Search

Background: Inactivating HRR gene mutations can lead to HRR deficiency (HRD) and predict response to PARPi therapy in patients with breast cancer. Copy number loss and large genomic rearrangements (LGR) can result in HRD but are challenging to detect in ctDNA. Here, we present the analytical validation of homozygous deletions, loss of heterozygosity (LoH) and LGR detection on the Guardant360 (G360) liquid biopsy panel, previously validated for detection of small variants, copy number amplifications, and fusions. We present real-world outcomes of BRCA1/2-mutant PARPi-treated patients to demonstrate the clinical validity of the detected variants. Methods: Analytical validation was performed using the G360 83-gene ctDNA panel. Cell line DNA and clinical patient cfDNA were titrated into matched normal cell line DNA or healthy donor cfDNA to establish the limit of detection (LoD) and precision for copy number loss and LGRs, respectively. Accuracy results were compared to those from an orthogonal, externally validated tissue and ctDNA panel. De-identified, longitudinal, claims data were linked to the cancer genomic profiles in Guardant INFORM, a clinical-genomics database. Advanced PARPi treated breast cancer patients with an inactivating or reversion BRCA1/2 mutation detected by G360 were assessed. Results: The analytical sensitivity (95% LoD) for detecting homozygous and LoH deletions for deletion sizes >10MB was established at tumor fractions (TF) of 12.5% and 25%, respectively. The 95% LoD for LGRs was 0.2% variant allele fraction (VAF). The per-sample false positive rate for copy number loss and LGRs was 1000 advanced breast cancer patients was 1.8%, 16.6% and 0.25% respectively, compared to 2.4%, 56.7% and 0.3% in TCGA. To verify the clinical impact of cfDNA-detected HRR alterations, overall survival was determined for PARPi-treated patients with >1 BRCA1/2 germline or somatic SNV, indel or LGR reversion mutation to be 23.2 months [16.4, 30, CI, n=75] compared to 54.4 [28, NA, CI, n=14] months for BRCA1/2-mutant patients without a reversion (p-value=0.049). Conclusion:. This analytical validation demonstrates that G360 detection of inactivating mutations, including copy number loss and LGRs, is highly sensitive, reliable and robust. Real-world evidence analysis confirmed worse survival outcomes in PARPi treated patients harboring a BRCA1/2 reversion compared to BRCA1/2-mutant patients with no reversion. This data further supports ctDNA as a compelling non-invasive means to identify potential PARPi sensitizing and resistance mutations in patients with advanced breast cancer. Citation Format: Jennifer Yen, Leylah Drusbosky, Caroline Weipert, Nicole Zhang, David Hanna, Catalin Barbacioru, Hao Wang, Alex Artyomenko, Arielle Yablonovitch, Yu Fu, Aaron Hardin, Nagesh Alla, Robert Foley, Max Maligska, Bhargavi Panchangam, Phil Yen, Jane Meisel, Neelima Vidula, Massimo Cristofanilli, Jeremy Force, Michael Dorschner, Martina Lefterova, Elena Helman, Becky Nagy, Darya Chudova, AmirAli Talasaz. Analytical and clinical validation of a ctDNA assay for detecting copy number loss and structural rearrangement variants contributing to homologous recombination and repair (HRR) deficiency [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr P5-13-29.

Table S1 is an Excel sheet that provides a detailed listing of primary BRCA and TP53 mutations as well as BRCA reversion mutations detected in pretreatment and postprogression circulating cell-free DNA and tumor biopsy samples

A key resistance mechanism to platinum-based chemotherapies and PARP inhibitors in BRCA-mutant cancers is the acquisition of BRCA reversion mutations that restore protein function. To estimate the prevalence of BRCA reversion mutations in high-grade ovarian carcinoma (HGOC), we performed targeted next-generation sequencing of circulating cell-free DNA (cfDNA) extracted from pretreatment and postprogression plasma in patients with deleterious germline or somatic BRCA mutations treated with the PARP inhibitor rucaparib. BRCA reversion mutations were identified in pretreatment cfDNA from 18% (2/11) of platinum-refractory and 13% (5/38) of platinum-resistant cancers, compared with 2% (1/48) of platinum-sensitive cancers (P = 0.049). Patients without BRCA reversion mutations detected in pretreatment cfDNA had significantly longer rucaparib progression-free survival than those with reversion mutations (median, 9.0 vs. 1.8 months; HR, 0.12; P < 0.0001). To study acquired resistance, we sequenced 78 postprogression cfDNA, identifying eight additional patients with BRCA reversion mutations not found in pretreatment cfDNA.Significance:BRCA reversion mutations are detected in cfDNA from platinum-resistant or platinum-refractory HGOC and are associated with decreased clinical benefit from rucaparib treatment. Sequencing of cfDNA can detect multiple BRCA reversion mutations, highlighting the ability to capture multiclonal heterogeneity.This article is highlighted in the In This Issue feature, p. 151

Patient-level clinical outcome data (overall survival) and bTMB scores

Mutations Analyzed

GuardantOMNI assay success rates

Purpose:The role of plasma-based tumor mutation burden (pTMB) in predicting response to pembrolizumab-based first-line standard-of-care therapy for metastatic non–small cell lung cancer (mNSCLC) has not been explored.Experimental Design:A 500-gene next-generation sequencing panel was used to assess pTMB. Sixty-six patients with newly diagnosed mNSCLC starting first-line pembrolizumab-based therapy, either alone or in combination with chemotherapy, were enrolled (Clinicaltrial.gov identifier: NCT03047616). Response was assessed using RECIST 1.1. Associations were made for patient characteristics, 6-month durable clinical benefit (DCB), progression-free survival (PFS), and overall survival (OS).Results:Of 66 patients, 52 (78.8%) were pTMB-evaluable. Median pTMB was 16.8 mutations per megabase (mut/Mb; range, 1.9–52.5) and was significantly higher for patients achieving DCB compared with no durable benefit (21.3 mut/Mb vs. 12.4 mut/Mb, P = 0.003). For patients with pTMB ≥ 16 mut/Mb, median PFS was 14.1 versus 4.7 months for patients with pTMB < 16 mut/Mb [HR, 0.30 (0.16–0.60); P < 0.001]. Median OS for patients with pTMB ≥ 16 was not reached versus 8.8 months for patients with pTMB < 16 mut/Mb [HR, 0.48 (0.22–1.03); P = 0.061]. Mutations in ERBB2 exon 20, STK11, KEAP1, or PTEN were more common in patients with no DCB. A combination of pTMB ≥ 16 and absence of negative predictor mutations was associated with PFS [HR, 0.24 (0.11–0.49); P < 0.001] and OS [HR, 0.31 (0.13–0.74); P = 0.009].Conclusions:pTMB ≥ 16 mut/Mb is associated with improved PFS after first-line standard-of-care pembrolizumab-based therapy in mNSCLC. STK11/KEAP1/PTEN and ERBB2 mutations may help identify pTMB-high patients unlikely to respond. These results should be validated in larger prospective studies.

pTMBand PD-L1 expression levels.

Purpose:Tumor mutational burden (TMB) has been shown to be predictive of survival benefit in patients with non–small cell lung cancer (NSCLC) treated with immune checkpoint inhibitors. Measuring TMB in the blood (bTMB) using circulating cell-free tumor DNA (ctDNA) offers practical advantages compared with TMB measurement in tissue (tTMB); however, there is a need for validated assays and identification of optimal cutoffs. We describe the analytic validation of a new bTMB algorithm and its clinical utility using data from the phase III MYSTIC trial.Patients and Methods:The dataset used for the clinical validation was from MYSTIC, which evaluated first-line durvalumab (anti–PD-L1 antibody) ± tremelimumab (anticytotoxic T-lymphocyte-associated antigen-4 antibody) or chemotherapy for metastatic NSCLC. bTMB and tTMB were evaluated using the GuardantOMNI and FoundationOne CDx assays, respectively. A Cox proportional hazards model and minimal P value cross-validation approach were used to identify the optimal bTMB cutoff.Results:In MYSTIC, somatic mutations could be detected in ctDNA extracted from plasma samples in a majority of patients, allowing subsequent calculation of bTMB. The success rate for obtaining valid TMB scores was higher for bTMB (809/1,001; 81%) than for tTMB (460/735; 63%). Minimal P value cross-validation analysis confirmed the selection of bTMB ≥20 mutations per megabase (mut/Mb) as the optimal cutoff for clinical benefit with durvalumab + tremelimumab.Conclusions:Our study demonstrates the feasibility, accuracy, and reproducibility of the GuardantOMNI ctDNA platform for quantifying bTMB from plasma samples. Using the new bTMB algorithm and an optimal bTMB cutoff of ≥20 mut/Mb, high bTMB was predictive of clinical benefit with durvalumab + tremelimumab versus chemotherapy.

Development and validation of a plasma-based panel for mutation detection and assessment of TMB.

Purpose: Tumor mutational burden (TMB) has been shown to be predictive of survival benefit in patients with non–small cell lung cancer (NSCLC) treated with immune checkpoint inhibitors. Measuring TMB in the blood (bTMB) using circulating cell-free tumor DNA (ctDNA) offers practical advantages compared with TMB measurement in tissue (tTMB); however, there is a need for validated assays and identification of optimal cutoffs. We describe the analytic validation of a new bTMB algorithm and its clinical utility using data from the phase III MYSTIC trial. Patients and Methods: The dataset used for the clinical validation was from MYSTIC, which evaluated first-line durvalumab (anti–PD-L1 antibody) ± tremelimumab (anticytotoxic T-lymphocyte-associated antigen-4 antibody) or chemotherapy for metastatic NSCLC. bTMB and tTMB were evaluated using the GuardantOMNI and FoundationOne CDx assays, respectively. A Cox proportional hazards model and minimal P value cross-validation approach were used to identify the optimal bTMB cutoff. Results: In MYSTIC, somatic mutations could be detected in ctDNA extracted from plasma samples in a majority of patients, allowing subsequent calculation of bTMB. The success rate for obtaining valid TMB scores was higher for bTMB (809/1,001; 81%) than for tTMB (460/735; 63%). Minimal P value cross-validation analysis confirmed the selection of bTMB ≥20 mutations per megabase (mut/Mb) as the optimal cutoff for clinical benefit with durvalumab + tremelimumab. Conclusions: Our study demonstrates the feasibility, accuracy, and reproducibility of the GuardantOMNI ctDNA platform for quantifying bTMB from plasma samples. Using the new bTMB algorithm and an optimal bTMB cutoff of ≥20 mut/Mb, high bTMB was predictive of clinical benefit with durvalumab + tremelimumab versus chemotherapy.

