This talk outlines the general steps for project management in rapid development (design to data in two weeks) of a novel digital PCR assay to validate and quantify low frequency variants discovered by sequencing (NGS) of a targeted comprehensive cancer gene panel by the Ion Torrent PGM on PDX models of metastatic colon cancer and spheroid (3-D) cell cultures. Goal: If the potential driver mutation is validated, treat both (PDX model & cell culture) with small molecule drugs, investigate coincident response.

1. Assay Development in Digital PCR
Orthogonal validation of low-frequency variants
discovered by NGS.
4/4/2017
Kirsten Copren, Ph.D.
Genome Analysis Core

2. Introduction
 Feasibility/Protocol Review
 Type of Assay
 Type & Quality of Sample
 Level of Detection/Sensitivity
 Precision/Accuracy
 Replication/Technical Variation/Reproducibility
 Timeline
 Staffing
 Cost
Project Management Workflow
4/4/20172 Assay Development in dPCR

3. Type of Sample
PDX Models of metastatic colon cancer and spheroid (3-D) cell cultures
Goal: treat both with small molecule drugs, investigate coincident response.
Dr. Robert Warren brought us a potential gene target to validate.
4/4/2017Assay Development in dPCR3

4. Project:
NGS Results. Comprehensive Cancer Panel. Ion Torrent PGM. Data from
SANOFI. Identified two variants in 23 tumor samples <10%, false positives or
real mutations?
4/4/2017Assay Development in dPCR4

5. Gene TSC1, Tuberous sclerosis 1 (TSC1), tumor growth
suppressor. Not previously found in colon cancer.
If validated, single assay could screen thousands of samples.
Gene
Chromosome
Location
Base Change
(Forward Strand)
Codon Change
(Reverse Strand)
Amino Acid
Change
TSC1 g.chr9:135772640 A>T c.(2905-2907)TTG>TAG p.L969*
TSC1 g.chr9:135781427 G>A c.(1537-1539)CCA>CTA p.P513L
4/4/2017Assay Development in dPCR5
Sample ID Sample Type Mutation
Sequencing Mutation
Rate
40006 FFPE Tissue L969* 7.2%
727 Frozen Tissue P513L 9.7%

6. Level of Detection/Sensitivity
 NGS and Cancer
• Ion Torrent PGM first sequencer that
clinically validated a Comprehensive
Cancer Panel (400 genes)
‒ Validated a variant in a tumor as 10%
of sample in a tumor/normal
comparison. Clinical approval
accepted.
‒ Oncologists want to detect down to
the first occurrence of a mutation
(0.1%).
‒ NGS is sensitive, but also has a high
error rate. Accuracy below 10%
begins to interact with false positives.
Research (Rx) accepts higher variation than Diagnostics (Dx)
4/4/20176 Assay Development in dPCR

7. Digital PCR
4/4/2017Assay Development in dPCR7
• dPCR counts individual DNA molecules to
allow absolute quantitation without the need
for a standard curve.
• Sensitive to 1-10% rare allele detection, can
be optimized down to detect rare mutants at
a prevalence of 0.1%.
• Taqman® chemistry with specific dye-labelled probes
(Vic or Fam) are used to detect sequence-specific
targets, e.g. WT or mutant.
• Following PCR each well is individually analyzed to
detect presence or absence of an end-point signal.
• The fraction of negative reactions is used to generate an
absolute count of the number of target molecules in the
sample.

8. Assay Design:
Gene
Chromosome
Location
Base Change
(Forward Strand)
Codon Change
(Reverse Strand)
Amino Acid
Change
TSC1 g.chr9:135772640 A>T c.(2905-2907)TTG>TAG p.L969*
TSC1 g.chr9:135781427 G>A c.(1537-1539)CCA>CTA p.P513L
4/4/2017Assay Development in dPCR8
Primer Design for P513L Mutation in TSC1 (Forward Strand)
Sequence Length
T
M
Forward GGCTGCCGAGTGGGTCTT 18 60
Reverse TGACTCTCCCTTTTACCGAGACA 23 59
Probe 1 CTGAGAACCTAGGAGA 16 65
Probe 2 CGCTGAGAACCTGGGA 16 66
Primer Design for L969* Mutation in TSC1 (Reverse Strand)
Sequence Length
T
M
Forward CCAGGTGTTTGAATTGGAGATCT 23 58
Reverse CCCAAGGTCATGAATCAGTTCTT 23 58
Probe 1 ATTTATATGGCAGGTAGG 18 65
Probe 2 TGGCAGGTTGGAGAA 15 67
1. Obtain 150 bp flanking sequence (300 bp)- genome.ucsc.edu
2. Mask repetitive elements: Repeatmasker.org
3. Mask SNPs: NCBI BLAST
4. Default QPCR parameters: Primer Express 3.0
5. Verify amplicon: In-Silico PCR- Genome Browser

9. Synthetic Oligos: Positive Controls
 500bp gBlocks mimic the WT and Mut gene sequences
 Feasibility testing
1. Specificity: QPCR using gBlocks
2. Specificity & Sensitivity: dPCR using gBlocks
3. Validation: Test samples.
4. Limits of detection & technical variation.
gBlocks® Gene Fragments are double-stranded, sequence-verified genomic blocks for affordable
and easy gene construction or modification. IDTDNA
4/4/2017Assay Development in dPCR9

