Comprehensive Analyses of Circulating Cell- Free Tumor DNA Boston, MA June 28th, 2016 Derek Murphy, Ph.D. Scientist, Research and Development Personal Genome Diagnostics
Acquisition of Somatic Alterations Throughout Tumorigenesis Sci Transl Med. 2010 February 24; 2(20): 20ra14. Nat Med. 2008 September; 14(9): 985 990. B Vogelstein et al. Science 2013;339:1546-1558 2
Circulating Cell-Free Tumor DNA The presence of cell-free DNA in the circulation was first formally described by Mandel and Metais in 1948. ctdna can be detected across many common solid tumor types in patients in both early and late stages of cancer, with levels ranging from less than one to greater than 100,000 mutant DNA fragments per milliliter of plasma. DNA fragments of 160-180bp with half life of several hours Genetic alterations are specific to the tumor Real-time, non-invasive, multi-lesions Unlike other tumor biomarkers, the specificity of somatic alterations is derived from the fact that mutations are present in the genome(s) of tumor cells but not in the genome of matched normal cells. Nat Rev Cancer. 2011 Jun;11(6):426-37. 3
Circulating Tumor DNA is Detectable Across Tumor Types with Rearrangements and Sequence Mutations Sci Transl Med. 2014 Feb 19;6(224):224ra24. 4
Applications of Circulating Tumor DNA Sci Transl Med. 2014 Feb 19;6(224):224ra24. 5
Technologies for Detection of Genomic Alterations in cfdna Single-Base Sequence Alterations Digital PCR Enrichment of Mutant DNA Ice-Cold PCR, PCR-Clamping, etc Specific Domains or Regions of Genes for Detection of Sequence Mutations Amplicon-Based, Molecular Barcoding and Next-Generation Sequencing Large Panels for Detection of Sequence Mutations, Copy Number Alterations and Translocations In-Solution Capture, Molecular Barcoding and Next-Generation Sequencing 6
PlasmaSelect Platform *Recurrently mutated exons were evaluated through a comprehensive analysis of COSMIC across all tumor types 7
Identification of Somatic Mutations with PlasmaSelect SureSelect Target Enrichment 8
Observed Allele Fraction (%) Detection of Sequence Mutations at Varying Levels of Tumor Contribution Using PlasmaSelect In-Solution Capture Technology 102.40% 51.20% 25.60% 12.80% 6.40% 3.20% 1.60% 0.80% 0.40% 0.20% 0.10% Limit of Detection = 0.10% 0.05% 0.05% 0.10% 0.20% 0.40% 0.80% 1.60% 3.20% 6.40% 12.80% 25.60% 51.20% 102.40% Expected Allele Fraction (%) A breast cancer tumor was diluted with matched wild-type DNA to 0.10%, 0.20%, 0.50%, 1.0%, 10%, 25%, 50% and 100% tumorderived DNA (0.10%, 0.20%, and 0.50% were evaluated in triplicate). For each sequence mutation, the observed allele fraction for the mutations identified in each case are plotted against the expected mutation allele fraction for that case. 9
Identification of Translocations at Varying Levels of Tumor Contribution 100% 10% 1% 1% 0.2% 0.2% 0.2% 0.1% 0.1% The chronic myeloid leukemia cell line 562 was titrated with wild-type DNA to 0.10%, 0.20%, 0.50%, 1.0%, 10%, 25%, 50% and 100% tumor-derived DNA (0.10%, 0.20%, and 0.50% were evaluated in triplicate). 10
A Case Study with Epizyme Development and application of a 62 gene panel for assessment of somatic sequence and structural variants in tumor DNA derived from non-hodgkin lymphoma.
