• Technology pipeline for methylation biomarker development • High throughput DNA methylation-qPCR workflows • Liquid biopsy – cfDNA methylation testing

1. High through-put DNA methylation analysis of
LungCancer - plasma cfDNA for biomarker
development
From tissue based discovery to cell free DNA methylation testing
Andreas Weinhaeusel, Ram Vinay Pandev, Walter Pulverer, Matthias Wielscher, Lisa
Milchram, Silvia Schönthaler, Gabriel Beikricher, Manuela Hofner, Christa Noehammer,
Klemens Vierlinger and Stephan Pabinger
AIT Austrian Institute of Technology, Vienna Austria
5th qPCR and Digital PCR Congress
4.-5. Dec 2017, London- UK

2. AIT Austrian Institute of Technology
➢ Largest Applied Research Institute in Austria
➢ Employees > 1150
➢ 8 Centers
214.12.2017

3. AIT Austrian Institute of Technology
AIT-Molecular Diagnostics

4. Biomolecules
• DNA
• Genotyping
• DNA-methylation
• Genomic Aberrations
• RNA
• Gene Expression
• miRNA
• ncRNA
• Protein
• auto antibody profiling
• DNA Microarray
• whole genome
• targeted
• Next Generation Sequencing
• qPCR
• Microfluidic qPCR
• Protein Arrays
• Peptide Arrays
• Biostatistics
• Bioinformatics
• Software Development
Technologies
Biomarker technology expertise
Thyroid CancerBreast Cancer Lung Cancer Leukemias
Infectious
Disease
Fibroprolif.
Diseases
Indications
Thyroid CaBreast Cancer Lung Cancer LeukemiasColon Cancer
Fibroprolif.
Diseases
AIM: improving minimal invasive diagnostics
Prostate Ca
DNA METHYLATION
auto antibody profiling

5. Cancer Statistics - Europe
Incidence (newly diagnosed/y) Mortality (deaths / y)
➢ 3.2 mio new cancers per year; 1.7 mio deaths from cancer
➢ ~ 30% of all deaths; ageing society

6. Source: WHO statistics - cancer death per year (2012)
http://www.who.int/mediacentre/factsheets/fs297/en/ NSCLC
≈ 87%
38%
38%
11%
13% squamous cell lung
carcinoma
adenocarcinoma
large-cell lung
carcinoma
small-cell lung
cancer (SCLC)
Lung Cancer - Facts (1)
Lung
1370000
Stomach
736,000
Liver
695,000
Colorect.
608,000
Breast
458,000
Cervival
275,000
others
3458000

7. Lung Cancer - Facts (2)
▪ Very high mortality rate
▪ Five-year survival rate: 20%
▪ Late diagnosis – causes high
mortality
▪ Laborious, uncomfortable and
insensitive diagnostic tools
Chest X-ray (78,3% sensitivity)
 Early diagnosis improves survival rates

8. techniques
exposure to
radiation
invasive sensitivity
chest X-ray ✓ detects only large
tumours
CT scan ✓ ✓ -
PET ✓ ✓ -
Bronchoscopy ✓✓ -
cytology (performed on lavage fluid
obtained during bronchoscopy)
✓✓ 43%
biopsy ✓✓✓ 90%
(but too invasive)
 Need for minimal invasive techniques with increased sensitivity/specificity
Lung Cancer Diagnosis – State of the Art

9. Lung Cancer - Molecular Diagnostic Tests
914.12.2017
Company Application Analyte Drawback Conclusion
Orion
Genomics
(US)
screening test DNA methylation
in sputum
specificity/sensitivity
not published
not yet fully
developed
Epigenomics
(Germany)
diagnostic test DNA methylation
of SHOX2 using
Bronchial Lavages
Bronchoscopy/
anaesthesia is
required to perform
test
Blood based
test not yet
commercially
available
Qiagen
(Netherlands)
companion diagnostics
of an anticancer drug
for NSCLC patients
mutation status of
EGFR in plasma
limited to a specific
use (identifying
NSCLC patients
sensitive to
anticancer drug)
Test for therapy
stratification
NOT for
diagnosis
Oncimmune
(UK)
early detection of lung
cancer
autoantibodies in
blood
low sensitivity (41%) not suited as
screening

