New technologies are disrupting all industries, including health care and drug development, said Amir Kalali, head of the Neuroscience Center of Excellence at Quintiles. Although they have yet to make a major impact in clinical trials, technological innovations offer the potential to improve efficiency and productivity through the use of novel outcomes, increased patient engagement, reduced patient burden, and improved trial management. Indeed, Kalali said that he believes technology can enable
the field’s ethical duty to conduct efficient next-generation clinical trials. Yet the expanded use of new technologies also raises regulatory and operational concerns as well as barriers to implementation. Moreover, Drew Schiller cautioned against replacing the entire human element of clinical trials with technology. Although there are some aspects of trial design where efficiencies can be gained by more rapidly enrolling and prequalifying large numbers of participants and gathering data that show whether a treatment works or not, humans are better at developing research objectives and analyzing data.
Exploiting new technologies in the design of clinical trials will require the drug development community to apply the lessons learned from other industries, including an increasing focus on the consumer, said Schiller. Most study participants today come into clinical trials with a high level of experience and comfort using technology in their personal life, said Kalali, and consumer technology companies have been built largely around the concept of improving the user experience. The R&D and business models upon which these companies have been built may be adaptable to pharmaceutical development, said Glen de Vries, president and co-founder of Medidata Solutions.
Kalali also mentioned other technologies that are likely to disrupt the drug development enterprise such as synthetic biology—so-called exponential technologies because they are developing at an exponential rate. Virtual reality and the Internet of Things are another two emerging paradigms that are transforming consumer-based technology development, said John Reites, head of Digital Health Acceleration at Quintiles. Given the rapid advancement of technology, William Potter, senior advisor to the director at NIMH, cautioned that companies may be reluctant to commit to a single technology for improved signal detection because future technologies may prove to be better.
Reites focused his remarks on direct-to-patient research, in which patient communities are built based on a clinical trial or study, but only later connected with the investigator. For example, if one is interested in evaluating prevention treatments for Alzheimer’s disease (AD), one might establish a community of interested individuals who are invited to complete cognitive scales and contribute data from mobile health devices.
Then, when a trial opportunity arises, they could be invited to contribute data to a trial, participate in the trial, or test a new drug or device.
This focus on patients in trials of the future was emphasized by many workshop participants. Bringing patients into the process—for example, understanding what matters to them and what the trial should measure—would help ensure that research is addressing clinically relevant questions, said Carlos Rodarte, chief executive officer of Health Rhythms. In addition, listening to what patients say about how they feel may provide clues about unexpected aspects of the treatment response, said Perry Nisen. He said the breakthrough in developing immunotherapy and checkpoint inhibitors in oncology came about when a patient reported feeling much better and was thus kept in the trial even though imaging studies suggested the therapy was not working. Only later were investigators able to develop new imaging techniques that visualized the immunological benefits.
Karl Kieburtz, Robert J. Joynt Professor in Neurology and director of the Clinical and Translational Science Institute at the University of Rochester Medical Center, said that regardless of the changes brought about by the introduction of new technologies, three aspects of traditional drug development will endure: (1) enabling the participation of patients and families in meaningful research opportunities, (2) generating robust data, and (3) drawing valid inferences. He predicted that the model of how clinical trials are conducted will change radically in the future, shifting from distributed loci to a central focus or single center, with investigators (academic, industry, or foundation based) starting off trials by directly reaching out to patients and families and with assessments carried out remotely or through telemedicine.
Kieburtz cited one Parkinson’s disease (PD) study based solely on telemedicine versus live, in-person visits (Dorsey et al., 2015). Individuals from 39 states completed all visits in the trial and reported high levels of satisfaction. Moreover, people from 49 countries expressed interest in participating, highlighting the pent-up and entirely untapped demand for patients and families to participate in research. However, trials that capture participants from many sites must deal with the additional challenge of understanding how regional and cultural variability may impact outcomes. For example, if a trial is testing an intervention for mood disorders, patients and controls in sunny California may respond differently than those in the wintry Northeast, due to both differences in light exposure and physical activity, commented Rodarte.