Background: Melanoma is a prime example of how basic and translational research can improve cancer prognosis The tumor immune microenvironment (TIME) of melanoma is a dynamic, heterogeneous mixture of multiple immune cells and malignant and nonmalignant cells. Tissue resident memory T cells (TRM) may be prognostic for improved melanoma survival and proliferate during treatment with immune checkpoint inhibitors. The role of TRM in melanoma and rare melanoma subtypes is unknown. Method: Our goal was to examine the relationship between TRM abundance and melanoma prognosis. We generated TRM signatures using multiple independent single-cell RNA sequencing data from human melanoma cells to infer TRM abundance. Lastly, we built a LASSO Cox regression model that predicts patient survival. The TCGA dataset was used as a test group. Model accuracy was validated using Kaplan-Meier survival analysis, receiver operating characteristic curves, principal component analysis, and t-distributed stochastic neighbor embedding. Findings: We found a significantly longer overall survival in patients with high TRM infiltration. Furthermore, we found that TRM can interact with the tumor microenvironment and genomic features of melanoma, making TRM abundance positively correlate with the prognosis of melanoma patients. Melanoma abundance was also significantly associated with several common genomic aberrations, including the TP53 mutation. In addition, we developed a 20-gene risk score that significantly classifies patients into low- and high-risk categories. We found that in several independent datasets, patients with high risk scores had significantly worse overall survival than patients with low risk scores. Conclusions: Altogether, our findings suggest that TRM abundance may positively correlate with melanoma survival. We developed a high-precision LASSO Cox regression model based on the new TRM abundance characteristics. Our study shows that this model effectively predicts the prognosis of melanoma patients and may provide new therapeutic targets and measure the effectiveness of melanoma-directed therapies. Keywords: Tissue resident memory T cell, tumor immune microenvironment, melanoma, prognosis Citation Format: Chongming Jiang, Jianrong Li, Elena Helman, Xiling Shen, Chao Cheng. A novel tissue resident memory T cell (TRM) signature predicts prognosis and tumor microenvironment of patients with melanoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 3119.

Introduction: The ability to detect genome-wide epigenetic changes, such as DNA methylation, has expanded translational applications in oncology settings. Because these changes occur early in carcinogenesis, they can be used for early cancer detection when genomic technologies fall short due to lower sensitivity, and in the early and late-stage cancer setting for minimal residual disease detection, disease monitoring and therapy selection. In this analysis, we demonstrated our detection of differential methylations that classify cancer from healthy normals, as well as the quantification of promoter methylation, using a highly sensitive targeted assay that simultaneously captures both genomic alterations and methylation signatures in cell-free DNA (cfDNA). Methods: Methylation signals were profiled with a broad genomic panel (15.2 Mb) targeting regions that are unmethylated in plasma cfDNA from cancer-free donors. The panel covers the promoter regions of 925 out of 1,217 known tumor suppressor genes (TSGs) (e.g. TP53, APC, RB1, PTEN), homologous recombination and repair (HRR) genes (e.g. ATM, BRCA1/2, CDK12, RAD51C/D). We applied our genomic and epigenomic assay on cfDNA from 1,968 colorectal cancer (CRC) patients, 480 patients with other 6 common cancers, and 2,037 cancer-free donors. To test the sensitivity of our epigenomic assay, we generated an in-silico dataset by computationally mixing reads from the cancer patients with those from cancer-free donors at a low tumor fraction (TF) of 0.1%. Results: Among 62 clinically relevant TSG and HRR genes, 56 (90%) were differentially methylated (Wilcoxon p Conclusion: We demonstrate that our assay can accurately detect cancer-driven DNA methylation across the genome in clinical plasma samples. Detection of differential methylation in cancerous versus non-cancerous tissues could allow for early cancer detection, leading to better survival and prediction of recurrence prior to imaging. The highly sensitive detection of promoter methylation shown with TSG, HRR, and MLH1 genes may provide orthogonal information to oncologists in therapeutic selection with high confidence. Citation Format: Sai Chen, Shile Zhang, Tingting Jiang, Jennifer Yen, Yupeng He, Ariel Jaimovich, Yvonne Kim, Dustin Ma, Giao Tran, Daniel P. Gaile, Rebecca J. Nagy, Elena Helman, Han-Yu Chuang. Detection of tumor-associated gene inactivation in clinical blood draws via cell-free DNA methylation profiling [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 3763.

BackgroundTumor mutational burden (TMB), defined as the number of somatic mutations per megabase of interrogated genomic sequence, demonstrates predictive biomarker potential for the identification of patients with cancer most likely to respond to immune checkpoint inhibitors. TMB is optimally calculated by whole exome sequencing (WES), but next-generation sequencing targeted panels provide TMB estimates in a time-effective and cost-effective manner. However, differences in panel size and gene coverage, in addition to the underlying bioinformatics pipelines, are known drivers of variability in TMB estimates across laboratories. By directly comparing panel-based TMB estimates from participating laboratories, this study aims to characterize the theoretical variability of panel-based TMB estimates, and provides guidelines on TMB reporting, analytic validation requirements and reference standard alignment in order to maintain consistency of TMB estimation across platforms.MethodsEleven laboratories used WES data from The Cancer Genome Atlas Multi-Center Mutation calling in Multiple Cancers (MC3) samples and calculated TMB from the subset of the exome restricted to the genes covered by their targeted panel using their own bioinformatics pipeline (panel TMB). A reference TMB value was calculated from the entire exome using a uniform bioinformatics pipeline all members agreed on (WES TMB). Linear regression analyses were performed to investigate the relationship between WES and panel TMB for all 32 cancer types combined and separately. Variability in panel TMB values at various WES TMB values was also quantified using 95% prediction limits.ResultsStudy results demonstrated that variability within and between panel TMB values increases as the WES TMB values increase. For each panel, prediction limits based on linear regression analyses that modeled panel TMB as a function of WES TMB were calculated and found to approximately capture the intended 95% of observed panel TMB values. Certain cancer types, such as uterine, bladder and colon cancers exhibited greater variability in panel TMB values, compared with lung and head and neck cancers.ConclusionsIncreasing uptake of TMB as a predictive biomarker in the clinic creates an urgent need to bring stakeholders together to agree on the harmonization of key aspects of panel-based TMB estimation, such as the standardization of TMB reporting, standardization of analytical validation studies and the alignment of panel-based TMB values with a reference standard. These harmonization efforts should improve consistency and reliability of panel TMB estimates and aid in clinical decision-making.

180 Background: ctDNA testing is a non-invasive means to identify genomic alterations in patients with aPC, including BRCA1/2 and other HRR gene alterations, which are targetable by PARP inhibitors (PARPi). However, classes of HRR inactivating mutations, such as copy number loss and large genomic rearrangements (LGRs) are challenging to detect by ctDNA. Here we describe a CLIA-validated plasma-based ctDNA test that identifies copy number loss and LGRs, in addition to SNV and Indels, in HRR genes. We report the landscape of HRR deficiency (HRD) in a clinical population of > 2,000 patients with aPC. Methods: 2932 samples from patients with aPC previously tested as part of clinical care using an 83-gene cfDNA NGS assay (Guardant360) within an 11-month period were reanalyzed for presence of HRD. HRD alterations were defined as copy number loss, LGR (i.e. intergenic or intragenic gene rearrangements) and deleterious SNV/Indel in 7 clinically focused HRR genes ( ATM, BRCA1, BRCA2, CDK12, CHEK2, PALB2, RAD51D). Clinical information, when available, was obtained from test requisition forms. Results: The median age of the analyzed cohort was 73 years (range: 31-100); when known, the majority of patients were tested at disease progression (81%, 2242/2750). In total, an inactivating BRCA1/2 variant was was identified in 26% (605/2932) of reanalyzed patients, comprising BRCA1/2 loss of heterozygosity (LOH) (13.4% of patients), deleterious SNV or Indels (5.5%), homozygous deletions (3.3%) and LGRs (1.1%). The prevalence of patients with any inactivating HRR alteration increased to 38.5% with the expanded 7 gene set. Overall, prevalence of loss-of-function mutations (SNV/Indel, homozygous deletion or LGR) in BRCA2 was 4.5X frequent than in BRCA1 and comparable to previously described prostate cohorts (8.4% and 1.8% compared to 8.6% and 1.1% in tissue, respectively). Conclusions: Genomic profiling by ctDNA identified an HRR alteration landscape comparable to previously characterized tissue cohorts, with enrichment in BRCA2 over BRCA1 inactivating mutations and frequent BRCA1/2 LOH. Patients with aPC may benefit from PARPi eligibility determination using rapid and reliable ctDNA assessment, particularly at progression. Future studies should assess PARPi outcomes for this population.

Purpose: The role of plasma-based tumor mutation burden (pTMB) in predicting response to pembrolizumab-based first-line standard-of-care therapy for metastatic non–small cell lung cancer (mNSCLC) has not been explored. Experimental Design: A 500-gene next-generation sequencing panel was used to assess pTMB. Sixty-six patients with newly diagnosed mNSCLC starting first-line pembrolizumab-based therapy, either alone or in combination with chemotherapy, were enrolled (Clinicaltrial.gov identifier: NCT03047616). Response was assessed using RECIST 1.1. Associations were made for patient characteristics, 6-month durable clinical benefit (DCB), progression-free survival (PFS), and overall survival (OS). Results: Of 66 patients, 52 (78.8%) were pTMB-evaluable. Median pTMB was 16.8 mutations per megabase (mut/Mb; range, 1.9–52.5) and was significantly higher for patients achieving DCB compared with no durable benefit (21.3 mut/Mb vs. 12.4 mut/Mb, P = 0.003). For patients with pTMB ≥ 16 mut/Mb, median PFS was 14.1 versus 4.7 months for patients with pTMB < 16 mut/Mb [HR, 0.30 (0.16–0.60); P < 0.001]. Median OS for patients with pTMB ≥ 16 was not reached versus 8.8 months for patients with pTMB < 16 mut/Mb [HR, 0.48 (0.22–1.03); P = 0.061]. Mutations in ERBB2 exon 20, STK11, KEAP1, or PTEN were more common in patients with no DCB. A combination of pTMB ≥ 16 and absence of negative predictor mutations was associated with PFS [HR, 0.24 (0.11–0.49); P < 0.001] and OS [HR, 0.31 (0.13–0.74); P = 0.009]. Conclusions: pTMB ≥ 16 mut/Mb is associated with improved PFS after first-line standard-of-care pembrolizumab-based therapy in mNSCLC. STK11/KEAP1/PTEN and ERBB2 mutations may help identify pTMB-high patients unlikely to respond. These results should be validated in larger prospective studies.

Background: Inactivating mutations in homologous recombination repair (HRR) genes are associated with a response to PARP inhibitors and platinum-based therapies. However, secondary reversion mutations in HRR genes remain a major mechanism of resistance. We sought to characterize the prevalence and spectrum of these reversions in advanced cancer patients in BRCA1/2 and other HRR genes, such as ATM and CDK12. Methods: We analyzed cfDNA from >75,000 late-stage cancer patients previously tested as part of routine clinical care using Guardant360®, a 73- to 74-gene panel. This assay detects single-nucleotide variants (SNVs), indels, fusions, and copy-number alterations in cfDNA with a reportable range of ≥0.04%, 0.02%, 0.04%, and ≥2.12 copies, respectively. Secondary reversion mutations were identified by re-analysis using a bespoke HRR deficiency (HRD) Module, which annotates SNV/indels and calls multi-exon deletions and copy-number loss with a 95% limit of detection of ≥ 0.1% and 10 to 25% tumor fraction, respectively. Results: Among >20,000 breast, ovarian, prostate and pancreatic cancer patients, deleterious germline or somatic mutations were detected in BRCA1 (2.7%), BRCA2 (4.8%), ATM (2.9%) and CDK12 (1.7%). The majority of reversions occurred in BRCA1/2-mut (8.7%), compared to 0.07% of CDK12-mut and none in ATM. Reversions showed selective gene and allele representation. In breast cancer, reversions in BRCA2-mut patients were twice as frequent as those in BRCA1-mut (12.9% vs 5.5%, p-value=0.0046), while the near opposite was true for ovarian (7.6% vs 12.8%, p-value=0.73). In pancreatic cancer (>7,500 patients), reversions presented exclusively in BRCA2-mut (8.4%, n=131 vs 0 in BRCA1-mut, n=63). CDK12 reversions were relatively rare: 1 detected in >5,000 prostate cancer patients. Across all genes, the majority of reversions were small indels while 18% of reversions removed the exon entirely. Pathogenic hotspots BRCA2 codons A938 and S1982 - accounted for ≥10% of reversions and were more likely to revert compared to other deleterious sites (p-value Conclusions: cfDNA has been previously shown to detect BRCA1/2 germline, somatic and polyclonal reversion mutations acquired over PARP and platinum-based therapy. We demonstrate that the Guardant HRD Module can robustly identify these reversions in the cfDNA of metastatic patients. We emphasize that reversions were frequent among BRCA1/2-mut breast, ovarian, prostate and pancreatic cancers, with preferential enrichment based on gene and tumor type. Understanding resistance patterns may aid in the treatment selection for progressing patients. Further work is needed to define these ideal treatment options. Citation Format: Yu Fu, Jennifer Yen, Alexander Aartyomenko, Sante Gnerre, Caroline Weiport, Elena Helman, Arielle Yablonovitch, Stephen Pettitt, Christopher Llord, Andrew Tutt, Rebecca Nagy, Justin Odegaard, Darya Chudova, AmirAli Talasaz. Reversion patterns in homologous recombination repair genes in metastatic cancer patients profiled by cell-free DNA [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 25.

