The Cancer Genome Atlas Cancer Research Articles

Pan-Cancer Analysis Reveals the Diverse Landscape of Novel Sense and Antisense Fusion Transcripts

Gene fusions that contribute to oncogenicity can be explored for identifying cancer biomarkers and potential drug targets. To investigate the nature and distribution of fusion transcripts in cancer, we examined the transcriptome data of about 9,000 primary tumors from 33 different cancers in TCGA (The Cancer Genome Atlas) along with cell line data from CCLE (Cancer Cell Line Encyclopedia) using ChimeRScope, a novel fusion detection algorithm. We identified several fusions with sense (canonical, 39%) or antisense (non-canonical, 61%) transcripts recurrent across cancers. The majority of the recurrent non-canonical fusions found in our study are novel, unexplored, and exhibited highly variable profiles across cancers, with breast cancer and glioblastoma having the highest and lowest rates, respectively. Overall, 4,344 recurrent fusions were identified from TCGA in this study, of which 70% were novel. Additional analysis of 802 tumor-derived cell line transcriptome data across 20 cancers revealed significant variability in recurrent fusion profiles between primary tumors and corresponding cell lines. A subset of canonical and non-canonical fusions was validated by examining the structural variation evidence in whole-genome sequencing (WGS) data or by Sanger sequencing of fusion junctions. Several recurrent fusion genes identified in our study show promise for drug repurposing in basket trials and present opportunities for mechanistic studies.

Abstract 2468: Towards precise and cost-effective fusion discovery: A landscape of druggable gene fusions across TCGA cancers

Abstract Gene fusions are one of the hallmarks of cancer and are among the most powerful biomarkers and drug targets in translational cancer genomics. We deploy sMACHETE (scalable MACHETE), a precise and sensitive fusion detection algorithm, particularly engineered for mining massive cancer sequencing databases, to provide a landscape of fusions across human primary cancers. sMACHETE consists of two main computational components: MACHETE-based component and Sequence Bloom Tree (SBT) checkpoint. MACHETE (Hsieh et al., 2017) is a precise fusion algorithm, which employs a statistical model to identify fusion junctions. The first component in sMACHETE is built on MACHETE and has undergone major algorithmic and computational improvements, such as the inclusion of well-known cancer fusions and a cloud-based implementation in Common Workflow Language, which makes the pipeline a good fit for large-scale studies. To control for false positives due to multiple testing in large datasets, the fusions called by the first component are then queried via SBT (Solomon and Kingsford, 2016), which is a kmer-based query algorithm. The fusions whose detection frequencies by MACHETE and SBT are statistically consistent could pass the checkpoint and are called by sMACHETE. sMACHETE achieved 100% positive predictive value, higher than any other top performing algorithm and comparable sensitivity on simulated benchmarking datasets. We have used sMACHETE to systematically analyze fusions in The Cancer Genome Atlas (TCGA) RNA-seq data datasets. sMACHETE calls 31,546 highly confident fusions in 9,946 TCGA tumor samples spanning 33 cancer types. Sarcoma (10 fusions per sample) and Esophageal Carcinoma (8 fusions per sample) have the highest abundance of fusions. We found 525 recurrent fusions, observed in at least 2 samples within a cancer type, in 12% of tumor samples. Our statistical analysis reveals a signature of selection for recurrent fusions and also for recurrent genes, which partner with more than one gene in fusions and are observed in 40% of samples, suggesting an evidence for their oncogenic role in tumorigenesis. Thyroid, Ovarian, Esophageal, and Lung Adenocarcinoma have rates of kinase fusions that exceed expectation by chance, strong evidence that they are unappreciated drivers of the disease. Having integrated our detected fusions with OncoKB database (Chakravarty et al., 2017), we detected druggable fusions in 3% of tumor samples. Our systematic and functional analysis highlights the substantial role of fusions as cancer drivers and their clinical implication in cancer treatment. Citation Format: Roozbeh Dehghannasiri, Milos Jordanski, Donald E. Freeman, Gillian L. Hsieh, Jonathan M. Howard, Erik Lehnert, Julia Salzman. Towards precise and cost-effective fusion discovery: A landscape of druggable gene fusions across TCGA cancers [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 2468.