10. • methylation targets = CpG dinucleotides; clustered in CpG Island
• 88% of human genes have a CpG islands - associated promoter (Kim,TH et al., 2005 Nature)
• human genome: 28000 CpG islands
Cross-species Tissue specific mCpG-patterns
mouse mouse mouse
brain liver spleen
Irrizary et al. 2009, NatGen
mC - „the 5th base“
➢ heritable pattern
➢ tissue specific / cross-species
➢ chemically as stable as DNA
➢ early event in cancerogenesis
➢ detectable in cell free serum
➢ PCR based detection
 DNA methylation = „ideal“ biomarker
DNA-Methylation Biomarkers

11. DNA methylation biomarker development
for minimal invasive diagnostics
„plasma“ markers early diagnosis, monitoring patients under therapy
Tissue
• Screening: malignant – bening – healthy / blood
Tissue
• Candidate marker – confirmation (include blood DNA)
• qPCR
PLASMA
• evaluation of markers for minmal invasive testing
• 10ng cfDNA / ml – background = blood DNA
• ≤1% is TU derived

12. Bisulfite
conversion
WGA
Hybridisation on methylation
specific probes
Assay for Discovery: Infinium 450/850k array

13. GENOME-wide discovery - Lung tissues
Illumina 450k methylation analyses
bioinformatic analysis for biomarker-
selection
univariate - cut off: adj. p-value < 0,005 &
methylation difference >7.5%; & AUC >0,85
multivariate – using class prediction analyses
LuCa, 18
IPF, 30
HP, 8
COPD, 42
normal
LU tissue,
34
blood
DNA, 24
Lung cancer (n=18) - AdenoCa (9), SqCCL.(9)
IPF - Idiopathic pulmonary fibrosis (n= 30)
HP - Hypersens. pneumonia (n= 8)
COPD (n=42) - GOLD 0-3, - n = 8-13 per group
Normal lung tissue (n=34)
Healthy blood DNA (n=24)
n=156

14. Methylome – e.g. Lung disease / tissue
5% (24300) most variant probes plotted
DNA methylation changes: HIGH  Cancer > Fibrosis >> COPD > LOW
CNV & chromosomal structural aberrations exclusively in cancer
Fibrosis (n=30)
Cancer (n=18)
COPD (n=42)
healthy (n=34)
1) QC - Picogreen DNA quantif
2) SO3 deamin  qMSP – 4 markers …GNAS-ME, GNAS-UM, NESP
3) Illumina 450k (now 850k EPIC) protocol
4) Bioinformatics

15. Assay for Validation: MSRE* HTqPCR
*methylation sensitive restriction enzyme
Advantages:
• Low input: < 10ng DNA (<1ml of plasma)
• No bisulfite conversion
• Multiplexing: 48-96plex
• useful for plasma cfDNA qPCR testing

16. HT-PCR methylation assay design tools
Pandey, et al. (2016). MSP-HTPrimer: a high-throughput primer design tool to
improve assay design for DNA methylation analysis in epigenetics. Clin.
Epigenetics 8, 101.
https://sourceforge.net/p/msrehtprimer/wiki/MSRE-HTPrimer_Manual/
Pandey, R.V., et al. (2016). MSRE-HTPrimer: A high-throughput and genome-wide
primer design pipeline optimized for epigenetic research. Clinical Epigenetics 8, 26.

17. IonTorrent – TDBS* confirmation of array data
*…targeted deep bisulfite amplicon sequencing
FP7: EurHealthAgeing
➢ 62 bisulfite PCRs established & bisulfite sequencing conducted on IonTorrent technology used
➢ implementation of analyses workflow for TDBS-analyses - use IonT Mapper in BISMARK  TMAP
o Pabinger et al. (2016): Analysis and Visualization Tool for Targeted amplicon Bisulfite Sequencing on Ion Torrent
sequencers; PLoS One. 2016 July 28;11(7).
Setup (384 well –MSP) Workflow
▪ 40 samples
▪ 62 targets
Correlation 450k array vs TDBS
Median correl: 0.91
DNA
Bisulfite
conversion
Target
enrichment
Pooling of
targets per
sample
Barcode
and adapter
annealing
Sequencing
microarray data
sequnecingdata

18. MSREqPCR-Confirmation
LungCa-tissue-methylation marks
▪ Design – pipeline: 276 different qPCRs assays  222 working assay
▪ optimization of 48-plex a/o 96-plex
▪ qualification according MIQE guidelines

