Search

Cancer cells can metabolize glutamine to replenish TCA cycle intermediates, leading to a dependence on glutaminolysis for cell survival. However, a mechanistic understanding of the role that glutamine metabolism has on the survival of glioblastoma (GBM) brain tumor stem cells (BTSC) has not yet been elucidated. Here, we report that across a panel of 19 GBM BTSC lines, inhibition of glutaminase (GLS) showed a variable response from complete blockade of cell growth to absolute resistance. Surprisingly, BTSC sensitivity to GLS inhibition was a result of reduced intracellular glutamate triggering the amino acid deprivation response (AADR) and not due to the contribution of glutaminolysis to the TCA cycle. Moreover, BTSC sensitivity to GLS inhibition negatively correlated with expression of the astrocytic glutamate transporters EAAT1 and EAAT2. Blocking glutamate transport in BTSCs with high EAAT1/EAAT2 expression rendered cells susceptible to GLS inhibition, triggering the AADR and limiting cell growth. These findings uncover a unique metabolic vulnerability in BTSCs and support the therapeutic targeting of upstream activators and downstream effectors of the AADR pathway in GBM. Moreover, they demonstrate that gene expression patterns reflecting the cellular hierarchy of the tissue of origin can alter the metabolic requirements of the cancer stem cell population. Significance: Glioblastoma brain tumor stem cells with low astrocytic glutamate transporter expression are dependent on GLS to maintain intracellular glutamate to prevent the amino acid deprivation response and cell death.

cBioPortal for Cancer Genomics is an open-source platform for interactive, exploratory analysis of large-scale clinico-genomic data sets. cBioPortal provides a suite of user-friendly visualizations and analyses, including OncoPrints, mutation “lollipop” plots, variant interpretation, group comparison, survival analysis, expression correlation analysis, alteration enrichment analysis, cohort and patient-level visualization. The public site (https://www.cbioportal.org) is accessed by >35,000 unique visitors each month and hosts data from >350 studies spanning individual labs and large consortia. In addition, at least 74 instances of cBioPortal are installed at academic institutions and companies worldwide. To better support all users, we unified our documentation (https://docs.cbioportal.org) and added a user guide and an ongoing series of ‘how-to’ videos to address common questions. In 2022 we added 32 studies (>38,000 samples) to the public site. In addition, we added a nonsynonymous tumor mutation burden (TMB) value for all samples and enhanced the TCGA PanCancer Atlas studies with DNA methylation and treatment data. All data is available in the cBioPortal Datahub: https://github.com/cBioPortal/datahub. We also host a dedicated instance for AACR Project GENIE, enabling access to the GENIE cohort of >165,000 clinically sequenced samples from 19 institutions (https://genie.cbioportal.org). The GENIE Biopharma Collaborative (BPC) enables the collection of comprehensive clinical annotations, including response, outcome, and treatment history. The first BPC cohorts are now available: ~2,000 non-small cell lung cancer samples and ~1,500 colorectal cancer samples. Support for multimodal data analysis has been a major focus, including several new integrations with external tools. Single cell data is now available in the CPTAC GBM study and can be visualized throughout cBioPortal, and via integration with cellxgene. On the patient page, H&E and mIF images can be visualized via integration with Minerva, and the genomic overview now integrates IGV. We continue to enhance existing features. In the study view, users can now add charts comparing categorical vs continuous data, and the plots tab includes a heatmap option. We replaced the existing fusion data type with a generalized structural variant data type that supports detailed information including breakpoints and orientation, to enable new visualizations and analyses. Pathway level analysis has been extended with a new integration with NDEx. cBioPortal is fully open source (https://github.com/cBioPortal/). Development is a collaborative effort among groups at Memorial Sloan Kettering Cancer Center, Dana-Farber Cancer Institute, Children’s Hospital of Philadelphia, Princess Margaret Cancer Centre, Caris Life Sciences, Bilkent University and The Hyve. We welcome open source contributions from others in the cancer research community. Citation Format: Ino de Bruijn, Tali Mazor, Adam Abeshouse, Diana Baiceanu, Stephanie Carrero, Elena Garcia Lara, Benjamin Gross, David M. Higgins, Prasanna K. Jagannathan, Priti Kumari, Ritika Kundra, Bryan Lai, Xiang Li, James Lindsay, Aaron Lisman, Divya Madala, Ramyasree Madupuri, Angelica Ochoa, Yusuf Ziya Özgül, Oleguer Plantalech, Sander Rodenburg, Baby Anusha Satravada, Robert Sheridan, Lucas Sikina, Jessica Singh, S Onur Sumer, Yichao Sun, Pim van Nierop, Avery Wang, Manda Wilson, Hongxin Zhang, Gaofei Zhao, Sjoerd van Hagen, Ugur Dogrusoz, Allison Heath, Adam Resnick, Trevor J. Pugh, Chris Sander, Ethan Cerami, Jianjiong Gao, Nikolaus Schultz. cBioPortal for Cancer Genomics. [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 4256.

Introduction: Clinical benefit from Immune Checkpoint Blockade (ICB) is a function of local T-cell specificity for tumor-associated antigens. However, overcoming local T-cell dysfunction necessitates systemic immunity engagement. Therefore, studying the dynamics of both local and peripheral T-cell repertoires in response to ICB is required to identify features of T-cell repertoires associated with pathological response. Methods: We conducted TCRβ-sequencing on tumor-residing T-cells (n=59), Peripheral Blood Mononuclear Cells (PBMCs) (n=306) and cell-free DNA (cfDNA) (n=73) from pre- and multiple on-ICB timepoints collected from patients enrolled in the pan-cancer INvestigator-initiated Phase II Study of Pembrolizumab Immunological Response Evaluation (INSPIRE; NCT02644369) trial. To assess specificity-agnostic shifts in TCR repertoires, we first compared TCR diversity and clonal expansion in longitudinal tumor and PBMC samples. Then, to temporally track the specificity-associated features of local and systemic TCR repertoires, we leveraged a Graph Neural Network (GNN) model that took in unique TCRβ chains as nodes. The connectivity between the nodes was defined by multi-relational edges that represented VJ-gene usage and GLIPHII-identified (Grouping Lymphocyte Interactions by Paratope Hotspots) specificities derived from a compendium of TCR sequences with empirically confirmed specificities. Results: While absolute diversity and clonal expansion values in baseline tumor (n=33) were not associated with response to ICB, changes in these values were informative between pre- and on-ICB tumors. All patients (n=4) with low baseline tumor TCR diversity and lack of clonotypic re-structuring in tumor TCR repertoire on-ICB had either progressive or short-term stable disease. Furthermore, pairwise comparison of pre- and on-ICB tumors for each patient (n=17) revealed that all the patients, irrespective of their pathological response, experienced emergence of new TCR clonotypes (i.e., clonal replacement) in response to ICB, suggesting only a minority of these TCRs might consist of tumor-associated clonotypes. Patients with clinical benefit also had higher degree of GLIPHII-identified clustering at baseline tumor, highlighting the role of both specificity-agnostic and specificity-centric TCR analysis in determining the response to ICB. Analysis of TCR sequences in blood plasma found cfDNA contains a small number of TCR sequences (median 32, range 12-89) enriched for TCRs found in matched tumor tissues, suggesting that cfDNA TCR repertoire may provide an indirect measurement of tumor-residing T-cells. Conclusions: TCR diversity and functional clonal annotation are emerging biomarkers of ICB response and cfDNA TCR repertoire can potentially be exploited for clinical diagnostics and monitoring. Citation Format: Shirin Soleimani, Ben X. Wang, Stephanie Pedersen, Jenna Eagles, Jacob Brick, Marcus O. Butler, Scott V. Bratman, Lillian L. Siu, Pamela S. Ohashi, Trevor J. Pugh. Pan-cancer assessment of tumour and peripheral T-cell receptor repertoire dynamics in patients treated with immune checkpoint inhibitors [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 6667.

cBioPortal for Cancer Genomics is an open-source platform for interactive, exploratory analysis of large-scale cancer genomics data sets. cBioPortal provides a user-friendly interface that integrates genomic and clinical data, and provides a suite of visualizations and analyses, including OncoPrints, mutation “lollipop” plots, variant interpretation, group comparison, survival analysis, expression correlation analysis, alteration enrichment analysis, cohort and patient-level visualization. cBioPortal also integrates external tools including CIViC, Cancer Digital Slide Archive, Next-Generation Clustered Heat Map, IGV and Bioconductor to facilitate interpretation. The public site (https://www.cbioportal.org) is accessed by ~35,000 unique visitors each month and hosts data from >325 studies spanning individual labs and large consortia. In addition, >67 instances of cBioPortal are installed at academic institutions and pharmaceutical/biotechnology companies worldwide. In 2021 we added data from 32 studies, totaling >24,000 samples, to the public site. All data is also available in the cBioPortal Datahub: https://github.com/cBioPortal/datahub/. We also host a dedicated instance for AACR Project GENIE, enabling access to the GENIE cohort of >135,000 clinically sequenced samples from 19 institutions (https://genie.cbioportal.org). In addition, the GENIE Biopharma Collaborative (BPC) enables the collection of comprehensive clinical annotations, including response, outcome, and treatment histories. The first BPC release contains data from >1,800 non-small cell lung cancer samples and will be released in early 2022. The growing GENIE cohort and the BPC clinical data have driven a number of recent developments, including performance improvements (the load time for the GENIE cohort was reduced from minutes to seconds). To leverage the BPC clinical data, we enabled sample selection based on treatment status, extended support for outcome analysis, and enhanced the patient timeline representation to incorporate response data. Additional development work has focused on improvements to variant interpretation, enhancements to the Mutations tab, and support for novel molecular assays via the ‘generic assay’ data type. Documentation on these new features and many others is available at https://www.cbioportal.org/news. cBioPortal is fully open source (https://github.com/cBioPortal/) under a GNU Affero GPL license. Development is a collaborative effort among groups at Memorial Sloan Kettering Cancer Center, Dana-Farber Cancer Institute, Children’s Hospital of Philadelphia, Princess Margaret Cancer Centre, Bilkent University and The Hyve. We welcome open source contributions from others in the cancer research community. Citation Format: Jianjiong Gao, Tali Mazor, Ino de Bruijn, Adam Abeshouse, Diana Baiceanu, Ziya Erkoc, Elena Garcia Lara, Benjamin Gross, David M. Higgins, Prasanna K. Jagannathan, Priti Kumari, Ritika Kundra, Xiang Li, James Lindsay, Aaron Lisman, Divya Madala, Ramyasree Madupuri, Angelica Ochoa, Oleguer Plantalech, Sander Rodenburg, Baby A. Satravada, Robert Sheridan, Lucas Sikina, Jessica Singh, S. Onur Sumer, Yichao Sun, Pim van Nierop, Avery Wang, Manda Wilson, Hongxin Zhang, Gaofei Zhao, Sjoerd van Hagen, Kees van Bochove, Ugur Dogrusoz, Allison Heath, Adam Resnick, Trevor J. Pugh, Chris Sander, Ethan Cerami, Nikolaus Schultz. cBioPortal for cancer genomics [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 1155.