NHL Study Background Tazemetostat is a small molecule inhibitor of the histone methyltransferase EZH2 and is being developed for the treatment of patients with non-hodgkin lymphoma and for patients with certain genetically defined solid tumors. Activating EZH2 mutations present in NHL patients have been implicated to predict response to EZH2 inhibition (Knutson et al. Nat Chem Biol. 2012) A Phase 1 clinical trial of tazemetostat demonstrated clinical responses in both EZH2 mutant and wild-type patients (NCT01897571) Epizyme and PGDx collaborated to build a custom panel for somatic mutation detection in ctdna of NHL patients for discovery of candidate biomarkers of response to tazemetostat Daigle et. al., 2016 AACR 12
NHL Study Design Development of a 62 gene panel designed to include previously identified variants occurring in NHL at >5% frequency. Analytically validated panel using a tumor cell line containing a subset of the variants of interest. Archive and ctdna collected from sixteen relapsed refractory NHL patients treated in the phase 1 clinical trial with tazemetostat. DNA was isolated from unstained FFPE slides and plasma derived from peripheral blood. Captured libraries were screened utilizing the Illumina HiSeq 2500 platform with 100 bp pairedend reads and analyzed using the PlasmaSelect pipeline. Daigle et. al., 2016 AACR 13
Lymphoma PlasmaSelect Panel Design Custom Lymphoma PlasmaSelect Sequence & Structural Gene List Gene Name Region(s) Included Gene Name Region(s) Included PRDM1 Full Coding Sequence KIT Specific Exon(s) EZH2 Full Coding Sequence KRAS Specific Exon(s) KDM6A Full Coding Sequence MEF2B Specific Exon(s) KMT2D Full Coding Sequence MYC Specific Exon(s) ARID1A Specific Exon(s) MYD88 Specific Exon(s) ATM Specific Exon(s) NOTCH1 Specific Exon(s) B2M Specific Exon(s) NOTCH2 Specific Exon(s) BCL2 Specific Exon(s) NRAS Specific Exon(s) BCL6 Specific Exon(s) PIK3CA Specific Exon(s) BCL7A Specific Exon(s) PIM1 Specific Exon(s) BRAF Specific Exon(s) POU2F2 Specific Exon(s) BTG1 Specific Exon(s) PTEN Specific Exon(s) CARD11 Specific Exon(s) PTPN1 Specific Exon(s) CCND3 Specific Exon(s) PTPN11 Specific Exon(s) CD58 Specific Exon(s) PTPN6 Specific Exon(s) CD79B Specific Exon(s) PTPRD Specific Exon(s) CDKN2A Specific Exon(s) RB1 Specific Exon(s) CREBBP Specific Exon(s) S1PR2 Specific Exon(s) EP300 Specific Exon(s) SGK1 Specific Exon(s) FOXO1 Specific Exon(s) SMARCB1 Specific Exon(s) GNA13 Specific Exon(s) SOCS1 Specific Exon(s) HIST1H1B Specific Exon(s) STAT6 Specific Exon(s) HIST1H1C Specific Exon(s) TBL1XR1 Specific Exon(s) HIST1H1E Specific Exon(s) TNFAIP3 Specific Exon(s) IKZF3 Specific Exon(s) TNFRSF14 Specific Exon(s) IRF4 Specific Exon(s) TP53 Specific Exon(s) ITPKB Specific Exon(s) XPO1 Specific Exon(s) Gene Name Region(s) Included Gene Name Region(s) Included ALK ALK_NM_004304_Intron19 CIITA Entire Gene BCL2 BCL2_MCR_Breakpoint_Region MYC Entire Gene + 40kbp upstream BCL2 BCL2_MBR_Breakpoint_Region CD274 (PDL1) Entire Gene BCL6 Entire Gene PDCD1LG2 (PDL2) Entire Gene Custom Lymphoma PlasmaSelect Amplification Gene List Gene Name BCL2 CD274 (PDL1) FOXP1 JAK2 KDM4C PDCD1LG2 (PDL2) REL 54 Target Genes for Sequence Mutations 8 Target Genes for Structural Alterations 7 Target Genes for Amplifications Daigle et. al., 2016 AACR 14
Mutation Allele Fraction Development of Cell-Free DNA Lymphoma Panel for Sequence Mutation Analyses Pre-Correction 100.00% Background Error Rate 25% Tumor Specimen 10% Tumor Specimen 1% Tumor Specimen 0.2% Tumor Specimen (Replicate 1) 0.2% Tumor Specimen (Replicate 2) 0.1% Tumor Specimen (Replicate 1) 0.1% Tumor Specimen (Replicate 2) 10.00% 1.00% 0.10% 0.01% Position in Targeted Region of Interest (bp) Daigle et. al., 2016 AACR 15