Abstract 2481: Developing analysis platform for pan-cancer study of DNA methylation, mirna and lncrna expression based on tumor subtypes using TCGA data

Abstract Introduction: The Cancer Genome Atlas (TCGA) consortium performed high-throughput sequencing of thousands of tumor and normal tissues across 33 cancer types. This has led to molecular characterization of different cancers and many novel discoveries. Such diverse data offers an excellent opportunity to further address the questions associated with tumor heterogeneity. Systematic exploration of epigenetic features and non-coding gene expression could lead clinicians/cancer researchers towards unearthing new diagnostic biomarkers and therapeutic targets. Previously our group developed a cancer transcriptome web portal, UALCAN [http://ualcan.path.uab.edu, Google: UALCAN)] to study expression and survival profile of protein coding genes among TCGA cancers based on various subgroups and molecular subtypes of cancer. Although many data portals facilitating easy access and in-depth analysis of TCGA data exists, there is need for a user-friendly resource facilitating comprehensive DNA methylation profile and gene expression analysis of lncRNA and miRNA within tumor subgroups based on factors such as stage, tumor grade, age, sex, race and integration with the gene expression. Methods: TCGA Level 3 RNA-seq, miRNA-seq and Illumina Infinium HumanMethylation450K data were downloaded using TCGA-Assembler 2 tool and Genomic Data Commons (GDC). The data were processed, organized based on tumor subtypes using custom R scripts. DNA methylation and gene expression analysis was performed using in-house PERL scripts, while statistical significance was estimated using “Statistics::TTest” module. The analyses results were stored as flat file database and hosted on a web portal developed using Apache2.0, PERL-CGI and HighChartJS JavaScript libraries. Results: The user friendly web portal 1) allows users could obtain list of top over-/under-expressed miRNAs and lncRNAs in each of 33 TCGA cancer 2) facilitates analysis of promoter methylation level and gene expression for given set of miRNA/ lncRNAs in various tumor sub-groups based on individual cancer stages, tumor grade, race, body weight or other clinicopathologic features, 3) enables researchers to download high resolution graphics such as heatmap, boxplots, dotplots as analyses outputs. The data can be downloaded in multiple output formats for use by researchers. Conclusion: The current resource aids cancer researchers in identifying DNA methylation mediated epigenetic modification of gene expression in tumor subgroup specific manner. Furthermore, our platform will facilitate analyses of long non-coding RNAs and microRNAs, thereby help in discovery of novel biomarkers and better understanding of the tumor biology. Citation Format: Darshan Shimoga Chandrashekar, Chad J. Creighton, Israel Ponce-Rodriguez, Sooryanarayana Varambally. Developing analysis platform for pan-cancer study of DNA methylation, mirna and lncrna expression based on tumor subtypes using TCGA data [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 2481.

Abstract LB-216: Tumor heterogeneity and therapy resistance analyzed at the single-cell level

Abstract Every tumor, either primary or metastatic is composed of many types of cancer cells, each with their own genomic and transcriptional makeup. Together with normal cells, mostly stroma, and immune cells, defining the tumor microenvironment, both populations create pathologic organ of unique spatial heterogeneity. Cancer therapies are selected based on analysis of the tumor cells historically using pathology evaluation, and more recently supported by mutational analyses through high throughput sequencing of panels of cancer genes or whole exomes. However, in many cases, less prominent and therapy-resistant cancer cell populations are missed due to bulk sequencing, and as such rare clones are not selected for, and can become resistant over time and create tumor spread. Here, we analyzed single-cell transcriptomic (scRNA-Seq) datasets of the tumor microenvironment of primary and metastatic samples of six tumor types: breast, colon, glioma, melanoma, head and neck, and kidney cancers. Using scRNA-Seq data from tumor and normal cells, we performed comprehensive analyses of tumor hallmarks, focusing on cancer stemness, molecular subtypes, and tumor heterogeneity. When available, we analyzed single-cells from longitudinal studies to understand mechanisms of resistance responsible for the failure of neo-adjuvant and adjuvant therapies. In total, we have performed an integrative analysis of gene expression of primary and metastatic tumors of 4202 bulk samples from The Cancer Genome Atlas (TCGA) and 96k cells belonging to 105 tumor samples from published single cell RNA-Seq datasets. TCGA datasets were analyzed by combining a recent BREW-based methodology and OCLR machine learning algorithm. OCLR trained on TCGA cancer specific samples estimated weights for each subtype and assigned the subtype to the highest correlated group for each individual cell. Following a similar procedure, we derived mRNAsi, a stemness index based on RNA-Seq data, for each individual single cell. Publicly available single-cell datasets and drug response data were utilized to test the association between stemness scores and drug sensitivity. Results confirmed that cell lines exposed to drug treatment resulted in lower stemness score compared to resistant cell lines. We detected intra-tumor stemness heterogeneity, and the presence of distinct molecular subtypes, suggesting that a unique target therapy might not be effective to eradicate all tumor cells. In case of breast cancer, therapy resistance was associated with higher mRNAsi. For low-grade glioma (LGG) we observed that most of the cells resulted in a codel subtype. In particular, mRNAsi score was able to stratify tumor subtypes; with higher scores corresponding to subtype G-CIMP-LOW and a lower mRNAsi score corresponding to Classic Like or LGm6-GBM. Moonlight dynamic recognition analysis showed upregulation of MYC Target and DNA_repair, while down-regulation of epithelial to mesenchymal (EMT) and TGF_beta signaling in cancer cells with higher mRNAsi. The results of our studies will help us to identify cancer cell phenotypes and tumor microenvironment composition that confer resistance to cancer therapies and may pave the way to implement scRNA-Seq analyses as diagnostic tools in clinical oncology. Citation Format: Antonio Colaprico, Francesca Petralia, Elena Papaleo, Olivier Gavaert, Xi Chen, Karol Szkudlarek, Maciej Wiznerowicz. Tumor heterogeneity and therapy resistance analyzed at the single-cell level [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 LB-216.

