Biomarkers Consortium Digital Monitoring Technologies
Goals of the Workshop
• Bring together diverse stakeholders in the field to reach consensus on the use of a single vocabulary that will be understood consistently in the regulatory context
• Identify areas of high medical need that could be addressed using digital system technologies
• Ensure stakeholder alignment and application of an evidence-based framework for the use of digital health technologies for therapeutic research and development
The digital tower of Babel
• Language confusion hinders medical practice and drug development o Misinterpretation of evidenceo Misunderstanding of
evidentiary requirements and regulations
o Failure of clinical trials o Delayso Potential harm to patients
• BEST resource used as remedy for biomarkerso Publicly available at
http://www.ncbi.nlm.nih.gov/books/NBK326791/
• Pubmed citationso “digital biomarker” – 8o “wearable” – 12,096o “digital health technology” – 61o “remote sensing technology” –
2,933o “digital measure” – 12 o “mobile application” – 1,419
Disclaimers for workshop lexicon
The use of the term “digital measure” (i.e. digital monitoring technologies, mobile applications) is defined as:• Objective, quantifiable, physiological, functional, and behavioral data collected and measured through
the use of wearables, ingestibles, implantables, and mobile technologies for the remote capture of datao Consistent with definition of mobile technology proposed by CTTI
o Important to level set the group to ‘how this will be used in the Workshop’ – recognizing others may have different definitions or use slightly different lexicon.
The use of the term “device” is to be consistent with the FDA’s definition for a “medical device”
When the tool is not a cleared device (e.g., smartphone, fitbit), we will use the term “technology”
Analysis grid for prioritizing case studies
5
Study or Specific Measure Name Measurement Thearpeutic area Measure status STD
Biomarker Type BEST
Algorithm availability STD
Reg path standard
Sensor Deploy Standard
Actigraphy as an assessment of performance status in patients with advanced lung cancer Performance status (ECOG) assessment via actigraphy in NSCLC Cancer existing measure prognostic black box Commercial wearableHome-Based Wearable Continuous ECG Monitoring Patch on Detection of Undiagnosed Atrial Fibrillation ECG patch for asymptomatic Atrial fibriallation Cardio existing measure diagnostic black box FDA Cleared wearable
The HUAWEI Heart study - JACC 2019 & Apple Heart Study Design Wrist wearable for A atrial fibrillation detection Cardio existing measure monitoring black box FDA Cleared wearableEvaluation of Wearable Digital Devices in a Phase I Clinical Trial; Clin Transl Sci (2018) Wearable pulse rate for Tachycardia Cardio existing measure monitoring black box FDA Cleared wearableContinuous Monitoring Using a Wearable Device Detects Activity-Induced Heart Rate Changes After Administration of Amphetamine
monitoring heart rate (HR) and respiratory rate (RR), via single-lead electrocardiogram (ECG) recordings; and mobility and sleep Cardio existing measure response black box FDA Cleared wearable
Home monitoring of HF patients at risk for hospital readmission using a novel under-the-mattress piezoelectric sensor
Heart rate and respiration monitoring during sleep for heart failure and asthma decompensaion Cardio novel measure monitoring black box Commercial In bed
Myocardial Infarction, COmbined-device, Recovery Enhancement Study (MiCORE)
Promoting self management, adherence to guided directed therapy Cardio novel measure prognostic black box Commercial wearable
VERKKO 3G-capable wireless glucose meter Diabetes existing measure monitoring black box unknown wearable
All of Us Research Program CRF Validation Study Smartphone-based measurement of heart rate and VO2max Cardio novel measure diagnostic open sourceResearch-grade tool