Introduction: Durvalumab immunotherapy has rapidly emerged as standard treatment for stage 3 NSCLC patients following definitive chemoradiotherapy (CRT). Multiple novel immunotherapeutic strategies are in development to enhance the chance of cure in this setting as well. There exists a critical need to identify blood-based biomarkers capable of predicting clinical benefit from adjuvant immunotherapy as well selecting patients at high-risk of relapse for further drug development. Cell-free DNA (cfDNA)-based analysis of both minimal residual disease (MRD) and T-cell receptor (TCR) clonality have immense potential to predict and monitor response to adjuvant immunotherapy. In this study, we have combined innovative cfDNA measures of MRD (CAPPseq), TCR clonality (CapTCR-seq) and methylation (cfMeDIPseq) as potential predictive biomarkers of disease progression in stage 3 NSCLC patients treated with CRT and durvalumab. Methods: Stage 3 NSCLC patients undergoing CRT and durvalumab were recruited prospectively to undergo serial blood collections at baseline, pre- and post- durvalumab. CAPPseq and cfMeDIPseq were performed as measures of MRD. TCR repertoire analysis (CapTCR-seq) was performed on cfDNA using hybrid-capture TCR sequencing and TCR diversity/clonality was estimated using the Shannon's index. Correlations between MRD, TCR clonality, response and progression-free survival (PFS) were examined using logistic/cox regression. Results: 79 stage 3 NSCLC patients have been prospectively recruited and undergone serial blood collection. CAPPseq, cfMeDIPseq and capTCR-seq have been completed in 22 patients (5 primary progression on CRT, 17 received durvalumab). Tumor cfDNA was detectable by CAPPseq at baseline in 14 patients. High correlation between tumor cfDNA detected by CAPPseq and cfMeDIPseq was found (R=0.68, p Conclusions: Failure to clear MRD and decreasing TCR clonality as assessed by cfDNA was highly correlated with increased risk of recurrence and reduced PFS with consolidation durvalumab. This innovative approach has significant potential to define a new biomarker for the use and development of adjuvant immunotherapy. Citation Format: Sally CM Lau, Shirin Soleimani, Jinfeng Zou, Justin Burgener, Shelley Kuang, Stephanie WY Wong, Malcolm Ryan, Ben X. Wang, Stephanie Pedersen, Devalben Patel, Penelope A. Bradbury, Geoffrey Liu, Natasha Leighl, Ming S. Tsao, Pamela S. Ohashi, Scott V. Bratman, Trevor Pugh, Frances A. Shepherd, Adrian G. Sacher. cfDNA-based analysis of minimal residual disease and T-cell receptor clonality as predictors of relapse in stage 3 NSCLC treated with chemoradiotherapy and durvalumab [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 563.

The cBioPortal for Cancer Genomics is an open-source software platform that enables interactive, exploratory analysis of large-scale cancer genomics data sets with a biologist-friendly interface. It integrates genomic and clinical data, and provides a suite of visualization and analysis options, including OncoPrint, mutation diagram, variant interpretation, survival analysis, expression correlation analysis, alteration enrichment analysis, cohort and patient-level visualization, among others. The public site (https://www.cbioportal.org) hosts data from more than 280 studies from diverse sources including individual labs and large consortia. All data is also available in the cBioPortal Datahub (https://github.com/cBioPortal/datahub/). Data from 40 studies, totaling more than 10,000 samples, was added in 2019, including the latest release from the Cancer Cell Line Encyclopedia and the Pediatric Preclinical Testing Consortium. The site is accessed by over 30,000 unique visitors per month. cBioPortal also supports AACR Project GENIE with a dedicated instance hosting the GENIE cohort of 80,000 clinically sequenced samples from 19 institutions worldwide (http://genie.cbioportal.org). In addition, more than 40 instances are installed locally at academic institutions and pharmaceutical/biotechnology companies. In support of these local installations, cBioPortal now has improved documentation and simplified installation via container technologies such as Docker and Kubernetes. Building on our successful refactoring of the code base, we have released a variety of new features and enhancements to cBioPortal over the past year. Most notably, we released a group comparison feature, enabling users to define groups of interest based on any clinical or genomic features. User-defined groups can be compared simultaneously across genomic and clinical data, including survival analysis and genomic alteration enrichment analysis. Additional new features include: integration of mutation annotations from dbSNP, ClinVar and gnomAD; support for waterfall plots to enable treatment response analysis; saving user preferences for chart layout on study view. The cBioPortal remains under active development. The portal is fully open source (https://github.com/cBioPortal/) under a GNU Affero GPL license. Development is a collaborative effort among groups at Memorial Sloan Kettering Cancer Center, Dana-Farber Cancer Institute, Children's Hospital of Philadelphia, Princess Margaret Cancer Centre, and The Hyve. Ongoing and future development is focused on: (1) building the open source community; (2) continued performance improvements; (3) expanding user support, documentation and training resources; (4) supporting longitudinal data analysis and visualization; (5) developing novel features to support immunogenomics and immunotherapy; (6) enhancing individual variants and overall patient interpretation; (7) supporting single cell data visualizations and analysis. Citation Format: Jianjiong Gao, Tali Mazor, Adam Abeshouse, Ino de Bruijn, Benjamin Gross, Karthik Kalletla, Priti Kumari, Ritika Kundra, Xiang Li, James Lindsay, Aaron Lisman, Pieter Lukasse, Ramyasree Madupuri, Angelica Ochoa, Oleguer Plantalech, Sander Rodenburg, Fedde Schaeffer, Robert Sheridan, Lucas Sikina, Jing Su, S. Onur Sumer, Yichao Sun, Paul van Dijk, Sjoerd van Hagen, Pim van Nierop, Avery Wang, Manda Wilson, Hongxin Zhang, Gaofei Zhao, Kelsey Zhu, Kees van Bochove, Ugur Dogrusoz, Allison Heath, Adam Resnick, Trevor J. Pugh, Chris Sander, Ethan Cerami, Nikolaus Schultz. The cBioPortal for Cancer Genomics [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 3209.

The brain tumor Glioblastoma (GBM) is a virtual death sentence for anyone who is diagnosed, with a median survival time of 12-15 months with standard treatments1 and a 5 year survival rate of under 10% 2. There is a strong body of evidence supporting the existence of stem like cells, termed glioma stem cells (GSCs), that can repopulate the tumor after removal and therapy application3-6. Thus, GSCs present a tantalizing target for potential therapies against GBM. However, studies of GSCs have shown heterogeneity at the level of the transcriptome and drug response5,7, suggesting that there is significant biological variation that translates into differential sensitivity to various drugs. A full characterization of biological heterogeneity may aid in the search for targeted therapies. Here, we profile the transcriptomes of 72 patient derived GSC cultures, and obtain scRNA-seq on 29 cultures from 26 patients (> 69,000 cells) plus 5 GBM tumors (> 14,000 cells). With this data, we find two anticorrelated transcriptional programs in the GSC cultures, one associated with immune or injury response related pathways and the other with neural developmental pathways. We then compare the GSC cultures to patient tumors in the scRNA-seq data, and find that a portion of GBM tumor cells are similar to GSC cultures. We find that a gradient between GSC and astrocyte programs separates cells with stemness properties from those that are not stem-like, and that within stem-like tumor cells and non stem-like tumor cells a gradient between the neural developmental and immune related programs exists as was seen for the cultured GSCs. In GSCs, we also find epigenetic variation in DNA methylation associated with the developmental and immune related programs. Overall these data suggest variation between two biological programs manifests at the level of gene expression and epigenetic regulation in GSCs in tumors. 1. Stupp, R. et al. Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma. N. Engl. J. Med. 352, 987-996 (2005). 2. Brennan, C. W. et al. The somatic genomic landscape of glioblastoma. Cell 155, 462-477 (2013). 3. Singh, S. K. et al. Identification of human brain tumour initiating cells. Nature 432, 396-401 (2004). 4. Chen, J. et al. A restricted cell population propagates glioblastoma growth after chemotherapy. Nature 488, 522-526 (2012). 5. Lan, X. et al. Fate mapping of human glioblastoma reveals an invariant stem cell hierarchy. Nature 549, 227-232 (2017). 6. Patel, A. P. et al. Single-cell RNA-seq highlights intratumoral heterogeneity in primary glioblastoma. Science 344, 1396-1401 (2014). 7. Meyer, M. et al. Single cell-derived clonal analysis of human glioblastoma links functional and genomic heterogeneity. Proc. Natl. Acad. Sci. U. S. A. 112, 851-856 (2015). Citation Format: Owen K. Whitley, Laura M. Richards, Florence Cavalli, Paul Guilhamon, Fiona Coutinho, Michelle Kushida, H. Artee Luchman, Samuel Weiss, Mathieu Lupien, Peter Dirks, Trevor Pugh, Gary Bader. Revelation of shared transcriptional gradients in glioblastoma tumors and cultured glioma stem cells [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 5724.

Background: Emergence of the EGFR T790M mutation accounts for acquired first generation EGFR TKI resistance in over half of patients with EGFR mutant NSCLC. In patients without emergent T790M, resistance mechanisms are less well understood. We explored the impact of DNA methylation status and TKI treatment failure in these patients. Methods: Using a prospective cohort of patients with acquired TKI resistance, tumour tissue samples pre/post TKI exposure were identified. DNA was extracted from FFPE tissue using the Qiagen AllPrep DNA/RNA FFPE Extraction Protocol, and subsequently analyzed using the Illumina Infinium EPIC array. Raw microarray data files were processed using the software package minfi for data normalization (Illumina method) and extraction of methylation levels (M-values). Samples were split into two groups according to the T790M status of each sample (T790M + or T790M-). The set of most informative probes, those whose M-value profiles align most closely with the T790M status of the study samples, was generated by selecting the 1,000 probes with lowest ANOVA's p-value. The stability of the resulting sample clustering was assessed by hierarchical clustering (Euclidean distance), classification with internal cross-validation (SVM leave-one-out), and non-parametric dimensional reduction (t-SNE). Results: 40 samples from 36 EGFR mutant NSCLC patients were successfully profiled. Pre TKI samples were available in 10 patients with an EGFR mutation of which 4 had matched post TKI tissue (3 T790M+, 1 T790M-). The remaining 26 samples in post TKI patients included 17 T790M + and 9 T790M- cases. A DNA methylation-based signature was developed by selecting the array probes that best discriminated T790M+ from T790M- cases. Group membership was stable, as shown by cross-validation by three different methods (hierarchical clustering, SVM leave-one-out and t-SNE). The 1,000 probe cut-off was arbitrarily selected; however, identical sample clusters were obtained using 500 or 2,000 methylation array probes. When analyzing the genomic location of the set of probes that form the signature, we found broad distribution across all chromosomes, thus, ruling out the possibility of selection bias due to focal or chromosome-level aberrations. Several genes contained a higher number of the selected probes, including EGFR, whose expression levels are known to be regulated at the methylation level in certain cancer types. Cluster analysis using the 1,000-probe signature revealed a high degree of concordance between EGFR T790M and DNA methylation status. All post-TKI (n=20) T790M+ samples concentrated within epi-group 2, whereas 8/10 T790M- samples were found within epi-group 1. Of the 4 patients with matched samples, 2 had baseline samples within epi-group 2 and went on to develop EGFR T790M post TKI. Of the 2 with baseline samples within epi-group 1, one went on to develop T790M (post-TKI epigroup 2) and one did not (post-TKI epigroup 1). Conclusions: We observed a concordance between T790M status and epi-group suggesting that the development of resistance to EGFR-TKIs may be associated with distinct DNA methylation signatures. This signature may be present at baseline and predict for subsequent emergence of T790M. Citation Format: Grainne M. O'Kane, Alberto J. León, Muqdas Shabir, Jennifer H. Law, Penelope A. Bradbury, Geoff Liu, Adrian Sacher, Frances A. Shepherd, Dax Torti, Tracy L. Stockley, Ming Tsao, Trevor J. Pugh, Natasha B. Leighl. Methylation signatures associated with T790M status in progressive NSCLC [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 1869.

