FeatureHow soon will digital endpoints become a cornerstone for future drug development?
Introduction
Pharmaceutical innovation needs to become more patient centric, and digital applications are set to revolutionize clinical research, delivering a wealth of individual patient data. This is at least the way recent industry and consultancy publications frame it [1]. In a similar recent statement, the FDA commissioner sees a role of wearables to better understand a patient’s well-being and needs [2]. Typically, endpoints accepted by the FDA include how a patient feels or functions and/or if they live longer. In selected cases, surrogate endpoints [e.g., validated biomarkers, such as blood pressure or low-density lipoprotein (LDL) cholesterol concentration] can replace relevant clinical endpoints, such as mortality [3]. The approvals granted based on surrogate medical evidence are often subject to post-approval Phase 4 studies to confirm and/or strengthen the evidence. Furthermore, health technology assessment (HTA) agencies want to evaluate evidence regarding clinical and cost effectiveness, safety, and overall patient benefit of the pharmaceutical intervention. Currently, the pharmaceutical industry is challenged by declining research and development (R&D) productivity, with the return on investment (ROI) in the industry at an all-time low of 3.7% in 2016 [4]. Scannell and co-workers named this phenomenon ‘Eroom’s law’, in reverse to the widely known Moore’s law on computational processor power, in that pharmaceutical R&D becomes slower and more expensive over time, despite advances in biomedical sciences [5]. One distinct problem that can be identified is the increasing number of so-called ‘late-stage failures’. These are mostly Phase 3 studies in which primary efficacy endpoints were not met or safety signals emerged. Hwang and co-workers, using public sources, showed that, between 1998 and 2015, out of 640 novel therapeutics, 57% failed because of inadequate efficacy and not meeting the studied endpoints [6]. A recent publication from MIT, using two large data sets, showed a success rate of only 40% for Phase 3 studies [7]. Of course, at the time point at which these late-stage failures occur, large investments have already been made. This translates into a diminishing pipeline and portfolio value and reduced ROI of the overall pharma R&D activities. A question that needs to be asked is whether preclinical scenarios are disconnected from real clinical outcomes If yes, no new endpoint could help to reduce the number of late-stage failures. If no, the reason for Eroom’s law could be a failure to translate meaningful changes in disease state in currently used endpoints.
Section snippets
Clinical endpoints hardly reflecting patient burden
In general, clinical endpoints, such as mortality, might only reflect one aspect of a disease burden because mortality can be rather low because of the benign nature of the disease and, therefore, not applicable as a clinical endpoint. Nevertheless, through its impacting signs and symptoms, the disease can significantly limit the patient’s quality of life. Furthermore, current clinical study designs, where only intermittent evaluations occur, often only allow for snapshot assessments of
Digital endpoints as new armamentarium for clinical development reflecting patient burden
A hypothesis that recently gained more traction across industry in different disease areas is the use of technology to gain new endpoints to quantify disease improvement by measuring physical parameters in a quality that is acceptable from a regulatory point of view. Recently, the Prescription Drug User Fee Act (PDUFA VI) was implemented. FDA is now required to consider the ‘patient’s experience’ to be an integral part of the benefit versus risk of new drugs to facilitate successful product
Pioneering work of the MOBILISED-D consortium
To address the current gap between consumer use of activity monitors and regulatory requirements, a group of pharmaceutical and technology companies recently sponsored an Innovative Medicines Initiative (IMI) program called MOBILISED-D in the EU [25]. MOBILISED-D is designed to establish a robust, device-agnostic, publically available algorithm to detect RWS and test its predictive power for endpoints of regulatory interest in three populations of frail or chronically ill patients. If
Digital endpoints are not limited to clincal development
Going forward, the use of digital endpoints is not limited to clinical studies. Whereas it was not previously possible to measure clinical outcomes continuously and assessments were instead done by retrospective analysis of large clinical studies, this setting is changing through the advent of the IoMT. Now, outcomes can be measured on an individual level continuously in real-time together with a new medical intervention before (i.e., in clinical studies as outlined above) or after its market
Concluding remarks
As the IoMT and, along with it, the possibilities to capture digital endpoints emerge, a new world of medical data interpretation, management, and collection will open to the pharmaceutical industry. Activity tracking and its application in clinical studies is only the beginning. In addition, the potential of these technologies is far from being fully exploited by the pharma industry and, therefore, initiatives such as MOBILISED-D have a major role in the transformation of the pharmaceutical
References (30)
iCap: instrumented assessment of physical capability
Maturitas
(2015)Efficacy of mobile health care application and wearable device in improvement of physical performance in colorectal cancer patients undergoing chemotherapy
Clin. Colorectal Cancer
(2018)The digitised clinical trial
Lancet
(2017)Digital R&D: Transforming the Future of Clinical Development
(2018)Statement from FDA Commissioner Scott Gottlieb, M.D., on FDA’s Efforts to Enhance the Patient Perspective and Experience in Drug Development and Review
(2018)Lipoprotein changes and reduction in the incidence of major coronary heart disease events in the Scandinavian Simvastatin Survival Study (4S)
Circulation
(1998)R&D returns continue to fall
Nat. Rev. Drug Discov.
(2016)Diagnosing the decline in pharmaceutical R&D efficiency
Nat. Rev. Drug Discov.
(2012)Failure of investigational drugs in late-stage clinical development and publication of trial results
JAMA Intern. Med.
(2016)Machine-Learning Models for Predicting Drug Approvals and Clinical-Phase Transitions
(2017)
Exploring new endpoints for patients with heart failure with preserved ejection fraction
Circ. Heart Fail.
Prognostic value of usual gait speed in well-functioning older people—results from the Health, Aging and Body Composition Study
J. Am. Geriatr. Soc.
Gait speed and survival in older adults
JAMA
Technology in Parkinson’s disease: challenges and opportunities
Mov. Disord.
Free-living monitoring of Parkinson’s disease: lessons from the field
Mov. Disord.
Cited by (26)
Artificial intelligence-driven drug development against autoimmune diseases
2023, Trends in Pharmacological SciencesDigital biomarkers for post-licensure safety monitoring
2022, Drug Discovery TodayCitation Excerpt :The combination of these data types has the potential to facilitate more robust safety monitoring systems. Although digital biomarker data are currently commonly collected in drug development programmes for enhanced efficacy assessment,10–12 there is for now little use of digital biomarkers for safety monitoring in routine healthcare delivery. The objective of this paper is to discuss how digital biomarkers might be used more broadly in routine healthcare delivery to enable the provision of more granular information to support safety evaluations.
A New Digital Value Chain Model with PLC in Biopharmaceutical Industry: The Implication for Open Innovation
2022, Journal of Open Innovation: Technology, Market, and ComplexityArtificial intelligence-enhanced drug design and development: Toward a computational precision medicine
2022, Drug Discovery TodayCitation Excerpt :During trial monitoring, AI helps to capture in a remote fashion patient-reported measurements and outcomes generated by wearable sensors or devices. It is also applied to mine such digital biomarkers providing useful information regarding symptoms, pain, cognitive function, motricity, or sleep patterns, to support diagnostic or therapeutic decisions made by the physician.65,66 AI and ML are also being used to analyze data from successful, but also failed studies to generate models capable of predicting simultaneously the evolution of multiple and multimodal clinical parameters.67
How to Design AI-Driven Clinical Trials in Nuclear Medicine
2021, Seminars in Nuclear Medicine
- ‡
These authors contributed equally.