Background: NTRK rearrangements, though rare in common cancers, are clinically actionable targets with two FDA-approved drugs for pan-cancer indications, larotrectinib and entrectinib. Fusion detection from ctDNA provides an opportunity to facilitate screening for this biomarker, though technical challenges such as heterogeneity in fusion partners remain. To address this, we developed an assembly-based de novo fusion algorithm that does not rely on a fixed set of partner genes and applied it to >18,000 clinical samples to detect NTRK1 fusions. Methods: A cohort of 18,867 patients across multiple cancer types (lung adenocarcinoma, breast cancer, and colorectal adenocarcinoma), plus 276 healthy control samples were previously tested with Guardant360(R), a CLIA-validated 74-gene cfDNA NGS-based assay. The median unique molecule coverage was approximately 3,000 molecules sequenced to 15,000x read depth. Samples were reanalyzed in silico using the de novo fusion algorithm: in brief, reads aligned to candidate fusion breakpoints were assembled into de Bruijn graphs. Resulting contigs were aligned to the reference and analytical filters were applied to achieve high specificity. Results: NTRK1 fusions were detected in 0.13% of patients, a similar prevalence to what has been previously described in tissue (Rosen et al., 2020). No fusions were identified in 276 healthy control samples. When examining intergenic fusions, 53% of fusions included previously characterized partner genes (TPM3, TPR, LMNA, and TP53), while the remaining fusions contained novel partners.The majority of NTRK1 fusion partners (76%) were detected only once, consistent with previous studies. KRAS G12D and BRAF V600E were detected in 17% and 13% of NTRK1 fusion-positive patients, respectively, and occurred predominantly in the colorectal cohort; most variants (60%) were subclonal to the NTRK1 fusion, and have been previously shown to be associated with resistance to TRK inhibitors through downstream activation of the MAPK pathway. The on-target resistance variant NTRK1 G595R was detected in 8% of patients positive for NTRK1 fusions, all of which were subclonal to the fusion. EGFR L858R was detected in 13% of NTRK1 fusion-positive patients, all of which were clonal to the fusion and in the lung cohort, in accordance with previous studies showing NTRK1 fusions as a resistance mechanism to TKIs in EGFR-positive NSCLC (Xia et al., 2020). Conclusions: NTRK1 fusions were detected in cfDNA at a similar prevalence to tissue NGS, demonstrating high sensitivity of plasma-based assays to detect these fusions. NTRK1 fusion partners were diverse, with the majority of partner genes observed only once across the cohort. Both clonal and subclonal NTRK1 rearrangements were detected, affirming that this biomarker can emerge as an oncogenic driver or as a mechanism of resistance. Citation Format: Arielle Yablonovitch, Sante Gnerre, Jennifer Yen, Scott Shell, Elena Helman, Stephen Fairclough, Rebecca J. Nagy, Justin Odegaard, Darya Chudova, AmirAli Talasaz. NTRK1 fusion detection from clinical cfDNA NGS using a de novo fusion caller [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 537.

Background: Molecular response (MR) estimated as a change in circulating tumor (ctDNA) load between an early on treatment sample (usually 2-9 weeks post treatment start) and pre-treatment baseline has been shown to predict patient response and outcomes across solid tumors and therapy types in many retrospective studies. There is no consensus, however, regarding the best method for assessing molecular response. Therefore, we aimed to assess several molecular response calculations and determine the optimal method for predicting outcomes in individual advanced cancer patients. Method: Aggregate results of > 4,000 patient sample pairs (3-10 weeks apart), > 1000 patient sample technical replicates, > 100 contrived sample dilutions, and in silico simulations were analyzed with Guardant360TM or GuardantOMNITM. Baseline and on-treatment paired patient samples were collected from advanced cancer patients with over 12 tumor types, including lung, colon, and breast. MR calculations included variant allele fractions (VAFs) of somatic SNVs, indels and fusions. Methods including Ratio treatment/baseline (R) of Maximum VAF (maxVAF), Ratio of Mean VAF (mVAF), and Mean of VAF Ratios were compared, with consideration of VAF precision. Analytical accuracy, reproducibility and limit of detection (LoD) were assessed. Results: Comparison of methods for calculating net change in ctDNA load on > 1500 sample pairs showed high correlation (ρ ranged from 0.93 to 0.98) and categorical agreement split by the median (93%). Therefore selecting an optimal method based on outcome prediction would require prohibitively large patient cohorts. Analytical evaluation and in silico simulations can predict the behavior of each method. Simulations of changes in tumor fraction of real pre-treatment samples found that RmVAF or RmaxVAF are more accurate than mVAFR, which can be skewed by low VAF ratios. Almost 25% of sample pairs have a tumor driver or resistance mutation that is not the maxVAF, suggesting tumor dynamics are better captured by mVAF than maxVAF. Newly-detected on-treatment variants can be an important signal of rising ctDNA levels, impacting MR in approximately 2% of sample pairs. Importantly, MR accuracy for all methods decreases as maxVAF approaches or falls below the variant LoD, due to both stochastic detection and higher CV of variants at low VAF. Thus the assay variant LoD is a key determinant of the fraction of patients who can receive MR evaluation. Technical replicates identified the variant criteria at which a 50% change in tumor fraction differs significantly from technical variation, and could define analytical reporting limits. Conclusions: Comparison of MR methods in a large set of patient samples and simulations supports RmVAF with inclusion of newly-detected mutations. Clinical validation of these methods will support patient-specific MR prediction of outcomes. Citation Format: Allysia J. Mak, Katie J. Quinn, Carin Espenschied, Kyle Chang, Han-Yu Chuang, Elena Helman, Darya Chudova, Justin Odegaard, Becky Nagy, AmirAli Talasaz. Comparison of molecular response calculations for prediction of patient outcome [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 401.

Background: PARP inhibition can cause synthetic lethality and increased therapeutic sensitivity in patients with HRR deficiency (HRD), which can be detected through the molecular profiling of HRR genes. Prostate cancer has a high prevalence of HRD (20-25%, Athie 2019). High failure rates for tissue biopsy in metastatic prostate cancer patients (25-75%) (Ross 2005, Spritzer 2013) pose challenges for HRD profiling, underscoring the need for a non-invasive, ctDNA alternative. Copy number loss, a frequent cause of HRD, is further difficult to call due to signal dilution by cell-free leukocytic DNA. We developed a pipeline that detects loss-of-function SNV/Indels, structural rearrangements, and gene deletions to identify HRD on GuardantOMNITM, a 500-gene liquid biopsy panel. We present its performance across >620 prostate cancer GuardantOMNITM samples. Methods: Samples from 627 prostate cancer patients were processed on GuardantOMNITM RUO, with median unique coverage of 4982 molecules sequenced to 20,000x read depth. Somatic and germline SNVs and small indels were called using the Guardant bioinformatics pipeline (Helman 2018). A novel HRD module was developed to annotate pathogenic SNV/Indels and identify structural rearrangements, gene-level homozygous deletions, loss-of-heterozygosity (LOH) and genome-wide LOH, comprising of a novel CNV (Barbacioru 2019) and de-novo fusion caller (Gnerre, submitted). Loss-of-function variants were analyzed in 24 HRD genes. Results: Pathogenic alterations in HRD genes were called in 260/620 (42%) prostate cancer samples with ctDNA detected: 28% of all samples had a pathogenic somatic or germline SNV/Indel, 20% had a homozygous deletion, 3.4% had a rearrangement involving an HRD gene. The majority of SNV/Indels occurred in BRCA2 (31% of all 158 SNV/Indels) and ATM (22%), similar to tissue (Dahwan 2016), but mutations also occurred across an additional 21 genes, including CDK12 (8%), CHEK2 (5%) and NBN (3.8%). Of prostate patients with a germline BRCA1/2 SNV/Indel and sufficient tumor shedding for LOH detection (max MAF>10%), 10/19 (52%) also had LOH, compared to 86% in tissue (Jonsson 2019). Homozygous deletions were enriched in BRCA2 (6.9% of all samples), ATM (4.4%) and CHEK1 (2.3%). Rearrangements, including fusions and multi-exonic deletions, accounted for 6.5% of inactivating HRD mutations detected. In total, 24% of prostate samples had a biallelic inactivation involving an SNV, Indel or deletion. Conclusion: We demonstrate in a prostate cancer cohort that GuardantOMNITM ctDNA profiling calls all classes of mutations contributing to HRD, with relative prevalence of alterations consistent with those in tissue. CfDNA presents a potential alternative for identifying patients who may benefit from PARP or cisplatin/platinum therapies, expanding the prevalence from 28% using small variants to 42% with the complete HRD biomarker set. Citation Format: Jennifer Yen, Sante Gnerre, Catalin Barbacioru, Elena Helman, Ohad Manor, Arielle Yablonovitch, Ravi Vijaya Satya, Leo Liu, Jennifer Saam, Stephen Fairclough, Becky Nagy, Richard Lanman, Darya Chudova, AmirAli Talasaz. Landscape of homologous recombination repair (HRR) mutations in prostate cancer profiled by ctDNA next-generation sequencing [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 729.

3545 Background: FGFR2/ 3 rearrangements are promising therapeutic targets, especially in advanced urothelial cancer (aUC) with FDA-approved erdafitinib. Liquid biopsy is an attractive non-invasive method to identify these fusions, but detection in cfDNA is technically challenging due to low tumor shedding levels, short molecules, and wide variation in gene partners. To address this, we developed an assembly-based fusion detection algorithm to call rearrangements in a de novo fashion without reliance on a fixed partner set and applied it to > 15,000 clinical samples. Methods: A cohort of 15,218 patients with mixed cancer types (including 698 aUC patients, as well as breast, cholangiocarcinoma, colorectal, and gastric), plus 276 healthy control samples were previously tested with Guardant360(R), a clinical 74-gene cfDNA NGS-based assay. The median unique molecule coverage was approximately 3,000 molecules sequenced to 15,000x read depth. Samples were reanalyzed in silico using the novel algorithm: in brief, reads aligned to candidate fusion breakpoints were assembled into de Bruijn graphs. Resulting contigs were aligned to the reference and filters were applied to remove technical artifacts. Results: The majority of FGFR2 (86%) and FGFR3 fusion partners (73%) in the mixed cancer cohort were observed only once, consistent with previous reports (Helsten 2016). FGFR3-TACC3 was the most common fusion, occurring in 72% of FGFR3 fusion-positive patients. In 37% of FGFR2 fusion positive patients, the de novo caller detected partners not previously described. In the aUC cohort, FGFR3 fusions were detected in 3.3% of patients, with 8/10 (80%) partner genes/intergenic regions occurring only once, which is in line with previous reports (Nassar 2018). No fusions were identified in 276 healthy control samples. In the mixed cancer cohort, common mutations co-occurring with FGFR2 fusions were FGFR2 N549K, PIK3CA H1047R, and TP53 R175H (5.6% each); KRAS Q61H was observed in 28% of patients with FGFR3 fusions. Conclusions: FGFR2/ 3 fusion partners detected by a highly specific assembly-based de novo fusion caller were heterogeneous and individually rare, highlighting the importance of a de novo approach. We observed an FGFR3 fusion prevalence in cfDNA from aUC patients that is comparable to previous reports for tissue testing, demonstrating an ability to capture targetable genomic rearrangements with plasma-based NGS in this patient population.