Introduction: Immune checkpoint inhibition is a novel approach of restimulating immune cells to overcome immunosuppressive microenvironment in tumors and has shown significant clinical benefit in 20-40% of adult patients. Although this success in adult cancers has resulted in a plethora of immunogenomic efforts to delineate mechanisms of response, such studies in pediatric cancers are lacking. The main objective of this study is to uncover the overall characteristics of immune microenvironment in pediatric solid tumors. Methods: We leveraged transcriptomes of 1,268 pediatric primary solid tumors acquired from multiple international initiatives, including Therapeutically Applicable Research to Generate Effective Treatments (TARGET, n = 269), International Cancer Genome Consortium (ICGC, n = 216) and Children Brain Tumor Tissue Consortium (CBTTC, n = 803). We compared our results with those of 7 adult cancer types from The Cancer Genome Atlas (TCGA, n = 2720). We applied methods of immune deconvolution, repurposed RNAseq data to recover infiltrating T- and B-cell clonotypes and studied checkpoint gene expression in pediatric tumors. In collaboration with Gabriela Miller’s Kids First Data Resource Centre, all analyses were performed and shared on CAVATICA computational platform (cavatica.sbgenomics.com). Results: Embryonal tumors with multilayered rosettes (ETMR) and medulloblastomas (MB) harbored the lowest levels of immune infiltration in our pediatric cohort. Immune deconvolution analysis revealed that neuroblastomas (NBL) had the highest T-cell infiltration amongst pediatric cancers. We found atypical teratoid/rhabdoid tumors (ATRT) had highest levels of CD8+ T cells among pediatric CNS tumors. While tumor mutational burden (TMB) was associated with immune cell infiltration in adult lung cancers and melanomas, we found no significant association in pediatric cancers. We found specific genetic alterations such as BRAF, H3F3A, and CTNNB1 had significant impact on the composition of the immune microenvironment. Analysis of T-cell repertoire revealed an inverse correlation between clonal diversity and TMB, suggesting a T-cell clonal expansion in high TMB samples. Finally, while the majority of NBL samples expressed LAG3, ~10% of samples had elevated levels of TIM3 gene, suggesting a distinct mode of immunosuppression in this subset. 2% of pediatric CNS tumors, mostly consisted of ATRTs, harbored high expression of PRF1 and GZMA genes suggestive of functionally capable immune cells in these cases. Conclusions: We report characteristics of the tumor microenvironment in pediatric tumors at primary diagnosis. We found that specific gene mutations, rather than mutational load, shape the composition of immune cell types. Furthermore, hypermutant samples exhibit evidence of clonal T-cell expansion. Our results suggest immune activity in 2% of CNS tumors and uncover a subtype of NBL expressing TIM3 checkpoint gene, calling for investigations to evaluate the effectiveness of immunotherapy. This abstract is also being presented as Poster A74. Citation Format: Arash Nabbi, Pengbo Sun, Sudhaman Sumedha, Kelsey Zhu, S.Y. Cindy Yang, Joseph N. Paulson, Marcel Kool, Komal Rathi, Karthik Kalletla, Pichai Raman, Yuankun Zhu, Adam C. Resnick, David T.W. Jones, Natalie Jäger, Stefan M. Pfister, Trevor J. Pugh. The immunogenomic landscape of pediatric primary solid tumors [abstract]. In: Proceedings of the AACR Special Conference on the Advances in Pediatric Cancer Research; 2019 Sep 17-20; Montreal, QC, Canada. Philadelphia (PA): AACR; Cancer Res 2020;80(14 Suppl):Abstract nr PR06.

More than 85% of patients with Li-Fraumeni syndrome harbor germline TP53 mutations. The spectrum of mutations and the heterogeneity of tumor presentation (age of onset and type) within and between families is remarkable and indicates that modifiers must play an important role in defining the phenotype of each patient and each family. We recently reported the feasibility and utility of a clinical surveillance protocol for early detection of tumors in TP53 mutation carriers (Villani et al., Lancet Oncol 2016). Results from deep sequencing of the genome and epigenome of germline and tumor tissues of TP53 mutation carriers provide exciting new insight into the role of modifiers on both the functional activity of p53 in the germline as well as their influence on tumor onset, tumor spectrum, and clinical outcome/response to therapy. Data will be presented to support the creation of molecular algorithms that may be used in a precise manner to predict cancer onset, as well as to better define the molecular landscape of cancers that arise in TP53 mutation carriers. It is anticipated that the combined use of this germline and somatic landscape data can more effectively refine and guide the creation of personalized surveillance and chemoprevention strategies for these patients. Citation Format: David Malkin, Nicholas Light, Valli Subrasi, Benjamin Brew, Sangeetha Paramathas, Arash Nabbi, Sergei Prykhozhij, Trevor Pugh, Jason Berman, Anna Goldenberg, Adam Shlien. Novel strategies for early cancer detection and prevention: The Li-Fraumeni syndrome story [abstract]. In: Proceedings of the AACR Special Conference on the Advances in Pediatric Cancer Research; 2019 Sep 17-20; Montreal, QC, Canada. Philadelphia (PA): AACR; Cancer Res 2020;80(14 Suppl):Abstract nr IA02.

Brain tumours remain a largely incurable disease, with glioblastoma multiforme (GBM) representing the predominant form of malignancy. GBMs contain a rare population of cells, termed brain tumour stem cells (BTSCs) that drive tumour growth and disease relapse. BTSCs can be isolated and expanded in culture from primary GBMs and retain self-renewal and proliferative capacity in vitro. However, the functional diversity within the BTSC fraction, as well as relationship to the tumour bulk remains unknown. To investigate genotypic and phenotypic states within the stem-like compartment of GBM, we profiled 54,442 patient-derived BTSCs isolated from 25 GBMs using single cell RNA-sequencing (10X Genomics Chromium). Within each BTSC population, we observed diverse transcriptional clonotypes ranging from 1-7 subpopulations per sample. BTSCs grown as adherent monolayers displayed greater heterogeneity (median 4 clusters, range 2-7) compared to sphere-based cultures (median 2 clusters, range 1-3). Across samples, clusters were commonly associated with cell cycling, glial cell development and DNA replication pathways. Within multiple BTSC cultures, we identified transcriptional subclones with variable stemness properties, such as expression of ASCL1, suggesting a hierarchy of differentiation within the tumor-initiating fraction of GBM. Additionally, inference of CNVs from scRNA-seq data revealed subclonal somatic CNVs within BTSCs correspond to distinct transcriptional clonotypes that may partially explain stemness phenotypes and the formation of subpopulations during culture. To nominate targeted and combination therapies against BTSC populations in GBM, we next mapped transcriptomic features of BTSC subpopulations to high-throughput drug-screening data from central nervous system cell lines aggregated by the PharmacoDB platform.This approach nominated FDA-approved therapies that may be effective against subpopulations within each BTSC culture. In some cases, subpopulations were predicted to be sensitive to different drugs, opening a path for testing combination therapies nominated from single cell RNA-seq in vitro. Overall, these data illustrate the extent to which BTSC cultures functionally recapitulate the stem fraction in bulk primary tumours, as well as define networks of self-renewal, therapeutic resistance and targetable vulnerabilities in glioblastoma. Ongoing efforts include single nuclei RNA-seq of primary GBM tumours to enable comparison of transcriptional programs of BTSC cultures with cellular populations found in primary tissues. Citation Format: Laura M. Richards, Owen Whitley, Florence M. Cavalli, Zhaleh Safikhani, Fiona Coutinho, H. Artee Luchman, Benjamin Haibe-Kains, Samuel Weiss, Peter Dirks, Gary Bader, Trevor J. Pugh. Single-cell transcriptomics uncovers clonal heterogeneity linked to drug response and cellular phenotype in adult brain tumor stem cells [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 967.

Introduction: Immune repertoire is a highly diverse pool of B and T cell receptors (B/TCR) that determine boundaries of immune surveillance. Pre-existing immune clones within Tumor MicroEnvironment (TME) suggest existence of a functional antigen presenting machinery and recognizing T cells, which fail to eliminate tumor, likely due to immunosuppressive environment. Shifts in composition of immune repertoire upon immune checkpoint blockade have been reported in adult cancers and shown to be associated with clinical response. As immunotherapy is being introduced to pediatric oncology, there is a need to better understand immunogenomic aspects of TME in childhood cancers. Objective: We hypothesize that characteristics of TCRs in conjunction with gene expression profile of immune cells are key determinants of immune response in pediatric patients. Methods: We compiled a pan-pediatric cohort and associated RNAseq datasets through multiple research initiatives. Participating programs include NCI TARGET (n ~ 270), International Cancer Genome Consortium (ICGC, n ~ 250) and Children’s Brain Tumor Tissue Consortium (CBTTC, n ~ 790). As comparator, we analyzed 7 adult cancer types as well as data from constitutive mismatch repair deficiency (CMMRD) consortium. We devised a simple and reliable index to estimate immune diversity. We used CIBERSORT tool to infer immune fraction of TME and investigated immune-related gene expression. In partnership with Gabriela Miller’s Kids First Data Resource Centre, analyses were performed on CAVATICA computational platform. Results: While median number of RNAseq reads were comparable across TARGET, CBTTC and TCGA datasets (~ 67-90 million reads), ICGC dataset contained a median of ~ 208 million reads. However, number of reads mapped to immune loci were comparable across all datasets and appeared to be confounded by infiltration extent rather than technical discordance. Our preliminary results indicate Neuroblastoma harbored the highest median of inferred diversity across pediatric cancers (TRβ=59.38). Inferred immune content indicated a range of infiltration from 0.59 in Teratoma/Germinoma to 0.15 in Medulloblastoma. Immune checkpoint gene expression across pediatric cancers show overall downward trend compared to adult counterparts. Conclusions: Our comprehensive study will serve as a common ground for researchers to delineate immune component of pediatric tumor microenvironment. Covered in this study are common solid tumors as well as rare malignancies, characterization of which will aid rational design of immunotherapy-related clinical trials of pediatric cancers. Note: This abstract was not presented at the meeting. Citation Format: Arash Nabbi, Natalie Jäger, Sumedha Sudhaman, Pengbo Sun, S. Y. Cindy Yang, Kelsey Zhu, Marcel Kool, Komal Rathi, Karthik Kalletla, Pichai Raman, Yuankun Zhu, Joseph N. Paulson, David T. Jones, Uri Tabori, Adam C. Resnick, Stefan M. Pfister, Trevor J. Pugh. Landscape of infiltrating immune repertoire in pediatric solid tumors [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 2891.

