Patrick Tan, MD PhDProfessor, Duke-NUS Medical School

Deputy Executive Director, Biomedical Research Council,Agency for Science, Technology and Research (A*STAR)

ESMO World Congress on Gastrointestinal CancerBarcelona, June 2018

Unravelling the Molecular Taxonomies of Gastroesophageal Cancers

Tay et al., Cancer Research (2003)

Molecular and Clinical Heterogeneity in Gastric Cancer

Tay et al., (2003) Cancer Research

TCGA Study (~3-4 Major GC Genomic Subtypes)

USA TCGA (2014) Nature

A) ChromosomalInstability (CIN)

B) Microsatellite Instability (MSI)

C) Genome Stable(GS)

D) Epstein-BarrVirus (EBV)

Today’s Topics

1) What are the most Frequent Genetic Events in Gastric Cancers?- Mutations in the *Non-Coding* Genome

2) What are the Key Determinants for Development of Gastric Cancer? - Pre-Malignant Condition (Intestinal Metaplasia)

Genetic Mutations in GC Have Largely Focused on Protein Coding Genes

Amaro Taylor-Weiner (USA TCGA)

Genetic Mutations in GC Have Largely Focused on Protein Coding Genes

Amaro Taylor-Weiner (USA TCGA)

Non-Coding Driver Mutations are Still Unknown for Many Cancer Types (Including GC)

“Driver Mutations” are MutatedMore Frequently than theBackground Mutation Rate

Whole-Genome Analysis of Gastric Cancer

Anders Skanderup, GIS

Uniform Mutation Callingon 212 GC WGS

Many Factors (Covariates) Affect Whole-Genome Mutation Rates

Single-Nucleotide Variants(SNVs)

Mut

atio

n C

ount

Indels

Mutation Count

Statistically Significant Non-Coding Mutation Hotspots are Enriched in CTCF-Binding Sites (CBSs)

** Significance Remains Even After Correcting for Elevated CBS Mutation Rates

CTCF: Regulator of 3D GenomeOrganization

Wikipedia

CBS Hotspot Mutations are enriched in GCs with copy-number instability (CIN)

Overall CBS Hotspot Mutation Frequency = 25% TP53 Mutations = 50%ARID1A Mutations = 14%

CBS hotspot mutations are associated with local chromosomal breaks

CBS hotspots Frequently and Specifically Mutated in Gastrointestinal Cancers

Top-4• Gastric• Colorectal• Liver• Pancreas

Clinical Use-Case : Use of Non-coding Hotspots in Liquid Biopsies

Gene Sensitivity (%) Size (bp) Sensitivity / bp Cumulative Sensitivity

KRAS 6 15 0.4 6.0%Hotspots 54 720 0.08 56.8%

TP53 50 750 0.07 78.4%RHOA 6 180 0.03 79.7%

• Cancer cfDNA fraction very low in blood• Non-coding hotspots represent densely mutated regions (=TP53)• Disease monitoring of gastrointestinal cancers

Summary SlideWhole-genome analysis of GC reveals frequent mutations at CTCF Binding Sites (CBSs)

CBS Hotspots are significantly mutated even after adjusting for covariates, consistent with positive selection

CBS Hotspot Mutations Occur at Frequencies Exceeded Only by TP53 Mutations, in GCs with Chromosomal Instability

CBS Hotspot Mutations are Associated with Local Chromosomal Breaks and Regional Changes in Gene Expression

CBS Hotspot Mutations are Specifically Observed in Gastrointestinal Malignancies

Guo et al., (2018) Nature Communications

Today’s Topics

1) What are the most Frequent Genetic Events in Gastric Cancers?- Mutations in the *Non-Coding* Genome

2) What are the Key Determinants for Development of Gastric Cancer? - Pre-Malignant Condition (Intestinal Metaplasia)

Gastric Cancer and Intra-Patient Heterogeneity

Sundar and Tan (2018) Cancer Discovery

Most Gastric Cancers* Follow a Multi-step Carcinogenesis Sequence

Yeoh and Tan (2015) Gastroenterology

*Diffuse-type GC does not involve metaplasia

Gastric Cancer Epidemiology Program(GCEP)