Mutation Allele Fraction Development of Cell-Free DNA Lymphoma Panel for Sequence Mutation Analyses Post-Correction 100.00% Background Error Rate 25% Tumor Specimen 10% Tumor Specimen 1% Tumor Specimen 0.2% Tumor Specimen (Replicate 1) 0.2% Tumor Specimen (Replicate 2) 0.1% Tumor Specimen (Replicate 1) 0.1% Tumor Specimen (Replicate 2) 10.00% 1.00% 0.10% 0.01% Position in Targeted Region of Interest (bp) Daigle et. al., 2016 AACR 16
Lymphoma PlasmaSelect Results NGS panel able to detect molecular amplifications, translocations and sequence mutations down to variant allele frequencies of 2% and 0.1% for archive and ctdna respectively. Sequencing of phase 1 NHL patients revealed a complex genetic landscape with epigenetic modifiers CREBBP and KMT2D representing the most frequently mutated genes in this sample set. 100% concordance with previously reported somatic mutations detected in archive tumor demonstrates panels reproducibility and utility. Demonstrates the ability to determine molecular profiles using ctdna, potentially enabling patient characterization where archive tumor is absent or limiting. Daigle et. al., 2016 AACR 17
Detection of Focal Gene Amplification in cfdna Identification of MET amplification through analyses of plasma DNA. The amplification of each gene is indicated as fold copy gain, along with the genomic rearrangements detected through whole-genome analyses of cell-free DNA obtained from plasma. Rearrangements associated with focal copy number alterations, including amplification of MET, are highlighted in red. (taken from Diaz et. al., Oncotarget. 2013 Oct;4(10):1856-7.). 18
Evaluation of MET Amplification in Dilution Series (Rearrangements) 100% (6,305x) 10% Replicate (3,781x) 10% Replicate (2,376x) 10% Replicate (2,922x) 5% (2,573x) 2% (2,832x) 1% (3,457x) 0.25% (4,485x) 19
Sequencing Run 2 Sequencing Run 2 Comparison of MET Copy Number Analyses Across Sequencing Runs 11.00 Sequencing (Amplified Samples) 9.00 7.00 5.00 3.00 R² = 0.9994 1.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00 11.00 Sequencing Run 1 Sequencing (WT Samples) 1.25 1.15 1.05 0.95 0.85 R² = 0.7276 0.75 0.75 0.85 0.95 1.05 1.15 1.25 Sequencing Run 1 20
Evaluation of MET Amplification in Dilution Series (Copy Number) Global MET Copy Number (Fold) 10.00 9.00 8.00 7.00 6.00 5.00 4.00 3.00 2.00 1.00 R² = 0.9892 0.25% (4,485x) 0.00 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Tumor DNA Fraction (%) 21
A Case Study with Amgen Inc. Detection of Aurora Kinase A (AURKA) Focal Amplification in Plasma Samples of Patients With Recurrent Ovarian Cancer
Detection of Aurora Kinase A (AURKA) Focal Amplification in Plasma Samples of Patients With Recurrent Ovarian Cancer The Aurora family of serine-threonine protein kinases (Aurora A, B, and C) regulates key processes in cell division; the genes for these kinases are AURKA, AURKB, and AURKC Aurora kinases are associated with acquired resistance to therapy in human cancers. Hypothesis; AURKA amplification is a late event in ovarian cancer (OC) progression and that the frequency of amplification might be underestimated if assessed in tumor samples collected at the time of initial diagnosis To test this hypothesis, we determined AURKA amplification levels in patients with recurrent Ovarian Cancer (ROC) and compared those to AURKA amplification levels in the same patients at the time of initial diagnosis. Juan G et. al., 2016 AACR 23
Study Design Development and analytical validation of a custom AURKADetect Panel Samples were collected from 33 patients with taxane- and platinum-resistant Ovarian Cancer. Patients were enrolled in a phase 1 clinical study that evaluated monotherapy with AMG 900, an orally administered pan-aurora kinase inhibitor (NCT00858377) AURKA FISH testing in formalin-fixed, paraffin-embedded (FFPE) tumor tissues collected at time of primary diagnosis. Use of AURKADetect to identify AURKA amplifications in plasma samples at the time of study enrollment. Juan G et. al., 2016 AACR 24
