EMIF-AD Project Background

With multiple clinical trials in dementia failing over the past years, attention has recently shifted to subjects with Alzheimer’s disease (AD) who have not reached the stage of dementia yet (i.e., ‘pre-dementia’). Since these subjects have limited brain damage and are suggested to be more likely responsive to treatment than subjects with full-blown dementia, they are an important target group for future treatment studies.

EMIF-AD Project Objectives

Prof Simon Lovestone, Oxford UniversityTrial design in pre-dementia AD, however, is challenging because subjects with pre-dementia AD are difficult to identify and limited information is available on their outcome. The lack of reliable diagnostic and prognostic markers for pre-dementia AD can be explained by the availability of only small-scale ongoing biomarker studies and longitudinal cohorts including these subjects. EMIF has linked this information and unlocked the true potential of these studies.

By connecting relevant cohort studies across Europe, EMIF-AD has set up a pan-European platform for large-scale research on biomarkers and risk factors for neurodegenerative disorders. The biomarker discovery activities in EMIF-AD were driven by an extreme phenotype approach, in which decline or biomarker status was used as the end point for biomarker discovery, rather than a clinical diagnosis. Doing so, EMIF-AD has developed new treatment targets, multimodality/omics diagnostic tools and qualification level biomarker datasets suitable for presentation to regulatory authorities prior to approval for use in clinical trials and practice.

Finally, prediction rules for cognitive decline in presymptomatic and prodromal AD were developed which will not only improve clinical diagnosis and prognosis, but equally support subject selection and stratification in future clinical trials. These achievements have been possible because EMIF-AD combined both large-scale patient cohorts, linkage with EHR data, and cutting edge biomarker discovery expertise.

Work Packages Overview

Placeholder image


EMIF-AD Work Packages Validate new biomarkers andselect subjects for prevention trials WP4 Biomarker discovery withextreme phenotypes as endpoint WP3 Characterise study populationand define extreme phenotypes WP2 Define study population and data collection WP1

EMIF-AD Project Achievements

WP 1 Definition of study population data requirements and data collection

Pieter Jelle Visser (UM) – Michael Arrighi (Janssen)



The main objective is:

The overall aim of this WP is to collect data required for the development and validation of new biomarkers for predementia AD into an overarching EMIF. The data collected will be used for further analysis in WP 2 to WP 4.

The sub-objectives are:

  1. To select cohorts that can be used for biomarker discovery and validation
  2. To pool data from the cohort studies in a Private Remote Research Environment
  3. To link data from research cohorts to EHR
  4. To collect additional data of key AD biomarkers and cognitive markers in subjects from ongoing longitudinal cohort studies


  • Symposium on data-sharing at the Alzheimer Association International Conference in Washington DC July 2015
  • Joint IMI AD-platform publication in Alzheimer’s and Dementia (2015)
  • Presentations for the European Alzheimer’s Disease Consortium (EADC) (2013-2015)
  • 40 publications, including papers in JAMA, Lancet Neurology, and Brain
  • Presentation preclinAD study on Dutch television and radio (January 2016)


Set-up catalogue/cohort finder for AD and aging cohorts (task 1.1)

  • Information available on 44 cohorts
  • Cohorts have data from over 50.000 subjects
  • Search performed to select cohorts for 1000 cohort in WP3

Pipeline developed for harmonised data upload in tranSMART (task 1.2)

  • 9 Cohorts uploaded with close to 3000 subjects
  • >60 variables standardised
  • Data used in analysis for WP2

70% baseline data collection preclin AD cohort (task 1.4)

  • Baseline completed in 170 cognitively normal subjects (including 68 monozygotic twin pairs) and 60 scheduled
  • First analysis performed (figure shows amyloid PET scan in a monozygotic twin pair)
amyloid PET scan in a monozygotic twin
amyloid PET scan in a monozygotic twin


  • The catalogue has been used by DP-UK and IMI-EPAD and will be used by IMI-ROADMAP, IMI-MOPEAD, IMI-PRISM, IMI-PHAGO, IMI-ADAPTED, JPND-EADB, the INTERDEM network, and the EADC network.