Four Singapore Hospitals : NUH, SGH, TTSH, CGHFunded by National Medical Research Council

GCEP Translational Study (“TransGCEP”) Selection of High-Risk IM Patients (n=148)

56% moderate/marked IM

All Chinese

All positive for Hp serology(Previous Infection)

Normal Mucosa

Mild IM (<30%)

IM (≥30% cellularity)

p value (IM vs Normal)

n=43 n=22 n=83Age (year), mean ±SD 62±7 60±7 62±7 0.17

RaceChinese 43 (100) 22 (100) 83 (100) --

Gender (%)Male 22 (51) 12 (55) 42 (51) 1Female 21 (49) 10 (45) 41 (49)

Smoking (%) 0.048Current/ Ex-Smoker 9 (21) 4 (18) 32 (39)

Non-smoker 34 (79) 18 (82) 51 (61)Alcohol consumption (%) 5 (12) 5 (23) 20 (24) 0.1

Family history of GC in first-degree relative (%)

7 (16) 3 (14) 13 (16) 1

Hp serology positivity (%) 43 (100) 22 (100) 83 (100) --

Chronic gastritis (%) 38 (88) 22 (100) 83 (100) 0.004Atrophic gastritis (%) 0 (0) 0 (0) 67 (81) --Low Grade Dysplasia (%) 0 (0) 0 (0) 2 (2) --

EGN (%) 0 (0) 0 (0) 4 (5) --Endoscopy surveillance (months), mean ±SD

56±12 58±8 49±18 0.04

DNA Mutations – Point Mutations and Indels(MAF>4%)

TP53 and ARID1AClonal Mutations are Rare

(TP53 – 2%; ARID1A – 3%)

Laser Capture Microdissection Confirms Mutations in IM Cells

IMs Exhibit Low Mutation Burdens Compared to GC

NL

IM

GC

Copy Number Alterations and Telomere Erosion

MYC

NL IM10% of IMs have sCNAs

Most sCNAs target Chr 8q

IMs have significantly shorter Telomeres (Genome Instability)

Sequencing Detects More Hp-Infected IMs Compared to Histology

All 15 have Hp DNA(100%)

33 cases have Hp DNA(27%)

Genomic Sequencing Can Detect Active Low-Level Hp Infection

Histology-confirmed HP casesShow no Hp reads after eradication

(ie Hp DNA is transient)

Giemsa staining confirms Hp infection inSequencing positive cases

IMs Exhibit Global DNA Methylation Alterationsand a Subgroup is Hypermethylated

IMs Exhibit IncreasedDNA Methylation

Normal Intestinal Metaplasia

Integration with Clinical Outcome(Regression, Persistence, Progression)

TransGCEPSamples

LGD/HGD /EGC

Regression criteria based on Rugge et al (2003)Progression includes both LGD and HGD, as both have higher GC risk than IM

Factors influencing IM regression and progression

Mutation Burden Telomere Length

DNA Methylation Copy Number Alterations

Regression

Summary Slide: What our Data SupportsGenomic profiling reveals that IMs exhibit low mutational burdens compared with GCs

In general, TP53 and ARID1A mutations are rare in IM

Some IMs have FBXW7 mutations, chromosome 8qamplifications, or shortened telomeres

Sequencing detects more IM patients with active H. pylori infection than histology

(Epi)genomic alterations in IM predict subsequent GC progression or regression

Huang et al., 2018 Cancer Cell

AcknowledgementsChang XuAngie Lay Keng TanMinghui LeeSuting TayKakoli DasManjie XingAliya FatehullahSyed Muhammad

Fahmy AlkaffTony Kiat Hon LimJonathan LeeKhek Yu HoSteven George RozenBin Tean TehNick BarkerChung King Chia

Christopher KhorChoon-Jin OoiKwong Ming FockJimmy SoWee Chian LimKhoon Lin LingTiing Leong AngAndrew Wong

Khay Guan YeohMing Teh

Kie Kyon HuangKalpana RamnarayananFeng ZhuSupriya Srivastava

Amanda Yu GuoMei Mei ChangWeitai HuangWen Fong OoiManjie Xing

Anders Skanderup

Andrea RajnakovaLee Guan LimWai Ming YapSoh Ee LeeJia Wei Lee