An early diagnosis kit is being developed that combines genetic and immunological tests to diagnose multiple sclerosis (MS) with high accuracy through a simple blood test. The kit leverages proven technologies from MS Genetics and Louphius that have shown accuracy rates of 83% and 91% respectively in differentiating MS from other conditions. By combining results from tests measuring gene expression, genetic risk alleles, and anti-glycan antibodies through conditional probability, the kit aims to outperform individual tests to enable early and differential diagnosis of MS with a high positive predictive value. A Horizon 2020 project is proposed to further optimize and validate the individual tests and their combined use in an integrated classification system to facilitate early treatment and reduce disability.

1. 1
Early Diagnosis of
Multiple Sclerosis
Made Possible
- March 2014-

2. 2
MS Genetics
Proven technologies
based on long terms
clinical results.
Investor backed by
Goldman Hirsh Partners
Patented: 2 approved
,7 pending and 1
provisional
Dr. Julia Rothman
(CEO)
Prof. Anat Achiron
(CSO)
Prof. Itzhak Haviv
(CTO)
Dr. Michael
Gurevich
(Clinical Director)
LeadershipStatus
1
2
3

3.  Inflammatory disease
 Destruction
of myelin sheaths around
the axons(=demyelination)
 brain and spinal cord
 broad spectrum of signs and
symptoms
MULTIPLE
SCLEROSIS

4. PRMS Progressive Relapsing MS
SPMS Secondary Progressive MS
PPMS Primary Progressive MS
RRMS Relapsing/ Remitting MS (85%)
CLASSIFICATION OF MULTIPLE SCLEROSIS
Gradual progression of the disease from its
onset with no relapses or remissions
Unpredictable attacks which may or may not leave
permanent deficits followed by periods of remission
Initial RRMS that suddenly begins to decline without
periods of remission and relapses.
Steady decline since onset with super-imposed
attacks.

5. 5
2.5M people
are affected
by MS
The leading cause of non-
traumatic neurological
disability among
young adults.

6. MS
Young. Women. Families.
~3% of first
degree relatives
pre-disposed to
develop disease
20-40 years
of age at
onset
67%
of affected
are women

7. Clinical Tools for MS Diagnosis :
McDonald Criteria

8. 85% of Patients with CIS Develop MS
Reprinted from Trapp BD, et al. Neuroscientist. 1999;5:48-57, with permission from Sage Publications.
Attempts to identify disease activity earlier than clinically definite relapsing-remitting MS (CDMS/RRMS)
has resulted in the identification of “McDonald Criteria” MS,
clinically isolated syndrome (CIS), and radiologically isolated syndrome (RIS) based around MRI.

9. Timing of Therapy vs level of Disability
First Clinical Attack
Time (years)
Clinical threshold
Axonal loss
Demyelination
Time
window for
early
treatment
Relapsing-Remitting Transitional
Secondary
Progressive
First
Demyelinating
Event
Pre-
clinical
Inflammation

10. 2 Long Years From Onset to Diagnosis
Patients treated
early showed less
attacks and lower
levels of disability
No available
laboratory test can
prove or rule out MS

11. Pharmacy and Therapeutics (P&T) committees Formulary
used by health plans, institutions, and hospital systems
Managed care aspects of managing multiple sclerosis.
Am J Manag Care. 2013 Nov;19(16 Suppl):s307-12.

12. Late Diagnosis:
A Huge Economic Burden
Annual cost of treating a single MS patient
Diagnosed
at severe stage:
€ 36,500
Diagnosed
at mild stage:
€3,600

13. Comprehensive Diagnostic Kit
for Early Diagnosis of MS
 Simple blood test
 Combining genetic &
immunological tests
 High PPV
GENE
expression
GWAS
Genotyping
OtherAnti-Glycan
antibodies

14. The power of
Conditional Probability
Combining a series of diagnostic tools,
each based on different measurements,
should outperform each tool, by itself.
Conditional probability measures the probability of
an event given another event has occurred

15. Gene Expression
Classifiers translating altered
gene expression signatures into
diagnostic prediction enable:
• Early CIS diagnosis (78%)
• Highly Accurate MS
differentiation (83%)
GENE
expression
WP1-Affymetrix into PCR