WP 2 Characterisation of the study population and definition of extreme phenotypes

M. Gordon (BI), G. Novak (Janssen), S Engelborghs (UA), B. Dubois (UPMC)

BI, Janssen, UA, UPMC, VTT, VUmc, Maastricht, UM, GSK, Pfizer, Roche, CamCog, Newcastle, KCL, Oxford


The main objectives are:

  • To characterize the study population and
  • To define extreme phenotypes based on subjects selected in WP 1

The sub-objectives are:

  1. To provide summary statistics of the cohorts selected in WP 1
  2. To operationalize criteria for pre-symptomatic AD and prodromal AD
  3. To define extreme phenotypes based on rate of decline
  4. To develop a multivariate prediction model for cognitive decline
  5. To define extreme phenotypes based on AD biomarkers
  6. To define extreme phenotypes based on resilience to dementia at old age


Peer-reviewed articles

  • Alexander, et al. Age-stratified prevalence of mild cognitive impairment and dementia in European populations, a systematic review. JAD 2015;48(2):355.
  • Vos, et al. Prevalence and prognosis of Alzheimer's disease at the mild cognitive impairment stage. Brain 2015 May;138(Pt 5):1327-38.
  • Jansen, et al. Prevalence of cerebral amyloid pathology in persons without dementia: a meta-analysis. JAMA 2015;313(19):1924.
  • Tang, et al. Current developments in dementia risk prediction modelling: An updated systematic review. PLoS ONE 10(9) September 3, 2015;1-31.

Scientific congress presentations(numerous each year). For July 2016 AAIC:

  • Perera, et al. Dementia prevalence and incidence in a combination of European electronic health record databases – the EMIF-AD EHR Resource.
  • Bos, et al. Prevalence of vascular and lifestyle risk factors in different stages of prodromal Alzheimer’s disease and its influence on cognitive decline.
  • Liedes, et al. Predicting change in hippocampal volume using MRI features from baseline.
  • Legdeur, et al. Resilience to clinical dementia at old age: The European Medical Information Framework (EMIF) 90+ Study.


Assess prevalence of dementia in EU and compare findings in literature and electronic health records

  • Systematic analysis of 26 European studies reveals continuously increasing prevalence of dementia from 60 years, reaching 44.7% at >95 years
  • Based on EHR in 6 cohorts, prevalence of dementia increased at a similar rate by age stratum, but ranged from 16-28% of that reported in the systematic review
prevalence of dementia in EU

Identify distinct trajectories of cognitive decline in SLaM EHR derived cohort

  • 3441 subjects with > 3 MMSE assessments and up to 5y follow-up
  • identified 6 trajectories of decline and covariates which differed among trajectories (e.g. #3 slower decline, more mild cognitive problems, use of donepezil vs. #4 faster decline, more psychosis and behavioral difficulties.
Identify distinct trajectories of cognitive decline in SLaM EHR derived cohort

The prevalence of risk factors in patients with prodromal AD and their influence on cognitive decline

  • Smoking was a risk factor in individuals with prodromal AD
  • Hypertension was a risk factor in individuals without prodromal AD
  • Hypercholesterolemia was protective factor in those without prodromal AD
prevalence of risk factors in patients with prodromal AD and their influence on cognitive decline-1
prevalence of risk factors in patients with prodromal AD and their influence on cognitive decline-2
prevalence of risk factors in patients with prodromal AD and their influence on cognitive decline-3


  • WP1 Summary Statistics
  • Platform: dementia prevalence / incidence, treatment pathways, inflammation in AD

WP 3 Biomarker Discovery

Simon Lovestone (UOXF) – Johannes Streffer (Janssen)



The main objective is:

The objective of this WP is to discover new biomarkers and genetic markers for the diagnosis and prognosis of predementia AD using the ‘extreme phenotypes’ defined in WP 2.

The sub-objectives are:

  1. To perform proteomics in CSF and plasma
  2. To perform GWAS analyses, exome sequencing, assessment of epigenomic profiles, and transcriptome profiling
  3. To perform metabolomics in CSF and plasma
  4. To perform voxel based MRI analyses
  5. To develop multivariate and class prediction algorithms
  6. To generate assays for new biomarkers


Presentation at ARUK

Protein biomarker results presented at conference

Presentation at AAIC

Both oral and poster presentations of protein biomarker work including EMIF500 accepted


Background to EMIF500 accepted by Journal of Alzheimer’s disease and EMIF500 paper in preparation