Smartphone sensor-based
mPower Parkinsons Study Smartphone-based measurement of tremor Neurodeg novel measure response open sourceResearch-grade tool
Smartphone sensor-based
Several validation studies onDuchenne Muscular Dystrophy patients Stride velocity 95th Percentile DMD novel measure response open source Commercial wearable
Digital biomarkers of mood disorders and symptom change actigraphy Neuropsych existing measure diagnostic published Commercial wearableCircadian rest-activity patterns in bipolar disorder and borderline personality disorder circadian rhythm, rest-activity cycles Neuropsych existing measure monitoring published Commercial wearableRelapse Prediction in Schizophrenia through Digital Phenotyping: A Pilot Study behavior sensing Neuropsych novel measure prognostic published Commercial
Smartphone sensor-based
REMOTE A series of modules (some worked; some didn't) Renal existing measure response published Commercial Smartphone sensor-based
Indoor Air Quality Sensor on Perceptions of and Behaviors Toward Air Pollution Indoor air quality monitoring Health existing measure diagnostic published Commercial In homeAchieving Self-directed Integrated Cancer Aftercare (ASICA) in Melanoma (ASICA) Promoting self exmination of skin amongst melanoma patients Cancer novel measure prognostic unknown unknown
Smartphone sensor-based
Digital Health Feedback System (DHFS) for Longitudinal Monitoring of ARVs Used in HIV Pre-exposure Prophylaxis (PrEP)
Longitudinal Monitoring of ARVs Used in HIV Pre-exposure Prophylaxis (PrEP) Infection novel measure monitoring unknown FDA Cleared
ingestible and a skin patch
Evaluation of a 'Hand-held' Fluorescence Digital Imaging Device for Real-Time Advanced Wound Care Monitoring (JDRTC/UHN) (JDRTC/UHN) Biological and molecular information of a wound Injury repair novel measure diagnostic unknown Commercial Hand HeldTelehealth Management of Parkinson’s Disease Using Wearable Sensors: An Exploratory Study;
motion sensor-based monitoring system(Kinesia™, Great Lakes NeuroTechnologies, Cleveland, OH) Neurodeg existing measure monitoring unknown FDA Cleared wearable
AT-HOME PD Change in Smartphone tapping (score) Neurodeg existing measure monitoring unknown Commercial Smartphone sensor-based
Smartphone-Based Testing to Generate Exploratory Outcome Measures in a Phase 1 Parkinson's Disease Clinical Trial
Smartphone based activity, transitions and movement measurements Neurodeg novel measure monitoring unknown Commercial
Smartphone sensor-based
Filter grid used by planning team
Strategy for selection of case studies
6
Existing Measure
Wearable
Commercial
black box Cancer
published Neuropsych
FDA Clearedblack box Cardio
unknown NeuroD
Unknown black box Diabetes
Smartphone sensor-based Commercial
Published Renal
Unknown NeuroD
Novel Measure
Wearable Commercial
Black box Cardio
Open source DMD
Smartphone sensor-based
Commercial
Published Neuropsych
Unknown NeuroD
Research-Grade Tool Open source
Cardio
NeuroD
Unknown unknown Cancer
MEASURESTATUS
SENSORDEPLOYMENT
REGULATORYPATH
ALGORITHMAVAILABILITY
THERAPEUTICAREA
MEASURESTATUS
SENSORDEPLOYMENT
REGULATORYPATH
ALGORITHMAVAILABILITY
THERAPEUTICAREA
1
2
4
1
1
1
1
1
1
1
1
1
1
1
7
Priority criteria for case studies
• At least 2 neuro cases• 1 commercial and 1 cleared• 1 novel and 1 existing• 1 black box and 1 open source• Multimodal vs single measurement type
Workshop case studies
• Mobile Parkinson Observatory for Worldwide, Evidence-based Research (mPower) and Parkinson’s Disease
o Diane Stephenson (Critical Path Institute), Dan Karlin (HealthMode), Abhi Pratap (SageBionetworks), Ninad Amondikar (MJFF), Xuemei Cai (Pfizer)
• Cardiac Monitoring in Phase 1 Clinical Trials
o John Wagner (Foresite Capital), Elena Ismailova (Koneksa Health), Vadim Zupunnikov (Johns Hopkins Univ.)