Optimization of Population Frequency Cutoffs for Filtering Common Germline Polymorphisms from Tumor-Only Next-Generation Sequencing Data

Clinical next-generation sequencing assays are often run on tumor specimens without a matched normal specimen, which complicates the differentiation of germline from somatic variants. In tumor-only testing, population data are often used to infer germline status, though no consensus exists on the exact population frequency (PF) cutoff above which a variant should be considered likely germline. In this study, five population databases plus the Catalog of Somatic Mutations in Cancer were used to demonstrate the impact of changing the PF cutoff on assignment of variants as germline versus somatic. The 1% to 2% PF cutoffs widely used in bioinformatic pipelines resulted in high sensitivity for classification of somatic variants, but unnecessarily reduced sensitivity for germline variants. Using optimized PF cutoffs, the source of variants in The Cancer Genome Atlas (TCGA) data could be predicted with >95% accuracy. Further exploration of four TCGA cancer data sets indicated that the optimal cutoff is influenced by both cancer type and the assay region of interest. Comparing TCGA data to data generated from a clinical, hybridization capture test (approximately 615 kb capture space) showed that PF cutoffs may not be transferable between assays, even when the gene set is held constant. Thus, filtering approaches need to be carefully designed and optimized, and should be assay-specific to support tumor-only testing until tumor-normal testing becomes routine in the clinical setting.

TRNA Fragments Show Intertwining with mRNAs of Specific Repeat Content and Have Links to Disparities

tRNA-derived fragments (tRF) are a class of potent regulatory RNAs. We mined the datasets from The Cancer Genome Atlas (TCGA) representing 32 cancer types with a deterministic and exhaustive pipeline for tRNA fragments. We found that mitochondrial tRNAs contribute disproportionally more tRFs than nuclear tRNAs. Through integrative analyses, we uncovered a multitude of statistically significant and context-dependent associations between the identified tRFs and mRNAs. In many of the 32 cancer types, these associations involve mRNAs from developmental processes, receptor tyrosine kinase signaling, the proteasome, and metabolic pathways that include glycolysis, oxidative phosphorylation, and ATP synthesis. Even though the pathways are common to multiple cancers, the association of specific mRNAs with tRFs depends on and differs from cancer to cancer. The associations between tRFs and mRNAs extend to genomic properties as well; specifically, tRFs are positively correlated with shorter genes that have a higher density in repeats, such as ALUs, MIRs, and ERVLs. Conversely, tRFs are negatively correlated with longer genes that have a lower repeat density, suggesting a possible dichotomy between cell proliferation and differentiation. Analyses of bladder, lung, and kidney cancer data indicate that the tRF-mRNA wiring can also depend on a patient's sex. Sex-dependent associations involve cyclin-dependent kinases in bladder cancer, the MAPK signaling pathway in lung cancer, and purine metabolism in kidney cancer. Taken together, these findings suggest diverse and wide-ranging roles for tRFs and highlight the extensive interconnections of tRFs with key cellular processes and human genomic architecture. SIGNIFICANCE: Across 32 TCGA cancer contexts, nuclear and mitochondrial tRNA fragments exhibit associations with mRNAs that belong to concrete pathways, encode proteins with particular destinations, have a biased repeat content, and are sex dependent.

CHASMplus Reveals the Scope of Somatic Missense Mutations Driving Human Cancers.

