Abstract Conclusion Uterine and ovarian cancers are the fourth and fifth, respectively, most common cancers in women in the United States and combined are estimated to account for nearly 25,000 deaths in the US each year. The use of circulating, cell-free tumor DNA (cfDNA) to monitor disease burden during and post treatment have implications for effectively treating this disease. Collection followed by targeted sequencing of a patient’s cfDNA over time would allow assessment of tumor-related mutations from a blood draw to detect tumor cell DNA presence and aid in defining treatment strategies which may be designed to target specific observed mutations. We performed a pilot study to retrospectively study the cfDNA collected from 11 women with gynecological cancers. Each had undergone tumor resection and the tumor was sequenced to determine its mutational profile. Tumor mutation allele frequencies were detected between 10–83% and 1–3 mutations were identified per tumor. cfDNA was collected and extracted at a minimum of two time points ranging from 7 to 64 months apart. Samples were sequenced using a next generation sequencing amplicon panel covering 56 oncology-related genes. We observed a correlation between tumor-specific mutation frequency in the cfDNA and survivability. In 9 of 11 women, no reappearance of the primary tumor mutations was observed and these women (6) remain in remission or living with disease (3). In 2 of 11 women, while the frequency of the primary tumor mutation in the first time point was less than 1%, a resurgence of the mutation(s) detected in the primary tumor was observed at the second time point, with mutant allele frequencies ranging from 5 – 78%. This increase in mutation frequency corresponded to morbidity or mortality. This study demonstrates the use of cfDNA in amplicon-based sequencing assays to assess tumor burden in a minimally-invasive manner in the research setting. Such techniques to monitor disease progression and treatment efficacy promise to be instrumental in reducing the lethality associated with gynecological cancers. Accel-Amplicon ™ 56G Oncology Panel v2 cfDNA Mutational Profiling – a Longitudinal Study Ultra-Low Frequency Variant ID Unique Sample Identification Targeted next-generation sequencing of cell-free tumor DNA to longitudinally monitor cancer burden and progression Jonathan Irish 1 , Cassie A Schumacher 1 , Navya Nair 2 , Olga Camacho-Vanegas 2 , Sukhinder Sandhu 2 , Laurie Kurihara 1 , Peter Dottino 2 , Melissa Schwartz 2 , Timothy Harkins 1 , John Martignetti 2 , Vladimir Makarov 1 Swift Biosciences, Inc. 58 Parkland Plaza, Suite 100 • Ann Arbor, MI 48103 wwwswiftbiosci.com © 2017, Swift Biosciences, Inc. The Swift logo and Accel-Amplicon are trademarks of Swift Biosciences. This product is for Research Use Only. Not for use in diagnostic procedures. MiSeq and Illumina are registered trademarks of Illumina, Inc. Qubit is a registered trademark of ThermoFisher Scientific Inc. Control 17-1439 04/17 www.swiftbiosci.com 1 Swift Biosciences, 58 Parkland Plaza, Suite 100, Ann Arbor, MI 48103 2 Icahn School of Medicine at Mount Sinai, Departments of Genetics and Genomic Sciences and Obstetrics/Gynecology & Reproductive Sciences, 1425 Madison Avenue New York, NY 10029 AACR 2017: # 5392 • Single-tube assay • 2-hour workflow • 10ng input DNA • Contiguous, overlapping coverage of 56 oncology- relevant genes • v2 panel adds coverage of selected germline SNP targets to track sample “fingerprint” • > 95% coverage uniformity • > 95% of aligned reads on-target ABL1 CSF1R FBXW7 GNAS KIT NPM1 STK11 AKT1 CTNNB1 FGFR1 HNF1A KRAS NRAS SMAD4 ALK DDR2 FGFR2 HRAS MAP2K1 PDGFRA SMARCB1 APC DNMT3A FGFR3 