16. Early diagnosis
Economic meaning- example
14% false positive 7% false negative
1,000,000,000 western Europe

17. Proprietary classifier differentiates MS
from Non – MS .
234 patients study
Training set= 86 patients
(63MS,23NONMS)
Test set validation =148 patients
(115MS,33NONMS)
Test cohort -61 patients CIS
MS to non MS :
86%±3% sensitivity;
76% ±7% specificity.
CIS: ~79% diagnosis accuracy.
GWAS MS
susceptibility
Genes (88) Data analysis =Partek
genomics solution software

18. GWAS Genotyping
Genome-Wide Association Studies (GWAS):
297 alleles associated with elevated risk of MS.
The International Multiple Sclerosis Genetics Consortium (IMSGC) Evidence for polygenic susceptibility to multiple sclerosis--the shape of things to come. Am J Hum Genet. 2010;86:621–625.
H2020measure those alleles for all
individuals, using PCR microfluidics (Life
Technologies™), and integrate the results
Individual Prediction of phenotype.
GWAS
GenotypingAllele-specific expression (ASE) analysis
major advantage:
assessing expression within an individual rather than
across subjects avoiding major sources of error and variation
Abraham, G., Kowalczyk, A., Zobel, J. & Inouye, M. SparSNP: fast and memory-efficient analysis of all SNPs for phenotype prediction. BMC Bioinformatics 13, 88 (2012).
Abraham, G., Kowalczyk, A., Zobel, J. & Inouye, M. Performance and robustness of penalized and unpenalized methods for genetic prediction of complex human disease. Genet Epidemiol 37, 184-95 (2013).
Potentiality - Actuality

19. Anti Glycan Antibodies (Glicominds)
Measures (ELISA) the plasma level of specific
glycosylation products naturally restricted to the
cerebrospinal fluid, and only circulate in the blood,
following severe damage to the brain blood barrier, a
process that occurs in MS.
Glicominds tests:
• gMSPROEDSS
identifying the risk level for disability
progression in CIS/newly Dx MS
• gMSDx
Differentiating MS from NONMS
Clinical Data :
Specificity of 91%;
Sensitivity of 33%.
Anti-Glycan
antibodies
CLIA compliant

20. Test each patient
with CIS for early
diagnosis.
Screening test
for families
and relatives
3 Years:
CIS Patients
Future:
Families
Go to Market Strategy

21. Development
Roadmap & Requirements
Patented
technology
Device: manufacturing
production,
marketing
Patients networks
and feedback
Reinforcement of
clinical evidence
Now 3 Years
Preliminary
clinical Data Diagnosis
guidelines
2014 2015 2016 2017

22. Innovation Horizon 2020 project
Assessment of added value of
combining multiple technologies to
make an integrated classification, with
maximal positive predictive value
one stop solution:
early and differential
diagnosis
Validation and
Dissemination
MS PATIENTS

23. Horizon 2020 :
plan for Louphius and MS Genetics
Common collection and sample processing for all methods
PTx= MS non MS
Gene
expression
Louphius
Precision recall curve and classifier
performance
Bootstrapping,
LOO, multiple
classifiers
Training set
Test set
Glicominds
GWAS-
derived
genotyping

24. Each of the methods needs :
Phase one -
• Optimizing: sample storage, freeze/thaw protocols
(every company has different material, such as live
PBMC, plasma protein, DNA, or RNA).
• Evaluating current pool of antigens (for Louphius) or
genes (MS genetics)
• Reevaluating new target structures (antigens for
Louphius or genes for MS genetics)
• New/additional/combination markers (initially on only
within each company and method, and later in an
integrated manner)
• Improving the assay sensitivity, reproducibility, and
robustness (customer friendly read out systems)
- for MS genetics changing platform from Affymetrix
arrays, where the observations were made, to RT-
PCR. The platform transition is planned on the basis
of RNA-Seq on Ilumina HiSeq2500 platform.
- Louphius  to perform whole RNA-Seq on the T-cell
culture stimulation??
On phase II:
• Validation of assays on additional
independent samples
• Optimizing parameters to maximize
positive predictive value.
• Creating a large dataset with all
measurements and repeating the
classifier screen with support vector
machine learning using all method-
measurements.
• Or, alternatively, generating a decision
tree that starts with the classifier with
the best performance (lowest false
positive or false negative), then adding
the classifier results from another
method, to properly classify the
problematic samples.

25. The best way to
predict the future
is to create it.
THANK YOU!