Large-scale cancer sequencing studies of patient cohorts have statistically implicated many genes driving cancer growth and progression, and their identification has yielded substantial translational impact. However, a remaining challenge is to increase the resolution of driver prediction from the level of genes to mutations because mutation-level predictions are more closely aligned with the goal of precision cancer medicine. Here, we present CHASMplus, a computational method that is uniquely capable of identifying driver missense mutations, including those specific to a cancer type, as evidenced by significantly superior performance on diverse benchmarks. Applied to 8,657 tumor samples across 32 cancer types in The Cancer Genome Atlas (TCGA), CHASMplus identifies over 4,000 unique driver missense mutations in 240 genes, supporting a prominent role for rare driver mutations. We show which TCGA cancer types are likely to yield discovery of new driver missense mutations by additional sequencing, which has important implications for public policy.

RHPCG: a database of the Regulation of the Hippo Pathway in Cancer Genome.

The Hippo signaling pathway is a highly conserved pathway controlling organ size, cell proliferation, apoptosis and other biological functions. Recent studies have shown that Hippo signaling pathway also plays important roles in cancer initiation and progression. However, a database offering multi-omics analyses and visualization of Hippo pathway genes in cancer, as well as comprehensive Hippo regulatory relationships is still lacking. To fill this gap, we constructed the Regulation of the Hippo Pathway in Cancer Genome (RHPCG) database. Currently, RHPCG focuses on analyzing the 21 core Hippo-protein-encoding genes in over 10 000 patients across 33 TCGA (The Cancer Genome Atlas) cancer types at the levels of genomic, epigenomic and transcriptomic landscape. Concurrently, RHPCG provides in its motif section 11 regulatory motif types associated with 21 core Hippo pathway genes containing 180 miRNAs, 6182 lncRNAs, 728 circRNAs and 335 protein coding genes. Thus, RHPCG is a powerful tool that could help researchers understand gene alterations and regulatory mechanisms in the Hippo signaling pathway in cancer.

Insights from multidimensional analyses of the pan-cancer DNA methylome heterogeneity and the uncanonical CpG-gene associations.

Although the DNA methylome profiles have been available in large cancer cohorts such as The Cancer Genome Atlas (TCGA), integrative analysis of the DNA methylome architectures in a pan-cancer manner remains limited. In the present study, we aimed to systematically dissect the insightful features related to the inter-tumoral DNA methylome heterogeneity in a pan-cancer context of 21 cancers in TCGA. First, pan-cancer clustering of the DNA methylomes revealed convergence of cancers and, meanwhile, new classifications of cancer subtypes, which are often associated to prognostic differences. Next, within each type of cancer, we showed that the transcription factor (TF) genes tend to bear more dynamic promoter DNA methylation profiles than the other genes, which serves as a potential source of the transcriptome heterogeneity in cancers. Finally, we found unanticipated significant numbers of the non-canonical promoter CpG sites that are positively correlated with the gene expression. Distribution patterns of these CpG sites in the CpG islands, ChIP-seq, DNaseI-seq, PMD regions and histone modification landscapes suggested against a pervasive mechanism of transcriptional activation due to mCpG-dependent binding of TFs, which is not in complete agreement with previous hypothesis. In summary, our deep mining of the highly heterogeneous DNA methylome data in a pan-cancer context generated novel insights into the architecture of cancer epigenetics and provided a series of resources for further investigations in the related fields of cancer genomics and epigenetics.

Characteristics of The Cancer Genome Atlas cases relative to U.S. general population cancer cases

BackgroundDespite anecdotal reports of differences in clinical and demographic characteristics of The Cancer Genome Atlas (TCGA) relative to general population cancer cases, differences have not been systematically evaluated.MethodsData from 11,160 cases with 33 cancer types were ascertained from TCGA data portal. Corresponding data from the Surveillance, Epidemiology, and End Results (SEER) 18 and North American Association of Central Cancer Registries databases were obtained. Differences in characteristics were compared using Student’s t, Chi-square, and Fisher’s exact tests. Differences in mean survival months were assessed using restricted mean survival time analysis and generalised linear model.ResultsTCGA cases were 3.9 years (95% CI 1.7–6.2) younger on average than SEER cases, with a significantly younger mean age for 20/33 cancer types. Although most cancer types had a similar sex distribution, race and stage at diagnosis distributions were disproportional for 13/18 and 25/26 assessed cancer types, respectively. Using 12 months as an end point, the observed mean survival months were longer for 27 of 33 TCGA cancer types.ConclusionsDifferences exist in the characteristics of TCGA vs. general population cancer cases. Our study highlights population subgroups where increased sample collection is warranted to increase the applicability of cancer genomic research results to all individuals.

