Epigenetics

:
A Scientific Frontier of
Precision & Preventative
Medicine
John B. Kisiel, MD
May 19, 2015

©2010 MFMER | slide-1

Disclosures
• Exact Sciences (Madison WI)
• Mayo Clinic
• Minor equity investor
• John B. Kisiel
• Scientific advisor
• Research support
• Potential royalties

©2010 MFMER | slide-2

Goals and Objectives
• Epigenetics and human disease
• Epigenetics terms and concepts
• Clinical applications overview
• Clinical application focus: early cancer detection

©2010 MFMER | slide-3

Epigenetics and Disease
• 10 leading causes of death in the US









Heart disease*
Cancer*
COPD*
Stroke*
Accidents
Alzheimer’s disease*
Diabetes*
Nephritis(-osis) *
Influenza & pneumonia
Suicide (major depression*)
*disease with an epigenetic component
©2010 MFMER | slide-4

What does the term “Epigenetics” mean?
• Changes in gene expression without a change
in DNA sequence
• Gene expression is the process by which
genetic code stored in DNA is “interpreted”
• Modifications to DNA or supporting proteins
which allow genes to be turned-on or off
• Normal part of human development
• Skin cell vs. blood cell have same genes but
different appearance and function (phenotype)

• Aberrancy in many diseases
©2010 MFMER | slide-5

Gene Expression: The Central Dogma

• Change in DNA sequence (mutation) might change the protein
• Altered proteins may cause disease
"Genetic code" by Madprime -. Licensed under CC BY-SA 3.0 via Wikimedia Commons-

©2010 MFMER | slide-6

Regulation of Transcription: A Closer Look
• Transcription factors:
• Proteins that initiate the process
• Must bind to the DNA upstream of transcribed code
• This “promotor” region must be accessible

Transcription factors

"Simple transcription elongation1" by Forluvoft - Own work. Licensed under Public
Domain via Wikimedia Commons ©2010 MFMER | slide-7

“Epigenetics”
• Regulation of gene expression without change
in DNA sequence
• Block access to promotor to prevent expression
• Example: DNA methylation

http://missinglink.ucsf.edu/lm/genes_and_genomes/methylation.html
©2010 MFMER | slide-8

“Accessible” DNA
• DNA in one cell is 9 feet long; must fit in small package
• Folded and condensed around histones (scaffolding
proteins) to form chromatin
• Uncondensed DNA is where active transcription occurs
Genes under active transcription

Less active genes

"Chromatin Structures" by Original uploader was Richard Wheeler Licensed under CC BY-SA 3.0 via Wikimedia Commons ©2010 MFMER | slide-9

Functional Role of Chromatin
• Regulates the >25,000 protein coding genes
• Each gene is tissue-specific and time-sensitive
• Example:

Cell division genes must be ON for
growing embryo

Same genes ON in adult
fuel cancer growth
"Colon cancer 2" by Emmanuelm at en.wikipedia. Licensed
under CC BY 3.0 via Wikimedia Commons –

"Tubal Pregnancy with embryo" by Ed Uthman, MD
(Flickr, Wikipedia)
©2010 MFMER | slide-10

“Epigenetics”
• Regulation of gene expression without change
in DNA sequence
• Open or close chromatin by modifying histones
• Example: Histone acetylation
Less transcription

More transcription

http://missinglink.ucsf.edu/lm/genes_and_genomes/acetylation.html
©2010 MFMER | slide-11

Epigenomic Targets in Cancer Therapy

Hamm and Costa (2015) Pharmacology and Therapeutics
DNMT, DNA methyltransferase; HDAC, Histone de-acetylase
©2010 MFMER | slide-12

Epigenetic Biomarkers of Cancer
Patient symptoms
Need for
biomarker

Precursor cell

Tumor cell

Early cancer

Invasive cancer

Carcinogenesis
Screening
Early detection
Cancer diangosis
Tumor profiling
Therapeutic option
Prognosis

Application of
DNA
methylation

Delpu, et al. (2013) Int. J. Mol. Sci.
©2010 MFMER | slide-13

Applied Epigenetics: Colorectal Cancer
(CRC) Screening
• Remains a major killer worldwide, #2 in USA
• Shift toward right colon
• Now ~50% R-side in US
• Olmsted county: ~60% R-side

• Conventional screening tools
• Reduce mortality
• Underutilized
• Biased towards left-side

• Imperative to improve
Siegel et al. CA Cancer J Clin 2015;65:5
Gupta et al. Clin Gastroenterol Hepatol. 2005;3:150
Shapiro et al. Cancer Epidemiol Biomarkers Prev. 2012;21:895

14
©2010 MFMER | slide-14

Stool DNA Testing
Biological Basis

• Exfoliation
• Abundant
• Continuous
• Cancer > normal
• DNA as marker
• Signature changes
• Stable
• Amplifiable