A key resistance mechanism to platinum-based chemotherapies and PARP inhibitors in BRCA-mutant cancers is the acquisition of BRCA reversion mutations that restore protein function. To estimate the prevalence of BRCA reversion mutations in high-grade ovarian carcinoma (HGOC), we performed targeted next-generation sequencing of circulating cell-free DNA (cfDNA) extracted from pretreatment and postprogression plasma in patients with deleterious germline or somatic BRCA mutations treated with the PARP inhibitor rucaparib. BRCA reversion mutations were identified in pretreatment cfDNA from 18% (2/11) of platinum-refractory and 13% (5/38) of platinum-resistant cancers, compared with 2% (1/48) of platinum-sensitive cancers (P = 0.049). Patients without BRCA reversion mutations detected in pretreatment cfDNA had significantly longer rucaparib progression-free survival than those with reversion mutations (median, 9.0 vs. 1.8 months; HR, 0.12; P < 0.0001). To study acquired resistance, we sequenced 78 postprogression cfDNA, identifying eight additional patients with BRCA reversion mutations not found in pretreatment cfDNA. Significance: BRCA reversion mutations are detected in cfDNA from platinum-resistant or platinum-refractory HGOC and are associated with decreased clinical benefit from rucaparib treatment. Sequencing of cfDNA can detect multiple BRCA reversion mutations, highlighting the ability to capture multiclonal heterogeneity. This article is highlighted in the In This Issue feature, p. 151

Introduction The genomic alterations driving resistance to third-generation EGFR tyrosine kinase inhibitors (TKIs) are not well established, and collecting tissue biopsy samples poses potential complications from invasive procedures. Cell-free circulating DNA (cfDNA) testing provides a noninvasive approach to identify potentially targetable mechanisms of resistance. Here we utilized a 70-gene cfDNA next-generation sequencing test to interrogate pretreatment and progression samples from 77 EGFR-mutated non-small cell lung cancer (NSCLC) patients treated with a third-generation EGFR TKI. Patients and Methods Rociletinib was evaluated in advanced or metastatic (second line or higher) disease with EGFR T790M-positive NSCLC in the TIGER-X ( NCT01526928 ) and TIGER-2 ( NCT02147990 ) studies. Plasma samples were collected at baseline and at the time of systemic progression while receiving rociletinib. The critical exons in 70 genes were sequenced in cfDNA isolated from plasma samples to elucidate a comprehensive genomic profile of alterations for each patient. Results Plasma-based cfDNA analysis identified 93% of the initial EGFR activating and 85% of the EGFR T790M resistance mutations in pretreatment samples with detectable tumor DNA. Profiling of progression samples revealed significant heterogeneity, with different variant types (eg, mutations, amplifications, and fusions) detected in multiple genes (EGFR, MET, RB1) that may be driving resistance in patients. Novel alterations not previously described in association with resistance to third-generation TKIs were also detected, such as an NTRK1 fusion. Conclusion cfDNA next-generation sequencing identified initial EGFR activating and secondary T790M resistance mutations in NSCLC patients with high sensitivity, predicted treatment response equivalent to tissue analysis, and identified multiple novel and established resistance alterations.

With the advent of next-generation sequencing techniques, tumor mutations that guide therapeutic decisions can be identified on a large scale. The ability to accurately detect these mutations is often impaired by obstacles such as sequencing coverage gaps, small sample amount, low cellularity, and tumor heterogeneity. Two methods commonly used to interrogate the genomic profile of a tumor are: i) cancer gene panels, where a small number of genes (in the range of ~5-500, depending on the specific panel design) is assessed, and ii) cancer exome sequencing, where most human genes are evaluated. Additionally, tumor RNA can be studied to elucidate gene fusions and expression features that may impact treatment. Depending on the goals of the investigator, both methods have certain advantages and disadvantages. Differences between these assays must be clearly understood in order for cancer researchers to make the best decision to achieve their objective. We developed a set of unique sequencing assays to augment and improve detection of major cancer mutations by enhancing coverage over known sequencing gaps and GC-rich regions. We utilize these protocols to create an accuracy and content enhanced whole-exome (ACE Exome) as well as an extended, augmented cancer gene panel (ACE Extended Cancer Panel), targeting over 1,300 cancer genes and 200 miRNA genes at very high depth. We compared these ACE assays to standard exome and gene panels currently available. We then compared the ACE panel to the ACE exome using a set of well-characterized cancer cell lines, cell line mixtures with known genomic perturbations, and primary tumor and matched normal pairs involved in drug resistance. We also performed whole-transcriptome and targeted RNA sequencing as companion assays to the DNA exome and panel, respectively. All data were analyzed using a cancer bioinformatics pipeline optimized for detection of low-representation small variants and indels, as well as somatic copy-number aberrations, gene expression and fusions. The ACE assays were able to detect somatic variants with higher accuracy than standard assays across regions targeted by both. For example, a low-representation BRAF mutation detected by standard exome was confirmed as a false positive using the ACE exome due to 90 times more coverage over that region. Between the ACE assays, we found that 88% of somatic variants identified through ACE exome were also identified with our comprehensive panel within regions targeted by both. Conversely, more than half of the somatic variants discovered in the ACE panel were missed in the ACE exome due to insufficient read depth. For instance, the panel detected two EGFR variants known to be present at 1% that were missed in the ACE exome. About 20 times more somatic variants across some ~18,000 genes were detected in whole-exome that are not targeted on the augmented gene panel. These genes have not yet been associated with cancer in current literature, but may be of potential interest for discovery research. Copy number profiles, although more comprehensive in the exome, were recapitulated using the panel, despite large distances between panel targets. Finally, the integration of RNA analysis to the DNA panel was able to detect expression and fusion patterns. In particular, we detected a DNA mutation associated with resistance in a post-treatment specimen, but accompanying RNA data proved this mutant unexpressed and thereby unlikely to be the driver of resistance. For NGS-based cancer analysis to effectively guide clinical decisions and research, accuracy across all relevant cancer genes is of paramount importance. An enhanced exome may offer a solution for researchers interested in novel cancer associations who desire broad, even coverage. A comprehensive, high-depth panel targeting a large set of genes associated with cancer, accompanied by RNA analysis, may accomplish research goals across a wide range of translational cancer research. Citation Format: Elena Helman, Michael James Clark, Sean Boyle, Richard Chen, Shujun Luo, Christian Haudenschild, Jason Harris, Gabor Bartha, Deanna Church, John West. Augmented targeted NGS in cancer diagnostics: Comparing gene panels and whole exome sequencing for accurate detection of driver mutations. [abstract]. In: Proceedings of the AACR Special Conference on Translation of the Cancer Genome; Feb 7-9, 2015; San Francisco, CA. Philadelphia (PA): AACR; Cancer Res 2015;75(22 Suppl 1):Abstract nr A2-39.

Retrotransposons constitute a major source of genetic variation, and somatic retrotransposon insertions have been reported in cancer. Here, we applied TranspoSeq, a computational framework that identifies retrotransposon insertions from sequencing data, to whole genomes from 200 tumor/normal pairs across 11 tumor types as part of The Cancer Genome Atlas (TCGA) Pan-Cancer Project. In addition to novel germline polymorphisms, we find 810 somatic retrotransposon insertions primarily in lung squamous, head and neck, colorectal, and endometrial carcinomas. Many somatic retrotransposon insertions occur in known cancer genes. We find that high somatic retrotransposition rates in tumors are associated with high rates of genomic rearrangement and somatic mutation. Finally, we developed TranspoSeq-Exome to interrogate an additional 767 tumor samples with hybrid-capture exome data and discovered 35 novel somatic retrotransposon insertions into exonic regions, including an insertion into an exon of the PTEN tumor suppressor gene. The results of this large-scale, comprehensive analysis of retrotransposon movement across tumor types suggest that somatic retrotransposon insertions may represent an important class of structural variation in cancer., National Cancer Institute (U.S.) (grant U24CA143867), National Cancer Institute (U.S.) (grant U24CA126546)

Background: In addition to BRCA1/2 germline and somatic inactivating mutations, loss of heterozygosity (LOH) and biallellic somatic copy number loss are associated with BRCA1/2 loss of function. Patients with cancers harboring these somatic features may benefit from treatment with PARP inhibitors. ctDNA analysis has proven a viable alternative to tumor tissue genotyping, especially in tissue- or time-limited clinical scenarios. However, ctDNA assessment of LOH and biallelic copy number loss is challenging given that ctDNA is diluted by cell-free leukocytic DNA making it difficult to confidently call these genomic events. We developed a method to identify somatic biallelic copy number loss in ctDNA using targeted sequencing of cell-free (cfDNA) across a wide range of cancer types. Methods: A novel statistical model was developed using coverage profiles and single nucleotide polymorphism (SNP) allelic frequencies estimated from plasma samples to determine the presence of LOH and biallellic copy number loss in BRCA1 or BRCA2. For each gene of interest, the model uses observed coverage and fragment size distribution, together with allele frequency of germline SNPs. The model was applied to plasma samples from 28,000 patients with advanced solid tumors sequenced using a 73-gene next generation sequencing ctDNA panel (Guardant360®, Guardant Health, Redwood City, CA). Results: The model was analytically validated using in-silico simulations in order to assess both the limit of blank and limit of detection. This method shows 95% sensitivity in detecting LOH and bi-allellic copy number loss for samples with a maximum somatic variant allele frequency as low as 9%. Sensitivity is mainly driven by the panel size and the depth of coverage of the gene of interest. The model was then applied to the 28,199 patient cohort. BRCA1 and BRCA2 LOH was observed in 2.4% (134/5568) and 7.4% (415/5568) of classic homologous recombination deficient (HRD) cancers including breast, ovarian, prostate, and pancreas. BRCA1 and BRCA2biallelic copy number loss was observed in 0.3% (19/5568) and 0.6% (31/5568) of this same group of HRD cancers. Discussion: In this cohort of 5,568 patients with classic HRD associated cancers, somatic LOH and biallelic copy number loss was detected in BRCA1 in 2.7% of samples and in BRCA2 in 8.0% of samples, which is aligned with previously reported tissue prevalence. BRCA1/2 somatic LOH and biallelic copy number loss can accurately be detected in ctDNA utilizing likelihood-based scores based on coverage and germline SNPs allele frequencies. The ability to identify this therapeutically targetable genomic alteration through a non-invasive ctDNA assessment has significant clinical implications. Citation Format: Catalin Barbacioru, Victoria M. Raymond, Marcin Sikora, Elena Helman, Sante Gnerre, Stephen Fairclough, Darya Chudova, Richard B. Lanman, AmirAli Talasaz. Cell-free circulating tumor DNA (ctDNA) detects somatic copy number loss in homologous recombination repair genes [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 435.