Elucidating therapeutic molecular targets in premenopausal Asian women with recurrent breast cancers

Breast cancer is an increasing problem in Asia, with a higher proportion of premenopausal patients who are at higher risk of recurrence. Targeted sequencing was performed on DNA extracted from primary tumor specimens of 63 premenopausal Asian patients who relapsed after initial diagnosis of non-metastatic breast cancer. The most prevalent alterations included: TP53 (65%); PIK3CA (32%); GATA3 (29%); ERBB2 (27%); MYC (25%); KMT2C (21%); MCL1 (17%); PRKDC, TPR, BRIP1 (14%); MDM4, PCDH15, PRKAR1A, CDKN1B (13%); CCND1, KMT2D, STK11, and MLH1 (11%). Sixty of the 63 patients (95%) had at least one genetic alteration in a signaling pathway related to cell cycle or p53 signaling. The presence of MCL1 amplification, HIF-1-alpha transcription factor network pathway alterations, and direct p53 effectors pathway alterations were independent predictors of inferior overall survival from initial diagnosis. Comparison with non-Asian premenopausal tumors in The Cancer Genome Atlas (TCGA) revealed a higher prevalence of TP53 mutations among HER2-positive cancers, and more frequent TP53, TET2, and CDK12 mutations among hormone receptor-positive HER2-negative cancers in our cohort. Given the limited number of non-Asian premenopausal breast cancers that had relapsed in TCGA, we compared the frequency of mutations in our cohort with 43 premenopausal specimens from both TCGA and International Cancer Genome Consortium that had relapsed. There was a trend toward higher prevalence of TP53 mutations in our cohort. Certain genomic aberrations may be enriched in tumors of poor-prognosis premenopausal Asian breast cancers. The development of novel therapies targeting these aberrations merit further research.

489 (PB-122) - Elucidating therapeutic molecular targets in premenopausal Asian women with recurrent breast cancers

Breast cancer is an increasing problem in Asia, with a higher proportion of premenopausal patients who are at higher risk of recurrence. Targeted sequencing was performed on DNA extracted from primary tumor specimens of 63 premenopausal Asian patients who relapsed after initial diagnosis of non-metastatic breast cancer. The most prevalent alterations included: TP53 (65%); PIK3CA (32%); GATA3 (29%); ERBB2 (27%); MYC (25%); KMT2C (21%); MCL1 (17%); PRKDC, TPR, BRIP1 (14%); MDM4, PCDH15, PRKAR1A, CDKN1B (13%); CCND1, KMT2D, STK11, and MLH1 (11%). Sixty of the 63 patients (95%) had at least one genetic alteration in a signaling pathway related to cell cycle or p53 signaling. The presence of MCL1 amplification, HIF-1-alpha transcription factor network pathway alterations, and direct p53 effectors pathway alterations were independent predictors of inferior overall survival from initial diagnosis. Comparison with non-Asian premenopausal tumors in The Cancer Genome Atlas (TCGA) revealed a higher prevalence of TP53 mutations among HER2-positive cancers, and more frequent TP53, TET2, and CDK12 mutations among hormone receptor-positive HER2-negative cancers in our cohort. Given the limited number of non-Asian premenopausal breast cancers that had relapsed in TCGA, we compared the frequency of mutations in our cohort with 43 premenopausal specimens from both TCGA and International Cancer Genome Consortium that had relapsed. There was a trend toward higher prevalence of TP53 mutations in our cohort. Certain genomic aberrations may be enriched in tumors of poor-prognosis premenopausal Asian breast cancers. The development of novel therapies targeting these aberrations merit further research.

BRAFwild papillary thyroid carcinoma has two distinct mRNA expression patterns with different clinical behaviors.

Using a large set of genomic data from The Cancer Genome Atlas (TCGA), we classified BRAFwild papillary thyroid carcinomas (PTCs) into 2 subtypes with distinct molecular patterns and different clinical behaviors. We also suggested gene signatures (RAS-score) to predict molecular subtypes and clinical behaviors of BRAFwild PTC. Integrated genomic analysis was done using all genomic data of PTC in TCGA data portal (https://tcga-data.nci.nih.gov) and cancer browser (https://genome-cancer.ucsc.edu). Using Gene Ontology and a logistic regression test, we selected gene signatures (RAS-score) and applied this prediction model to the validation cohort (GSE60542). When we performed multiplatform genomic analysis, BRAFwild PTCs were divided into 2 molecular subtypes. Each subtype showed distinct molecular patterns and clinical behaviors. Gene signatures successfully predicted molecular subtype in another validation cohort. We found that BRAFwild PTCs were divided into 2 molecular subtypes and each subtype showed distinct molecular patterns, different activated pathways, and different clinical behaviors.

