Alzheimer s disease The future of clinical trials: big problem, Big Data, big solution?
The Problem More than 36m people with dementia Consumes 1% global GDP Serial trials failure
Targets for therapy
Nothing works! b secretase a secretase g secretase
January 2014
The solution Diversify target development More drugs! Trials in prodromal or preclinical disease Earlier treatment! Rapid trials with early read-outs Fail quicker!
Agenda What is Big Data? An integrated platform for translation Example studies Data driven proof of concept trials Data driven biomarkers for early detection Data driven experimental medicine Data driven mechanisms and drug discovery Looking forward
Big Data Volume
Big Data Variety
Big Data Velocity
Big Data Complexity
Too few targets; too slow progress Secondary prevention by 2025 Dementia research Institute ARUK Drug Discovery Institute Dementias Platform UK European Medical information Framework Translational Research Collaboration in Dementia European Prevention of Alzheimer s Disease Target identification Target development Trials ready cohorts and data Proof of concept trials Pre-competitive and collaborative
Too few targets; too slow progress Secondary prevention by 2025 Dementia research Institute ARUK Drug Discovery Institute Dementias Platform UK European Medical information Framework Translational Research Collaboration in Dementia European Prevention of Alzheimer s Disease Target identification Target development Trials ready cohorts and data Proof of concept trials
Prevention of Alzheimer s disease Secondary prevention Symptomatic therapy Adapted from Sperling et al (2011) Alzheimer s and dementia 7 280-92
EPAD Registry integration Cohort A Cohort B Cohort C Readiness cohort Selection criteria Trial cohort placebo Rx 1 Rx 2 Rx n Adaptation by change in intermediate phenotype Adaptation on cognition outcomes
Global Alzheimer s Prevention C-PAD? J-PAD US-PAD EPAD
The challenge of clinical trials volume variety velocity complexity Big data solution Add complexity : statistical modelling and dynamic trial design
Too few targets; too slow progress Secondary prevention by 2025 Dementia research Institute ARUK Drug Discovery Institute Dementias Platform UK European Medical information Framework Translational Research Collaboration in Dementia European Prevention of Alzheimer s Disease Target identification Target development Trials ready cohorts and data - Identification of biomarkers Proof of concept trials
Markers to enable secondary prevention Selection and stratification markers Secondary prevention Progression markers Symptomatic therapy Adapted from Sperling et al (2011) Alzheimer s and dementia 7 280-92
European Medical Information Framework
Biomarkers predicated on case-ness Systematic review of case-control blood based biomarkers for Alzheimer s disease 1 cohort; 109 proteins 21 studies, 3469 subjects 163 proteins Limited replication Methodological variation 6 cohorts; 4 proteins 4 cohorts; 6 proteins 3 cohorts; 6 proteins 2 cohorts; 38 proteins Kiddle et al. (2014) Candidate blood proteome markers of Alzheimer's disease onset and progression: a systematic review and replication study. J Alzheimers Dis 38 (3), 515-531
EMIF catalogue making data visible
Available research cohort data EMIF-AD Number of subjects Controls SCI MCI All 31219 3173 11976 Plasma 3537 1764 3961 DNA 8763 3163 5731 RNA 1637 484 2128 CSF 846 1336 3757 Urine 2307 1116 2349 MRI 3449 1035 5586 FDG-PET 551 308 1942 Amyloid PET 335 143 492 EEG 301 773 1173 Either access to results from analysis or access to samples/scans
Stratification by pathology
Biomarkers predicated on in vivo pathology Study 1 In two independent cohorts FGG predicted PET amyloid burden Study 2 Training set (AIBL) Test set (UCSF) EMIF 500 n=500 FGG is significantly associated with CSF amyloid Study 3 Baltimore Longitudinal study of ageing FGG associated with PET amyloid
Biomarkers for secondary prevention of Alzheimer s the deep and frequent phenotyping study Aims to identify optimal set of markers for : patient stratification and selection for pre-clinical secondary prevention trials Markers of change for target engagement and early proof of concept trials Deep phenotyping Stratification markers Progression markers Adapted from Sperling et al (2011) Alzheimer s and dementia 7 280-92 Molecular, imaging electrophysiology and cognitive markers Frequent phenotyping Approximately 2 month interval repeat measures Feasibility study Participant acceptability Multicentre practicability in the NIHR TRC-D
The challenge of biomarkers volume variety velocity complexity Big data solution Add volume : increase the size of studies Add complexity: combine different markers
Dementias Platform UK
UK Biobank Enhancements for Dementia Research Web-based questionnaires for additional exposures and outcomes (cognition, mental health, occupation..) Wrist-worn accelerometers mailed to 100,000 participants to measure physical activity Multimodal imaging in 100,000 Repeat Neuroimaging in 10,000 Genotyping of all participants (820,000 SNPs) Repeat cognition, samplling Connectivity to EMRs for mental health
Use of electronic medical records for research - Case Records Interactive Search (CRIS) SLaM CRIS South London and Maudsley NHS BRC implementation D-CRIS Cambridge & Peterborough, Oxford Health, West London, Camden and Islington 1 million plus patients UK-CRIS 10 site extension Connectivity to UK BioBank 2 million patients?
MMSE score 16 18 20 22 24 EMR-data re-use for research (Donepezil/Arricept and Alzheimer s disease) Phase IV of ACHeI > 2500 patient years of therapy > 8 fold dataset compared to Cochrane Costs and effectiveness precompetitive collaboration with pharma Text mining derivation of service utilisation and costs Predictors of response Biomarkers and clinical -1 0 1 2 3 time(years) Data from Robert Stewart, KCL
The challenge of mechanistic epidemiology volume variety velocity complexity Big data solution Add volume : massively deep and extensive testing Add variety : using coded (numbers) and narrative (text) data Add complexity : working across datatypes
Too few targets; too slow progress Secondary prevention by 2025 Dementia research Institute ARUK Drug Discovery Institute Dementias Platform UK European Medical information Framework Translational Research Collaboration in Dementia European Prevention of Alzheimer s Disease Target identification Target development Trials ready cohorts and data Proof of concept trials
Informatics driven targets JAK-STAT signalling common to Alzheimer s, macular degeneration and inflammatory disease JAK-STAT signalling correlates with risk Alejo Nevado-Holgado Proof of concept in experiments - In vitro and in vivo
Big Data and the dementia challenge? volume variety velocity complexity
Big Data challenges Data management Data analysis Data governance
Too few targets; too slow progress Secondary prevention by 2025 Dementia research Institute ARUK Drug Discovery Institute Dementias Platform UK European Medical information Framework Translational Research Collaboration in Dementia European Prevention of Alzheimer s Disease Target identification Target development Trials ready cohorts and data Proof of concept trials Big Data
With thanks to. 50 executive members 150 principal investigators 500 scientists 1500 collaborators 50,000 volunteers 50,000,000 individuals contributing data 200,000,000 euros from funders
Many thanks!