19. 2114.12.2017
Established assays WITH & WITHOUT PreAmplification
PCR- efficiency: 80-120%
PCR Perfomance – upon Pre-Amplifikation (48x)

20. Results – MSREqPCR (96x) vs 450k chip- methylation
156
samples
healthy
blood (24)
lung tissue
(34)
diseased
lung tissue
cancer(18)
IPF (30)
HP (8)
COPD (42)
222 assays x 156 Tissue DNA samples  65% confirmed
 best 96 selected for cfDNA testing

21. cfDNA methylation test analysis workflow

22. cfDNA testing (400µl plasma): Cancer vs other
Healthy N=61 (29.9%) COPD N=42 (20.5%)
healthy n=27 (13.2%) GOLD 1-2 n=31 (15.1%)
COPD at risk n=34 (16.6%) GOLD 3-4 n=11 (5.4%)
Cancer N=33 (16,1%) Fibrosis N=68 (33.3%)
low stage (I-II) n=9 (4.4%) UIP n=27 (13.2%)
High stage (III-
IV)
n=15 (7.3%) NSIP n=11 (5.4%)
11 AdCa, 8 SqCC, 7 SCLC, 7 LCLC HP n=22 (10.7%)
correct classification:
Wielscher M., et al.: EBioMedicine 2
(2015) 927–934

23. 14.12.2017 25
Class prediction
(lasso penalized logistic regression - from R package glmnet)

24. 14.12.2017 27
23 lung cancer vs 23 healthy controls
Validation-set: n=46 independent cfDNA samples

25. cfDNA MSREqPCR workflow
Genome wide (e.g. 450k / 850k array)
– tissue
3x96-plex confirmation -
tissue
96 plex transfer to
plasma
48plex assay
further clinical
validation & applications

26. Summary
▪ MSRE-qPCR
▪ enables confirmation, validation and classifier-refinement
• works with bisulfite based discovery
• extremely low LOD, able to detect 0,1 – 1% of tumor DNA in 10ng of cfDNA
▪ efficient tool (costs & time) for qualification of methylation markers
and applicabel for plasma cfDNAmethylation testing
▪ Lung-Cancer Methylation markers
▪ new biomarkers discovered in genomewide methylation screening
▪ biomarkers were validated with MSRE-qPCR in plasma testing with AUC
calssifiers ≈0.8-0.95
• plasma samples up to 15y old are useful (age had no effect)

27. Potential use & need for further validation
▪ alternative to „wait and watch“ approach:
▪ CT-scan „suspicious“ - when no biopsy is feasible
▪ early-diagnostics
▪ avoid biopsy in CT-false positives
▪ early cancer detection in at risk persons: heavy smokers or IPF & COPD
patients == alternative to CT-scan
▪ screening test – requires validation in prediagnostic patient samples – serial
samples collected some years before diagnosis
▪ therapy monitoring
▪ patient cohorts in clinical studies or in standard treatment
▪ therapy response prediction ?

28. ….The technologies are available…
cooperations & partnering welcome….
analyse 92 markers
from 1µl of serum
Oncology
Cardio-Vascular / CVD
Inflammation
Biomarker Technology platforms

29. Acknowledgement
DNA Methylation work
Matthias Wielscher, Manuela Hofner, Elisabeth Reithuber, Gabriel Beikircher, Walter Pulverer, Christa Noehammer
Immunoprofiling work: Regina Soldo, Peter Hettegger, Istvan Gyurjan, Johana Luna, Stefanie Brezina, Roman Kreuzhuber,
Lisa Milchram, Sandra Rosskopf, Ronald Kulovics, Gabriel Beikircher, Silvia Schönthaler, Johannes Söllner, Michael
Stierschneider
Assay design pipelines and bioinformatics
Ram Vinay Pandey, Stefan Pabinger, Fabian Schröder, and Klemens Vierlinger
Clinical Cooperations:
Funding & Support:
T. Liloglou, J. Field; Roy Castle Lung Cancer Foundation,
Liverpool, UK
M. Hajduch, Molecular and Translational Medicine, University
Hospital Olomouc, Czech Republic
C. Singer, A. Gsur. & P. Hofer, R. Ziesche
Medical Univ. Vienna
F. Längle, J. Hofbauer, LKH Wr. Neustadt
G. Leeb & K. Mach, LKH Oberpullendorf
C. Jungbauer, Austrian Red Cross
J. Söllner, Emergentec