1. Comprehensive Genomic Profiling of Breast Cancer By Massively Parallel Sequencing Reveals New Routes To Targeted Therapies
JS Ross, CE Sheehan A Parker, M Jarosz, S Downing, R Yelensky, D Lipson, P Stephens, G Palmer, M Cronin
Department of Pathology and Laboratory Medicine
Albany Medical College
Albany, NY
Foundation Medicine, Inc.
Cambridge, MA

2. Targeted Therapies for Cancer
Molecular profiling is driving many new targeted cancer therapeutics
~500 compounds hitting ~140 targets in development
Growing number of newly identified potential targets
All – this is the source data for slide 6, just for your reference. Just in case anyone asks about the chart, here’s the methodology: we took the list of targets on the old slide and combined them with a list of targets from Phil, which yielded 141 unique targets, then used the BioCentury Online Intelligence Database to find out how many candidates for those targets are in the pipeline for development industry-wide. Of the 141 targets we looked at, 97 have products in development or on the market and there were a total of 521 compounds.  
Subset of analyzed targets listed; data from BioCentury Online Intelligence Database

3. Background (1)
Next Generation DNA Sequencing (NGS) has recently been applied to FFPE cancer biopsies and major resections (Ross JS et al. J Clin Oncol 29: 2011)
Current Hot-Spot Genotyping only detects:
Mutations restricted to specific exons and codons
NGS detects:
Whole exome mutations in numerous cancer related genes
Insertions and deletions
Translocations and fusions
Copy number alterations (amplifications)

4. Background (2)
Biomarker testing has driven the selection of therapy for breast cancer for longer than any other solid tumor
ER/PR testing introduced in 1971
HER2 testing/trastuzumab approved in 1998
Oncotype DxTM mRNA profiling in 2004
Currently, “hot-spot” DNA sequencing is driving the selection of targeted therapies for other solid tumors, but not for breast cancer:
EGFR genotyping for tyrosine kinase inhibitor use in NSCLC in 2005
KRAS genotyping for anti-EGFR antibody use in CRC in 2007
BRAF genotyping for BRAF inhibitor use in melanoma in 2011
The emergence of comprehensive genomic profiling by NGS has led investigators to question whether more thorough gene sequencing techniques could discover potential targets for the treatment of metastatic breast cancer not currently searched for in current routine practice

5. Design
DNA was extracted from 4 x 10 m FFPE sections from an initial study-set of 15 primary invasive ductal breast cancers
The exons of 145 cancer-related genes were fully sequenced using the Illumina HiSeq 2000 (Illumina, San Diego, CA) and evaluated for point mutations, insertions/deletions (indels), specific genomic rearrangements and copy number alterations (CNA)
A total of 606,676-bp content was sequenced and selected using solution phase hybridization, to an average coverage of 253×, with 84% of exons being sequenced at ≥100× coverage
The NGS assay captures and sequences 2,574 coding exons representing 145 cancer-relevant genes (genes that are associated with cancer-related pathways, targeted therapy or prognosis), plus 37 introns from 14 genes that are frequently rearranged in cancer
To maximize mutation-detection sensitivity in heterogeneous breast cancer specimens, the test was validated to detect base substitutions at a ≥10% mutant allele frequency with ≥99% sensitivity and to detect indels at a ≥20% mutant allele frequency with ≥95% sensitivity, with a false discovery rate of <1%

6. Cancer Genome Profiling Workflow
QC Sample, Isolate DNA, QC DNA
<14-21 days

7. Deep coverage is required for clinical grade samples
Increasing Coverage To 500x Allows For >99% Sensitivity To Detect Mutant Alleles >5%, With No False Positive Mutation Calls
Sensitivity vs Allele Frequency at 500X Coverage (1Mb panel) 5%
10%
Deep coverage is required for clinical grade samples

8. Lower Coverage Misses Relevant Mutations
The data is from the MSKCC study.
The set has 43 Lung, 26 CRC, and 3 Melanoma.
Includes known and likely somatic mutations (both Subs and Indels)
0% 0 0%
5% 18 11%
10% 36 21%
15% 19 11%
20% 20 12%
25% 20 12%
30% 16 9%
35% 12 7%
40% 3 2%
45% 7 4%
50% 6 4%
55% 4 2%
60% 3 2%
65% 1 1%
70% 2 1%
75% 1 1%
80% 1 1%
85% 0 0%
90% 0 0%
95% 0 0%
100% 0 0%
Mutant Allele frequency spectrum of known mutations found in a series of clinical samples
Fraction of mutations <5%
Fraction of mutations <10%
Fraction of mutations <20%
Fraction of mutations <25%
Fraction of mutations <50%
Fraction of mutations <100%
11%
32%
55%
67%
93%
100%

9. Genomic Alteration Categories
Highly Actionable
“Page 1”
Category A:  Approved / standard alterations that predict sensitivity or resistance to approved / standard therapies
Category B:  Alterations that are inclusion or exclusion criteria for specific experimental therapies 
Category C: Alterations with limited evidence that predict sensitivity or resistance to standard or experimental therapies 
Category D: Alterations with prognostic or diagnostic utility
Category E: Alterations with clear biological significance in cancer (i.e. driver mutations) without clear clinical implications
Actionable in Principle
“Page 2”
Prognostic
“Page 3”
Biologically Significant
“Page 4”

10. Results (1)
15/16 (94%) of the tumors revealed 33 total somatic genomic alterations
Mean of 2.2 alterations per tumor with a range of 0 to 4 alterations per sample
Standard of care alterations consisted of 3(19%) tumors with HER2 copy number increases
The NGS HER2 copy number measurements by NGS in the HER2 amplified cases averaged 80% of the counted HER2 copies on FISH assessment of the same tumor block
Genes co-amplified with HER2 included RARA