Epigenetic silencing of lncRNA MORT in 16 TCGA cancer types

We have previously described a hominid-specific long non-coding RNA, MORT (also known as ZNF667-AS1, Gene ID: 100128252), which is expressed in all normal cell types, but epigenetically silenced during cancer-associated immortalization of human mammary epithelial cells. Initial analysis of The Cancer Genome Atlas (TCGA) showed that 15 of 17 cancer types, which represent the 10 most common cancers in women and men, display DNA methylation associated MORT silencing in a large fraction of their tumors. In this study we analyzed MORT expression and DNA methylation state in the remaining 16 TCGA cancer types not previously reported. Seven of the 16 cancer types showed DNA methylation linked MORT silencing in a large fraction of their tumors. These are carcinomas (cervical cancer, and cancers of esophagus, stomach, and bile duct), and the non-epithelial tumors mesothelioma, sarcoma, and uterine carcinosarcoma. Together with the findings from our previous report, MORT expression is silenced by aberrant DNA methylation in 22 of 33 of TCGA cancer types. These 22 cancers include most carcinoma types, blood derived cancers and sarcomas. In conclusion, results suggest that the MORT gene is one of the most common epigenetic aberrations seen in human cancer. Coupled with the timing of MORT gene silencing during in vitro epithelial cell immortalization and its occurrence early in the temporal arc of human carcinogenesis, this provides strong circumstantial evidence for a tumor suppressor role for MORT.

Epigenetic silencing of lncRNA MORT in 16 TCGA cancer types.

We have previously described a hominid-specific long non-coding RNA, MORT (also known as ZNF667-AS1, Gene ID: 100128252), which is expressed in all normal cell types, but epigenetically silenced during cancer-associated immortalization of human mammary epithelial cells. Initial analysis of The Cancer Genome Atlas (TCGA) showed that 15 of 17 cancer types, which represent the 10 most common cancers in women and men, display DNA methylation associated MORT silencing in a large fraction of their tumors. In this study we analyzed MORT expression and DNA methylation state in the remaining 16 TCGA cancer types not previously reported. Seven of the 16 cancer types showed DNA methylation linked MORT silencing in a large fraction of their tumors. These are carcinomas (cervical cancer, and cancers of esophagus, stomach, and bile duct), and the non-epithelial tumors mesothelioma, sarcoma, and uterine carcinosarcoma. Together with the findings from our previous report, MORT expression is silenced by aberrant DNA methylation in 22 of 33 of TCGA cancer types. These 22 cancers include most carcinoma types, blood derived cancers and sarcomas. In conclusion, results suggest that the MORT gene is one of the most common epigenetic aberrations seen in human cancer. Coupled with the timing of MORT gene silencing during in vitro epithelial cell immortalization and its occurrence early in the temporal arc of human carcinogenesis, this provides strong circumstantial evidence for a tumor suppressor role for MORT.

Abstract 11: APOBEC is a major source of mutations in head and neck squamous cell carcinoma

Abstract Background: Cancers arise from an accumulation of somatic mutations. Numerous intrinsic and extrinsic mutagenic sources exist, yet our understanding of which processes are active in a given cancer genome is limited. Specific mutagenic processes have mutational preferences, called signatures. Large-scale analyses of cancer genomes/exomes from The Cancer Genome Atlas (TCGA) have identified 30 unique signatures. APOBEC is a family of cytosine deaminases which function in the immune response to viral infection. A signature for APOBEC3 activity has been identified in a subset of cancers, including head and neck squamous cell carcinoma (HNSCC). The specifics of APOBEC activity in HNSCC remain poorly understood. Methods: We utilized bioinformatic approaches to explore the role of APOBEC mutations in tumor exomes, transcriptomes and germline exomes from 511 HNSCC patients in TCGA. Results: 58% of HNSCC are statically enriched for the APOBEC3 signature compared to all other cancers in TCGA. HPV+ HNSCC are statistically enriched for the APOBEC3 signature compared to HPV- samples (p < 0.0013). MutSig was used to test the prevalence of APOBEC3 mutations in known driver genes compared to passengers. While the mean rate of APOBEC3 mutations was similar between MuSig genes and all other genes (19.4% vs 21.9%), PIK3CA demonstrated a 20-fold increase above all other genes. When evaluated by HPV status, 87% of PIK3CA mutations in HPV+ samples were APOBEC3 mutations compared to 55% of mutations in HPV- samples (p <0.025). All APOBEC3 mutations in HPV + samples occurred at two known hotspots, while HPV- mutations were distributed throughout the gene, strongly arguing against a random occurrence in HPV+ samples. Isoform analysis revealed a slight predominance of APOBEC3A mutations compared to other APOBEC3 isoforms. RNA-Seq analysis of 511 transcriptomes showed a weak correlation between APOBEC3 mutation rates and APOBEC3 expression. RNA-Seq Pathway Analysis demonstrated an overrepresentation of immune pathway dysregulation in APOBEC3 mutation rich samples. Conclusions: APOBEC3 is a major source of mutations in HNSCC. HPV+ HNSCC have the highest enrichment of APOBEC3 mutations. APOBEC3 mutations are strongly overrepresented in PIK3CA at two known hotspots, suggesting these mutations are of functional importance. APOBEC3A is the most likely APOBEC to be active in HPV+ HNSCC; however, expression levels of APOBEC isoforms do not correlate well with APOBEC mutational rates, suggesting a temporal dissonance between enzyme activity and tumor sample harvest. APOBEC3 activity appears to correlate with upregulation of immune signaling pathways, supporting the hypothesis that APOBEC is acting as part of the innate response to viral infection, although its activity in HPV- cancer suggests other mechanisms of upregulation as well. Citation Format: Daniel L. Faden, Sean Thomas, Jeffrey Myers, Nishant Agrawal, Joseph DeRisi. APOBEC is a major source of mutations in head and neck squamous cell carcinoma [abstract]. In: Proceedings of the AACR-AHNS Head and Neck Cancer Conference: Optimizing Survival and Quality of Life through Basic, Clinical, and Translational Research; April 23-25, 2017; San Diego, CA. Philadelphia (PA): AACR; Clin Cancer Res 2017;23(23_Suppl):Abstract nr 11.