Muco-cellular layer
Cancer

Normal

Ahlquist et al. Hum Pathol 2000

Key Technological Advances
• Preservative buffer, assay sensitivity, marker selection

DNA mutations are
too heterogeneous

Olson et al. Diagn Mol Pathol 2005;14:183
Ahlquist et al. Ann Int Med 2008;149:441
http://thebluenose.com/blog/2015/02/03/
the-cell-a-molecular-approach-downloadfree/
©2010 MFMER | slide-16

Key Advances: DNA Methylation Markers

Just 4 DNA methylation
markers in tissue
samples detect:
•100% colon cancers
•96% advanced polyps
•100% specificity

Zou et al. Clin Chem 2012; 58:375
©2010 MFMER | slide-17

Multi-target Stool DNA Test

Optimized & Automated
• Simple device for
collection & mailing
• Preservative buffer

• Targets multiple markers



Methylated BMP3 & NDRG4
Mutant KRAS
β-actin (human DNA)
Hemoglobin (FIT)

• Sensitive multiplex DNA assay (QuARTS)

Multi-target Stool DNA Test
Addresses all elements for effective detection
• High sensitivity for CRC & greatest-risk
precancer
Sensitivity
• Unaffected by tumor site
• Operator independent (automated)
• Noninvasive
• No cathartic preparation
Compliance
• No diet or medication restriction
• Off-site collection
Access
• Widely accessible

DeeP-C study
•10,000 asymptomatic patients,
•91 North American centers,
•Multi-target stool DNA prior to blinded colonoscopy
N Engl J Med 2014; 370:1287-97

CRC Sensitivity by Stage
MT-sDNA 94%
FIT
70%

N 29

29

21

MT-sDNA

21

10

10

NO Difference between
left and right sided CRC

60

60

4

4

CC - 21

Advanced Adenoma Sensitivity by Size
MT-sDNA

Highest risk to
progress

N

66

66

577

574

79

79

38

38

DNA Methylation Beyond the Colon…
• Highest mortality for GI cancers
• 50,000 CRC
• 100,000 foregut cancers combined

• Majority are de novo
• No population screening upper GI cancers
• Symptoms = advanced stage disease
• Novel applications for powerful marker class?

©2010 MFMER | slide-23

Pancreas Cancer: A Critical Need
• Pancreatic cancer (PanC): #2 cancer killer by 2020
• Hypothesis: DNA sequencing would identify novel
and accurate candidate methylation markers
• Aim: Test marker discrimination of PanC from
• Benign pancreatic tissues
• Other gastrointestinal epithelia

©2010 MFMER | slide-24

Preliminary Data in Pancreas Cancer:
Discovery, technical validation
• Discriminant detection (Pancreas cancer vs normals)
• >500 markers met selection criteria (top 38 shown)
• Many not previously reported in cancer biology

Kisiel et al. DDW 2013
©2010 MFMER | slide-25

Clinical Pilot in Pancreatic Juice Aspirates

Raimondo et al. DDW 2013
©2010 MFMER | slide-26

Extended Implications of Sequencing Data
• Many strong DNA methylation markers without
known cancer role
• Not reported in colon cancer
• Not reported in pancreatic cancer
• Not reported in other upper GI tract cancers

• Could new markers predict cancer type?

©2010 MFMER | slide-27

Tumor Site Prediction by DNA Methylation

Training set

Validation set
Kisiel et al. DDW 2013
©2010 MFMER | slide-28

Summary of Site Prediction Accuracy
Predicted by Model
UGIC
LGI*
Normal
UGI Cancer (66)
Pancreas Cancer (38)

35
13
10

0

3

0

6

0

0

Colon Cancer (20)

1

0

Colon Adenoma (18)

2

19
14

Pancreas Normal (15)

0

0

Colon Normal (20)

0

0

Buffy Coat Normal (10)

0

0

15
20
10

Esophagus Cancer (18)
Stomach Cancer (10)

LGI* (38)
2

Controls (45)

* LGI = CRC + Adenoma > 1 cm

}
}
}

88%

(58/66)

87%

(33/38)

100%

(45/45)

Χ2 p < 0.0001
Kisiel et al. DDW 2013
©2010 MFMER | slide-29

Extended Implications of Sequencing Data
• Many strong DNA methylation markers without
known cancer role
• Not reported in colon cancer
• Not reported in pancreatic cancer
• Not reported in other upper GI tract cancers

• Could new markers predict upper vs lower GI
cancers?
• Could sequencing other cancers give us
anatomic location of primary cancer?