Background: Clonal hematopoiesis (CH) is the acquisition of mutations in hematopoietic progenitor cells that can lead to clonal expansion. Recent studies suggest that CH-derived mutations can confound interpretation of cell-free DNA (cfDNA) sequencing results. To better understand the contribution of CH to cfDNA analysis in the metastatic cancer setting, we characterized CH-associated alterations observed in the cfDNA of late-stage cancer patients. Methods: We analyzed somatic variants from cfDNA profiles of over 62,000 patients in four late-stage cohorts: lung (>35,000), gastro-intestinal (>14,000), urogenital (>4,700), breast (>8,600). Matched white blood cell (WBC) and cfDNA was obtained for a subset of patient samples. Plasma cfDNA was processed using a 73-gene (150Kb, Guardant360TM) or 500-gene (2Mb, GuardantOMNITM) panel and sequenced to average depth of ~5000 molecules, with a 95% limit of detection (LOD) of 0.3-0.4% and 0.15-0.6% variant allele fraction (VAF) for SNVs, and 0.2%-0.7% and 0.4-0.8% for indels (for G360 and OMNI, respectively). Results: In samples processed on the 500-gene panel, DNMT3A, TET2, PPM1D, ASXL1 and SF3B1 were the most frequently mutated CH-associated genes (75% of samples). While the majority (73%) of these mutations were at low variant allele fractions (VAF) of 60,000 samples, JAK2 V617F, a known CH mutation, was observed in 927 samples with VAFs between 0.29% and 98%. Across the combined cohort, we found that in samples with a known CH variant (n=7,717), half of samples had a max CH VAF less than 50% of max tumor VAF. Interestingly, in >25% of samples, the max CH VAF was higher than the max tumor VAF. Initial comparisons of matched plasma and WBC DNA show that 100% (16/16) of clinically relevant and 92% (84/91) of variants of unknown significance (VUSs) across the 73-gene panel were found exclusively in plasma DNA and not in WBC DNA testing. Conclusions: We characterize the distribution of CH mutations across a large number of late-stage plasma samples and show that within highly pre-treated metastatic patients, the VAFs of CH mutations often differ from the tumor and can surpass the level of tumor shedding in circulation. Further investigation of these variants will enable improved understanding of CH in metastatic disease and its differentiation from the circulating tumor DNA. Citation Format: Jennifer Yen, Katie Quinn, Elena Helman, Andrey Chursov, Tracy Nance, Ariel Jaimovich, Kimberly Banks, Aleksandra Franovic, Kristin Gleitsman, John Latham, Arielle Yablonovitch, Marcin Sikora, Stephen Fairclough, Darya Chudova, Richard B. Lanman, AmirAli Talasaz. Analysis of clonal hematopoiesis-associated mutations in the cell-free DNA of advanced cancer patients [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 2509.

Background: Tumor mutational burden (TMB) has emerged as a predictive biomarker of response to immune checkpoint inhibitor (ICI) therapy. Current panel-based TMB algorithms aggregate signal from certain types of somatic variants (e.g. non-synonymous coding SNVs); however, delineating the contributions of these and other types of mutations may refine TMB calculation from gene panels. Moreover, early studies suggest other possible genomic correlates of patient outcome to ICI which may be complementary to TMB. Here, we explore the landscape of mutations comprising TMB and other genomic features correlating with TMB on a subset of several thousand late-stage plasma samples run on GuardantOMNITM (OMNI), a highly sensitive 500-gene cfDNA sequencing platform. Methods: We developed a cfDNA-based TMB algorithm which is robust to variable tumor shedding levels and presence of clonal hematopoiesis. We assessed cfDNA-based TMB in over 1,000 plasma samples across six tumor types, including lung and prostate. We examine the contribution of nonsynonymous, synonymous, intronic SNVs, and indels to TMB score. We investigate correlations between TMB and additional genomic features, including chromosomal instability, loss of HLA-bearing chromosome 6p, microsatellite instability (MSI), and common oncogenic and resistance mutations. Results: We found that the distribution of WES-calibrated TMB scores across this cohort of samples is consistent with TCGA, with median 10 mutations/Mb and upper-tertile of 14 mutations/Mb across tumor types. The number of non-synonymous coding SNVs per sample correlated highly with synonymous coding SNV and intronic SNV counts (Pearson’s r > 0.7 for each). Including this additional signal in TMB calculation improves clinical sensitivity by up to 5%. In MSS samples, indels were highly correlated with SNVs, indicating that both likely arise from a similar underlying mechanism. We found no clear correlation between high TMB and chromosomal instability, with high TMB samples exemplifying a range of tumor ploidies. TMB association with oncogenic drivers is consistent with existing literature, with lower median TMB in EGFR-driven lung tumors (p < 0.01), but little to no correlation between TMB and KRAS or PIK3CA driver status, or STK11 loss of function (p > 0.05), suggesting these latter events could be independent clinical biomarkers to TMB. Conclusions: Panel-based TMB scores can leverage synonymous and non-coding mutations to strengthen the signal of exome-wide mutation load. As more patient outcome data becomes available, TMB algorithms and orthogonal biomarkers of tumor genome immunogenicity will evolve further for improved guidance of patient response to immunotherapy. Sequencing panels with high sensitivity for TMB, via large panel space, and the ability to detect copy-number variations and MSI-status, will be important for biomarker development and clinical applications. Citation Format: Katie Quinn, Elena Helman, Tracy Nance, Jennifer Yen, John Latham, Kristin Gleitsman, Ravi Vijaya-Satya, Carlo Artieri, Alex Artyomenko, Marcin Sikora, Darya Chudova, Richard B. Lanman, AmirAli Talasaz. Landscape and genomic correlates of ctDNA-based tumor mutational burden across six solid tumor types [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 3404.

Background: Late stage therapy selection is advancing beyond the detection of classical driver alterations (SNV, Indel, CNV, Fusion) to the use of aggregate biomarkers comprising combinations of SNVs, Indels, and fusions, that reflect the underlying mutational process, indicate the cellular dysfunction, and suggest the neoantigen signature. GuardantOMNI (OMNI), is a highly sensitive 500-gene cfDNA sequencing test designed to detect primary somatic mutations as well as aggregate biomarkers including TMB and DDR pathway defects. The newest version of OMNI introduces MSI biomarker detection, which is an indication for the first FDA-approved pan-cancer therapy. Here we present the first analytical validation study of OMNI MSI-H detection. Methods: Probabilistic models for MSI detection were designed to interrogate instability across >1300 loci, differentiating PCR and sequencing induced artifacts from true biological signals using molecular barcoding as implemented in the Guardant digital sequencing platform. Cut-offs were selected based on the analysis of the training healthy donor cohort and locked prior to validation. Analytical performance was assessed using pre-characterized cell lines, cancer-free donor-samples, and clinical patient samples. Qualitative and quantitative orthogonal confirmation was provided by next-generation sequencing (NGS) and published information. Results: 350 samples were processed for this study, using both 5ng and 30ng cfDNA input levels, and all samples passed sequencing QC metrics established prior to testing. Cell line-based dilution studies performed in triplicate spanning 5ng and 30ng input range as well as in silico simulation demonstrated 95% limit of detection (LoD) at 0.1% Analysis of 121 healthy donor and 16 MSS advanced cancer patient samples identified no false positives (FP). In 240 patient samples with orthogonal confirmation of MSI status from tissue or panel NGS, GuardantOMNI demonstrated 100% analytic sensitivity (3/3) and 100% analytic specificity (237/237) with a PPV of 1 and and NPV 1. Retrospective analysis of 2000+ clinical samples detected MSI high in 16% of colorectal cancers, 4% of prostate cancers, and 0.6% of non-small cell lung cancers. Conclusions: Here we have demonstrated the performance of MSI-H detection with GuardantOMNI (OMNI), which has the potential to accelerate clinical trial enrollment, research and discovery with a single, non-invasive blood sample. Citation Format: Jing Zhao, Alexander Artyomenko, Carlo Artieri, John Latham, Stephen R. Fairclough, Catalin Barbacioru, Elena Helman, John Strickler, Darya Chudova, Richard Lanman, AmirAli Talasaz. Analytical validation of MSI High detection with GuardantOMNI [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 1675.

Major international projects are underway that are aimed at creating a comprehensive catalogue of all the genes responsible for the initiation and progression of cancer. These studies involve the sequencing of matched tumour-normal samples followed by mathematical analysis to identify those genes in which mutations occur more frequently than expected by random chance. Here we describe a fundamental problem with cancer genome studies: as the sample size increases, the list of putatively significant genes produced by current analytical methods burgeons into the hundreds. The list includes many implausible genes (such as those encoding olfactory receptors and the muscle protein titin), suggesting extensive false-positive findings that overshadow true driver events. We show that this problem stems largely from mutational heterogeneity and provide a novel analytical methodology, MutSigCV, for resolving the problem. We apply MutSigCV to exome sequences from 3,083 tumour-normal pairs and discover extraordinary variation in mutation frequency and spectrum within cancer types, which sheds light on mutational processes and disease aetiology, and in mutation frequency across the genome, which is strongly correlated with DNA replication timing and also with transcriptional activity. By incorporating mutational heterogeneity into the analyses, MutSigCV is able to eliminate most of the apparent artefactual findings and enable the identification of genes truly associated with cancer.

Aberrant DNA hypermethylation of tumor suppressor genes is thought to be an early event in tumorigenesis. Many studies have reported the methylation status of individual genes with known involvement in cancer, but an unbiased assessment of the biological function of the collective of hypermethylated genes has not been conducted so far. Based on the observation that a variety of human cancers recapitulate developmental gene expression patterns (that is activate genes normally expressed in early development and suppress late developmental genes), we hypothesized that the silencing of differentiation-associated genes in cancer could be attributed in part to DNA hypermethylation. To this end, we investigated the developmental expression patterns of genes with hypermethylated CpG islands in primary human lung carcinomas and lung cancer cell lines. We found that DNA hypermethylation primarily affects genes that are expressed in late stages of murine lung development. Gene ontology characterization of these genes shows that they are almost exclusively involved in morphogenetic differentiation processes. Our results indicate that DNA hypermethylation in cancer functions as a selective silencing mechanism of genes that are required for the maintenance of a differentiated state. The process of cellular de-differentiation that is evident on both the microscopic and transcriptional level in cancer might at least partly be mediated by these epigenetic events. Our observations provide a mechanistic explanation for induction of differentiation upon treatment with DNA methyltransferase inhibitors.