• Continuous Glucose Monitoring & Remote Digital Monitoring (VERKKO Study)
o Roberto Calle (Pfizer), Nadir Ammour (Sanofi)
• Stride Velocity 95th Centile 2o Endpoint in Duchenne’s Muscular Dystrophy
o Pat Furlong (PPMD), Laurent Servais (Univ. of Oxford), Francesca Cerreta (EMA)
• Remote Assessment of Disease and Relapse in Major Depressive Disorder (RADAR – MDD)
o Matthew Hotopf (King’s College), Vaibhav Narayan, Linda Brady (NIH/NIMH)
Evidentiary criteria framework
General• Characterization of Relationship Between the
Biomarker and Clinical Outcome• Biological Rationale for Use of Biomarker (If
Known)• Type of Data and Study Design (i.e.
Prospective, Retrospective, etc.)• Independent Data Sets for Qualification• Comparison to current standard• Assay performance• Statistical Methods to Use
Leptak, Menetski, Wagner, et al. Sci Transl Med. 9(417), 2017
General evidentiary criteria document development
Evidentiary CriteriaFramework [Drafts]
Workshop Framework & Case Studies
Case Study 1: Markersof Drug-Induced Kidney Injury
Case Study 2: Hepatotoxicity (GLDH)
Case Study 3: Markers of Drug-Induced Vascular Injury
M-CERSI Analytical Validation Workgroup Input
PhRMA Focus Group Review and Input
FDA Focus Group & Medical Policy Council Input
M-CERSI StatisticalWorkgroup Input
Final Framework
Workshop FeedbackPhRMA Focus Group
Review and Input
FDA Focus Group & Medical Policy Council Input FNIH Biomarkers
Consortium Website,STM publication;
assisted FDA Guidance
What does the framework provide?
• A clear set of steps needed for working toward Biomarker Qualification• Identify key areas for defining biomarker need• Specify and limit biomarker development focus to allow successful generation of
appropriate evidence• Provide consistent set of characteristics to describe and define the biomarker development
program with the regulatory agency
Primary Assumption:• A clearly defined goal to the project will provide a better view of a path to ultimate drug
development decision making and regulatory approval.• The framework provides a context for the discussion between sponsor and the agency
The Specific Context of Use for a Biomarker Drives the Extent of Evidence Needed for Qualification
Analytical Validation(establish performance and acceptance characteristics of the biomarker assay)
Clinical Validation
(establish that the biomarker acceptably identifies, measures, or predicts the
concept of interest)
Reference Ranges/
Decision Points
Pre-Analytical and Assay
Performance Characteristics
Analytical Rigor/ Reproducibility
Study Design Acceptability
Clinical Meaningfulness/Decision Points
Benefit/Risk Assessment
Analytical assay and clinical validation considerations in biomarker qualification
12
Sample Handling/ Stability
Adapted from:Peter Stein, M.D.Deputy DirectorOND, CDER, FDA
Flow of data for digital monitoring technologies
13
Raw data
Raw processing Processed
data
Final device
processingTransmit
Data
Data security
Data Storage Data
Analysis
Data Interpretation
Tool use
In Device
To Analysis
From Device
Each step needs evidence!