The cBioPortal for Cancer Genomics is an open-source software platform that enables interactive, exploratory analysis of large-scale cancer genomics data sets with a biologist-friendly interface. It integrates genomic and clinical data, and provides a suite of visualization and analysis options, including OncoPrint, mutation diagram, variant interpretation, survival analysis, expression correlation analysis, alteration enrichment analysis, cohort and patient-level visualization, among others. The public site (http://www.cbioportal.org) hosts data from more than 200 studies from individual labs and large consortia, including the newly added TCGA Pan-Cancer Atlas data and the Count Me In project. These studies can be explored and queried individually or combined together into “virtual studies”. Users are now allowed to login and save virtual studies for query and analysis. The site is currently accessed by approximately 30,000 unique visitors per month. The software is also installed locally at dozens of academic institutions and pharmaceutical/biotechnology companies. A notable instance is the cBioPortal for AACR GENIE (http://www.cbioportal.org/genie/) hosting 60,000 clinically sequenced samples from multiple institutions. Over the past year, the code base has been fully refactored, resulting in a more responsive and interactive website. A new web API is in beta facilitating easier programmatic access to data. In addition, all public studies are available for download from the new datahub (https://github.com/cBioPortal/datahub/). The cBioPortal remains under active development. The portal is fully open source (https://github.com/cBioPortal/) under a GNU Affero GPL license. Development is a collaborative effort among groups at Memorial Sloan Kettering Cancer Center, Dana-Farber Cancer Institute, Children’s Hospital of Philadelphia, Princess Margaret Cancer Centre, and The Hyve. Ongoing and future development is focused on: (1) building the open source community; (2) continued performance improvements; (3) expanding user support, documentation and training resources; (4) developing novel features to support immunogenomics and immunotherapy; (5) enhancing individual variants and overall patient interpretation; (6) creating a simplified query interface; and (7) enabling comparative analysis of user-defined patient cohorts. Citation Format: Jianjiong Gao, Tali Mazor, Adam Abeshouse, Ersin Ciftci, Ino de Bruijn, Benjamin Gross, Karthik Kalletla, Priti Kumari, Ritika Kundra, James Lindsay, Aaron Lisman, Pieter Lukasse, Ramyasree Madupuri, Angelica Ochoa, Oleguer Plantalech, Pichai Raman, Fedde Schaeffer, Robert Sheridan, Jing Su, S. Onur Sumer, Yichao Sun, Sander Tan, Sjoerd van Hagen, Avery Wang, Manda Wilson, Hongxin Zhang, Gaofei Zhao, Kelsey Zhu, Kees van Bochove, Ugur Dogrusoz, Trevor J. Pugh, Adam Resnick, Chris Sander, Ethan Cerami, Nikolaus Schultz. The cBioPortal for cancer genomics [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 910.

Background Aberrant DNA methylation contributes to cancer initiation and progression. In preclinical models of solid tumors, chronic low dose administration of DNA hypomethylating agents can induce T-cell mediated immune activation responses by stimulating expression of endogenous retroviral elements, culminating in an IFN-mediated response. The addition of physiological levels of vitamin C may potentiate viral mimicry and increase anti-tumor immune priming. Immunologically ‘cold’ tumors were selected to evaluate whether these strategies can enhance their otherwise poor responses to immune checkpoint blockade. Methods PD-L1/PD-1 inhibitor naïve patients (pts) with advanced microsatellite stable colorectal cancer (MSS CRC); platinum resistant ovarian cancer (OC); estrogen receptor positive, HER2 negative breast cancer (ER+HER2- BC) were enrolled in this single institution, multi-cohort, investigator-initiated trial. The initial regimen (regimen A) consisted of oral CC-486 300mg QD Days 1-14 (cycles 1-3 only) in combination with durvalumab 1500mg IV Day 15, in 28 day-cycles. A protocol amendment (regimen B) after 19 patients changed CC-486 to 100 mg QD Days 1-21 (cycles 1-3), added oral vitamin C 500mg QD continuously and kept durvalumab 1500mg IV Day 15, in 28-day cycles. Adverse events (AEs) assessed by CTCAEv4.03 grade (G); tumor response by RECIST 1.1 every 2 cycles; paired tumor biopsies at baseline and cycle 2 days 10-14; and serial peripheral blood mononuclear cells (PBMCs) for immune-profiling (IP) and epigenetic analysis. PD-L1 by IHC was assessed using SP263 assay, positivity defined by TC>25%. Results A total of 28 pts with MSS CRC (n=15), OC (n=4), ER+HER2- BC (n=9) were enrolled, 19 pts treated on regimen A, 9 on regimen B. Median age was 56 (range 36-78), ECOG 0:1=7:21, 100% had ≥3 prior lines of therapy, all tumors were PD-L1 negative. Best response was SD (3/28 pts received 3, 3 and 4 cycles respectively) with DCR 7.1%. Median follow-up of 4.7 months, mPFS was 1.9 months (95% CI 1.5-2.3) and mOS was 5 months (95% CI 4.5-10). Three patients (all regimen A) experienced DLTs (2 G4 neutropenia and 1 G3 anemia). Fifteen patients (54%) experienced G1/2 fatigue, 46% experienced G1/2 nausea, vomiting or diarrhea. Toxicity was comparable with addition of vitamin C. Initial analysis of PBMCs by flow cytometry in 3 of 4 OC pts demonstrated an increase in PD-1 and Ki67 expression in CD8 T cells only following administration of durvalumab. EPIC methylation array on 4 pts’ paired tumors and LINE 1 assay on 19 pts’ serial PBMCs (regimen A) demonstrated minimal change in global methylation. Conclusion No meaningful clinical responses to CC-486 plus durvalumab were observed. Tumor tissue and PBMCs both demonstrate minimal global DNA demethylation, with or without physiological dose vitamin C. Incremental immune activation beyond PD-L1 blockade was not demonstrated. Clinical trial information: NCT02811497 Citation Format: Kirsty Taylor, Helen Loo Yau, Ben X. Wang, Philippe L. Bedard, Albiruni R. Razak, Aaron R. Hansen, Anna Spreafico, Dave Cescon, Marcus O. Butler, Amit M. Oza, Stephanie Lheureux, Neda Stjepanovic, Lisa Wang, Brendan Van As, Sarah Boross-Harmer, Trevor Pugh, Lillian L. Siu, Daniel D. De Carvalho. A Phase II basket study of hypomethylating agent oral cc-486 and durvalumab in advanced solid tumors (METADUR) [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 CT190.

Overall outcomes for HPV-negative head and neck squamous cell carcinoma (HNSCC) remain poor with 5-year overall survival rates of 50-60%. Oral squamous cell carcinoma (OSCC), the most common subtype of HPV-negative HNSCC, is typically treated with surgery, and clinico-pathologic features are used to identify patients in need of adjuvant therapies such as radiation therapy (RT), or radiation plus concurrent chemotherapy (CRT). It is clear from the rate of loco-regional or distant failures that more accurate methods of risk stratification would greatly improve outcomes for OSCC patients. This requires biomarkers to identify patients that will benefit from adjuvant RT or CRT but currently there are no validated molecular biomarkers that have been clinically implemented for the personalized treatment of OSCC. In addition to biomarkers for better risk stratification, there is also a need for novel therapeutic strategies leading to improved outcomes. Recently patient-derived xenografts (PDXs) have been shown to faithfully recapitulate human tumor biology and predict drug responses, supporting their relevance as preclinical models for new drug development. Upon subcutaneous implantation of HNSCC specimens into NOD/SCID/IL2Rγ-/- mice, 161 of 243 samples (66%) successfully formed patient-derived xenografts (PDX). Using univariable and multivariable analyses, the ability to form a PDX correlated significantly with adverse clinical outcomes, and specifically, patients with palpable PDX-formation within 8 weeks experienced particularly poor outcomes (hazard ratio for overall survival = 3.0). A cohort of engrafting and non-engrafting samples were sequenced using a targeted sequencing panel designed for both mutational and copy number alteration detection. The overall frequency of somatic genomic alterations detected was similar to The Cancer Genome Atlas cohort and interestingly, successful engraftment correlated to amplification of the CCND1 gene. Twelve PDX models were treated with the CDK4/6 inhibitor, abemaciclib; 5 of 6 models with CCND1 amplifications and/or CDKN2A mutations responded to abemaciclib treatment, whereas only 1 of 6 models lacking these alterations responded. These results demonstrate the potential of using PDX models to identify novel targeted therapies for HNSCC patients who have the poorest outcomes. In the future, PDX avatars could also be exploited to individualize treatment for patients at high risk of relapse following definitive treatment. Citation Format: Christina Karamboulas, Jeffrey P. Bruce, Andrew J. Hope, Jalna Meens, Shao Hui Huang, Jie Su, Fei-Fei Liu, Trevor J. Pugh, Scott V. Bratman, Wei Xu, Laurie E. Ailles. Patient-derived xenografts for prognostication and personalized treatment for head and neck squamous cell carcinoma [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 1033.

AACR Project Genomics Evidence Neoplasia Information Exchange (GENIE) is a multi-phase, multi-year, international data-sharing consortium whose goal is to generate an evidence base for precision cancer medicine by integrating and linking clinical-grade cancer genomic data with clinical outcome data for tens of thousands of cancer patients treated at multiple institutions worldwide. The project fulfills an unmet need in oncology by providing the statistical power necessary to identify novel therapeutic targets, to understand genomic determinants of response to therapy, to design new biomarker-driven clinical trials and ultimately, to improve clinical decision-making and the care delivered to patients. Here we describe the goals, structure and data standards of the GENIE consortium and conclusions from a high-level analysis of the first public release of genomic and limited clinical data from approximately 19,000 patients treated at eight cancer centers obtained during this initial phase of the project. We also explore the clinical utility of these genomic data by examining rates of clinical actionability across multiple cancer types and by estimating patient enrollment rates to the NCI MATCH Trial. Based on yearly rates of sequencing at each of the eight founding institutions, together with the planned addition of new members, we estimate the GENIE database could grow to >100,000 samples within five years. Consistent with the goals of the proposed Cancer Moonshot National Cancer Data Ecosystem, GENIE is committed to the principles of generating interoperable, open access data that can be widely shared across the entire scientific community. Citation Format: Ethan Cerami, Alexander S. Baras, Justin Guinney, Eva Lepisto, Trevor J. Pugh, Nikolaus Schultz, Thomas Stricker, Shawn M. Sweeney, Laura J. van't Veer, Gerrit A. Meijer, Fabrice Andre, Victor E. Velculescu, Kenna R. Shaw, Mia A. Levy, Philippe L. Bedard, Barrett J. Rollins, Charles L. Sawyers, on behalf of the AACR Project GENIE Consortium. Landscape analysis of the initial data release from AACR Project GENIE [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-102. doi:10.1158/1538-7445.AM2017-LB-102

Background: Despite recent advances in personalized cancer therapeutics, little impact has been made on curing advanced melanomas over the last several decades. Patients with metastatic melanoma generally have a poor prognosis; survival is limited and typically the 5-year survival rate is less than 15% in patients with metastatic disease. Dysregulation of BRAF signaling has been shown to be one of these key drivers of the pathogenesis of disease in a large subpopulation (∼40%) of patients. Although targeted therapy with BRAF and MEK inhibitors is the mainstay of therapy for recurrent/metastatic melanomas, this treatment modality provides a temporary respite. Often resistant tumor clones(carrying mutations in oncogenes) emerge, followed by a rapid progression, causing the patients to succumb to the disease. As such, in this study, we aim to understand inter- and intra- tumor heterogeneity that leads to resistance to therapy. Methods: We conducted deep exome sequence analysis (250X coverage) of a collection of 8 metastatic melanoma tumor samples collected by rapid autopsy from a patient who became resistant to Vemurafinib despite harboring sensitizing BRAF p. V600E mutations. We used whole exome sequence analysis to detect somatic mutations and copy number variants in every metastatic tumor. To infer subclonal populations present within each metastasis, we applied a combination of clonal inference tools to determine subclonal structures based on somatic point mutations (PyClone) and copy number alterations (TITAN). Subsequently, we applied clustering algorithms to the inferred populations to construct inter-tumor phylogenies. Results: We found that 89% of somatic point mutations were shared across all metastatic sites and were primarily clonal. Conversely, the cellular prevalence inferred from copy number alterations showed a greater degree of subclonal evolution. We found a clear relationship between the physical locations of tumors and shared subclonal patterns, whereby metastases from lung, brain and omentum metastases clustered distinctly from those in rectum, small intestine, peritoneum, diaphragm and mesentery. To look for shared patterns of subclonal variation underlying metastatic disease and resistance to therapy, we are currently extending our analysis to an additional six patients, each with tissues from 7-9 metastatic sites collected by rapid autopsy. Conclusion: Our analysis of multiple metastatic sites from a single patient by deep exome sequence analysis found most somatic mutations were shared across tumor sites and subclonal evolution appears to be driven primarily by acquisition of copy number alterations. Tumors in near proximity shared greater genomic similarity compare to distant sites. Ongoing analysis of additional patients is required to uncover recurrent mechanisms of resistance to targeted therapies in this disease. Citation Format: Soroush Samadian, Prashant Bavi, Dianne Chadwick, Arnavaz Danesh, Mark Dowar, Tiantian Li, Madura Siva, Michael Roehl, Anthony m. Joshua, Trevor J. Pugh. Inferring subclonal relationships between multiple metastases from rapid autopsy of a single melanoma patient. [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 167.