Background: Translational research and enrollment in clinical trials are limited by the rarity of individual mutations and lack of sufficient tissue for comprehensive testing. To address these limitations, we developed GuardantOMNI (OMNI), a highly sensitive 500-gene cfDNA sequencing test requiring as little as 2 mL of plasma and designed for broad genomic detection of somatic single-nucleotide variants (SNVs) and small indels in 497 genes, copy number amplifications (CNAs) in 106 genes, and fusions in 21 genes. Additionally, the OMNI panel enables assessment of tumor mutational burden (TMB), and DNA damage and mismatch repair, with coverage of over 30 genes associated with the DDR pathway. Here we present the first analytical validation study of OMNI. Methods: Analytical performance was assessed as per Nex-StoCT Working Group guidelines using precharacterized cell lines and healthy normal donor-derived samples. Qualitative and quantitative orthogonal confirmation was provided by exome sequencing, microarrays, and data from published compendia. Results: Seventy-three validation and 150 development plasma samples were processed for this study, using both 5ng and 30ng cfDNA input levels, and all samples passed sequencing QC metrics established prior to testing. Reportable ranges for SNVs were ≥0.04% variant allele fraction (VAF), ≥0.02% for indels, ≥2 supporting molecules for fusions, and ≥2.18 copies for CNAs. Cell line-based dilution studies demonstrated 95% limits of detection (LoD) of 0.24-0.6% VAF for SNVs (depending on known cancer association), 0.4-0.8% for non-homopolymeric indels (depending on clinical relevance), 0.1-0.2% for fusions, and 2.2-2.9 copies for 90% of CNA genes targeted. Comparison of diluted cell line and healthy donor samples to orthogonal sequencing and published genotype data demonstrated accuracies of 98.7% for SNVs, 97.2% for indels, and 100% for CNAs and fusions across the reportable range. The analytical false-positive rate per sample measured across 24 healthy donors was 0.25 for SNVs, 0.04 for indels, and 0 for CNAs and fusions, with positive predictive values (PPVs) of 97.5% for SNVs, 98% for indels, and 100% for CNAs and fusions. Quantitative correlation of allele fraction with confirmatory methods was high (r2 > 0.99). Conclusions: To our knowledge, OMNI is the largest comprehensive ctDNA cancer gene panel available. It detects alterations in genes under study in over 98% of current clinical trials with sensitivity, specificity, and accuracy similar to currently available targeted ctDNA sequencing tests. OMNI has the potential to accelerate clinical trial enrollment, research and discovery with a single, noninvasive blood sample. Citation Format: Elena Helman, Carlo Artieri, James V. Vowles, Jennifer Yen, Tracy Nance, Marcin Sikora, Joshua Gourneau, Mohit Goel, Stefanie Mortimer, Darya Chudova, Justin Odegaard, Richard B. Lanman, AmirAli Talasaz. Analytical validation of a comprehensive 500-gene ctDNA panel designed for immuno-oncology and DNA damage research [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 5603.

Background: A recent large study found that over 2% of advanced cancer patients have unidentified germline alterations found incidentally during next-generation sequencing (NGS) for targetable somatic alterations. However, tissue-based NGS cannot definitively distinguish germline from somatic mutations without comparison to normal tissue. Because somatic variants typically occur in plasma at allele fractions 1-2 orders of magnitude lower than germline, liquid biopsy often enables differentiation of incidental germline mutations. Not uncommonly, high tumor burden or allelic imbalance from copy number variation or loss of heterozygosity complicates simplistic differentiation based on assumption of germline 50% allele frequency. We developed a novel statistical method for discriminating between somatic and germline variants in deep sequencing data, and applied to data generated from the Guardant 360 targeted cfDNA assay. Our model dynamically incorporates prior knowledge of germline SNP variants and observed allele frequencies within a sample, and gains power in larger panels like the GuardantOmni 500-gene assay. Methods: Our method uses common heterozygous SNPs to model local germline allele count behavior, and calls variants somatic if they deviate significantly from the observed germline mutant allele fraction (MAF). A betabinomial model is well suited to this problem, because it models both the mean and variance of mutant allele counts at common SNPs. This is important since simpler methods like fixed MAF cutoffs or Poisson models may not represent the variance in molecule counts appropriately. Results: We evaluated our method against variants from 361 clinical samples that were manually annotated by reviewers with expertise in cancer genomics. Of the 11,679 heterozygous variants considered in these samples, 7,221 (62%) had two or more common SNPs in the same gene, so that a local germline heterozygous MAF could be estimated and the betabinomial model could be applied. Using the manual annotation as ground truth, we created ROC curves for a classifier based on the betabinomial model with varying p-value cutoffs, as well as a classifier based on MAF cutoffs. The AUC for the betabinomial model was 0.996, as compared with an AUC of 0.985 for MAF cutoffs. The ROC curve allowed us to choose a betabinomial p-value cutoff that maintained a similar level of specificity as a 15% MAF cutoff (99.2% for MAF cutoff, 99% for betabinomial), but with significantly increased sensitivity (89% for MAF cutoff, 97% for betabinomial). Conclusions: We have developed a novel method for differentiation of somatic versus germline variants in a single plasma sample. Identifying incidental germline alterations in advanced cancer patients may identify new opportunities to apply targeted therapy such as PARP inhibitors for BRCA1/2, and may also serve to alert physicians and their patients to familial risk. Citation Format: Tracy Nance, Elena Helman, Carlo Artieri, Jennifer Yen, Thomas P. Slavin, Darya Chudova, Richard B. Lanman, AmirAli Talasaz. A novel approach to differentiation of somatic vs. germline variants in liquid biopsies using a betabinomial model [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 4272.

Background: Duplicated genomic regions pose a challenge to accurate variant calling in clinical sequencing applications as duplicate-specific variants may incorrectly be assigned to the target. One source of duplicated coding sequences are processed pseudogenes (PPGs) that can originate from LINE-mediated reverse transcription and genomic integration of processed mRNA, resulting in partial or complete copies of the original gene, lacking intronic sequences. False-positive variants resulting from pseudogenes found in the reference genome, such as those of PIK3CA and PTEN, have been well studied; however, the recent discovery of rare and even individual-specific cancer-related PPGs demonstrates a need for more systematic interrogation and mediation of PPG-related clinical artifacts on a sample-by-sample basis. Methods: We analyzed germline and somatic alterations in over 15,000 clinical circulating cell-free DNA samples sequenced using a 73-gene panel (Guardant Health, CA) focused primarily on clinically-relevant somatic alterations. All targeted genes were interrogated for two key hallmarks of PPGs: 1) split sequencing reads mapping to adjacent exons without intervening intronic sequence and 2) atypically increased sequencing coverage of exons relative to adjacent introns.Results: Applying a strict cutoff of ≥5 unique molecules supporting any putative intron-skipping event, we observed evidence of 94 PPGs in 91 of 15,315 analyzed samples (0.59%), affecting SMAD4 (57 samples), CDK4 (20), GNAS (6), HRAS (2), KRAS (1), TP53 (1), ESR1 (1), STK11 (1), RB1 (1), MAPK3 (1), MAPK1 (1), NF1 (1), and GATA3 (1). To determine the impact of PPGs on specificity, we applied a resampling test to assess whether somatic variant calling without PPG aware detection of SNVs results in over-representation of calls near splice junctions in samples harboring PPGs. Indeed, SMAD4 somatic SNVs were identified in 30% of samples containing SMAD4 PPGs as compared to an expected 0.7% of non-PPG samples (two-tailed p < 10E-4). Similarly, 10% of samples containing CDK4 PPGs also showed somatic CDK4 SNVs, whereas only 0.9% would be expected by chance (two-tailed p = 0.03). These results strongly suggest that PPG-related artifacts may be misclassified as true somatic events in these contexts. Algorithmically restricting variant-calling only to those reads containing intronic sequence can maintain high specificity with only minimal cost to sensitivity in most exons. Conclusions: Our analysis reveals the existence of rare PPGs derived from over a dozen common clinical cancer sequencing targets and provides strong evidence that these may produce false positive SNV calls in an appreciable number of cases. Importantly, we have also shown that PPGs can be detected in a sample-specific manner based on the presence of intron-skipping split reads, enabling algorithmic detection and elimination of artifacts. Citation Format: Carlo G. Artieri, Marcin Sikora, Alex Artyomenko, Elena Helman, Darya Chudova, Richard Lanman, AmirAli Talasaz. Accounting for processed pseudogene-related artifacts improves specificity of clinical cancer diagnostic sequencing [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 636.

Background: DCC-2618 is a potent switch control inhibitor of KIT and PDGFRα kinases maintains potent inhibition of mutant forms across all exon regions in preclinical models. Gastrointestinal stromal tumor (GIST) is an important disease to achieve a proof-of-concept due to the heterogeneity of resistance mutations in KIT which emerge on treatment with approved KIT inhibitors. In later lines of therapy resistance mechanisms independent of the KIT gene have also been described. Methods: The ongoing phase 1, PK-guided dose escalation study of DCC-2618 given orally BID [28-day cycle] tested doses from 20 mg to 200 mg in patients (pts) with advanced solid tumors including GIST (NCT02571036). We report preliminary longitudinal results of plasma cell free (cf) DNA sequencing by Guardant 360 collected throughout the study and levels of circulating tumor cells (CTCs) based on a viral telomerase promoter-driven GFP expression assay. Results: To date, 24 out of 31 enrolled pts had metastatic KITm GIST refractory to standard therapy. A high, total mean exposure of DCC-2618 and its active metabolite was achieved at 100 and 150 mg BID, affording steady state Cmax >5 µM in Cycle 1. Starting with 50 mg BID dose level, concentrations of total drug exceeding IC90 of the most resistant mutations to DCC-2618 were achieved. Next-generation sequencing of plasma cfDNA revealed a total of 40 KIT mutations in 16 of 18 GIST pts at baseline. DCC-2618 led to rapid decrease and/or clearance of the heterogeneous array of KIT mutations from plasma cfDNA including exons 9, 11, 13, 14, 17, and 18. Independent of suppressed KIT mutation burden, longitudinal monitoring of cfDNA revealed changes in non-KIT oncogenic mutations which may contribute to heterogenous mechanisms of resistance. KIT mutation burden will be correlated with metabolic response assessment by PET scans and exposure to DCC-2618. CTCs have been detected in blood from all GIST patients at baseline using a non-biased assay capable of identifying sarcoma cells. Preliminary result show that CTCs with immunofluorescent detection of KIT or PDGFRα, corresponding to their respective mutational status, show 1 of 3 patterns when compared to radiologic response: most pts show relatively stable low levels at stable disease (SD), a minority of pts with prolonged SD a decline in CTCs and 2 pts with progressive disease had significant increase in KIT positive CTCs. Conclusions: DCC-2618 and its active metabolite achieved high plasma concentrations sufficient to inhibit the most resistant KIT mutations at well-tolerated exposures. Monitoring of cfDNA KIT mutation frequency demonstrates rapid clearance of a broad spectrum of KIT mutations in this heavily pretreated GIST patient population and suggests candidate resistance genes that are independent of KIT. Our data provide a first signal that CTC monitoring might represent a potential marker for tumor control in KIT mutant GIST. Citation Format: Filip Janku, Albi Razak, Michael Gordon, David Brooks, Daniel Flynn, Anu Gupta, Michael Kaufman, Cynthia Leary, Bryan Smith, Deb Westwood, Neeta Somaiah, Elena Helman, Eric Gerstenberger, Oliver Rosen, Suzanne George. Translational research in a phase I proof-of-concept study supports that DCC-2618 is a pan-KIT inhibitor [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr LB-039. doi:10.1158/1538-7445.AM2017-LB-039

