Measured Momentum: Digital Biomarkers and Clinical Trials

Rhoda Au remembers well when her interest in using digital biomarkers for disease research was not shared among her peers. The Boston University anatomy and neurology professor recalled when she delivered a keynote speech at a conference, about seven years ago.

“I had [finished] my digital spiel,” says Au, who is also director of neuropsychology, Framingham Heart Study, at Boston University. “Someone stood up and said that everything I said was all nonsense.”

It is unlikely that anyone today would express such a sentiment, at least not as publicly and rudely as that. In the years since, some smart people have hypothesized, and shown, that digital biomarkers can fill in some blanks about disease progression, symptoms, and therapeutic effects of investigational molecules. The interest is strong in using these biomarkers with the latter.

“In general, digital biomarkers will teach early signs and symptoms we don’t know yet about diseases,” Antonella Santuccione Chadha, MD, chief medical officer, Altoida, tells Applied Clinical Trials. “In early diagnosis, in disease monitoring, in treatment response, and in active populations, they will allow us to learn” and characterize even new disease phases. Altoida has designed an augmented reality platform that measures functional and cognitive brain health.

From a business vantage point, says Chris Benko, CEO, Koneksa, the ability to know steps in disease progression, both physiological and behavioral, and to predict treatment response and dose selection can reduce the number of patients needed for a trial. “Unquestionably, when you get to a point where you can get a signal in half as many patients, the economic value is just staggering.” Koneksa develops digital biomarkers for monitoring, diagnostics, treatment responses, and clinical event prediction.

Traditional blood or cerebrospinal fluid (CSF)-extracted biomarkers show point in time but digital biomarkers can show a patient’s response to a cognition test multiple times a day, every day. It is sussing out the pathway to the response that is important, says Au, who added that digital data will forever be available, whereas blood must be used judiciously.

The data from a digital biomarker can be gathered with various devices, including smartphones and smart watches. There are also platforms for various diseases; Altoida has a platform for Alzheimer’s disease, and Koneksa is developing digital biomarkers for Alzheimer’s and Parkinson’s disease.

Many people in the digital biomarker world are primarily interested in identifying those signs and symptoms of disease before actual onset and the inevitable deterioration they cause. About 80% of all diseases in the US are chronic, notes Au. Such an illness has been “coming for a long time before anyone said there is a problem.”

Today, adds Jason Hassenstab, PhD, associate professor of neurology and of psychological and brain sciences, and director, Cognitive Technology Research Laboratory, Washington University in St. Louis, people understand that decline from neurological disease is not simple or linear. In early disease stages, the brain can reorganize itself a bit, allowing for the patient to cope for some time.

“We have a cognitive reserve, rerouting our skills and wiring in network, to compensate for the cellular level changes,” says Hassenstab, who also leads the Cognition Cores for the Dominantly-Inherited Alzheimer Network-Trials Unit (DIAN-TU) and the DIAN observational study. He and his team designed a smartphone app for use by these patients a few years ago. “We know so much more about cellular level changes,” notes Hassenstab. “It was just a matter of time to find behavioral measures to detect.”

What researchers have long hoped, he adds, is to find changes in the brain as early as possible. “I think digital biomarkers have their place as outcome measures,” he says. “It will be difficult to approve a drug for [Alzheimer’s] without testing behavior.”

The research

As Hassenstab and colleagues put it¹, “While diagnosis, disease monitoring, and intervention are traditionally addressed using distinctly separate tools, these different use cases may be overlapping or integrated in single solutions in the digital brain health space.” Sleep trackers, they wrote, can monitor quality of sleep; proffer advice for improving it; and possibly detect changes that are early neurological-disease signs.

The tools of the trade are being used on measures such as voice, temperature, activity, gait, blood oxygen, heart rate, and touch. These are but a handful of examples of what researchers look for in biomarkers.² But there are literally hundreds more, depending on the disease.