The cBioPortal for Cancer Genomics provides intuitive visualization and analysis of complex cancer genomics data. The public site (http://cbioportal.org/) is accessed by more than 1,500 researchers per day, and there are now dozens of local instances of the software that host private data sets at cancer centers around the globe. We have recently released the software under an open source license, making it free to use and modify by anybody. The software and detailed documentation are available at https://github.com/cBioPortal/cbioportal. We are now establishing a multi-institutional software development network, which will coordinate and drive the future development of the software and associated data pipelines. This group will focus on four main areas: 1. New analysis and visualization features, including: a. Improved support for cross-cancer queries and cohort comparisons. b. Enhanced clinical decision support for precision oncology, including an improved patient view with knowledge base integration, patient timelines and improved tools for visualizing tumor evolution. 2. New data pipelines, including support for new genomic data types and streamlined pipelines for TCGA and the International Cancer Genome Consortium (ICGC). 3. Software architecture and performance improvements. 4. Community engagement: Documentation, user support, and training. This coordinated effort will help to further establish the cBioPortal as the software of choice in cancer genomics research, both in academia and the pharmaceutical industry. Furthermore, as the sequencing of tumor samples has entered clinical practice, we are expanding the features of the software so that it can be used for precision medicine at cancer centers. In particular, clean, web-accessible, interactive clinical reports integrating multiple sources of genome variation and clinical annotation over time has potential to improve clinical action beyond current text-based molecular reports. By making complex genomic data easily interpretable and linking it to information about drugs and clinical trials, the cBioPortal software has the potential to facilitate the use of genomic data in clinical decision making. Citation Format: Jianjiong Gao, James Lindsay, Stuart Watt, Istemi Bahceci, Pieter Lukasse, Adam Abeshouse, Hsiao-Wei Chen, Ino de Bruijn, Benjamin Gross, Dong Li, Ritika Kundra, Zachary Heins, Jorge Reis-Filho, Onur Sumer, Yichao Sun, Jiaojiao Wang, Qingguo Wang, Hongxin Zhang, Priti Kumari, M. Furkan Sahin, Sander de Ridder, Fedde Schaeffer, Kees van Bochove, Ugur Dogrusoz, Trevor Pugh, Chris Sander, Ethan Cerami, Nikolaus Schultz. The cBioPortal for cancer genomics and its application in precision 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 5277.

Background T-cell clonality assays (TCAs) support a variety of clinical and research interests including T-cell malignancy clone identification, minimal-residual disease (MRD) testing and T-cell receptor (TCR) repertoire characterization for the purposes of immunotherapy. Unfortunately, many TCAs are unable to interrogate alpha, beta, gamma, and delta TCR loci simultaneously. Aiming to overcome these limitations, we designed a T-cell clonality assay (the “NTRA”), capable of identifying T-cell gene rearrangements from all four TCR loci, using hybrid-capture followed by deep next-generation sequencing (NGS). A novel informatics package exploiting the Burrows Wheeler Alignment algorithm and finding of CDR3 seed sequences was also deployed. We then validated the assay using orthogonal methods and in a series of clinical T-cell malignancy specimens. Methods We used DNA probes for hybrid-capture of all V and J segments in the TCR loci, followed by NGS on the Illumina NextSeq 500 platform. Analytical validation used a series of 10 specimens, including 6 flow-cytometry characterized T-cell specimens of variable degrees of immunophenotypic uniformity and 4 cell-lines with known TCR rearrangements. PCR/gel electrophoresis and Sanger sequencing were used for orthogonal validation, using primer sets designed to test each specimen for all 90th-centile V-J configurations. Subsequently, 61 clinical T-cell lymphoma specimens were tested, each previously assessed for T-cell clonality using the clinical gold standard (i.e. PCR/ electrophoresis using the BIOMED-2 consensus primers for the TCR-beta (TRB) and TCR-gamma (TRG) loci). Results PCR confirmed the NTRA-identified V-J configurations with an area-under-the-curve (AUC) by receiver-operator characteristic (ROC) analysis of 0.91. Relative to single-strand Sanger sequencing, the NTRA CDR3 sequence results showed a ROC AUC of 0.83. In a DNA dilution series employing the results of a “clonal” specimen (a Jurkat cell line) spiked into a “polyclonal” specimen (a mononuclear peripheral blood specimen), we could identify “clonal” specimen-specific V-J combinations and error-corrected CDR3 sequences down to less than 10⁁-5. In the final clinical validation, the NTRA performed with a ROC AUC of 0.82 relative to the current TRB & TRG BIOMED-2 clinical assay. Conclusions Our novel assay can overcome the current TCR locus data-yield limitations resulting from primer-based TCAs in a single-tube. The NTRA shows comparable sensitivity and specificity relative to current standard assays and excellent performance relative to current MRD techniques when applied to clinical T-cell lymphoma specimens. Citation Format: Etienne Mahe, David Mulder, Mark Dowar, Mahadeo Sukhai, Linh Nguyen, Pamela Ohashi, Jan Delabie, Tracy L. Stockley, Trevor Pugh, Suzanne Kamel-Reid. Improving T-cell receptor clonotyping of T-cell lymphomas using hybrid-Capture and next-generation sequencing. [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 846.

Background: Genome sequencing of multiple myeloma (MM) tumors has revealed recurrent mutations that serve as a fertile ground for targeted therapies. Indeed, activating mutations of KRAS, NRAS and BRAF have been reported in approximately 27%, 24% and 4% of MM cases. Based on this observation, we initiated a Phase II NCI-CTEP sponsored clinical trial of trametinib in patients with MM (PHL-9460). Detection of mutations currently require bone marrow aspirates that are invasive and can yield suboptimal samples. Targeted, ultra-deep sequencing of circulating tumor DNA (ctDNA) is a promising tool for accessing the tumor genome that has not been well studied in MM. We set out to determine the feasibility of detecting ctDNA in MM and of identifying actionable mutations and mutational load using liquid biopsy through ctDNA analysis. Methods: MM patients enrolled onto PHL-9460 or those with heavy tumor burden were identified and consented to have their peripheral blood drawn for analysis. Where possible, matched tumor DNA was also obtained. Cell-free DNA was extracted from 7-15 mL of plasma isolated within ¬1 hour of blood draw using the QIAamp Circulating Nucleic Acid Kit and tagged with barcoded sequencing adapters for subsequent pooling. All exons of KRAS, NRAS, BRAF, PIK3CA and EGFR genes were isolated using a custom hybrid capture panel (IDT xGen Lockdown) and sequenced on an Illumina HiSeq 2000. Reads were aligned to the human genome reference (hg19) using bwa and somatic mutations were detected using muTect. Results: We have collected 25 samples from 23 patients, 7 from patients on PHL-9460, 5 newly diagnosed, and 13 from relapsed patients having received 3.3 median prior lines of therapy (range 1-7). To date, 11 samples from 10 patients have been sequenced. The sample with the lowest DNA yield failed due to low library complexity (range 16.6-3872 ng, median yield: 197 ng). From the remaining 10 samples, the mean target coverage ranged from 31,500 to 32,500. Somatic mutations in KRAS, NRAS, or PIK3CA genes were present in 5 of 9 patients with mutant allele frequencies ranging from 1.1% to 32% (3 KRAS and 2 NRAS of which 2 cases also had a low frequency PIK3CA mutation). We did not uncover mutations in BRAF or EGFR. For patients with matched tumor DNA, mutations in ctDNA concurred with those found in tumor DNA sequencing (4 of 4 tumors with known genotypes). Two patients with NRAS mutations enrolled onto PHL-9460 have responded to trametinib (1 partial and 1 minor response) and remain on therapy. Conclusion: ctDNA analysis in this cohort has identified key mutations in MM. The rate of RAS or RAF mutations in ctDNA compared to matched tumors and correlation to response to trametinib is ongoing and will be presented. Preliminary data suggest that ctDNA may be a reliable method of detecting mutations in MM and an alternative to bone marrow biopsy. Citation Format: Rayan Kaedbey, Olena Kis, Arnavaz Danesh, Mark Dowar, Tiantian Li, Zhihua Li, Jessica Liu, Mark Mansour, Mahadeo Sukhai, Tong Zhang, Suzanne Kamel-Reid, Trevor J. Pugh, Suzanne Trudel. Noninvasive diagnosis of actionable mutations by deep sequencing of circulating tumor DNA in multiple myeloma. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 615. doi:10.1158/1538-7445.AM2015-615