The PD study also highlighted the important enabling role of social media. Volunteers for this study were solicited through Fox Trial Finder, an online tool that matches people with PD to trials for which they may be eligible. One concern raised regarding social media is including people in a sample who have not been clinically diagnosed with a condition. Another concern is that individuals might falsely report data, or lend their wearable device to a friend. Rodarte, however, said that with a large enough sample size, it might be possible to identify the outliers; and Reites added that connecting directly with electronic medical records can provide added confirmation that participants are who they say they are.
Simple online assessments, data capture by wearable devices, remote monitoring, and virtual clinical visits offer additional advantages in terms of reducing patient burden, and this can have a substantial impact on recruitment and retention, said Schiller. For example, participants with child care and transportation needs may be able to remain in a study through the use of remote data capture technologies, he said. Indeed, added Rodarte, improving the patient experience in research studies is essential. The hospitality industry may provide lessons in this regard, he said.
Continuous measurement of activity and behavior is one approach that enables collection of precise and frequent information at a relatively low cost, as well as new types of information that could not be measured in the past, said Atul Pande. Smaller and more sophisticated sensors are driving the increasing use of these technologies beyond the consumer market, added Rodarte. Thus, they could be used to develop digital signatures that characterize how different populations behave, such as people with schizophrenia or bipolar disease, he added. Continuous data capture of an individual enrolled in a trial can provide insight into that person’s mental well-being and the stability of daily routines. Such measures could allow the reframing of behaviors beyond those included in the DSM-5 and could also enable the implementation of just-in-time interventions, said Rodarte.
Major challenges with regard to these devices include how to make sense of the enormous amounts of data that can be acquired, and how to leverage that data, including data about how an individual uses technology, to learn more about the disease itself and its progression, said Rodarte.
Differentiating the signal from the noise presents yet another challenge, although de Vries suggested that digital measures allow investigators to embrace the noise by identifying interesting signals embedded in noisy measures. However, while continuous monitoring such as this may be more objective than patient self-report measures, Thomas Laughren, director of Laughren Psychopharm Consulting, LLC, pointed out that conscious experience is an important component of psychiatric disease; this requires active measures such as self-report, in addition to more objective passive measures.
Other challenges for those hoping to use data from wearables and in-home monitoring devices are the standardization and normalization of data from many different types of devices and applications, noted Schiller. Standardization of measures is also very important, said S. Kapur, noting ADNI’s success in this regard. According to Schiller, an association of large consumer technology companies has established a standards committee that is tasked with creating standards for consumer devices that assess activity, sleep, electroencephalogram (EEG), and other measures. Rodarte added that the underlying technology for these devices is often very similar, which should make standardization somewhat easier. There is also a trend toward “make-your-own” devices, he added. For example, Biogen has publicly stated it is developing a device with enhanced sensitivity to the multiple sclerosis patient experience.
However, Greg Koski, chief executive officer, president, and co-founder of the Alliance for Clinical Research Excellence and Safety (ACRES), noted that although implementation of Clinical Data Interchange Standards Consortium (CDISC) standards reduces the time from start-up to finish of a study by 60 percent, adoption of CDISC standards has been slow and has not penetrated the entire research ecosystem of stakeholders.
Integrating these data, along with physiologic, genotypic, and phenotypic data from other sources, presents an additional challenge, said de Vries. However, he posited that integrated disease models will enable creation of interesting multivariate models in neuroscience and other disease areas, and thus will become a preferred type of model.
Wearables and other types of sensors may also be useful as tools to assess compliance with a study protocol, said Stephen Brannan, vice president of clinical research and medical affairs at Forum Pharmaceuticals. de Vries added that models can be developed through machine learning to identify fraudulent sites and participants. In addition, Kieburtz suggested that objective data such as how many steps a person takes a day,
or how many times he or she interacts with others, could provide good objective indicators of depression and other psychiatric conditions. Whatever novel treatment targets, endpoints, and trial designs are used, what regulators are most interested in is clinical meaningfulness, according to Tiffany Farchione, deputy director of the Division of Psychiatry Products in the Center for Drug Evaluation and Research (CDER) at the FDA. To what extent does the treatment affect how the patient feels, functions or survives; how do you measure this?