Background: First and second-generation EGFR tyrosine kinase inhibitors (TKIs) have benefited patients with EGFR-mutated non-small cell lung cancer (NSCLC), but resistance invariably develops after a median of 9-16 months. In ~60% of patients, resistance is mediated by a second mutation in EGFR, namely T790M. Hence, third-generation EGFR TKIs such as osimertinib and rociletinib were developed to target both activating EGFR mutations as well as T790M. Unfortunately, patients also develop resistance to these therapies through mechanisms that have not yet been thoroughly explored. Since repeat tissue biopsies pose potential complications from invasive procedures, circulating tumor DNA (ctDNA) testing is increasingly used in the clinical setting to identify potentially targetable mechanisms of resistance. Methods: Matched pre-treatment and progression plasma from 57 patients with EGFR-mutated NSCLC treated with rociletinib were profiled using a 70-gene ctDNA targeted next-generation sequencing panel (Guardant360) that detects somatic single nucleotide variants, short insertions and deletions, fusions, and copy number variants. Pre-treatment EGFR ctDNA allele fractions were also determined by BEAMing, a technique based on droplet digital PCR followed by flow cytometry. Pre-treatment tumor EGFR status was assessed by the therascreen EGFR test. Results: In all 57 pre-treatment samples profiled, plasma-based ctDNA analysis detected the initial EGFR driver and T790M resistance mutations that were identified in the matched tumor. Interestingly, we found that 12% (7/57) of patients had evidence of compound EGFR driver mutations at baseline, including E709A-L858R, K860I-L858R, and L718V-L858R. EGFR T790M mutations in plasma were observed subclonally (present on average at 40% of the allele fraction of the driver mutation), suggesting tumor heterogeneity at baseline. The correlation coefficients (r) between Guardant360 and BEAMing for EGFR L858R, Exon19Del, and T790M were 0.90, 0.92, 0.95, respectively. Upon progression on rociletinib, 5% of patients (3/57) developed the EGFR C797S resistance mutation, 5% (3/57) developed focal MET amplification, and 2% (1/57) developed a NTRK1 fusion that were not present in the matched baseline plasma. Additionally, 4 deleterious BRCA1/2 alterations (2 germline and 2 somatic) were identified, with the somatic alterations emerging at progression. In 14% (8/57) of the patients, mutations in genes involved in the RAS/RAF signaling pathway, including KRAS Q61H, KRAS K117N and NF1 Q1822*, emerged or increased at progression. Conclusions: Plasma ctDNA revealed heterogeneity and multiple mechanisms of resistance in rociletinib treated patients. Thus comprehensive ctDNA sequencing allows for the identification of potentially actionable alterations and may help inform the choice of next therapy for patients progressing on a third-generation EGFR TKI. Citation Format: Elena Helman, Andrew D. Simmons, Chris A. Karlovich, Thomas C. Harding, Mitch Raponi, Darya I. Chudova, Daniel A. Simon, Richard B. Lanman, AmirAli Talasaz. Comprehensive ctDNA sequencing reveals mechanisms of resistance to rociletinib in EGFR T790M-mutated NSCLC [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 1009. doi:10.1158/1538-7445.AM2017-1009

Summary Recent genomic analyses of pathologically defined tumor types identify "within-a-tissue" disease subtypes. However, the extent to which genomic signatures are shared across tissues is still unclear. We performed an integrative analysis using five genome-wide platforms and one proteomic platform on 3,527 specimens from 12 cancer types, revealing a unified classification into 11 major subtypes. Five subtypes were nearly identical to their tissue-of-origin counterparts, but several distinct cancer types were found to converge into common subtypes. Lung squamous, head and neck, and a subset of bladder cancers coalesced into one subtype typified by TP53 alterations, TP63 amplifications, and high expression of immune and proliferation pathway genes. Of note, bladder cancers split into three pan-cancer subtypes. The multiplatform classification, while correlated with tissue-of-origin, provides independent information for predicting clinical outcomes. All data sets are available for data-mining from a unified resource to support further biological discoveries and insights into novel therapeutic strategies.

Osteosarcoma is the most common primary bone tumor, yet there have been no substantial advances in treatment or survival in three decades. We examined 59 tumor/normal pairs by whole-exome, whole-genome, and RNA-sequencing. Only the TP53 gene was mutated at significant frequency across all samples. The mean nonsilent somatic mutation rate was 1.2 mutations per megabase, and there was a median of 230 somatic rearrangements per tumor. Complex chains of rearrangements and localized hypermutation were detected in almost all cases. Given the intertumor heterogeneity, the extent of genomic instability, and the difficulty in acquiring a large sample size in a rare tumor, we used several methods to identify genomic events contributing to osteosarcoma survival. Pathway analysis, a heuristic analytic algorithm, a comparative oncology approach, and an shRNA screen converged on the phosphatidylinositol 3-kinase/mammalian target of rapamycin (PI3K/mTOR) pathway as a central vulnerability for therapeutic exploitation in osteosarcoma. Osteosarcoma cell lines are responsive to pharmacologic and genetic inhibition of the PI3K/mTOR pathway both in vitro and in vivo.

The Cancer Genome Atlas (TCGA) Research Network has profiled and analyzed large numbers of human tumors to discover molecular aberrations at the DNA, RNA, protein and epigenetic levels. The resulting rich data provide a major opportunity to develop an integrated picture of commonalities, differences and emergent themes across tumor lineages. The Pan-Cancer initiative compares the first 12 tumor types profiled by TCGA. Analysis of the molecular aberrations and their functional roles across tumor types will teach us how to extend therapies effective in one cancer type to others with a similar genomic profile., National Cancer Institute (U.S.), National Human Genome Research Institute (U.S.)

Adenocarcinoma of the lung is the leading cause of cancer death worldwide. Here we report molecular profiling of 230 resected lung adenocarcinomas using messenger RNA, microRNA and DNA sequencing integrated with copy number, methylation and proteomic analyses. High rates of somatic mutation were seen (mean 8.9 mutations per megabase). Eighteen genes were statistically significantly mutated, including RIT1 activating mutations and newly described loss-of-function MGA mutations which are mutually exclusive with focal MYC amplification. EGFR mutations were more frequent in female patients, whereas mutations in RBM10 were more common in males. Aberrations in NF1, MET, ERBB2 and RIT1 occurred in 13% of cases and were enriched in samples otherwise lacking an activated oncogene, suggesting a driver role for these events in certain tumours. DNA and mRNA sequence from the same tumour highlighted splicing alterations driven by somatic genomic changes, including exon 14 skipping in MET mRNA in 4% of cases. MAPK and PI(3)K pathway activity, when measured at the protein level, was explained by known mutations in only a fraction of cases, suggesting additional, unexplained mechanisms of pathway activation. These data establish a foundation for classification and further investigations of lung adenocarcinoma molecular pathogenesis.

Tumor biopsies are often Formalin-Fixed and Paraffin-Embedded (FFPE) for histological staining, genetic testing and archival purposes. Formalin treatment preserves tissue by crosslinking proteins, but can lead to mutation of the nucleic acid bases that can be detected by next-generation sequencing (NGS) methods despite being unrelated to the cancer. Studies have shown that certain fixation protocols are compatible with high quality nucleic acid isolation and variant calling by NGS. These studies used tissue fixed with 10% neutral buffered formalin for 24 hours, a standard protocol in the pathology field. In practice, however, FFPE sample handling varies from site-to-site. As we processed FFPE samples from various sites, we found that the quality of the isolated nucleic acids and subsequent sequencing results varied substantially. We hypothesized that this is due to deviations from the standard protocol such as improperly buffered reagents, variation in the fixation time and temperature, microwaving, and varied storage conditions of the samples. To understand the role of formalin fixation on the quality of the isolated nucleic acids and subsequent NGS analyses, we subjected the widely studied cell line NA12878 to various formalin fixation conditions. We then performed an augmented target enrichment and sequencing assay on both the DNA and RNA isolated from fresh frozen (FF) and formalin fixed (FFPE) NA12878 samples. We assessed raw nucleic acid quality, library quality, alignment rate, duplication rate, on-target efficiency and variant concordance between FF and FFPE. We noted substantial negative effects on sequencing library quality associated with sample incubation in unbuffered formalin, at high temperatures, and long periods of time (e.g. 3 days rather than 1 day). These harshly treated samples also showed poor alignment qualities, higher duplication rates, and lower mapping qualities. At the small variant level, they showed an increase in global C>T deamination (CG->TA, 50% FFPE vs 35% in FF) and oxidation (CG->AT, 15% FFPE vs 10% FF) rates. Variants caused by formalin fixation were most commonly detected at low ( Using the results of this study, we intend to provide guidance on optimal fixation conditions for NGS applications. It is crucial to fix tissues in buffered formalin for less than 24hr at room temperature to preserve nucleic acid integrity and amenability for downstream NGS assays. We also demonstrate how a deeper understanding of the effects of formalin can improve variant calling from FFPE tissues, especially at lower AFs where formalin-related errors have the greatest impact. Citation Format: Ravi Alla, Shujun Luo, Elena Helman, Sean M. Boyle, Michael J. Clark, Kirk Scott, Parin Sripakdeevong, Mirian Karbelashvili, Deanna M. Church, Michael Snyder, John West, Richard Chen. Fix the fixation: effect of formalin fixation on targeted sequencing, variant calling and gene expression. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 3616.

Accurate detection of somatic variants is a staple of both research and clinical cancer analysis, with applications ranging from detecting new common driver mutations in large patient cohorts to selecting therapeutic small molecule treatment courses for an individual patient. Recent research into neoantigens and immunotherapy has shown great promise as a precision therapeutic, and somatic variant detection by next-generation sequencing represents an ideal method of identifying candidate neoantigens. Somatic variant detection typically involves assaying the DNA for changes in gene sequences without assessing whether those variants are actually expressed in RNA. However, the expression of small variants is key because only expressed peptides will be displayed as neoantigens on the cell surface. From a technical standpoint, detection of somatic variants in the RNA represents additional challenges above and beyond those of somatic detection in DNA. The widely varying expression levels of cancer genes, alternative splicing, and RNA editing are all features that make somatic variant calling in RNA uniquely challenging. However, accurately detecting variants directly from expressed transcripts is beneficial to neoantigen prediction, and therefore we sought to create and validate a method for somatic variant calling in RNA. We have designed a highly accurate expression-based somatic variant detection pipeline utilizing extensive discovery and filtering methods to overcome the challenges inherent in RNA somatic variant calling. We validated our pipeline using a combination of well-characterized cell lines, commercially available reference standards, and real world FFPE patient samples. To our knowledge, this is the most extensive validation of its kind to date, representing over 29,158 small variants across 39 samples. In testing, we measured our detection method at >99% sensitivity and >99% PPV using a combination of gold set small variants and orthogonal validation. This method, in combination with our validated DNA somatic variant calling pipeline (>99% sensitivity and >99% PPV), enables precise detection of variant expression levels in a given sample, even at low allele frequency (5%). After validating our RNA somatic variant calling method, we applied it to detect candidate neoantigens in patient tumor samples. We performed HLA typing for each sample using HLAssign software and predicted MHC presentation of the expressed somatic variants. In ongoing studies, we are validating our most promising putative neoantigens using orthogonal technologies and demonstrating our ability to detect the most promising clinically effective peptides for therapy. Citation Format: Sean M. Boyle, Michael J. Clark, Ravi Alla, Shujun Luo, Deanna M. Church, Elena Helman, Parin Sripakdeevong, John West, Rich Chen. Accurately identifying expressed somatic variants for neoantigen detection and immuno-oncology. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 533.