Cistrome Cancer: A Web Resource for Integrative Gene Regulation Modeling in Cancer.

Cancer results from a breakdown of normal gene expression control, so the study of gene regulation is critical to cancer research. To gain insight into the transcriptional and epigenetic factors regulating abnormal gene expression patterns in cancers, we developed the Cistrome Cancer web resource (http://cistrome.org/CistromeCancer/). We conducted the systematic integration and modeling of over 10,000 tumor molecular profiles from The Cancer Genome Atlas (TCGA) with over 23,000 ChIP-seq and chromatin accessibility profiles from our Cistrome collection. The results include reconstruction of functional enhancer profiles, "super-enhancer" target genes, as well as predictions of active transcription factors and their target genes for each TCGA cancer type. Cistrome Cancer reveals novel insights from integrative analyses combining chromatin profiles with tumor molecular profiles and will be a useful resource to the cancer gene regulation community. Cancer Res; 77(21); e19-22. ©2017 AACR.

High Expression of FGD3, a Putative Regulator of Cell Morphology and Motility, Is Prognostic of Favorable Outcome in Multiple Cancers.

PurposeIdentification of single-gene biomarkers that are prognostic of outcome can shed new insights on the molecular mechanisms that drive breast cancer and other cancers.MethodsExploratory analysis of 20,464 single-gene messenger RNAs (mRNAs) in the Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) discovery cohort indicates that low expression of FGD3 mRNA is prognostic for poor outcome. Prognostic significance of faciogenital dysplasia 3 (FGD3), SUSD3, and other single-gene proliferation markers was evaluated in breast cancer and The Cancer Genome Atlas (TCGA) cohorts.ResultsA meta-analysis of Cox regression of FGD3 mRNA as a continuous variable for overall survival of estrogen receptor (ER)–positive samples in METABRIC discovery, METABRIC validation, TCGA breast cancer, and Combination Chemotherapy in Treating Women With Breast Cancer (E2197) cohorts resulted in a combined hazard ratio (HR) of 0.69 (95% CI, 0.63 to 0.75), indicating better outcome with high expression. In the ER-negative samples, the combined meta-analysis HR was 0.72 (95% CI, 0.63 to 0.82), suggesting that FGD3 is prognostic regardless of ER status. The potential of FGD3 as a biomarker for freedom from recurrence was evaluated in the Breast International Group 1-98 (BIG 1-98; Letrozole or Tamoxifen in Treating Postmenopausal Women With Breast Cancer) study (HR, 0.85; 95% CI, 0.76 to 0.93) for breast cancer–free interval. In the Hungarian Academy of Science (HAS) breast cancer cohort, splitting on the median had an HR of 0.49 (95% CI, 0.42 to 0.58) for recurrence-free survival. A comparison of the Stouffer P value in five ER-positive cohorts showed that FGD3 (P = 3.8E-14) outperformed MKI67 (P = 1.06E-8) and AURKA (P = 2.61E-5). A comparison of the Stouffer P value in four ER-negative cohorts showed that FGD3 (P = 3.88E-5) outperformed MKI67 (P = .477) and AURKA (P = .820).ConclusionFGD3 was previously shown to inhibit cell migration. FGD3 mRNA is regulated by ESR1 and is associated with favorable outcome in six distinct breast cancer cohorts and four TCGA cancer cohorts. This suggests that FGD3 is an important clinical biomarker.