Analysis of protein markers indicative of CSF endophenotypes

  • EMIF 500 cohort assembled; 500 participants from diverse cohorts with pre-existing CSF measures of Abeta pathology
  • EMIF 500 samples obtained; plasma samples from parent cohorts identified and transferred to analysis team at Oxford
  • EMIF 500 proteomics; planned analysis of 21 previously identified protein markers of Abeta pathology underway
  • EMIF 500 results; first 20 protein analyses completed; confirmation of FGC as peripheral marker of Abeta pathology
  • EMIF 500 dissemination; presentation accepted at AAIC 2016 and paper in preparation

Establishing pipelines for multimodal biomarker (MMB) analysis

  • EMIF MMB dataset generated; new data acquisition of metabolomics
  • EMIF MMB dataset assembled; full multimodal dataset including MRI, proteomics, genomics and metabolomics assembled and made available on common platform
  • EMIF MMB workspace generated; wikipedia and Github analysis tracking generated
  • EMIF MMB workflows; analysis is underway and informaticians are sharing workflows

Extreme phenotype multimodal analysis in EMIF 1000

  • EMIF 1000 sample set in generation; ~750 samples in hand at central sample facility
  • EMIF 1000 analysis; modality teams completed protocol generation

WP 4 Validation of new biomarkers and identification and selection of individuals for pharmacological interventions

Hilkka Soininen (UEF) – Piotr Lewczuk (UKER) – Henrik Zetterberg (UGOT) – TBD (Janssen)



The main objective is:

The main objective of this WP is to cross-validate biomarkers discovered in WP 3 in independent samples and to develop recruitment strategies for AD prevention trials.

The sub-objectives are:

  1. To cross-validate the new plasma, CSF, genetic and MRI markers from WP 3 in an independent cohort
  2. To investigate the relation between genetic risk factors and amyloid pathology on neuropathological examination
  3. To investigate the relation between new AD plasma markers and long-term cognitive decline in the general population
  4. To develop stepwise screening algorithms for recruitment of subjects with presymptomatic AD or prodromal AD
  5. To analyse data from ongoing prodromal AD trials
  6. To select sites for selection and inclusion for subjects in predementia AD trials


Cross‐validation of new plasma, imaging, and CSF markers

  • New markers are expected from WP 3 in April 2016
  • In the meantime: Three interesting biomarkers were selected from other projects: Neurogranin, Neurofilament light and Abeta 40
  • A clinical validation and a technical validation (e.g. analytes, assay quality according to modalities) will be performed
  • The 1000 cohort
    • Validation of Neurogranin, Neurofilament light and Abeta 40 for those where available as well as in a small collective of independent samples
    • Abeta analysis is needed to be done – communications are ongoing with ‘1000 cohort’ partners that will provide CSF samples
    • EPAD / The use of a fully automated Roche assay for biomarkers
      • cross-validation of this with the EMIF 1000 cohort samples and with a small collective of independent samples
  • WP 4 (Lewczuk/Zetterberg) will also aim to validate oligomeric Ab assaydeveloped by Zetterberg / Blennow within WP 3
  • Protein panel cross-validation:
    • A collection of 500 plasma samples / Simon Lovestone
  • LipidiDiet: a possibility to cross-validation studies in prodromal AD
  • For later stage cross-validation:
    • A collection of data from 300 normal subjects is ongoing / Pieter Jelle Visser
    • A cohort of 90+ sample / Pieter Jelle Visser

Technical validation

  • Validation of the body fluid biomarkers (CSF, plasma, serum): the material is limited
  • A study on the stability of the candidate biomarkers under different storage conditionsand in different body fluids(the experience from BIOMARKAPD)
  • The preparation of QC samplesin different matrices (natural body fluids, spiked material, etc.)
  • The coordination of an external QC roundwith laboratories where a given technology is available
  • Technical cross-validation across different platforms(ELISA-Luminex-MSD) wherever available

Analysis of the relation of new genetic risk factors with amyloid pathology on neuropathology

  • Work can be done only later

Analysis of the relation of new plasma markers with cognitive decline in the general population

  • Work can be done only later

Develop stepwise screening algorithms

  • Work can be done towards the end of WP4
  • This overlaps with EPAD

Analysis of data from ongoing prodromal AD trials

  • Database now locked; looking at data, additional analyses (following Task 4.4) will be done later

Provide sites for selection and inclusion for subjects in predementia AD trials

  • There is a lot of overlap between EMIF and EPAD
  • An EPAD network exists (Netherlands, Italy, Switzerland, Spain, UK)
  • EMIF-AD has already contributed to the set-up of this network action through the catalogue by selecting parent cohorts and fingerprinting potential trial sites