IDH1 MET PIK3CA SMO ATM EGFR FLT3 IDH2 MLH1 PTEN SRC BRAF ERBB2 FOXL2 JAK2 MPL PTPN11 TP53 CDH1 ERBB4 GNA11 JAK3 MSH6 RB1 TSC1 CDKN2A EZH2 GNAQ KDR NOTCH1 RET VHL Figure 1. Accel-Amplicon 56G Oncology Panel v2 Specifications. The Accel-Amplicon workflow consists of two steps. Step 1 is a multiplex PCR step, containing all oncology-specific primer pairs as well as primer pairs for germline mutations to generate the genetic footprint. After a bead-based clean-up, the second step adds a unique index and the sequencing adapters to each library to allow multiplexing on the sequencing instrument. The table depicts the 56 genes that are represented in this panel. Contiguous, overlapping coverage of selected areas is included for these genes. Full coding exon coverage is included for this gene. Tumor Mutational Profiling Sequencing Metrics # Tumor/normal samples passing QC 55 Frequency of detected mutations 5-86% Average coverage Tumor: 5113X Normal: 139X Coverage uniformity Tumor: 95.2% Normal: 97.7% % on target Tumor: 95.4% Normal: 95.9% Mutation Summary # Tumors with no mutations 17 # Tumors with mutations detected 38 # Tumors with mutations in TP53 35 # Tumors with mutations in PIK3CA 8 # Tumors with mutations in PTEN 5 Other mutations detected in clinically- relevant genes ATM, BRAF, CTNNB1, EGFR, ERBB2, ERBB4, FBXW7, GNA11, KIT, KRAS, MAP2K1, MLH1, MSH6, PDGFRA, STK11 Figure 2. Preliminary Study to Determine Tumor Mutational Profiles. 57 tumor samples and corresponding normal samples derived from blood cells were obtained and sequenced using Accel-Amplicon 56G. 2 samples were excluded after not passing quality control. Tumors were sequenced on an Illumina® MiSeq® to > 5000x coverage for somatic variant calling; normal samples were sequenced to > 100x coverage for germline variant calling. Allele frequencies were determined bioinformatically using LoFreq. Indiv. Cancer Type Gene Mutation AF Tumor AF cfDNA #1 AF cfDNA #2 AF cfDNA #3 Outcome 338 Stage 3A, grade 1 endometrial adenocarcinoma PTEN Leu318_Thr319fs 15% 4-22-15 - 8-27-15 - N/A 6-16-16 Disease- free PTEN Tyr16fs 16% - - CTNNB1 Ser38Phe 13% - - 066 Stage 2A serous ovarian TP53 Tyr220Cys 53% 5-13-09 - 7-8-10 - N/A 7-1-16 Disease- free EGFR Ser715Ile 4% 0.8% 3% 175 Stage 3C serous ovarian TP53 Tyr175His 9% 8-1-11 0.3% 12-16-13 - 12-16-13 0.3% 6-11-14 Deceased PIK3CA Met1043Ile 8% - - 0.4% 194 Stage 3C serous ovarian TP53 Gly112fs 40% 4-5-12 0.1% 4-10-14 0.2% N/A 5-7-14 Deceased PIK3CA Glu545Lys 62% - 0.8% 067 Stage 4B serous ovarian TP53 Leu252fs 28% 9-14-09 0.4% 1-2-15 78% N/A 1-6-15 Deceased 158 Stage 3C serous ovarian TP53 Arg248Gln 31% 1-24-11 - 7-16-15 - N/A 5-25-16 Deceased 208 Stage 3C ovarian carcinoma TP53 Arg248Trp 18% 2-28-13 - 1-16-15 0.4% N/A 1-16-15 + Deceased 217 Stage 3C serous ovarian TP53 Arg158fs 71% 3-14-13 2% 2-14-14 - N/A 6-27-14 Deceased 078 Stage 3C serous ovarian KRAS Gly12Val 35% 3-25-10 - 11-17-14 - N/A Unknown 105 Stage 4B serous ovarian TP53 His193Arg 86% 1-11-10 1% 5-28-15 - N/A Unknown 247 Stage 4 uterine papillary serous PIK3CA His1047Arg 16% 12-19-13 1% 5-29-14 5% N/A Unknown TP53 Val216fs 11% - 6% Figure 4. Longitudinal Study of cfDNA. cfDNA was retrospectively isolated from blood samples at various times after tumor resection for 11 individuals from Figure 2. Accel-Amplicon 56G libraries were made from 10 ng of cfDNA (exceptions: blue, 7-9 ng and plum, 3 ng) and sequenced on a MiSeq to an average 15,000X coverage. Allele frequencies (AF) were determined using LoFreq and compared against the AFs detected in the tumor samples. All AFs below 1% were determined by visual inspection using the Integrated Genome Viewer (IGV) software from the Broad