dynamic, changing set of signals comes together and reflects that there’s a real change happening.”

Bonnie Wong, a neuropsychologist at Massachusetts General Hospital in Boston, agrees that highly structured, traditional neuropsychological testing tends to miss the early symptoms of FTD. Wong said such tests are not “ecologically valid,” meaning they fail to capture what goes on in day-to-day life. Digital technology may unveil the fuller picture, one that matches what patients and their families are actually experiencing.

For example, bvFTD symptoms such as apathy, compulsion, and agitation could be detected simply by monitoring activity patterns. While an apathetic person may become planted on the couch, a compulsive or agitated person may pace back and forth right in front of it.

Rigging a willing person’s home with sensors that record their movements and behaviors is one way to capture signs that something is amiss. At the summit, Zach Beattie of Oregon Health Sciences University, Portland, described the ORCATECH (Oregon Center for Aging and Technology) monitoring platform. Funded by the Collaborative Aging Research Technology (CART) initiative, ORCATECH involves the placement of multiple in-home sensors and wearable devices that together track everything from a person's movement patterns within and outside the home, walking speed, sleep, computer and medication use, social time, and more. This surveillance is coupled with weekly online questionnaires that keep tabs on events that may have influenced changes in the readings—such as visitors in the home, falls, or hospital visits.

So far, the platform has been focused on changes in the run-up to Alzheimer's disease. For example, a recent study found that computer usage dropped off dramatically over two years in people with MCI, while healthy controls kept up their use. However, the platform is agnostic to which sensors are used, and new ones can easily be added to meet different goals, Beattie said.

Say Yes to Surveillance? The ORCATECH platform integrates data from multiple sensors, including wearable devices as well as sensors placed strategically around a person’s home. [Courtesy of Zach Beattie, OHSU.]

Research centers in the United States, as well as in Europe, Australia, India, and Taiwan, are deploying ORCATECH in their cohorts. Beattie said efforts are underway to transfer its massive amounts of data onto the cloud, where it can be accessed by researchers around the world.

Beattie thinks this multi-sensor approach is well-suited to pick up symptoms of FTD. With Jeffrey Kaye at OHSU, Beattie is collaborating with scientists at the University of California, San Francisco, on a pilot study to try out ORCATECH in the ALLFTD cohort, a multicenter observational study of familial and sporadic FTD ongoing in the United States and Canada (Mar 2021 news).

As presented by UCSF’s Adam Staffaroni at the Holloway summit, the pilot study will deploy ORCATECH across the spectrum of FTD, in 10 people with bvFTD, five with PPA, and five with PSP/CBS, along with 20 study partner controls. Participants will be monitored continuously for a year. The researchers hope to pick up changes in motor function, cognition, sleep, speech, and language over that time, pegging digital markers that track with disease progression and overall function.

Besides passive monitoring, ALLFTD is also implementing an FTD-tailored mobile app developed by Datacubed Health, which hosts a traditional battery of neuropsychological tests in digital form. The app is currently used in more than 45 clinical trials spanning different disorders, Staffaroni said. It runs on Android and Apple phones, and researchers can customize it with questionnaires and tasks that pertain to the goals of a given study. For the purposes of ALLFTD, the app currently includes tests of executive function, spatial memory, speech and language, and motor function, with tests of naming and semantic judgment in the works. To entice participants to complete these tasks, each is gamified with avatars, “virtual gems,” and links to Amazon gift cards.

After a pilot study deemed the app easy to use, ALLFTD initiated a Phase 2 trial; it has enrolled 223 participants across 18 centers so far. At the summit, Staffaroni presented preliminary results on 207 participants, comprising people with different variants of FTD in a range of stages. The scientists found that scores on the digital tasks correlated with scores of disease severity as gauged by the in-clinic CDR+NACC FTLD. This version of the clinical dementia rating scale has been adapted to include questions relevant to FTLD (Mar 2021 conference news).

Staffaroni's finding held true not only for new digital measures of executive function and spatial memory, but also for motor function. The scientists validated a remote version of the finger-tapping test—in which participants tap their phones as many times as possible in a short time window—as well as remote measures of gait and balance. These tests agreed with their in-clinic versions, and they correlated with disease severity.

Jonathan Rohrer of University College London presented the current status of digital health technology development within the Genetic FTD Initiative (GENFI), a European-Canadian natural-history study of familial FTD. GENFI has done extensive work tracking behavioral, language, neuroimaging, and fluid biomarkers in its cohort, which includes both asymptomatic and symptomatic mutation carriers. Now, myriad tests are being deployed in digital form (see below).

Digital Palette. GENFI is validating digital tools that monitor cognition, language, activity, and motor function. [Courtesy of Jonathan Rohrer, UCL.]