For instance, Alzheimer’s. Published data show that the Altoida platform, which incorporates augmented reality and artificial intelligence, captures 800-plus multimodal features of the patient, through many different sensors.³ These features are then tested against fluid-based biomarkers. Eisai is testing the Altoida platform in a five-year trial with 3,500 patients to diagnose neurological diseases in their earliest stages.⁴

Au’s research is looking at voice as one potential biomarker in particular. As a Framingham researcher, Au started recording participants’ responses to neurological testing beginning in 2005; hand-written notes didn’t provide the objective consistency she wanted, nor would they provide the informational pathways that the participants used to arrive at those answers.

By 2017, the Au team had 5,200 recorded sessions. A subset of 146 participants’ recordings were used from those that showed normal cognition, mild cognitive impairment, or dementia. A panel of neurology experts verified each case via diagnostic process.⁵ The researchers looked for pauses, hesitations, word-finding difficulties, circumlocution, pitch, timbre, and more. “When you are trying to think about what to say, you will” pause, stutter, or repeat words, says Au. These are all indicators of the person’s cognitive state, she adds.

According to Benko, at Koneksa, the company spent five years developing its Parkinson’s platform with partner Sanofi. In November 2021, the pharma giant extended the collaboration with Koneksa into central nervous system trials, which includes the evaluation of tolebrutinib a novel Bruton’s tyrosine kinase inhibitor for multiple sclerosis.⁶ He declined to discuss current findings, but demonstrated, via Zoom, how one of the measurements worked. He clicked on “ready.”

“‘With the phone in your right hand,” a voice said, “‘turn your palm up and down as many times as you can for 20 seconds.’” The platform, says Benko, “can measure tremor better than my eyeball.” He adds that Koneksa tested the algorithm in healthy people and then in those with Parkinson’s.

Koneksa is also looking at voice as a biomarker in Parkinson’s, and is recruiting for an asthma trial, comparing at-home mobile spirometry to in-clinic spirometry in participants with moderate asthma who take a long-acting beta agonist.⁷

Not all attempts to find biomarkers are rooted in a smartphone. In order to find early detection of Alzheimer’s and related dementias, the National Institute of Aging and Indiana University are using a machine learning algorithm, embedded in electronic health record (EHR) systems at two sites, to monitor 4,000 Medicare patients using either the patients’ annual wellness visit by itself; the wellness visit with the digital marker; or the digital marker with the Quick Dementia Rating scale.⁸

Not all studies are focused on myriad physiologic and/or cognitive issues. A biomarker of major interest is gait. Mobilise-D is a large 34-partner, 10-country, 2,400-person trial in the planning stages that will look at Parkinson’s; chronic obstructive pulmonary disease; multiple sclerosis; and fall-related issues, diseases, and frailty in the real world.⁹

Other researchers are investigating how digital biomarkers can help with other serious conditions, such as death by suicide and alcoholism. In the latter work, which is testing whether problem drinkers can disassociate with Pavlovian cues, participants will use a smartphone, equipped with a GPS, every day to report on their drinking, activities, and moods.¹⁰

Outside help

Evidation is a consumer platform, five-million app-downloads strong, that allows people to get a new view of their health situation. “People can tell us what conditions they have, and whether they are using a smartphone, Fitbit, or CGM (continuous glucose monitor),” says Ernesto Ramirez, Evidation’s chief data scientist. “We use that data to help contextualize their health in everyday life.”

In clinical trials. Evidation has been involved in numerous studies, including the Boston University voice trial study. “We’ve been supporting trials for a long time,” notes Ramirez. Evidation says it has users in 97% of US zip codes and has enrolled over 600,000 participants in sponsored research, and with remarkable speed: The company cited a cardiology study involving US Veterans that attracted 800 veterans in 38 days. 