©2010 MFMER | slide-30

Cornerstone Project
Anatomic Site

Lesion Category

RRBS

Validation
by qMSP

Oropharynx, lung

Tonsil, tongue
Tracheal, bronchial

In progress

Partial

Esophagus

Normal
Barrett’s (+/- dysplasia
Adenocarcinoma, Squamous carcinoma

Complete

Complete

Complete

Stomach

Normal
Adenoma
Adenocarcinoma

Complete

Complete

In progress

Pancreas

Normal colon, WBC, Benign pancreas
Ductal adenocarcinoma

Complete

Complete

Complete

Pancreas precursors

Main, branch duct IPMN
PanIN I-III

Complete

In progress

Bile duct

Normal bile duct, liver
Intra-hepatic CCA, Peri-hilar CCA

Complete

iCCA complete
eCCA in-progress

Liver

Hepatoma
Cirrhotic, NASH controls

In progress

Small bowel

Normal
Adenocarcinoma

In progress

Colon

Normal
SSA > 1cm, Adenoma > 1 cm, CRC

Complete

IBD

Colitis
LGD, HGD, CRC

In progress

Stool

Normal stools

Complete

Peripheral blood

Buffy coat, plasma, cellular compartments

In progress

Complete

Clinical
pilot

In progress


Complete

©2010 MFMER | slide-31

Site-specificity: Expanding the scope
• Best markers from each site (95)
• Simultaneous multi-organ biological validation
• Controls (119 tissue samples)
• Neoplasia (266 tissue samples)

• Analysis



Cancer vs Normal
Upper vs CRC
Pancreatico-biliary vs gastroesophageal
Combined model of entire GI tract

©2010 MFMER | slide-32

Results: High Accuracy for GI Neoplasm
Chr12.133

Predicted by model
Neoplasia Control

BMP3
Any Neo
230/5

Any Neo
10/4

Neoplasia 253

13

0.049

Control

111

0.067

Accuracy

Any Neo Chr11.123
15/1

Error

8

0.945

CTRL
11/109

©2010 MFMER | slide-33

Results: Upper GI vs Colorectal Neoplasm
Chr7.258

Predicted by model
Colorectal

Upper

Error

Colorectal 89

8

0.082

Upper

167

0.012

QKI
Upper
133/4
Upper
30/5

Colorectal
6/88

Accuracy

2

0.944

©2010 MFMER | slide-34

Pancreaticobiliary vs Gastroesophageal Cancer
PDGFD

Predicted by model
Pancreatico- Gastrobiliary
esophageal

Error

Pancreaticobiliary

74

10

0.049

Gastroesophageal

7

78

0.067

ELOVL2
Pancreatico
-biliary

70/4
Gastroesophageal

1/70

Pancreatico
-biliary

8/0

PCPB3

Accuracy

0.941

Gastroesophageal

5/11

©2010 MFMER | slide-35

Overall Model
Predicted by model
Tissue
source

CTRLs

Colorectal
neoplasia

Pancreatico- Gastrobiliary cancer esophageal
neoplasia

CTRLs

116

9

10

2

Colorectal
neoplasia

1

86

1

7

Pancreatico- 1
biliary cancer

0

65

4

Gastroesophageal
neoplasia

2

8

72

1

Overall accuracy

0.88, P<0.0001, chi-square
Kisiel, et al. AACR 2015

©2010 MFMER | slide-36

Non-invasive Test: Proof of Concept
• 2 methylated markers (BMP3, QKI)
• Non-optimized assays
• 2 mL of blood plasma
Predicted by model
Tissue source

Controls

Colorectal neoplasia

Pancreatic cancer

Error

Controls (n=14)

12

0

2

0.14

Colorectal neoplasia (n=14)

4

8

2

0.43

Pancreatic cancer (n=14)

3

0

11

0.21

Overall accuracy

0.74 (p<0.0001)

©2010 MFMER | slide-37

Implications for the Future
• Stop Single Organ Screening
• Screen multiple cancers in a single test
• Aggregate incidence outweighs organ incidence
• The PERSON as the new unit of analysis
• Stool test: cancers of the GI tract, lung, ENT
• Blood test: all cancers

©2010 MFMER | slide-38

Conclusions
• Epigenetics a rich field in human disease
biology
• Important implications for prevention, tumor
diagnosis, and therapy
• The future is NOW: stool DNA for CRC
screening
• Novel diagnostic applications:
• Distinguish cancer from normal tissue
• Accurately classify GI tumors by location

©2010 MFMER | slide-39

Acknowledgements
• Jack and Maxine Zarrow Family Foundation of Tulsa Oklahoma
• Paul Calabresi Program in Clinical-Translational Research (NCI
CA90628).
• Carol M. Gatton endowment for Digestive Diseases Research.
• Mayo Clinic SPORE in Pancreatic Cancer (P50 CA102701),
• Lustgarten Foundation for Pancreatic Cancer Research
• Clinical Core of the Mayo Clinic Center for Cell Signalling in
Gastroenterology (P30DK084567).
• Reagents and RRBS sequencing costs were provided by Exact
Sciences (Madison WI)

©2010 MFMER | slide-40

GI Cancer Molecular Diagnostics Lab

©2010 MFMER | slide-41

Sign up to vote on this title
UsefulNot useful