11. Results (2)
10 (63%) of tumors harbored at least one alteration that potentially could have led to clinical trials of novel targeted therapies including copy number increases for:
IGF-1R in 2 (13%) tumors (IGF-1R inhibitors)
MDM2 in 1 (6%) tumor (nutlins)
CCND1 in 3 (19%) tumors (CDK inhibitors)
CCNE1 in 1 (6%) tumor (CDK inhibitors)
CDK4 in 1 (6%) tumor (CDK inhibitors)
FGF1R in 1 (6%) tumor (FGF1R inhibitors)
5 (31%) of tumors had 1 or more PIK3CA mutations (PIK3CA and mTOR inhibitors)
6 (38%) of tumors had alterations classically associated with adverse clinical outcome including:
TP53 and PTEN mutations
HER2 copy number increases.

12. Comprehensive Genomic Profiling of Breast Cancers (n=15)
Tumor Sample
Number of Alerations
Known and Likely Somatic Non-Synonymous Mutations (mutant allele frequency, sequence coverage depth) and Copy Number Gains (fold change over normal)
Potential Actionability
AB_1
None
AB_3 1
CDH1:NM_004360:c.2436_2439delTGAAG:frameshift(0.39,250)
AB_17 2
MEN1_c.207_207delC_p.D70fs*49(0.04,196), MEN1:NM_130801:c.1322G>A_p.W441*(0.24,202)
AB_19 4
SMARCA4_c.805delC_p.M272fs*31(0.03,95), TP53:NM_ :c.1028_1028delT:frameshift(0.24,365),
IRS2_gain(9x), IGF1R_gain(6x)
IGF-1R Inhibitors
AB_33
PTEN_c.370T>A_p.C124S(0.76,452)
AB_35
PIK3CA_c.3140A>G_p.H1047R(0.42,455), TP53_c.332T>C_p.L111P(0.49,299)
PIK3CA Inhibitors, mTOR Inhibitors
AB_49
TP53_c.809T>C_p.F270S(0.11,372)
AB_51 3
IGF1R_gain(18x), MDM2_gain(8x), CCND1_gain(4x)
IGF-1R Inhibitors, Nutlins, CDK inhibitors
AB_65
PIK3CA_c.3140A>G_p.H1047R(0.26,300), TP53_c.488A>G_p.Y163C(0.28,232)
AB_67
PIK3CA_c.3140A>G_p.H1047R(0.16,535), PIK3CA_c.316G>C_p.G106R(0.03,537)
AB_81
ERBB2_gain(6x), CCNE1_gain(3x)
Lapatinib, Trastuzumab, CDK Inhibitors
AB_83
CCND1_gain(4x)
5014A
TP53_c.752T>G_p.I251S(0.34,489),
ERBB2_gain(8x), MCL1_gain(5x), CDK4_gain(3x)
Lapatinib, Trastuzumab, CDK inhibitors
5016A
PIK3CA_c.3140A>T_p.H1047L(0.23,1549), TP53_c.396G>C_p.K132N(0.07,282), PAK3:NM_ :c.414G>A_p.M138I(0.06,889), LRP1B:NM_018557:c.11762C>G_p.S3921*(0.23,1528)
5018A
PIK3CA_c.1616C>G_p.P539R(0.47,603), PIK3CA_c.3140A>G_p.H1047R(0.53,782),
CCND1_gain(6x), FGFR1_gain(5x)
PIK3CA Inhibitors, mTOR Inhibitors, CDK inhibitors, FGFR1 inhibitors
Average:
2.2
Standard-of-care
Plausibly Actionable (in trials)
Resistance/Negative Predictors

13. Percentage Of Samples With Actionable Alterations Across Major Tissue Types (224 Total Cases)
71% cases carried ≥1 plausibly actionable alterations
32 % cases carried ≥2 plausibly actionable alterations
N= N= N= N= N=24

14. “Long Tail” Of Genomic Alterations Highlights Potential Benefits Of Comprehensive Profiling in Breast Cancer *

15. Novel Genomic Alterations
Novel Genomic Alterations* Discovered in Breast Cancer by NGS in an Expanded Cohort
Total Number of Sequenced Breast Cancers
Total Number of Novel Alterations
Types of Novel Alterations
327
7 (2%)
- Chromosomal Inversions (2)
In Frame Deletion (1)
Gene Truncation (1)
Chromosomal Rearrangement (1)
Tandem Duplications (2)
* Novel alterations discovered in tumor cell (somatic) sequence only as determined by comparison with the COSMIC database. Gene variants of undetermined significance which may represent germline variants are not included in this list.

16. HER2 Gene Copy Number Alteration Validation
ERBB2
RARA
Increased HER2 gene copies detected by NGS
HER2 FISH Positive Breast Invasive Duct Carcinoma Demonstrating High HER2 Copy Number
HER2 Protein 3+ Expression by IHC

17. Clinical Dilemmas Potentially Resolved by NGS
ER IHC+ with lack of benefit for hormonal therapy
ESR1 Mutation truncates estradiol binding site of the ER receptor protein
“Functional assay” is negative
HER2 IHC 3+ and FISH-
Activating mutation in the HER2 gene increases HER2 mRNA and HER2 receptor protein levels
No copy number increase

18. Conclusions
Deep massively parallel DNA sequencing of clinical breast cancer samples uncovers an unexpectedly high frequency of genomic alterations that could influence therapy selection for breast cancer
Deep sequencing of genomic DNA can provide a broad cancer-related gene survey at a depth of coverage that provides sensitive detection for all classes of genomic alterations, and when applied to breast cancer patients can reveal actionable genomic abnormalities that inform treatment decisions.