In addition to passive monitoring, accelerometers, and other devices that track movements and physiology, GENFI has developed a tablet-based app that takes stock of cognitive function. Called Ignite, it includes gamified tasks that measure executive function, social cognition, and semantic memory. The entire battery takes 20-25 minutes to complete, with each task lasting no longer than 90 seconds, Rohrer said. So far, Ignite has been tested in 2,000 healthy controls, who showed slippage on some of the tasks with increasing age.

After verifying test-retest reliability of the app, Rohrer said Ignite will be rolled out to people with FTD, initially within the United Kingdom and then later in the wider GENFI cohort. Besides Ignite, tasks of social cognition and speech that have been developed by GENFI researchers in the clinic are being translated into mobile form, Rohrer said. This includes an eye-tracking test that follows a person’s gaze to assess how well they can recognize emotions in the faces of others.

One major domain of FTD symptoms that is noticeably absent from ALLFTD’s Mobile App or GENFI’s Ignite is behavior. Behavioral changes are difficult to capture in a test. They tend to come up in subjective reports from family members, and often go unnoticed by the person with FTD. At the summit, Michael Gold of AbbVie and other industry scientists urged the field to focus on developing more tools to assess behavioral change early on. Passive monitoring is one way to document these changes.

Speech is another domain that is amenable to digital monitoring and analysis. Naomi Nevler of the University of Pennsylvania, Philadelphia, reported that natural speech patterns may serve as clinical biomarkers that can tell apart types of FTD. Speech is the product of multiple neural networks working together, including networks that facilitate language, social cognition, executive, and motor function. Digitally recording acoustic, lexical, and rhythmic features of a person’s speech as they freely describe a picture—the classic example being the “cookie theft scene” (see below)—can bring to the fore deficits in these neural networks, Nevler said.

For example, people with semantic variant PPA, who have difficulty naming objects, relied on pronouns or ambiguous nouns to describe objects in the picture that they were unable to name. People with non-fluent, agrammatic PPA paused often, and tended to utter more partial words (Cho et al., 2021). Even people with behavioral variant FTD—who do not suffer from a language disorder, per se—had a speech signature, in that they spoke within a narrow acoustic range. This fits with the observation that people with bvFTD tend to speak in a monotone voice. People with this variant also spoke in shorter segments, with longer pauses than healthy controls, in keeping with their symptoms of apathy and distraction. Nevler was able to tie all these different speech signatures to atrophy in corresponding regions of the brain.

The findings suggest that speech harbors a wealth of information about dysfunction in different regions of the brain and can even differentiate among clinical variants of FTD. Nevler said these speech measurements are currently being deployed via tablets and smartphones in the clinic and remotely via a virtual conference app. The recording sessions last but a few minutes. What’s more, automated speech-analysis methods are more objective and quantifiable than a human listener, suggesting they might work as biomarkers in clinical trials. “I see different digital-speech features supporting all elements of patient care,” Nevler told Alzforum. “This includes screening, diagnosis, monitoring, prognosis (risk assessment), and support for patient quality of life at home.”

Commercial speech-assessment tools have started making their way into clinical trials as exploratory measures. Jessica Robin of Winterlight, a Toronto-based company that develops tablet-based speech-assessment tools, reported that people with FTD have distinct speech patterns that correlate with clinical decline. The tablet-based Winterlight speech composite, which is in use as an exploratory endpoint in AD clinical trials, scores some 500 vocal features, including acoustics, speed, types of words used, vocabulary, fluency, and sentence construction. Robin previously reported that over one year of regular assessments, the app picked up changes in speech in people with FTD, including those with the behavioral variant (Dec 2021 news). Winterlight’s speech-assessment tool is being used as an optional, exploratory measure in Alector’s INFRONT-3 trial, which is evaluating the progranulin-boosting monoclonal antibody AL001 in people with FTD who carry a pathogenic mutation in the gene.

With so many digital gadgets and measures to choose from, how can the FTD field come together to push the best ones forward? In Part 2 of this series, read about how scientists in related neurodegenerative disease fields have fared, and the major challenges that stand in the way of worldwide implementation of digital biomarkers.—Jessica Shugart

Comments

No Available Comments

Make a Comment

To make a comment you must login or register.

References

News Citations

1. Merged Consortia Forge Path to Trials in Frontotemporal Dementia 13 Mar 2021
2. Moving Target: Can Standardized Tests Track Symptoms of FTD? 19 Mar 2021
3. The Rain in Spain: Move Over Higgins, AI Spots Speech Patterns 5 Dec 2021
4. Digital Tools Abound, Yet Remote Biomarkers for FTD Remain Exploratory 29 Jun 2022

Paper Citations

1. Cho S, Nevler N, Ash S, Shellikeri S, Irwin DJ, Massimo L, Rascovsky K, Olm C, Grossman M, Liberman M. Automated analysis of lexical features in frontotemporal degeneration. Cortex. 2021 Apr;137:215-231. Epub 2021 Feb 6 PubMed.

External Citations

1. GENFI

Further Reading

No Available Further Reading