I have no personal or financial interests to declare:

I have no financial support from an industry source at the current presentation.

Use the following slide to disclose any conflicts of interest

Form A: no conflicts of interest to declare.

대한혈액학회 Korean Society of Hematology

COI disclosureName of author ： Liran I Shlush

Liran I Shlush

Weizmann Institute of science Israel

Rambam Healthcare campus

Seoul Mar 2019

Predicting Leukemia Development from Preleukemic Clonal

Hematopoiesis

Definition of a preleukemic mutation

Mutant allele frequency: Shlush et.al Nature 2014

DNMT3amut

Leukemic blasts

T cells

PreL-HSC

X

X

X

X

XY XX

X

Age Related Clonal Hematopoiesis (ARCH)is driven by preleukemic mutaitons

Shlush LI Blood 2018

Can leukemia be diagnosed earlier?

ARCH

Preleukemic Mutation

Pre-LeukemiaLeukemia

(AML MPN MDS etc)

Aging

Non Age Dependent

CHGenetic drift Non- recurrent

genetic variations

Preleukemia versus ARCH using EPIC data (Precision medicine)

6

European Prospective Investigation into Cancer

and Nutrition (EPIC)

520,000 People

Targeted error corrected Seq of AML related genes

124 AML Cases677 Matched Controls

In Collaboration with George Vassiliou

ARCH more prevalent in pre-AMLs with higher VAFs

ARCH=Restricted Gene list and specific positions VAF>0.5%

pre-AML CasesMatched Controls

Pre-AML cases carry more mutations at younger age

pre-AML CasesMatched Controls

Somatic mutations in specific genes are more predictive

We NEVER found:

NPM1cFLT3/ITDCEBPA

Clonal dynamics in ARCH and Pre-AML

pre-AML casesMatched Controls

p-value = 0.2861

Pre-AML clones grow in the same rate as ARCH clones

pre-AML CasesMatched Controls

Mutations in specific genes contribute to AML risk differently

A molecular predictive model on a validation cohort from the Sanger institute (Vasilliue group)

Abelson S & Collord G (Nature 2018)

Desai et.al Nat Med 2018

Long term follow-up (median 4 years) of 30 AML cases

Provide evidence for parallel evolution of preleukemic clones

Chapal-Ilani N et.al unpublished

Summary So far

AgeYears

Preleukemia

Months Leukemia

Diagnosis

TherapeuticWindow

Pre-AML Control

30

60

67

FLT3-ITDNPM1cCEBPA

High penetranceMutations: IDH1/2SRSF2, U2AF1, TP53RUNX1

Low penetranceDNMT3a TET2

AML Prediction Based on RDW in EPIC

P=0.008

Electronic health record analysis

4.5 Million individual over 15

years1696 AMLs

Machine learning

Electronic health record (HER) basedAML prediction model

Mendelson Cohen N Niemeyer E Tanay A (Nature 2018)

Modelling blood for 4.5 million patients

Projecting Red Blood Cells on the multi-parameter space

Smokers/Runners

Aging

NormalHigh

Normal/ Average

Normal/Low W

High Value

CBC Map of Israel (3.2 Million people)

Modelling blood for 3.2 million patients

Projecting Hemoglobin on the multi-parameter space

Smokers/Runners

Aging/Anemia

NormalHigh

Normal/ Average

Normal/Low W

High Value

Modelling blood for 4.5 million patients

Mean Cell Volume

Thalassemia minorAverage Hb low MCV

Microcytic AnemiaMacrocytic AnemiaMacocytosis WO AnemiaAlcohol/Medications

NormalHigh

Normal/ Average

Normal/Low W

High Value

Projecting MCV on the multi-parameter space

NormalHigh RDW

AverageRDW

NormalLow RDWRDW

High RDW

Red Blood Cell Distribution Width (RDW)

Patients 1-15 years before AML

Iron Deficiency Anemia

Changes in Blood Counts Before AML

AML predictive model based on Electronic Health Records (HER)

Abelson S et.al Nature 2018

Clinical Trial E7820Healthy Individuals

Clalit 750,000ARCH/CCUS Clalit

N=200

EHR prediction model to identify high risk

N=1000

ARCH/CCUSHospitals

N=200

Genetic testing

100 Positive for SRSF2 U2AF1

E7820 for 3 month end point reduction

in VAF>5%

Abdel-waheb Shlush Feldman Stein Eisai

Acknowledgments

Sagi AbelsonJohn DickStanley Ng

John Dick Lab Shlush Lab

Collaborators

Mark Minden (PMH)Scott Bratman (PMH)Trevor Pugh (PMH)Lawrence Heisler (OICR)Philip Awadalla(OICR)Philip Zuzarte (OICR)Yogi Sundaravadanam (OICR)Paul Brennan (EPIC)Amos Tanay (WIS)Netta Mendelson-Cohen (WIS)Omer Weisboard (WIS)Stanly Ng (UOT)

Elisabeth NeimayerNoa Chapal-IlaniAviv De-MorganBarak OronNathali KaushanskyMax KushnirYoni MoskovitzAmos TuvalYoav WigelmanYael Morgenstern

GrantsLLS – quest for cureERC – Horizon 2020BIRAX

George Vassiliou LabGrace Collord (Sanger)

Moritz Gerstung

EMBL

Slide Number 1Predicting Leukemia Development from Preleukemic Clonal�HematopoiesisDefinition of a preleukemic mutationAge Related Clonal Hematopoiesis (ARCH)�is driven by preleukemic mutaitons Can leukemia be diagnosed earlier?Preleukemia versus ARCH �using EPIC data (Precision medicine)�ARCH more prevalent in pre-AMLs with higher VAFsPre-AML cases carry more mutations at younger ageSomatic mutations in specific genes are more �predictiveSlide Number 10Slide Number 11Mutations in specific genes contribute to AML risk differentlyA molecular predictive model on a validation cohort from the Sanger institute (Vasilliue group)Slide Number 14Long term follow-up (median 4 years) of 30 AML cases�Provide evidence for parallel evolution of preleukemic clones Summary So farAML Prediction Based on RDW in EPICSlide Number 18Slide Number 19Slide Number 20Slide Number 21Slide Number 22Changes in Blood Counts Before AMLAML predictive model based on Electronic Health Records (HER)Clinical Trial E7820Slide Number 26