Background The majority of high-risk neuroblastomas initially respond to chemotherapy, but over half of patients experience therapy-resistant relapses. The molecular defects driving relapse and drug resistance are unknown. Methods We performed Illumina or Complete Genomics whole genome sequencing of 23 paired diagnostic and relapsed neuroblastomas, and corresponding normal lymphocyte DNA, to define genetic alterations associated with relapse. A panel of 18 neuroblastoma cell lines was analyzed for the presence of RAS-MAPK mutations and sensitivity to small molecule inhibitors of this pathway. Results Neuroblastomas that relapsed after chemotherapy showed dramatic clonal evolution, with only 33% of primary tumor mutations also detected at relapse. In 21 out of 23 patients, more somatic coding mutations were observed at relapse (median: 29 unique to relapse, range: 4-129). Unbiased pathway analysis of the somatic mutations detected in the relapse tissues identified a strong enrichment in genes associated with RAS-MAPK signaling (p = 6.1×10−7). 18 of the 23 cases (78%) showed somatic mutations (N = 15) or structural alterations (N = 3) predicted to activate the MAPK pathway, and these were mutually exclusive: ALK (N = 10), NRAS (N = 1), KRAS (N = 1), HRAS (N = 1), BRAF (N = 1), PTPN11 (N = 1), FGRF1 (N = 1) and NF1 (N = 2). These RAS-MAPK mutations were clonally enriched at relapse and exist within clonal or major subclonal tumor populations. Seven of these RAS-MAPK mutations were detected only in the relapse tumor by whole genome sequencing (∼50X coverage), and only 2 of these 7 mutations were detectable in the primary tumor with targeted detection methods (104-105X coverage). Similar MAPK pathway mutations were detected in 11 of 18 human neuroblastoma-derived cell lines, and these lesions are predicted to be sensitive to small molecule inhibition of MEK in vitro (p Conclusions In this study of 23 neuroblastoma cases selected based solely on having diagnostic-relapse specimens available for analysis, MAPK pathway mutations were highly enriched in the relapsed genomes, providing a potential biomarker for new therapeutic approaches to chemotherapy refractory disease. The fact that several ALK-RAS-MAPK mutations were found in the relapse but not in the corresponding primary tumors favors a model in which rare subclones with secondary driver mutations expand over time. However, it remains to be determined whether these mutations occurred de novo after treatment, were present in rare subclones below detection limits, or were undetectable due to spatial heterogeneity of the primary tumor, which will impact the clinical utility of targeted sequencing at diagnosis. Our study provides strong rationale for performing biopsies on relapse neuroblastoma tumors in order to comprehensively characterize the molecular lesions that underlie treatment-refractory disease, determine their prognostic relevance, and guide treatment decisions for patients. Citation Format: Derek A. Oldridge, Thomas F. Eleveld, Virginie Bernard, Jan Koster, Leo C. Daage, Sharon J. Diskin, Linda Schild, Nadia B. Bentahar, Angela Bellini, Mathieu Chicard, Eve Lapouble, Valérie Combaret, Patricia Legoix-Né, Jean Michon, Trevor J. Pugh, Lori S. Hart, JulieAnn Rader, Edward F. Attiyeh, Jun S. Wei, Shile Zhang, Arlene Naranjo, Julie M. Gastier-Foster, Michael D. Hogarty, Malcolm A. Smith, Jaime G. Auvil, Thomas B. K. Watkins, Danny A. Zwijnenburg, Marli E. Ebus, Peter van Sluis, Anne Hakkert, Esther van Wezel, C. Ellen van der Schoot, Ellen M. Westerhout, Johannes H. Schulte, Godelieve A. Tytgat, M. Emmy M. Dolman, Isabelle Janoueix-Lerosey, Daniela S. Gerhard, Huib N. Caron, Olivier Delattre, Javed Khan, Rogier Versteeg, Gudrun Schleiermacher, John M. Maris, Jan J. Molenaar. Relapsed neuroblastomas show frequent RAS-MAPK pathway mutations. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 2980. doi:10.1158/1538-7445.AM2015-2980

Patient-derived xenografts and cell lines have been the underpinning of functional characterization and drug discovery efforts in cancer. The use of these models is often under the assumption that these systems are renewable, faithful representations of original progenitor tumor cell populations. To test this assumption, we performed whole exome (92X median coverage) and genome sequencing (36X median coverage) analysis of cell line and patient-derived mouse xenografts (PDXs) originating from 7 neuroblastoma patients. Data are from1 primary tumor, 3 PDXs, and 15 neuroblastoma cell lines cultured from tumor, bone marrow, or blood. The cell lines consisted of 4 pre-/post-therapy pairs and 3 pairs established and maintained in either hyperoxia (room air i.e. “standard” cell culture) or physiologic (bone marrow hypoxia = 5%) oxygen. 7 lymphoblastoid or fibroblast cell lines were used as matched normals to identify somatic mutations. Subclonal population structures were inferred from somatic mutation calls calibrated for copy number state and tumor purity. In all cell lines and xenografts, we observed 1-2 additional subclonal populations, primarily supported by deep coverage from exome sequencing. In nearly every case, we observed shifts in the proportional representation of genetic subclones and many subclones showed additional mutations not evident in the progenitor tissue or cancer line derived in parallel. Comparison of three cell line pairs established in bone marrow level hypoxia versus room air found only ∼40% of coding mutations in each line were shared (average 82 mutations per line), suggesting significant genetic impact of growing tumor cells in the two different culture conditions. Matched PDXs from these cases had only ∼17% of coding mutations shared across all three models. The greatest genetic similarity was seen between paired cell lines established from tissue obtained pre-/post-therapy from the same patient (36 coding mutations shared, 14 private to diagnosis and 13 private to progression). However, a second pre/post-therapy cell line pair did not share any coding mutations, although they did have 585 non-coding mutations in common (of 4,033 and 2,480 in each line), assuring that the relapse was derived from a diagnostic tumor clone. These results highlight a need for comprehensive subclonal analysis of human cancer laboratory models to better inform design and interpretation of biological and preclinical therapeutic studies. Citation Format: Maya Schonbach, Arnavaz Danesh, Jeff Bruce, Tito Woodburn, Tanja Davidsen, Leandro Hermida, Patee Gesuwan, Jaime Guidry Auvil, Oliver Hampton, David Wheeler, Julie Gastier-Foster, Malcolm Smith, Daniela Gerhard, John M. Maris, Patrick Reynolds, Trevor J. Pugh. Fidelity of subclonal representation in human neuroblastoma-derived cell line and patient-derived xenograft models: A report from the NCI-TARGET project. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 484. doi:10.1158/1538-7445.AM2015-484

Introduction Epidemiological data for rare malignancies can be difficult to obtain; population-based data sets are often abridged to group disparate rare malignancies into larger more manageable clusters. These problems are acute in Canada, where such epidemiological data are often estimated from “second-hand” US data (e.g. Surveillance, Epidemiology and End-Results (SEER) data). Recently, the Canadian Institute of Health Information (CIHI) has set out to make epidemiological data more readily accessible, permitting University-affiliated researchers to access the anonymized and codified CIHI Discharge Abstract Database (DAD). Methods We undertook to estimate the incidence, demographics, and outcome data relating to the various subtypes of peripheral T-cell lymphomas (PTCLs) in Canada (excluding Quebec and British Columbia, for which data was not collected). The CIHI DAD consists of a two fiscal-year anonymized 10% random sample of all hospital discharge abstracts in Canada. The DAD is indexed by a unique anonymous patient identifiers and includes relative date metrics, by which all dates originally present on the abstract are standardized to a unique but confidential CIHI DAD reference date. From these data we were able to identify all hematolymphoid diagnoses, isolate the PTCLs, separate new from historical diagnoses, and identify patient age range, gender and disposition data. When the disposition data were combined with the relative date metrics, a gross estimate of T-cell lymphoma overall survival (relative to all other hematolymphoid diagnoses) was generated. Population normalization was achieved using inter-censal estimates obtained from Statistics Canada. Results PTCL incidence was estimated at 0.72 cases per 100,000 per annum (comparable to recently published SEER data). We also estimated a prevalence of 21 PTCLs per 100,000 healthcare encounters. Most cases of PTCL originated from males (63%) and the distribution of age ranges was skewed toward older adults (median age by number of cases = 60 years). The most frequent diagnosis was PTCL, NOS (46%). By the Cox-proportional hazards method, there was a statistically significant difference in survival between the T-cell lymphomas and non T-cell hematolymphoid malignancies (regression co-efficient for PTCL vs. non-PTCL diagnosis p = 0.003) in favor of the latter; not surprisingly, age was also predictive of overall survival, regardless of the subtype of malignancy (regression co-efficient p = 0.004). Conclusions To our knowledge, the above is the first attempt to estimate the epidemiology of PTCLs in Canada. In addition, we present a unique approach to obtaining high-quality (albeit geographically incomplete) Canadian epidemiological data via the CIHI DAD database; this dataset may serve as a valuable resource in the context of rare diseases whose epidemiological data may not be widely or publicly available. Citation Format: Etienne R. Mahe, Princess Margaret Cancer Centre Advanced MolecularDiagnostics Laboratory, Trevor Pugh, Tracy Stockley, Suzanne Kamel-Reid. Filling the void of Canadian T-cell lymphoma epidemiology: Data from the canadian institute for health information discharge abstract database. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 3708. doi:10.1158/1538-7445.AM2015-3708

Introduction: The molecular characterization of patients who exhibit a unique response to treatments that are not effective in most patients, so called “exceptional responders”, can shed light on the biological underpinnings of the response and provide rationale for selection of future patients for the same treatment. Here we present the genomic characterization of a 22-year-old female patient who presented with an unclassifiable kidney cancer and diffuse bilateral lung metastases. The primary tumour was resected and platinum plus gemcitabine therapy achieved a complete resolution of the lung metastases. Approximately 2 years later the patient again developed bilateral lung metastases and an additional course of platinum plus gemcitabine once again achieved a complete radiological response. To better understand the genomic contributors to this dramatic and recurrent response, we performed whole exome sequence analysis of the diagnostic formalin-fixed paraffin-embedded tumor tissue and a matched blood sample. Methods and Results: The patient provided informed consent and was profiled through the Princess Margaret Cancer Centre Genomic Investigation of Unusual and Spectacular responders (GENIUS) program. From tumour and matched blood specimens, we isolated coding and untranslated regions followed by deep sequencing (∼250X median coverage) on an Illumina HiSeq 2000. Copy-number analysis using VarScan2 revealed 23 somatic copy-number alterations (12 gains, 11 losses) including loss of 9p and gains of 3q and 8q, consistent with reports on clear cell renal cell carcinomas. Mutational analysis using muTect found 60 somatic, coding mutations, consistent with mutation rates seen in other kidney cancers. Of these variants, only one was a previously reported hotspot mutation, GNAS p.R201H (NM_000516, also known as p.R844H on transcript NM_080425). This mutation causes constitutive activation of the G-protein complex and activates adenylate cyclase to produce cyclic-AMP (cAMP) that can activate oncogenic pathways. However, excess cellular cAMP levels have also been found to promote apoptosis. Furthermore, a germline SNP in GNAS (rs7121), thought to increase GNAS transcript stability, is associated with improved response to gemcitabine plus platinum in non-small-cell lung cancer patients. The current patient was found to be heterozygous for the rs7121 SNP, which may compound the chemosensitivity phenotype observed. Conclusion: We have identified a p.R201H mutation in GNAS that may both drive tumour progression and confer exceptional chemo-sensitivity in a patient with an unclassified kidney cancer. Further analysis is underway in vitro to validate the hypothesis that this mutation confers sensitivity to chemotherapy. Confirmation of this hypothesis may suggest that chemotherapy could be considered for patients harbouring this mutation. Citation Format: Jeff P. Bruce, Arnavaz Danesh, Neil Winegarden, Patrick Yau, Carl Virtanen, Suzanne Kamel-Reid, Michael H. Roehrl, Anthony M. Joshua, Jennifer Knox, Trevor J. Pugh. Exome analysis of an exceptional responder uncovers an activating GNAS mutation that may confer sensitivity to cytotoxic chemotherapy. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 3898. doi:10.1158/1538-7445.AM2015-3898