Targeted sequencing assays are increasingly used to identify tumor mutations that guide therapeutic decisions. Interpretation of a cancer variant's origin and therapeutic impact poses analytical challenges. Recent studies have indicated that jointly analyzing a tumor with its matched normal can accurately discriminate between tumor-specific (somatic) and inherited (germline) mutations. Moreover, a NHGRI/NCI Clinical Sequencing Exploratory Research Consortium Tumor Working Group just released a set of guidelines recommending that laboratories performing cancer sequencing tests should include germline variants. However, procurement of a matched sample is often logistically impractical. In the absence of a matched normal, large databases and analytical techniques are currently used to identify cancer variants in tumor sequencing data. Whether the benefits outweigh the additional burden of sequencing the matched normal for accurate detection of cancer-relevant mutations remains an open question. To compare tumor-only and tumor/normal analysis of cancer samples, we collected a set of >100 formalin-fixed (FFPE) and fresh frozen cancer samples of various tumor types, where matched normal blood or adjacent tissue was available. We performed augmented target enrichment sequencing (exome and large cancer gene panel) of both DNA and RNA. The data was analyzed using cancer bioinformatics pipelines that detect base substitutions, small insertions/deletions, copy number alterations, and gene fusions in both tumor-only and tumor/normal modes. Variants were annotated using described clinical actionability filtering strategies. Analysis of germline variants for secondary findings was performed. We find that 67% of mutations detected in tumor-only mode are reclassified as germline variants when analyzed together with the matched normal sample. These include mutations in hereditary cancer predisposition genes, such as BRCA1, VHL, and other genes with ACMG guidelines that warrant germline variant classification and appropriate management. Clinically actionable mutations may be miscalled as somatic when a matched normal is not available; however, we find the definition of ‘actionable’ can greatly impact the results of this analysis. Finally, the use of newly available large datasets, such as ExAC, substantially decreases the number of miscalled somatic variants in the absence of a matched normal. The effects of administering targeted therapies to patients with germline mutations in the relevant gene are largely unknown. Mutations of putative germline origin may be important for hereditary cancer knowledge and tumor treatment, and should be reported as such. For NGS-based cancer interpretation to guide clinical decisions in a practical and cost-effective manner, highly optimized tumor-only and tumor/normal analyses must be available with proper attention to germline consent, classification and education. Citation Format: Elena Helman, Michael J. Clark, Ravi Alla, Sean M. Boyle, Shujun Luo, Selene Virk, Deanna Church, Parin Sripakdeevong, Jason Harris, Mirian karbelashvili, Christian Haudenschild, John West, Richard Chen. The benefits and burdens of assaying matched normal tissue when sequencing cancer genomes. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 3169.

11561Background: Interpretation of a cancer variant’s origin and therapeutic impact poses analytical challenges. Recent studies have indicated that analyzing a tumor with its matched normal can dis...

The estrogen receptor (ER or ESR1) drives proliferation and growth of luminal type breast cancers. Endocrine therapies are highly effective in a majority of these cancers types; however, disease progression eventually occurs due to acquired resistance resulting in hormone-independent breast cancer. One mechanism of resistance is acquired mutations at codons 537 and 538 in the ligand binding domain of the receptor which are found in ∼12% of pretreated, ER+ patients. These mutations result in constitutive ER activation and hormone-independent progressive disease. Although ESR1 mutated breast cancers are insensitive to endocrine therapy, novel agents targeting these mutations may be effective; however few preclinical models of ESR1-mutant breast cancer are available for preclinical analysis. To this end we have established two patient derived xenograft (PDX) models harboring mutations at codon 537; ST941 with ESR1Y537S and ST1799 with ESR1Y537C. We have characterized these models using genomic analysis and compared with analysis from paired or serially collected clinical tissue and blood. Dependence on estradiol for model growth and sensitivity to endocrine therapies were also evaluated and compared with a hormone-dependent breast cancer models. The ESR1 mutations were not present in primary clinical tissue while several additional mutations were identified including in TP53 and PIK3CA genes. In vivo the ESR1 mutant models grew in the presence or absence of exogenous estradiol and demonstrated reduced sensitivity to endocrine therapies compared with hormone-dependent breast cancer models. ST941 treated with tamoxifen, letrozole or fulvestrant reported moderate tumor growth inhibition versus control while tumor stasis or regressions were reported in MCF-7 and hormone-dependent PDX breast models. Overall we have established and characterized two models of ESR1 mutant breast cancer which can be utilized for development of targeted therapies. Citation Format: Wick MJ, Helman E, Meade J, Clark MJ, Vaught T, Tolcher AW, Rasco D, Patnaik A, Lang A, Beeram M, Papadopoulos KP. Establishment and characterization of ESR1-mutant breast cancer PDX models. [abstract]. In: Proceedings of the Thirty-Eighth Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2015 Dec 8-12; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2016;76(4 Suppl):Abstract nr P3-03-04.

Lung squamous cell carcinoma is a common type of lung cancer, causing approximately 400,000 deaths per year worldwide. Genomic alterations in squamous cell lung cancers have not been comprehensively characterized, and no molecularly targeted agents have been specifically developed for its treatment. As part of The Cancer Genome Atlas, here we profile 178 lung squamous cell carcinomas to provide a comprehensive landscape of genomic and epigenomic alterations. We show that the tumour type is characterized by complex genomic alterations, with a mean of 360 exonic mutations, 165 genomic rearrangements, and 323 segments of copy number alteration per tumour. We find statistically recurrent mutations in 11 genes, including mutation of TP53 in nearly all specimens. Previously unreported loss-of-function mutations are seen in the HLA-A class I major histocompatibility gene. Significantly altered pathways included NFE2L2 and KEAP1 in 34%, squamous differentiation genes in 44%, phosphatidylinositol-3-OH kinase pathway genes in 47%, and CDKN2A and RB1 in 72% of tumours. We identified a potential therapeutic target in most tumours, offering new avenues of investigation for the treatment of squamous cell lung cancers., National Institutes of Health (U.S.) (Grant U24 CA126561), National Institutes of Health (U.S.) (Grant U24 CA126551), National Institutes of Health (U.S.) (Grant U24 CA126554), National Institutes of Health (U.S.) (Grant U24 CA126543), National Institutes of Health (U.S.) (Grant U24 CA126546), National Institutes of Health (U.S.) (Grant U24 CA126563), National Institutes of Health (U.S.) (Grant U24 CA126544), National Institutes of Health (U.S.) (Grant U24 CA143845), National Institutes of Health (U.S.) (Grant U24 CA143858), National Institutes of Health (U.S.) (Grant U24 CA144025), National Institutes of Health (U.S.) (Grant U24 CA143882), National Institutes of Health (U.S.) (Grant U24 CA143866), National Institutes of Health (U.S.) (Grant U24 CA143867), National Institutes of Health (U.S.) (Grant U24 CA143848), National Institutes of Health (U.S.) (Grant U24 CA143840), National Institutes of Health (U.S.) (Grant U24 CA143835), National Institutes of Health (U.S.) (Grant U24 CA143799), National Institutes of Health (U.S.) (Grant U24 CA143883), National Institutes of Health (U.S.) (Grant U24 CA143843), National Institutes of Health (U.S.) (Grant U54 HG003067), National Institutes of Health (U.S.) (Grant U54 HG003079), National Institutes of Health (U.S.) (Grant U54 HG003273)

Background: Anaplastic large cell lymphoma (ALCL) is an uncommon, aggressive CD30-positive T-cell lymphoma exhibiting or lacking chromosomal translocation involving the Anaplastic Lymphoma Kinase (ALK) gene and expression of ALK protein. ALK positive ALCL (ALK+-ALCL) typically has an improved prognosis compared with ALK-negative; however, patients often respond favorably to standard chemotherapy and CD30-directed immunoconjugates only to progress and are left with few treatment options. To help identify additional treatment options, we collected tumor tissue and established, characterized and evaluated in vivo an ALK+-ALCL patient-derived xenograft (PDX) model from a thirty one year old Caucasian female with relapsed ALCL following initial therapy with a pediatric S-phase specific regimen. Methods: The ST2698 ALK+-ALCL model was established in CB17 SCID mice using tissue collected from a lymph node tumor biopsy. Clinical tissue, patient blood and the PDX model were subjected to WGS sequencing using an augmented and content-enhanced exome. The augmented exome is optimized to detect major cancer mutations by enhancing coverage over known sequencing gaps and GC-rich regions across >1300 cancer and 200 miRNA genes. We also performed whole-transcriptome sequencing on the PDX model. All data were analyzed using a cancer bioinformatics pipeline optimized for high accuracy detection of small variants and indels, somatic copy-number aberrations, gene expression and fusions. Drug sensitivity studies were performed evaluating sensitivity of the model to patient's current clinical treatment and relevant targeted therapies; study endpoints included tumor volume and time from treatment initiation with tumor growth inhibition, delay and regression reported at study completion. Results: Exome analysis identified several variants which were confirmed by transcriptome data. Efficacy studies confirmed model sensitivity to the patient's current clinical course (CHOP). In addition, ST2698 was found moderately sensitive to the IMiD lenalidomide (T/C=47%). However, treatment with the ALK-inhibitors crizotinib or ceritinib resulted in tumor regressions including durable complete responses in some mice. Conclusion: We have established and characterized a patient-derived ALK+-ALCL xenograft model using DNA and RNA-based analysis. In addition we evaluated the model in vivo and confirmed sensitivity to the patient's current clinical course and identified two ALK inhibitors as active, including reported complete tumor regressions. Disclosures Rasco: Celgene: Research Funding; Asana BioSciences, LLC: Research Funding. Tolcher:Asana BioSciences, LLC: Consultancy, Research Funding; AbbVie: Consultancy; ArQule: Consultancy; Bayer: Consultancy; BioMed Valley Discoveries: Research Funding; Janssen R&D: Consultancy.

Background: A subset of non-small cell lung cancer (NSCLC) is driven by mutations or deletions in the epithelial growth factor receptor (EGFR). Several targeted therapies including gefitinib and erlotinib are initially effective towards EGFR-mutant NSCLC; however, resistance to therapy often occurs during the course of treatment. To better understand the mechanisms for resistance to targeted therapies in EGFR-mutant NSCLC we collected tissue from a patient prior to and following response and progression to erlotinib and established xenografts for preclinical drug screening. Clinical and PDX tissue were characterized and compared using DNA and RNA analysis and the models compared in vivo for sensitivity to erlotinib and other therapies. Methods: The NSCLC adenocarcinoma PDX models were established in CB17 SCID mice using tissue collected from a lung biopsy (ST551) or ascites (ST551C). Clinical tissue, patient blood and the PDX model were subjected to WGS sequencing using an augmented and content-enhanced exome. The augmented exome is optimized to detect major cancer mutations by enhancing coverage over known sequencing gaps and GC-rich regions across >1300 cancer and 200 miRNA genes. We also performed whole-transcriptome sequencing on the PDX model. All data were analyzed using a cancer bioinformatics pipeline optimized for high accuracy detection of small variants and indels, somatic copy-number aberrations, gene expression and fusions. Drug sensitivity studies were performed evaluating sensitivity of the model to patient's current clinical treatment and relevant targeted therapies; study endpoints included tumor volume and time from treatment initiation with tumor growth inhibition, delay and regression reported at study completion. In erlotinib efficacy studies the therapy was administered daily via oral gavage at 35 mg/kg for sixty days. Results: Genomic analysis identified several variants which were confirmed by transcriptome data including the previously identified EGFR deletion as well as an MGMT deletion. Efficacy studies reported activity of erlotinib towards the ST551 model with a tumor growth inhibition of 112% including tumor regressions during treatment. However erlotinib was found less activity towards the ST551C model with a tumor growth inhibition of 55% and no tumor regressions. Conclusion: We have established a pair of PDX models from the same patient representing erlotinib sensitive and resistant NSCLC and have characterized the model using comparative genomic analysis and in in vivo efficacy testing. These models can be utilized to better understand mechanisms of resistance to EGFR-targeted therapies and represent unique tools in developing novel therapies to EGFR resistance. Citation Format: Michael J. J. Wick, Elena Helman, Teresa L. Vaught, Monica Farley, Lizette Gamez, Alyssa Moriarty, Scott Ulmer, Anthony W. Tolcher, Drew W. Rasco, Michael Clark, Amita Patnaik, Kyriakos P. Papadopoulos. Establishment and characterization of serial patient derived xenograft models representing erlotinib sensitive/resistant non-small cell lung cancer. [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2015 Nov 5-9; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2015;14(12 Suppl 2):Abstract nr A18.