MHC class II associated stomach cancer mutations correlate with lack of subsequent tumor development.

The role of tumor cell expression of major histocompatibility class II (MHCII) has been controversial, with evidence indicating that tumor cell expression of MHCII may lead to an anti-tumor immune response and to tumor cell apoptosis and that MHCII has pro-tumorigenic functions. The cancer genome atlas (TCGA) indicates numerous deleterious mutations for the highly specific, MHCII transcriptional activation proteins, RFX5, RFXAP, RFXANK and CIITA. Also, mutations in the non-polymorphic, human leukocyte antigen (HLA)-DRA gene, which encodes the heavy chain for the most prominent human MHCII molecule, HLA-DR, are common. For many, if not most TCGA cancer datasets, the MHCII specific mutations do not associate with clinical outcomes. However, stomach carcinoma represents an exception, where the data indicate that MHCII-specific mutations are associated with a more favorable outcome. These data raise the question of whether stomach cancer mutations represent effective haploinsufficiency or whether mutations that are associated with a favorable outcome occur with other stomach cancer molecular features that limit the function of the two alleles that represent these MHCII-related proteins.

Abstract 5438: Knowledge of which miRNA isoforms are expressed in a sample can successfully discriminate amongst the 32 TCGA cancer types

Abstract Background: Next-generation sequencing (NGS) technologies made possible the generation of large amounts of data quickly and inexpensively process. As a result, the challenge nowadays is how best to manipulate and analyze Big Data. Among non-coding RNAs, microRNAs (miRNAs) are the best studied to date. MiRNAs have attracted a lot of attention because they are centrally involved in homeostasis and can regulate thousands of coding and non-coding transcripts. The complexity of miRNA-mediated regulation increased further with the recent discovery that miRNA isoforms (isomiRs) are produced constitutively and that isomiR expression depends on a person’s sex, population origin, race and also on tissue type, tissue state, and disease subtype. Methods: We computed and analyzed isomiR profiles across 10,271 short RNA-seq datasets from The Cancer Genome Atlas (TCGA) repository, which represent 32 cancer types. Focusing on the isomiRs of each dataset, we built a classifier that relied solely on “binarized” isomiR profiles: in a dataset, an isomiR was labeled ‘present,’ if its expression exceeded a threshold that depended on the depth of sequencing, and ‘absent’ otherwise. This differs radically from previous methods in that it attempts to classify by relying on the presence or absence of isomiRs, intentionally discarding all information about the isomiRs’ expression level. We also built a classifier that relied solely on “binarized” miRNA arm profiles: a miRNA arm was labeled ‘present’ if at least one of the isomiRs it produces is ‘present’ in the same dataset. Results: The classifier we built using binarized isomiR profiles discriminated among all 32 TCGA cancer types and classified datasets with sensitivity >90% and a false discovery rate (FDR) < 3.0%. The ability of this classifier to correctly distinguish among the 32 cancers persisted even after a 15x reduction in the number of isomiRs used. The classifier we built using binarized miRNA arm profiles could also discriminate among all 32 TCGA cancers albeit with a lower sensitivity (83%) and a slightly higher FDR (6%). We also examined how well the various miRNA loci have been studied over time by measuring the corresponding number of published articles. Surprisingly, the miRNAs whose isoforms have the highest ability to classify tumors are not those with the most articles in PubMed. Conclusions: The presence or absence of a specific isomiR provides adequate discriminatory power to accurately classify a sample as belonging to one of the 32 TCGA cancers. The findings suggest that isomiRs can be used as potent, novel, cancer-specific biomarkers. By ignoring the expression levels of isomiRs, we effectively show that the same miRNA produces different clouds of isoforms in each cancer type. And, since distinct isomiRs target distinct mRNAs, it follows that a miRNA locus will target different mRNAs in different cancers simply by expressing different clouds of isoforms each time. Citation Format: Aristeidis G. Telonis, Rogan Magee, Phillipe Loher, Inna Chervoneva, Eric Londin, Isidore Rigoutsos. Knowledge of which miRNA isoforms are expressed in a sample can successfully discriminate amongst the 32 TCGA cancer types [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 5438. doi:10.1158/1538-7445.AM2017-5438