10. Results: Specificity Using QPCR.
4/4/2017Assay Development in dPCR10
Wildtype Allele - T Mutant Allele - A
Sample Name Expected Observed Expected Observed
Wildtype
gBlock
+ + - -
Mutant
gBlock
- + + +
NTC - - - -
Wildtype Allele - T Mutant Allele - A
Sample Name Expected Observed Expected Observed
Wildtype
gBlock
+ + - -
Mutant
gBlock
- - + +
NTC - - - -
(a) L969*: Passes (b) PL513:Fails. Off target amplification
Scaling
considerations

11. Results: Specificity Using dPCR
copies/µL Precision Allele Frequency
Sample
Copy
Input
No. chips
for
analysis
Wildtype
T
Mutant
A
Wildtype
T
Mutant
A
Wildtype
T
Mutant
A
Wt gBlock 20000 1 1089.20 3.03 1.96% 33.10% 99.72% 0.28%
Mut bBlock 20000 2 8.54 954.42 12.82% 1.43% 0.89% 99.11%
50Wt/50Mu
t
20000 2 539.68 458.17 1.72% 1.83% 54.08% 45.92%
90Wt/10Mu
t
20000 1 946.13 95.25 1.99% 5.23% 90.85% 9.15%
NTC 0 1 3.45 2.08 30.90% 41.41% 62.37% 37.63%
(a) L969*: Passes
4/4/2017Assay Development in dPCR11
The precision indicates the probability of a false positive, and values <10% are desirable.
Background
10% Mutant
100% WT

12. Validation: Test Sample
copies/µL Precision Allele Frequency
Sample Copy Input
No. chips
for analysis
Wildtype T Mutant A Wildtype T Mutant A Wildtype T Mutant A
40006 20000 2 2225.90 3.61 1.28% 19.87% 99.84% 0.16%
(a) L969*: Mutation not detected
4/4/2017Assay Development in dPCR12
The precision indicates the probability of a false positive, and values <10% are desirable.
Sample ID Sample Type Mutation
Sequencing Mutation
Rate
40006 FFPE Tissue L969* 7.2%
727 Frozen Tissue P513L 9.7%

13. Reproducibility
copies/µL Precision Allele Frequency
Mixture
Date
Diluted
Date Run
Copy
Input
No. chips
for
analysis
Wildtype
T
Mutant A
Wildtype
T
Mutant A
Wildtype
T
Mutant A
1 11/7/2014 11/7/2014 20000 1 946.13 95.25 1.99% 5.23% 90.85% 9.15%
2
11/10/2014 11/10/2014 20000 1 1291.50 94.72 1.84% 5.21% 93.17% 6.83%
11/10/2014 11/14/2014
20000 2 1331.90 93.36 1.30% 3.71% 93.45% 6.55%
11/10/2014 11/14/2014
10000 2 636.57 48.63 1.68% 5.41% 92.90% 7.10%
3
11/14/2014 11/14/2014
20000 2 1101.70 36.38 1.35% 5.93% 96.80% 3.20%
11/14/2014 11/14/2014
10000 2 584.77 33.38 1.65% 6.24% 94.60% 5.40%
(a) L969* - 90% wildtype and 10% mutant gBock mixture
4/4/2017Assay Development in dPCR13
The precision indicates the probability of a false positive, and values <10% are desirable.

14. Sensitivity
copies/µL Precision Allele Frequency
Dilutio
n
Date
Diluted
Date Run
Copy
Input
No.
chips for
analysis
Wildtype
T
Mutant
A
Wildtype
T
Mutant
A
Wildtype
T
Mutant A
1 11/14/2014 11/14/2014 20000 2 1102.30 33.82 1.38% 6.27% 97.02% 2.98%
L969*: 95% wildtype and 5% mutant gBock mixture
4/4/2017Assay Development in dPCR14

15. Conclusions
 One assay failed: PL513
 One assay confirmed mutation in NGS was false positive: L969*
 Quick protocol. Quick pilot.
 Can build cost model based on probability of potential failures.
 Successful pilot. Low-cost ($200 for probes). Two week demo period. Vendor
collaboration.
 High DNA input for lower sensitivity. Not appropriate for all sample types.
 Lowest levels of sensitivity not reached. Require additional testing.
4/4/2017Assay Development in dPCR15

16. Re-visit
 Feasibility/Protocol Review: Very good
 Type of Assay: QPCR SNP
 Type & Quality of Sample: FFPE & Fresh Frozen Tissue
 Level of Detection/Sensitivity: 5-10%
 Precision/Accuracy: Acceptable. Not all assays will pass.
 Replication/Technical Variation: Requires additional testing.
 Timeline: 2 weeks
 Staffing: Highly skilled
 Cost: $200 for probes. Successful vendor collaboration.
Project Management Workflow: Successful collaboration. Funded grant.
4/4/201716 Assay Development in dPCR

17. Genome Core:
Kathryn Thompson, B.A., Geneticist,
CareDx
Jennifer Dang, B.S., Geneticist,
CureSeq
Cassandra Adams, Ph.D., Staff
Scientist
Department of Surgery:
Robert Warren, M.D., Professor &
Chief of Surgical Oncology
David Donner, Ph.D., Professor
Mary Matli, M.S., Staff Research
Associate III
Team
4/4/2017Assay Development in dPCR17