Ramirez discussed work with a current partner, Johnson & Johnson/Janssen, in a heart study with 43,244 patients.¹¹ (Apple is also a partner.) This three-year observational study will enroll the patients, via the app, into a heart healthy program. The study will keep tabs on patients already prescribed an oral anticoagulant for atrial fibrillation (AF), see how adherent they are, record if and when they are diagnosed with AF, and then look at the percentage of days the medication was taken, minus any time it was not. The controls will have no history of AF, but will be enrolled in the same heart healthy program. There is an ECG sensor in the watch.

Challenges of digital

One person who sees what’s new in the field is Roozbeh Ghaffari, PhD, editor of Digital Biomarkers; biomedical engineer and neuroscientist, and a research associate professor, Biomedical Engineering at Northwestern University; and director of the university’s Translational Research, the Querrey Simpson Institute for Bioelectronics. In the four-plus years that he has been in this role, Ghaffari has seen a few more submissions than when he started.

There are papers whose authors used devices in the course of the study; there are others submitting work on proposed novel biomarkers and biosensors for use remotely. Still others are “more holistic types of studies,” looking to determine the quantity of data needed to achieve an outcome. It’s a highly translational field, says Ghaffari, that needs cooperation from academics, pharma, and commercial interests. Adoption happens, he adds, when many studies validate the biomarkers, which clinicians can then embrace. At which point, it’s “no longer just cool technology,” notes Ghaffari.

Ramirez says he understood why someone would ask about bias, considering the technical sophistication a patient might need to use the Evidation app. A few years ago, he explains, people who used Fitbits and smartwatches were generally confined to two groups: the weekend warriors or the “worried well.” “I don’t think that is true anymore,” says Ramirez.

Use is becoming more widespread, so it crosses into different age groups, social backgrounds, and ethnicities. Data show that the elderly do have difficulty with the technology, but are more than willing to learn.¹²

One person acutely observant of bias is Santuccione Chadha, who is also pro bono CEO of the Women’s Brain Project. Brain diseases are more common in women, she notes, and women react differently to treatment; supporting evidence has shown a trend toward better effect in men as seen in the three anti-amyloid drugs and two immunotherapies for Alzheimer’s.

Last June, Santuccione Chadha and a large team of researchers wrote about these differences, based on findings with the Altoida platform. The cohort of 568 people was composed of those with mild cognitive impairment, brought on by Alzheimer’s, and healthy individuals. The platform predicted, with 75% accuracy, the sex of the healthy people and those with mild cognitive impairment.

“These results stress the need to integrate traditional approaches to dementia research with digital biomarker technologies and personalized medicine perspectives to achieve more precise predictive diagnostics, targeted prevention, and customized treatment of cognitive decline,” they wrote.²

Santuccione Chadha mused about the immunotherapy drugs approved for Alzheimer’s; the data, she says, all showed a trend that men responded better to the medication than women.

“If we had digital biomarkers embedded in those trials, there would have been a different level of learning than what we know today,” she believes, mentioning speech recognition and microtremors that could have been measured via a digital health technology. The existing scales, she adds, were developed decades ago.

Other challenges in using digital biomarkers? They are of the human kind.

Mixing technology with patients, especially patients who are aging, have trouble processing information, and who are remote from investigators, present whole new challenges to researchers who have to learn to anticipate human glitches in all kinds of ways.

“You have to go through the pain of learning,” says Ghaffari, recalling how elderly patients will tend to throw away their device after day one. “That happens all the time. But the solution, the data, the promise of it is certainly there. The more we do, the more we will learn.”

Adams and colleagues, in work published in April on comparing data gathered from various devices on Parkinson’s, found that patients wore the watch on different wrists and data were lost because permissions restrictions to use the data had been turned off, so data from the watch couldn’t be sent to the database.¹³

Benko notes that researchers have to think through all these questions, like Phase II trials involving therapy-naive patients, but Phase III trials not. Or that similar conditions in a disease spectrum will involve other similarities.

“For a drug developer, being able to distinguish why something is happening and define what is going on in a subgroup is very important,” he says. “You have to make sure that a tool doesn’t discriminate on something you didn’t think to question.”