Institute. Examples are in orange and green boxes and illustrated in Figure 5. +actual date of death was near but not on this date Figure 5. Ultra-low Frequency Variant Identification. For tumor mutations which LoFreq did not identify in the cfDNA, IGV was used to inspect the region. Left: 247 cfDNA around PIK3CA His1047Arg. Right: 067 cfDNA around TP53 Leu252fs. Reads identified with each base (and deletions) are shown in the yellow box. The upper region shows cfDNA #1; the bottom region shows cfDNA #2. Colors correspond to the boxes in Figure 4. PIK3CA 3:178952085 A>G His1047Arg TP53 17:7577518 TGATGGTGA>T Leu252fs Figure 6. SampleID Spike-In. In addition to the 56G primers, primers to generate a unique genetic fingerprint for each sample are spiked into the panel at a low percentage (2-4% of total reads), resulting in a mean depth of 200X coverage of SampleID targets for a panel with a somatic variant mean target coverage of 5000X. Confidently Track Samples • 10 ng of tumor DNA and normal DNA from blood can be used to identify somatic variants present in ovarian and uterine tumors. • Sequencing cfDNA derived from women with gynecological cancer can identify the same variants seen in the tumors and at various times after resection. • There is an observed correlation between resurgence of tumor mutations in the cfDNA of individuals with gynecological cancers and outcome. More and broader studies should be done with more frequent and consistent time points to further probe the utility of this technique. Other factors such as age, treatment, type of cancer, etc should also be considered when interpreting results. • Sample_ID is a critical tool to allow samples derived from the same individual to be properly paired during analysis. chr POS SNP ID REF ALT Normal 33 Tumor 35 Normal 34 Tumor 33 Normal 35 Tumor 34 Normal 37 Tumor 37 Normal 41 Tumor 41 Normal 57 Tumor 57 1 67861520 rs2229546 C A 48% 52% 54% 50% 100% 100% 46% 49% 51% 52% 100% 100% 1 158582646 rs2251969 T C 47% 47% 52% 49% 52% 49% 51% 53% 47% 45% 1 167849414 rs203849 A G 100% 100% 100% 100% 100% 100% 48% 46% 100% 100% 100% 100% 1 179520506 rs1410592 G A 100% 100% 46% 56% 55% 29% 100% 100% 100% 100% 49% 60% 1 209811886 rs2076356 T G 100% 100% 48% 34% 100% 100% 100% 55% 54% 1 209968684 rs2013162 C A 46% 50% 55% 37% 52% 47% 49% 1 228431095 rs1771455 A G 99% 98% 100% 47% 50% 45% 2 44502788 rs3738985 A C 100% 100% 100% 100% 100% 100% 100% 100% 100% 47% 52% 57% 2 49381585 rs1394205 C T 45% 18% 59% 47% 46% 53% 2 75115108 rs10194657 A G 100% 100% 46% 43% 51% 2 169789016 rs497692 T C 48% 48% 51% 59% 100% 100% 100% 100% 52% 2 170092395 rs2229267 A G 52% 16% 100% 99% 52% 63% 2 179454394 rs1560221 A G 52% 53% 51% 48% 42% 2 179455207 rs2163009 T C 41% 46% 52% 46% 45% 2 215820013 rs10498027 G A 46% 49% 46% 54% 46% 55% 49% 47% 2 219941063 rs897477 G A 45% 48% 51% 55% 48% 71% 100% 100% 100% 48% 100% 100% 2 227896976 rs10203363 C T 100% 100% 55% 36% 49% 80% 51% 53% 100% 42% 100% 100% 3 4403767 rs2819561 A G 100% 100% 100% 100% 51% 39% 100% 100% 100% 3 4712413 rs2306875 G A 50% 48% 51% 38% 46% 63% 48% 46% 100% 100% 100% 3 45989044 rs2234358 T G 53% 42% 50% 60% 100% 3 148727133 rs4938 G A 52% 44% 51% 54% 54% 54% 4 5749904 rs386594666 T C 52% 53% 58% 35% 45% 42% 43% 60% 48% 49% 4 86844835 rs6824722 A G 46% 56% 52% 61% 100% 50% 4 86915848 rs10003909 T C 50% 51% 56% 68% 44% 27% 49% 52% 47% 54% 4 88534235 rs2736982 A G 100% 100% 49% 39% 49% 71% 54% 50% 53% 53% 51% 51% chr POS SNP ID REF ALT Normal 33 Tumor 33 Normal 34 Tumor 34 Normal 35 Tumor 35 Normal 37 Tumor 37 Normal 41 Tumor 41 Normal 57 Tumor 57 1 67861520 rs2229546 C A 48% 50% 54% 100% 100% 52% 46% 49% 51% 52% 100% 100% 1 158582646 rs2251969 T