Receive Data
Data security
Raw sensor data
e.g. Heart rate or Blood O2
Packaging; or alerting
Convert to readout
Biology(Biomarker or COA)
Data transfer can occur at any step
FDA Qualification
Data Aggregation
Evidentiary Criteria
Evaluation
Characteristics of Types of EvidenceAs identified in Evidentiary Criteria Workshop – July 2018
• Universalityo to what extent is there evidence across drug mechanisms or across different populations
• Plausibilityo is the biology of the measure so compelling that it adds to the weight of evidence for
acceptance
• Causalityo is there a compelling case for it being causal so there is less of a need for evidence of
universality
• Proportionalityo to what extent does the measure explain the disease or the change in disease
• Specificity and potential for off target effects
Each biomarker class needs a different amounts and types of evidence
Caus
ality
Plau
sibili
tySp
ecifi
city
Prop
ortio
nalit
yU
nive
rsal
ity
Susceptibility/RiskAm
ount
of e
vide
nce
Surrogate Endpoint
Caus
ality
Plau
sibili
tySp
ecifi
city
Prop
ortio
nalit
yU
nive
rsal
ity
Amou
nt o
f evi
denc
ePrognostic
Caus
ality
Plau
sibili
tySp
ecifi
city
Prop
ortio
nalit
yU
nive
rsal
ity
Amou
nt o
f evi
denc
e
Caus
ality
Plau
sibili
tySp
ecifi
city
Prop
ortio
nalit
yU
nive
rsal
ity
Diagnostic
Amou
nt o
f evi
denc
e
The type and amount of evidence will change depending on risk and benefit
Enough evidence needed to convince a company who to include in a trial
Enough evidence needed to convince the FDA that the biomarker faithfully
predicts the biological outcome
Conclusion
• Alignment from multiple, diverse stakeholders • Consistent, comprehensive, semi-quantitative parameters for biomarker qualification• Greater degree of clarity, predictability, and harmonization• Broadly applicable across multiple categories of biomarkers and COUs• Since each category of biomarker and COU has unique factors to consider as part of the development process,
multiple modules are proposed to address these more specific issues
• …and keep in mind that the evidentiary criteria framework, the BEST resource, analytical validation, clinical validation, and qualification have all be discussed and designed for biomarkers, not digital monitoring technologies
Template slides for the case study presentations
17
The measurements are for the biology
18
Evidentiary criteria framework
General• Characterization of Relationship Between the
Biomarker and Clinical Outcome• Biological Rationale for Use of Biomarker (If
Known)• Type of Data and Study Design (i.e.
Prospective, Retrospective, etc.)• Independent Data Sets for Qualification• Comparison to current standard• Assay performance• Statistical Methods to Use
Statement of Need
• How is this needed in drug development?– Increase efficiency, safety, speed of drug development?
• Why take the path of digital measure vs. current modalities?• Ex. wearables, smartphone apps, etc.
Take several slides to describe the background that fits into the statement of need and
the known biology
Context of Use
• What decision is going to be made for drug development?
• What is the population involved?
• What factors will define the limits of the decision?
Benefit AssessmentTaken from the Evidentiary Framework Document (Oct 2016). Benefit with respect to the Context of Use (CoU)
• What is the relative perceived benefit of the new measure vs. the current standard (if there is one)?o If novel, there is no standard
• When in the drug development lifecycle is the measure intended to be used?
• How will the measure impact drug development and regulatory review?
• Is the benefit of the measure to the individual or society? o Separate discussions for both
o Does not include business issues
Risk AssessmentTaken from the Evidentiary Framework Document (Oct 2016). Risk with respect to the Context of Use (CoU)
• What is the potential consequence or harm if the measure’s performance is not aligned with expectations based on the COU?
• What is the severity of the disease or condition? What are the unmet needs of the population defined in the COU? What are the risks for mortality and morbidity in the absence of treatment?
• What is the relative overall perceived incremental risk vs. benefit of the new measure vs. the current standard?
• Is the risk of the measure to the individual or society? o Separate discussions for both
o Does not include business issues
• Risk mitigation strategy
State of EvidenceCurrent evidence available:
• Focus on the evidence relevant to the decision being made.
• Single or multi-component measure?
• Existing analytical and clinical validation?
Additional evidence needed to meet the minimum evidentiary standards given the benefit/risk assessment:
• Additional analytical and clinical validation needed
• Highlight gaps in the data that would need to be filled in order to make a confident decision
• Keep in mind that the decision maker is either the FDA reviewer or the industry/trial sponsor
Panel Discussions
• Did the framework help focus the case study discussion?• What was missing from the framework?• What needs to be modified in the framework to address unique questions around
remote monitoring?• Are there suggested modifications to the BEST vocabulary based on the case study?