The use of Next-Generation Sequencing (NGS) technologies is increasingly prevalent within diagnostic labs. As genomic regions are sequenced to greater depth in cancer diagnostics, it is critical to differentiate clinically actionable variants from artifacts arising from sequencing-errors, sample- processing or sample-age, and to identify samples that will be difficult to evaluate. We sought to determine whether strand-specific sequencing approaches, such as the TruSight Tumor Sequencing Panel (Illumina) could enable sample and variant triage in a clinical diagnostic settingTruSight Tumor Sequencing Panel allows for paired-end sequencing of individual strands of DNA and analyzing them either together (Paired) or separately (Pool A and Pool B). Variants identified in one pool, but not the other, are putative artifacts; variants identified in both pools are considered true calls. Combined analysis of both pools was performed in two ways: By summing variant calls across pools, and by informatically determining overlapping variant calls between pools. In a test cohort of 44 FFPE samples of varying age and tumor type, we assessed whether age of sample, strand bias, and fixation impacted the detection of high confidence variants using the TruSight Tumor Sequencing panel. Data were compared to the results of analysis of the same samples using the established Illumina TruSeq Amplicon Cancer Panel and/or Sanger Sequencing.Sample age, tumor cellularity, tumor type and template DNA quality were not found to be associated with quality of NGS output in our study. We also evaluated the overall transition/transversion (Ti/Tv) ratios for variants detected either uniquely in one pool or in combined analysis. Interestingly, for variants detected in both pools, the Ti/Tv ratio was 1.97, compared to 0.52-0.60 for those detected in only 1 pool (p < 0.001). Strikingly, samples that sequenced successfully but gave inconclusive and difficult to interpret variant lists were associated with%G>A:C>T transition > 62.5% and Ti/Tv ratios of > 4.0 (p < 0.001). G>A:C>T transitions were significantly over-represented in these samples. The overall%G>A:C>T transitions were equivalent (44-52%) in individual pools or in paired analysis. However, when inconclusive samples were accounted for, the%G>A:C>T transitions differed between the two analyses: 49.7% (paired) vs. 30.1-32.1% (individual pools). In summary, the Ti/Tv ratio can act as a critical determinant of variant call quality - Ti/Tv ratios ∼0.5 represent sequencing artifacts, while Ti/Tv ratios > 4.0 are indicative of inconclusive sequencing output.We conclude that variant Ti/Tv ratio as well as%G>A:C>T transition in variants detected by the TruSight Tumor Sequencing Panel may be helpful evaluators of quality and clinical utility of sequencing output for FFPE tumor samples tested in a clinical diagnostic setting. Citation Format: Swati Garg, Mahadeo A. Sukhai, Mariam Thomas, Michelle Mah, Tong Zhang, Trevor Pugh, Suzanne Kamel-Reid, Tracey L. Stockley. Determinants of quality of next-generation sequencing output from the strand-specific TruSight Tumor Sequencing Panel in a clinical diagnostic setting. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 628. doi:10.1158/1538-7445.AM2015-628

Recent cancer genome profiling studies have increased our understanding of the somatic mutation landscape of myeloid malignancies. A number of genes and variants are known to have prognostic/predictive utility in several myeloid malignancies, allowing for more accurate stratification, and enhanced patient management. This has led to consideration of NGS for detection of somatic mutations in myeloid malignancies in the clinical diagnostic setting, to supplant single-gene molecular testing assays in current use. Building on our prior work establishing a novel mass spectrometry-based high throughput mutation detection assay for hematologic malignancies, we investigated the application of NGS to myeloid malignancy diagnostics. To do this, we validated the Illumina TruSight Myeloid Sequencing Panel (54 genes, 568 amplicons) on 71 acute myeloid leukemia (AML), myelodysplastic syndrome (MDS), and myeloproliferative neoplasm (MPN) patient samples, alongside relevant controls. NGS Libraries were prepared using standard protocols, and sequencing performed on the Illumina MiSeq platform. Concordance among NGS calls, single gene tests, and Sanger verification was tested as part of the clinical validation process. Analysis of 41 cases tested in our previously developed mass spectrometry assay indicated 100% concordance (70/70) for reportable variants mutually covered in both assays. Single nucleotide variations detectable in lab-standard single gene assays were all found by NGS (100% concordance). 58/71 (82%) cases had at least one additional potentially clinically relevant variant that would not have been identified in the existing assay (mean 1.95 additional variants/case; range 1-9). Additionally, we determined that AMLs carrying IDH1, IDH2 or TET2 mutations had a higher mutation burden (mean 4.6 mutations/case, range 2-7), compared to AMLs wild-type for these three genes (mean 3.1 mutations/case, range 2-4; p = 0.008). Clinically relevant insertions (up to 33 bp) and deletions (up to 52 bp) associated with AML and MPN, including FLT3 ITD and CALR deletions, were detected in known positive cases. In these cases, our analysis was supplemented with a custom bioinformatics algorithm allowing for alignment against an artificial reference sequence to detect larger indels. Due to low coverage of the clinically actionable CEBPA, we supplemented the NGS assay with Sanger sequencing for this locus. Therefore, we report the validation of an NGS panel for high throughput detection of mutations in myeloid malignancies, and the development of a wet-bench and informatics workflow enabling maximal information benefit in the diagnostic setting. This pipeline allows the detection of variants that impact diagnosis and patient management, with significantly improved information benefit over current tests. Citation Format: Mariam Thomas, Mahadeo Sukhai, Tong Zhang, Djamel Harbi, Justin De Souza, Katherine MacDonald, Trevor Pugh, Mark Minden, Andre Schuh, Tracy L. Stockley, Suzanne Kamel-Reid. Clinical testing and implementation of the TruSight Myeloid Next Generation Sequencing (NGS) panel for identification of clinically relevant variants in hematological malignancies. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 4260. doi:10.1158/1538-7445.AM2015-4260

Despite improvement of survival using multimodal chemo- and immunotherapy, high mortality and morbidity is still substantial for pediatric patients with metastatic cancers. Recent large-scale sequencing studies of pediatric tumors including rhabdomyosarcoma (RMS) and neuroblastoma (NB) have been focusing on somatic mutations, and revealed a low somatic mutation rate and surprisingly few recurrently somatic mutated genes in these childhood tumors. Currently, only a small portion of pediatric cancer cases can be explained by somatic driver events; whereas the cause for the majority of these diseases remains unknown. Because both these two types of tumors are uncommon, here we hypothesize that infrequent germline mutations (frequency To test this hypothesis, we utilized sequencing data from two cancer patient cohorts consisting of RMS (n=133) and NB (n=222) patients, of which latter is a part of the Therapeutically Applicable Research to Generate Effective Treatments (TARGET) initiative for pediatric cancers. First, high-quality protein-coding changing single nucleotide variants (SNVs) were called in both paired germline and tumor genomic DNAs. Then we excluded common variants with frequency >5% in a normal human population using the 1000 Genomes data. Due to our interest in the enriched variants, we further required the frequencies of variants in our rhabdomyosarcoma and neuroblastoma patient cohorts are higher than those in the ESP dataset, a non-cancer control population comprising 6503 individuals. There are 63247 SNVs fulfilled these selection criteria. Among them, 1589 have been reported in these pediatric cancers or in other malignancies in the Cancer Genome Atlas (TCGA) project; and 1178 variants are present in the Human Gene Mutation Database (HGMD). Of these HGMD variants, 49 have been reported in human diseases and 34 of them are known disease-causing mutations for human cancers and genetic disorders including TP53, ALK, CHEK2, and PINK1. Interestingly, the most frequent germline mutations in these pediatric tumors were rarely found in the TCGA project which mostly consists of adult cancers. This observation suggests a very different genetic background of pediatric cancer patients from that of the adult cancers, and warrants a careful examination of germline mutations in these cancers. Furthermore, previous studies have highlighted the importance of expression of variant genes (including tumor suppressor genes) for identification of driver mutations in cancers. Therefore we will use 178 transcriptome sequencing experiments available for these tumors (RMS=84; NB=93) to identify expressed variants in tumor. Statistical and pathway analyses are currently underway to determine potential pathological or casual germline mutations associated with neuroblastoma and rhabdomyosarcoma. Citation Format: Jun S. Wei, Rajesh Patidar, John Shern, Shile Zhang, Trevor Pugh, Sharon J. Diskin, Sivasish Sindiri, Young K. Song, Hongling Liao, Xinyu Wen, Jianjun Wang, Stephen X. Skapek, James R. Anderson, Frederic G. Barr, Robert C. Seeger, John M. Maris, Douglas S. Hawkins, Javed Khan. Systematic identification of germline mutations in rhabdomyosarcoma and neuroblastoma using massively paralleled sequencing. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 5081. doi:10.1158/1538-7445.AM2014-5081

Medulloblastoma (MB) is the most common malignant brain cancer in children. Although MB can be categorized into four distinct subtypes, the underlying genetic changes required to promote MB remain unclear. We performed whole exome sequencing was performed on germ-line DNA from two affected cousins who were diagnosed at similar ages with classic WNT subtype MB, and their unaffected mothers, who are full sisters. We found cousins to be homozygous for a variant in a gene regulating WNT-signaling (variant 1), while their mothers were heterozygous. In a mouse model of WNT-type MB, incomplete penetrance and the requirement for inactivation of TP53 were observed. We, therefore, hypothesized that a deleterious mutation in the TP53 pathway may be present. Indeed, we found a single copy of a novel mutation in a regulator of TP53 shared between the affected cousins (variant 2). This allele segregated through the cousins’ maternal line as evidenced by both mothers being heterozygous for this allele. Finally, we identified a common homozygous variant in a gene involved in promoting RAS-mediated cell proliferation (variant 3). We found that mutations in the genes of variants 1 and 3 have been previously reported to be involved in sporadic MB, suggesting that what we observe in this family translates to sporadic disease. Thus, our model for MB development is that of a threshold effect in which combinations of both familial and common variants interact to increase susceptibility. We hypothesize that this model with incomplete penetrance may underlie cancer susceptibility in many families with apparent cancer predisposition syndromes but in whom no causative mutation has been identified. We also hypothesize that this multi-gene model with incomplete penetrance may also explain a proportion of apparently sporadic cases of cancer. This work underscores the advantages of analyzing sequence data from multiple well-chosen affected and unaffected individuals in well-chosen families for gene discovery in cancer and other complex diseases. It also puts forth a new model for analysis, incorporating both agnostic and hypothesis driven models using data from the literature and complementary datasets. Citation Format: Mark M. Sasaki, Andrew D. Skol, Trevor J. Pugh, Matthew Meyerson, Kenan Onel. Whole exome sequencing analysis of familial Medulloblastoma. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 4018. doi:10.1158/1538-7445.AM2013-4018

Pleuropulmonary blastoma (PPB) is a rare, aggressive childhood cancer arising from lung mesenchymal cells. In early stages, these cells reside in the walls of epithelial-lined cysts and may remain dormant or progress to thickened cysts and then highly malignant masses over several years. Familial germline loss-of-function DICER1 variants have been described, although carriers may not be affected. Therefore, we sought to uncover additional genetic events underlying tumor development through whole exome sequencing of 15 tumor/normal pairs. In addition to known loss-of-function germline variants found throughout DICER1, this analysis identified somatic missense mutations clustered in the DICER1 RNase IIIb domain in nearly every case. While the somatic mutations impact novel amino acid residues, expression analysis demonstrated depletion of 5p-derived miRNAs similar to that described in other DICER1-mutant tumors. While mutation rates were relatively high compared to other pediatric solid tumors, most mutations, except those in DICER1 and TP53, appeared to be subclonal. Early stage tumors (PPB Type II) had predominantly single-copy deletion of TP53 whereas late stage tumors (PPB Type III) had secondary TP53 mutations. Several late stage tumors also had activating mutations in RAS pathway members, including NRAS and BRAF. Therefore, compound DICER1 loss appears to be initiating and synergistic with incremental loss of TP53 and acquisition of additional driver mutations. We subsequently characterized this molecular pattern using deep sequencing and immunohistochemistry in an extension cohort of 45 additional tumors. This step-wise series of germline and somatic alterations illustrates a genetic mechanism consistent with the progressive nature of pleuropulmonary blastoma. Citation Format: Trevor J. Pugh, Weiying Yu, Jiandong Yang, Lauren Ambrogio, Scott L. Carter, Adam Kiezun, Aaron McKenna, Petros Giannikopoulos, Gad Getz, Yoav Messinger, Matthew Meyerson, D. Ashley Hill. Progressive biallelic loss of TP53 is associated with progression of pleuropulmonary blastoma initiated by germline loss and somatic mutation of DICER1. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 3806. doi:10.1158/1538-7445.AM2013-3806