C 47% 47% 49% 52% 52% 49% 51% 53% 47% 45% 1 167849414 rs203849 A G 100% 100% 100% 100% 100% 100% 48% 46% 100% 100% 100% 100% 1 179520506 rs1410592 G A 100% 56% 46% 29% 55% 100% 100% 100% 100% 100% 49% 60% 1 209811886 rs2076356 T G 100% 34% 48% 100% 100% 100% 100% 55% 54% 1 209968684 rs2013162 C A 46% 37% 55% 50% 52% 47% 49% 1 228431095 rs1771455 A G 99% 98% 100% 47% 50% 45% 2 44502788 rs3738985 A C 100% 100% 100% 100% 100% 100% 100% 100% 100% 47% 52% 57% 2 49381585 rs1394205 C T 18% 45% 59% 47% 46% 53% 2 75115108 rs10194657 A G 100% 43% 46% 100% 51% 2 169789016 rs497692 T C 48% 59% 51% 100% 100% 48% 100% 100% 52% 2 170092395 rs2229267 A G 16% 52% 100% 99% 52% 63% 2 179454394 rs1560221 A G 52% 53% 51% 48% 42% 2 179455207 rs2163009 T C 41% 46% 52% 46% 45% 2 215820013 rs10498027 G A 46% 54% 46% 49% 46% 55% 49% 47% 2 219941063 rs897477 G A 45% 55% 51% 71% 48% 48% 100% 100% 100% 48% 100% 100% 2 227896976 rs10203363 C T 100% 36% 55% 80% 49% 100% 51% 53% 100% 42% 100% 100% 3 4403767 rs2819561 A G 100% 100% 100% 39% 51% 100% 100% 100% 100% 3 4712413 rs2306875 G A 50% 38% 51% 63% 46% 48% 48% 46% 100% 100% 100% 3 45989044 rs2234358 T G 42% 53% 60% 50% 100% 3 148727133 rs4938 G A 52% 44% 51% 54% 54% 54% 4 5749904 rs386594666 T C 52% 35% 58% 42% 45% 53% 43% 60% 48% 49% 4 86844835 rs6824722 A G 46% 61% 52% 56% 100% 50% 4 86915848 rs10003909 T C 50% 68% 56% 27% 44% 51% 49% 52% 47% 54% 4 88534235 rs2736982 A G 100% 39% 49% 71% 49% 100% 54% 50% 53% 53% 51% 51% ? Figure 7. Using Sample_ID to Track Samples. Germline variants generated using Sample_ID from 6 tumor/normal pairs from Figure 2 are shown. Since each tumor/normal sample came from the same individual, it is expected that the same genetic footprint of germline SNPs will be generated. Top: allele frequency pairings of the samples as labeled. By highlighting frequencies at 100% in green and between 30-65% in gray, improper pairing appears to have occurred in samples 33, 35, 35, and 41, while the pairings match perfectly in samples 37 and 57. Bottom: By examining the frequencies, an accurate match exists between Normal 33 and Tumor 35, Normal 34 and Tumor 33, and Normal 35 and Tumor 34. An examination of pairs does not find a match for sample 41. Since the genetic identity of these samples did not match, sample 41 was excluded from the study. Only a subset of the variants in SampleID are shown; the full panel identifies 104 SNPs. cfDNA Quantification and QC Figure 3. Alu qPCR Assay. Accurate qPCR quantification of cfDNA is imperative for successful library preparation. Fluorometric methods such as Qubit® do not quantify amplifiable DNA and cannot distinguish cfDNA from high molecular weight (HMW) genomic DNA (gDNA). A qPCR assay targeting both 115bp and 247bp regions of the Alu repeats elements (shown above) can quantify amplifiable DNA. The Alu115 primers quantify both cfDNA and HMW gDNA, while the Alu247 primers quantify only HMW gDNA. The ratio of 247/115 determines a DNA integrity score; the expected score for HMW gDNA is 1, and the expected score for cfDNA is between 0.29 and 0.65, but can vary with cancer types. This assay is based on Hao et al, Br J Cancer 2014 Oct 14; 111(8); 1482-9. ALU Alu247 Alu115 Indiv. cfDNA Sample Qubit (ng/µl) Alu115 (ng/µl) Alu 247/115 338 1 0.42 0.18 0.35 2 0.76 0.72 0.56 066 1 3.95 4.92 0.27 2 1.75 1.72 0.34 175 1 1.42 1.06 0.13 2 1.98 1.49 0.38 3 0.66 0.50 0.33 194 1 0.54 0.42 0.49 2 0.89 0.41 0.21 067 1 1.60 1.57 0.57 2 10.5 5.10 0.11 158 1 6.30 3.44 0.44 2 0.31 0.17 0.59 208 1 0.45 0.10 0.33 2 1.07 0.71 0.45 217 1 1.41 1.28 0.46 2 1.71 1.45 0.27 078 1 4.05 4.94 0.79 2 1.52 1.39 0.54 105 1 1.49 0.93 0.42 2 1.69 1.76 0.32 247 1 3.02 2.50 0.40 2 0.63 0.30 0.15