AURKADetect Analytical Validation Data Case Type Experimental Tumor Purity (%) Average Total Coverage AURKADetect Result Plasma from Healthy Individual - 5,272 Wild-Type Plasma from Healthy Individual - 4,317 Wild-Type Plasma from Healthy Individual - 4,130 Wild-Type BT-474 (Cell Line) 100.00% 6,998 Amplified BT-474 (Cell Line) 100.00% 7,825 Amplified BT-474 (Cell Line) 100.00% 8,103 Amplified BT-474 (Cell Line) 100.00% 7,370 Amplified BT-474 (Cell Line) 10.00% 3,996 Amplified BT-474 (Cell Line) 2.00% 3,920 Amplified BT-474 (Cell Line) 1.00% 3,070 Amplified BT-474 (Cell Line) 0.50% 3,040 Amplified BT-474 (Cell Line) 0.20% 2,928 Amplified 10% 2% 1% 0.5% 0.2% Juan G et. al., 2016 AACR 25
AURKADetect Results Sample AURKA FISH Percent Mapped to Genome Percent Mapped to ROI Average Total Coverage AURKADetect Result Sample 1 Wild-type 97.7% 50.9% 3,476 Wild-Type Sample 2-97.8% 52.3% 3,294 Wild-Type Sample 3 Wild-type 96.7% 47.3% 3,204 Amplified Sample 4-97.6% 47.7% 3,105 Wild-Type Sample 5 Wild-type 97.3% 47.7% 3,527 Wild-Type Sample 6 Amplified 97.8% 50.0% 3,979 Wild-Type Sample 7 Borderline 97.7% 53.7% 4,114 Amplified Sample 8 Wild-Type 97.7% 64.9% 4,424 Wild-Type Sample 9 Wild-Type 97.2% 64.3% 3,609 Wild-Type Sample 10 F 97.7% 64.9% 4,470 Wild-Type Sample 11-97.3% 67.9% 4,922 Wild-Type Sample 12 F 97.3% 64.6% 3,965 Wild-Type Sample 13 F 97.3% 75.0% 4,462 Wild-Type Sample 14 F 97.6% 59.1% 4,224 Wild-Type Sample 15 Wild-Type 97.9% 66.9% 4,573 Amplified Sample 16 Wild-Type 97.7% 67.8% 4,972 Wild-Type Sample 17 F 97.6% 60.3% 4,172 Amplified Sample 18 Wild-Type 97.7% 56.1% 3,805 Wild-Type Sample 19 F 97.9% 64.7% 5,049 Wild-Type Sample 20 F 98.4% 69.4% 3,832 Amplified Sample 21 Wild-Type 98.0% 69.0% 5,798 Amplified Sample 22 Wild-Type 97.7% 61.0% 5,494 Wild-Type Sample 23-98.2% 65.8% 3,200 Wild-Type Sample 24 F 95.3% 56.9% 3,666 Wild-Type Sample 25 Wild-Type 98.3% 71.8% 4,903 Amplified Sample 26 Wild-Type 97.8% 69.4% 4,839 Wild-Type Sample 27-97.5% 64.9% 3,818 Wild-Type Sample 28-97.9% 64.7% 5,154 Amplified Sample 29 F 97.7% 62.7% 3,725 Wild-Type Sample 30 Wild-Type 96.4% 59.0% 2,129 Amplified Sample 31 Wild-Type 97.9% 64.8% 4,386 Amplified Sample 32 97.7% 65.1% 3,839 Wild-Type Sample 33 Fail 97.9% 60.2% 4,969 Amplified F= Assay Did not Pass QC (-) = No Sample Available or Not Run Juan G et. al., 2016 AACR 26
AURKA Amplification is a poor prognostic Factor in ROC PFS 10 weeks longer in WT vs Amp (P = 0.04) OS was 18 weeks longer in WT vs Amp (P = 0.013) Wild-type Amplified Wild-type Amplified Group Median Time Lower 95% CI Upper 95% CI Amplified 9.43 4.57 18.57 Wild-type 19.93 5.14 42.43 Combined 14.57 5.57 25.14 Group Median Time Lower 95% CI Upper 95% CI Amplified 9.42 4.57 18.57 Wild-type 27.64 6.42 43.43 Combined 16.143 8.28 30.14 Juan G et. al., 2016 AACR 27
AURKA Study Conclusions AURKA amplification more common in patients with a higher number of prior lines of therapy AURKA amplification likely to be a late event in ovarian cancer progression Detection of AURKA focal amplification in plasma is a highly sensitive method; cfdna is an appropriate material for this type of analysis Evaluation of AURKA amplification by sequencing of plasma DNA allows the noninvasive assessment of acquired resistance to cancer therapy. Observed a correlation between AURKA amplification and PFS and OS: development of AURKA amplification may be indicative of poorer patient outcomes, with shorter PFS and OS Juan G et. al., 2016 AACR 28
Acknowledgments Epizyme Scott R. Daigle Scott Ribich Heike Keilhack Peter T. Ho Stephen J. Blakemore Institut Bergonie, Bordeaux, France Antoine Italiano Insitut Gustave Roussy, Villejuif, France Vincent Ribrag Amgen Inc. Gloria Juan Katherine Paweletz Abraham Anderson Erick Gamelin Gregory Friberg Robert Loberg Florian D. Vogl Personal Genome Diagnostics Samuel Angiuoli Sian Jones Mark Sausen 29
Agilent products described are For Research Use Only. Not for use in diagnostic procedures. 30