Introduction: The International Cancer Genome Consortium (ICGC) is a worldwide network aiming to provide comprehensive molecular genetic profiles of 50 clinically and societally important tumors. A major focus of the ICGC PedBrain Tumor project is medulloblastoma (MB) - the most common malignant brain tumor in childhood. Importantly, recent work has shown that MB is not a single disease, but is in fact comprised of 4 distinct molecular subgroups (WNT, SHH, Group 3, Group 4). Using an integrative next-generation sequencing-based approach, we have investigated tumorigenic events underlying medulloblastoma, with the aim of identifying novel diagnostic or prognostic markers, and therapeutic targets. Methods: The whole genome of 50 tumor-normal DNA pairs, plus the whole exome or 2,500 target genes in a further 100 sample pairs from different histological and transcriptomic subgroups, was sequenced on an Illumina HiSeq platform. Sequencing-based high-resolution copy-number data was also generated for all cases. Results: Overall, a smaller number of non-synonymous somatic mutations was seen than in most adult malignancies. The number of somatic changes also clearly showed a positive correlation with patient age, suggesting either longer-term selection for more weakly transforming alterations, or the acquisition of additional passenger mutations in tumors which grew over a longer period. Whilst some point mutations and small InDels affected known MB or cancer-related genes (e.g. CTNNB1, PTCH1, MLL2), the vast majority were in genes which have not previously been implicated in medulloblastoma. Furthermore, even the most recurrently altered genes were mutated in Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 4872. doi:1538-7445.AM2012-4872

DDX3X is an ATP-dependent RNA helicase that belongs to an extended family of D-E-A-D motif (DEAD-box) containing proteins whose functions have been implicated in key biological processes ranging from RNA transcription and splicing to the regulation of translation initiation. Through whole exome sequencing of 94 primary medulloblasotmas, we have identified DDX3X as one of the most common recurrently mutated genes in medulloblastoma. These mutations predominantly occur in WNT pathway medulloblastomas which concurrently harbor stabilized forms of mutant beta-catenin. Interestingly, we have also identified SMARCA4 mutations in WNT pathway tumors that are mutually exclusive of DDX3X mutations. In addition, two WNT pathway tumors that have neither DDX3X nor SMARCA4 gene mutations have mutations in genes known to encode for direct binding partners of DDX3X and SMARCA4 (EIF3E and ACTL6A, respectively). This mutual exclusivity of DDX3X and SMARCA4 mutations suggests either a direct interaction between the DDX3X and SMARCA4 gene products or an overlap of the biological processes regulated by each. Here we reveal the direct interaction between DDX3X and SMARCA4 in normal physiological conditions and further investigate the impact of somatic mutations in DDX3X and SMARCA4 on this interaction in WNT pathway associated medulloblastomas. Our results point to a critical link between RNA helicases and the SWI/SNF chromatin remodeling complex in WNT/beta-catenin signaling and disease. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 1313. doi:1538-7445.AM2012-1313

Medulloblastoma is a pediatric brain tumor with poor overall survival and adverse long-term effects from current surgical and radiation treatments. Our group has recently used expression and copy number analysis to define six medulloblastoma subtypes, c1-c6. To assess whether these subgroups contain somatic mutations that encode potential therapeutic or diagnostic targets, we performed whole exome hybrid capture and Illumina sequencing of 94 tumor/normal pairs. For each sample, we sequenced 193,094 exons from 18,863 genes to 126X average coverage. Tumors contained a median 5 silent and 11 non-silent candidate mutations, corresponding to 0.34 non-silent mutations per megabase, a low mutation rate consistent with other pediatric tumors. Overall, 16 genes were mutated at statistically significant frequency and several clustered within subtypes. Eleven tumors harbored mutually-exclusive, likely loss-of-function missense mutations within the helicase domains of candidate oncogene DDX3X (n=7) or SMARCA4 (n=4). These mutations were found in five of seven c6 tumors (3 DDX3X, 2 SMARCA4) of which four had known beta-catenin mutations. Two SMARCA4 mutations were identical and two affected adjacent residues. None of the DDX3X mutations were recurrent, however several are proximal when mapped to a tertiary protein model. To confirm these variants, we are performing deep sequencing of these genes using multiplex PCR (Fluidigm) followed by single-molecule real-time sequencing (PacBio). We are also characterizing the functional effect of DDX3X and SMARCA4 mutations alone and in combination with beta-catenin mutations in medulloblastoma cell lines. Other genes with subtype-associated, loss-of-function mutations include DULLARD in 2 of 9 c1 tumors, PTCH1 in 5 of 16 c3 tumors, and MLL2 in 2 of 3 tumors without any copy number alterations as well as a c1 tumor and two c3 tumors. Several genes are mutated across subtypes, notably tumor suppressors TP53 (n=3), GPS2 (n=3) and SOCS4 (n=2). 19 chromatin remodeling genes including cancer genes BRCA2, KDM5/6A, CREBBP, EP300, BRD4, and MLL3/4 are mutated across 17 tumors from all subtypes. While we have uncovered several subtype-associated mutations, 88% of mutated genes are only altered in a single tumor. This analysis demonstrates the diversity of somatic mutation in medulloblastoma, even within copy number subtypes. To better understand infrequently mutated genes, we will attempt to assemble them into commonly altered gene sets and pathways. We also intend to investigate potentially pathogenic germline variations in each case. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 4868. doi:1538-7445.AM2012-4868

The landscape of somatic alterations in cancer is being rapidly uncovered by large-scale cancer genome sequencing projects in individual laboratories as well as consortia such as The Cancer Genome Atlas (TCGA) and the International Cancer Genome Consortium (ICGC). The decreasing cost of sequencing has resulted in an increase in the number of cancer genomes that are being sequenced, and thus the numbers of candidate somatic mutations being identified are rapidly expanding. A single cancer genome from a cancer with an average mutation background rate, such as ovarian or breast cancer, will have 3000-6000 mutations. A high mutation rate cancer such as melanoma or colon cancer can have as many as 100,000 mutations/genome. Multiplying these numbers by the dozens of sequenced genomes and hundreds of sequenced exomes across a wide spectrum of tumor types yields a staggering number of candidate mutations. However, approaches for rapidly validating these findings at scale have been lagging behind. Current approaches, such as custom designed hybrid capture arrays followed by sequencing, can take months to complete and cannot begin until the initial sequencing is complete and mutations are called. This can lead to long delays in the interpretation and publication of biological findings. Here we describe the results of several approaches for rapid and efficient validation on large-scale projects, such as TCGA. We will share our approach and results using Fluidigm and PacBio sequencing for validation and sample extension at a small scale with rapid turnaround where we have validated PIK3CA and TP53 mutations in breast cancer samples of varying purity as well as significantly mutated genes in medulloblastoma. Ideally one could validate mutations and gene fusions concurrent with their discovery. To that end, we will also describe our validation results using RNA-Seq data, commonly produced concurrently with genome and exome sequencing, in lung squamous and renal cancer. Finally, we will explore an approach for instant-validation through simultaneous sequencing of barcoded discovery and cross-validation libraries. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 3960. doi:1538-7445.AM2012-3960

Prostate cancer is the second most common cause of male cancer deaths in the United States, accounting for 200,000 new cases and 32,000 deaths per year. Chromosomal rearrangements comprise a major mechanism driving prostate carcinogenesis. For example, recurrent gene fusions that render ETS transcription factors under the control of androgen-responsive promoters are present in the majority of prostate cancers. Other types of somatic alterations, such as base substitutions, small insertions/deletions, and chromosomal copy number alterations, have also been described, yet the full repertoire of genomic alterations that underlie primary human prostate cancer remains incompletely characterized. We present here the most comprehensive genome sequencing effort in prostate cancer reported to date. We have characterized the complete genomes of 7 primary prostate cancers and patient-matched normal samples using massively parallel sequencing technology. We observed a mean mutation frequency of 0.9 per megabase, consistent with what has been reported for other tumor types. However, our results indicate that translocations and other chromosomal rearrangements are far more common than expected, with a median of 90 per prostate cancer genome. Several tumors contained chains of balanced rearrangements involving multiple loci associated with known cancer genes. We observed a striking and unexpected relationship between rearrangement breakpoints and chromatin structure, which differed for tumors harboring the ETS gene fusion TMPRSS2-ERG and tumors lacking ETS fusions. We also observed an enrichment of point mutations near rearrangement breakpoints. Three of seven tumors contained rearrangements that disrupted CADM2, a nectin-like member of the immunoglobulin-like cell adhesion molecules; recurrent CADM2 rearrangements were also detected in an independent cohort by fluorescent in situ hybridization (FISH). Four tumors harbored rearrangements disrupting either PTEN, a prostate tumor suppressor, or MAGI2, a PTEN interacting protein not previously implicated in prostate cancer. Together, these results illuminate potential avenues for target discovery and reveal the potential of complex rearrangements to engage prostate tumorigenic mechanisms. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 102nd Annual Meeting of the American Association for Cancer Research; 2011 Apr 2-6; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2011;71(8 Suppl):Abstract nr 3925. doi:10.1158/1538-7445.AM2011-3925

Neuroblastoma (NBL) is a cancer arising from the peripheral nervous system and is the most common solid tumor outside the brain in infants and children. Despite improved treatment over the past 20 years, only 45% of those with high-risk disease survive emphasizing the need for new molecular targets for therapy. To identify genetic mutations that encode such targets, we present an analysis of 162 whole exome sequences from 81 tumor/normal pairs from high-risk NBL patients (N=27 with MYCN amplification). From each sample, genomic DNA fragments encoding 185,961 exons from 18,380 genes were isolated using in-solution hybrid capture and sequenced using the Illumina sequencing platform to a median of 195X coverage. A wide diversity of somatic variation was observed with non-silent mutation counts ranging from 5 to 149 per tumor (median=24). Novel hyper-mutation phenotypes (> 125 candidate coding somatic mutations) were observed in four tumors including one case with both a deletion and a truncating mutation in MLH1. Several genes were mutated across the set at a statistically significant level including ALK (6 tumors, q=0.006) and PTPN11 (3 tumors, q=0.012). There are 27 other genes with candidate mutation frequencies in the 4-10% range, but no single gene identified to date with a mutation frequency >10%. Importantly, multiple paired blood samples (acquired at diagnosis) have sufficient numbers of circulating tumor DNA to potentially interfere with our SNV detection algorithm; refinement of the analytic pipeline using SNP-derived “contamination” metrics is ongoing. We anticipate validating candidate somatic candidates using an orthogonal genotyping method and further defining their true frequency in an expanded sample set that includes non-high-risk cases as well. Finally, we plan a concerted effort to catalogue potentially detrimental germline variations in order to further extend our understanding of the heritable component of NBL. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 102nd Annual Meeting of the American Association for Cancer Research; 2011 Apr 2-6; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2011;71(8 Suppl):Abstract nr 4756. doi:10.1158/1538-7445.AM2011-4756