Christine Bahls is a freelance writer for medical, clinical trials, and pharma information.

References

1. Raket, L.L.; Petcu, P.; Wac, K.; Hassenstab, J. Editorial: Digital Brain Health. Front Digit Health. 2023, 5 (2). https://www.frontiersin.org/articles/10.3389/fdgth.2023.1142897/full
2. Harms, R.L.; Ferrari, A.; Meier, I.B.; et al. Digital Biomarkers and Sex Impacts in Alzheimer’s Disease Management—Potential Utility for Innovative 3P Medicine Approach. EPMA Journal. 2022, 13 (2), 299–313. https://pubmed.ncbi.nlm.nih.gov/35719134/
3. Meier, I.B.; Buegler, M.; Harms, R.; et al. Using a Digital Neuro Signature to Measure Longitudinal individual-level change in Alzheimer’s disease: The Altoida Large Cohort Study. NPJ Digit Med. 2021, 4, 101. https://www.nature.com/articles/s41746-021-00470-z
4. Diagnosis and Monitoring of Disease Progression Using Deep Neuro Signatures (DNS). ClinicalTrials.gov. https://clinicaltrials.gov/ct2/show/NCT05153941?term=eisai+altoida&draw=2&rank=1
5. Tavabi, N.; Stϋck, D.; Signorini, A.:, et al. Cognitive Digital Biomarkers from Automated Transcription of Spoken Language. J Prev Alz Dis. 2022, 9 (4), 791-800. https://link.springer.com/article/10.14283/jpad.2022.66
6. Koneksa Broadens Multi-year CNS Research Collaboration with Sanofi. PR Newswire. November 3, 2021, https://www.prnewswire.com/news-releases/koneksa-broadens-multi-year-cns-research-collaboration-with-sanofi-301415238.html
7. Study Comparing At-home Mobile Spirometry to In-clinic in Moderate Asthma Participants Taking Long-acting Beta Agonist (LEARN). ClinicalTrials.gov. https://clinicaltrials.gov/ct2/show/NCT05757908?term=koneksa&cond=Asthma&draw=2&rank=1
8. Digital Detection of Dementia (D Cubed) Studies (Dcubed). ClinicalTrials.gov. https://clinicaltrials.gov/ct2/show/NCT05231954?term=NIA+Indiana+ehr&cond=alzheimer%27s&draw=2&rank=1
9. Mikolaizak, A.S.; Rochester, L.; Maetzler, W.; et al. Connecting Real-world Digital Mobility Assessment to Clinical Outcomes for Regulatory and Clinical Endorsement–the Mobilise-D Study Protocol. PLoS One. 2022, 17 (10), e0269615. https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0269615
10. Effects of a New Behavioral Intervention on Alcohol Craving and Drinking. ClinicalTrials.gov. https://clinicaltrials.gov/ct2/show/NCT02831049?term=smartphones+biomarker&recrs=adf&draw=3&rank=19
11. A Heart Health Study Using Digital Technology to Investigate if Early AF Diagnosis Reduces the Risk of Thromboembolic Events Like Stroke in the Real-world Environment. ClinicalTrials.gov. https://clinicaltrials.gov/ct2/show/NCT04276441
12. Nicosia, J.; Aschenbrenner, A.J.; Adams, S.L.; et al. Bridging the Technological Divide: Stigmas and Challenges With Technology in Digital Brain Health Studies of Older Adults. Front Digit Health. 2022, 4. https://www.frontiersin.org/articles/10.3389/fdgth.2022.880055/full#h12
13. Adams, J.L.; Kangarloo, T.; Tracey, B.; et al. Using a Smartwatch and Smartphone to Assess Early Parkinson’s Disease in the WATCH-PD Study. NPJ Parkinson’s Dis. 2023, 9 (1), 64. https://pubmed.ncbi.nlm.nih.gov/37069193/