Extras
26
Constructing a biomarker road map
Leptak, Menetski, Wagner, et al. Sci Transl Med. 9(417), 2017
Need statement and context of use (COU)
• Need statemento The nature and extent of the need, drug
development issue it addresses and target population
o The major challenge(s) and unique aspects of these challenges the project is to address
o The reasons and causes for the deficit being addressed
• COU statement – concise description of how a biomarker is intended to be used in drug development
• COU simplified to only 2 elements:o What class of biomarker is proposed and what
information content would it provide? o What question is the biomarker intended to
address? (“What is the biomarker’s specific fit-for-purpose use?”)
Biomarker Classes Susceptibility/Risk: Indicate potential for developing disease in an individual without clinically apparent disease
Diagnostic: Identify patients with a particular disease or a subset of the disease
Monitoring: Detect a change, over time, in the degree or extent of disease
Prognostic: Indicate likelihood of a clinical event, disease recurrence or progression, in the absence of a therapeutic intervention
Predictive: Identify patients likely to experience a favorable or unfavorable effect from a specific treatment
Pharmacodynamic: Indicate that a biological response has occurred in a patient who has received a therapeutic intervention. May become clinical trial endpoints and for a very small subset, surrogate endpoints.
Safety: Indicate toxicity to a therapeutic intervention
BEST Resource (Biomarkers, EndpointS, and other Tools) http://www.ncbi.nlm.nih.gov/books/NBK326791/
Examples of COU
A prognostic marker for disease progression to be used as an inclusion criteria in a Phase 2 clinical trial of a novel drug to enrich for the likelihood of organ transplantation.
BEST: identify likelihood of a clinical event
Clinical Trial Decision
A safety marker for organ toxicity to be used in a Phase 1 clinical trial of a novel drug in addition to a standard measure of organ toxicity to explore and refine the clinical trials stopping criteria.
BEST: response to an intervention or exposure.
Clinical Decision
Benefit and risk
• The benefit and risk profile, given that the COU is related to the biomarker’s value to drug development or clinical trials, is assessed from the perspective of patients
• Benefit assessment o What are the unmet needs of the population defined in the COU? o What is the mortality and morbidity of the disease’s natural history in the
absence of treatment?o What is the severity of the disease or condition? o What is the perceived benefit of the new biomarker vs. the current standard?
• Risk assessment o What is the potential consequence or harm if the biomarker’s performance is not
aligned with expectations based on the COU? o What is the perceived incremental risk, new biomarker vs. current standard?o When in the drug development lifecycle is the biomarker intended use?o What is the scope of the biomarker COU in terms of impacting drug development
and regulatory review?
Examples of benefit and risk analyses
• Favorable benefit and risk profile – lower level of evidenceo Stratification of patients to ensure equal distribution of biomarker positive and biomarker negative individuals in
the different arms of a clinical trialo If biomarker does not perform – loss of resources but not patient safety
• Less favorable benefit and risk profile – moderate level of evidenceo Safety biomarker used in addition to the traditional safety biomarkerso Degree of risk depends on the impact on decision-making in drug development and the risk to patients enrolled in
the trials
• Challenging benefit and risk profile – higher level of evidenceo Surrogate endpointo If the biomarker is not truly a surrogate endpoint for predicting clinical benefit, results invalid and inappropriate
approval decisions madeo Leads to potentially ineffective drugs marketed or patients denied access to effective therapy
Evidence map
• The evidence maps in this framework are inspired by, but not identical to, the one used by Altar et al. (2008)
• The COU choices made determine the overall relative level of benefit and risk• Benefit and risk determined as a result of the COU in turn determines the levels
of evidence needed to evaluate the biomarker for qualification• The evidence acceptable for satisfying evidentiary criteria in some cases may be
partially or entirely composed of retrospective, literature, or other “real world” types of evidence
• The levels of evidence required to qualify the marker can be described according to a series of variables
Altar et al. CPT, 83:368-371, 2008