The Cancer Genome Atlas Cancer Research Articles

A pan-cancer analysis of the oncogenic role of Keratin 17 (KRT17) in human tumors.

BackgroundAlthough new evidence from cells or animals suggests a relationship between Keratin 17 (KRT17) and cancer, no pan-cancer analysis is currently available.MethodsThe expression level of KRT17 in generalized carcinoma was detected by the Tumor Immune Estimation Resource, version 2 (TIMER2) database, and then verified the protein expression of KRT17 in different cancer species in UALCAN database, and analyzed the relationship between the expression level of KRT17 and the clinical stage and survival of different cancers. We further explored the genetic variation of KRT17 in different tumor types included in The Cancer Genome Atlas (TCGA) and the specific mutations in each domain. The changes of KRT17 protein phosphorylation levels and protein expression levels at different phosphorylation sites in different tumors were explored. TIMER2 database was used to explore the potential relationship between the infiltration level of different immune cells and KRT17 gene expression in different TCGA cancer types. Finally, the protein binding to KRT17 and genes related to KRT17 expression were explored by STRING database and TCGA database.ResultsKRT17 is overexpressed in most malignancies, and we observed a distinct relationship between KRT17 expression and tumor patient prognosis. Enhanced phosphorylation levels of S13, S24, S32, and S39 were observed in several tumors, such as lung adenocarcinoma (LUAD), colon and ovarian cancers, and uterine corpus endometrial carcinoma (UCEC). Intermediate filament cytoskeleton and keratinization may be simultaneously acting with KRT17 on tumor pathogenesis.ConclusionsOur pan-cancer analysis provides relatively complete information on the oncogenic functions of KRT17 in various cancers.

Chromoanagenesis Landscape in 10,000 TCGA Patients.

Simple SummaryChromoanagenesis is a single catastrophic event in which one or few chromosomes are shattered and disorderly reassembled. Chromoanagenesis is common in many types of cancers. In this study, we utilize data from The Pan-Cancer Analysis of Whole Genome (PCAWG) to build a machine learning algorithm that detects chromoanagenesis with high accuracy. We applied the algorithm on ~10,000 samples from The Cancer Genome Atlas (TCGA), thereby providing, for the first time, chromoanagenesis status labels for the complete data set. An in-depth analysis of somatic and clinical chromoanagenesis features is presented for 20 cancer types. Mutual exclusivity patterns between genes impaired in chromoanagenesis versus non-chromoanagenesis cases might imply at distinct pathways involved in chromoanagenesis-driven tumorigenesis.During the past decade, whole-genome sequencing of tumor biopsies and individuals with congenital disorders highlighted the phenomenon of chromoanagenesis, a single chaotic event of chromosomal rearrangement. Chromoanagenesis was shown to be frequent in many types of cancers, to occur in early stages of cancer development, and significantly impact the tumor’s nature. However, an in-depth, cancer-type dependent analysis has been somewhat incomplete due to the shortage in whole genome sequencing of cancerous samples. In this study, we extracted data from The Pan-Cancer Analysis of Whole Genome (PCAWG) and The Cancer Genome Atlas (TCGA) to construct and test a machine learning algorithm that can detect chromoanagenesis with high accuracy (86%). The algorithm was applied to ~10,000 unlabeled TCGA cancer patients. We utilize the chromoanagenesis assignment results, to analyze cancer-type specific chromoanagenesis characteristics in 20 TCGA cancer types. Our results unveil prominent genes affected in either chromoanagenesis or non-chromoanagenesis tumorigenesis. The analysis reveals a mutual exclusivity relationship between the genes impaired in chromoanagenesis versus non-chromoanagenesis cases. We offer the discovered characteristics as possible targets for cancer diagnostic and therapeutic purposes.

Machine learning analysis of TCGA cancer data.

In recent years, machine learning (ML) researchers have changed their focus towards biological problems that are difficult to analyse with standard approaches. Large initiatives such as The Cancer Genome Atlas (TCGA) have allowed the use of omic data for the training of these algorithms. In order to study the state of the art, this review is provided to cover the main works that have used ML with TCGA data. Firstly, the principal discoveries made by the TCGA consortium are presented. Once these bases have been established, we begin with the main objective of this study, the identification and discussion of those works that have used the TCGA data for the training of different ML approaches. After a review of more than 100 different papers, it has been possible to make a classification according to following three pillars: the type of tumour, the type of algorithm and the predicted biological problem. One of the conclusions drawn in this work shows a high density of studies based on two major algorithms: Random Forest and Support Vector Machines. We also observe the rise in the use of deep artificial neural networks. It is worth emphasizing, the increase of integrative models of multi-omic data analysis. The different biological conditions are a consequence of molecular homeostasis, driven by both protein coding regions, regulatory elements and the surrounding environment. It is notable that a large number of works make use of genetic expression data, which has been found to be the preferred method by researchers when training the different models. The biological problems addressed have been classified into five types: prognosis prediction, tumour subtypes, microsatellite instability (MSI), immunological aspects and certain pathways of interest. A clear trend was detected in the prediction of these conditions according to the type of tumour. That is the reason for which a greater number of works have focused on the BRCA cohort, while specific works for survival, for example, were centred on the GBM cohort, due to its large number of events. Throughout this review, it will be possible to go in depth into the works and the methodologies used to study TCGA cancer data. Finally, it is intended that this work will serve as a basis for future research in this field of study.

Abstract 2408: Targeted bioinformatics analysis identifies high ABAT transcript expression as a multi-cancer marker of favorable patient prognosis

Abstract Unsupervised analysis of γ-aminobutyric acid (GABA) system genes has identified some genes as correlated with indicators of severity in several cancers. Central to GABA system function is the GABA shunt, a series of reactions bypassing two steps of the tricarboxylic acid (TCA) cycle, which encompasses the synthesis and metabolism of the non-proteinaceous amino acid GABA. Here, we have undertaken a targeted bioinformatics study utilizing publicly available databases - including cBioPortal, UALCAN, the Human Protein Atlas, and the Human Cancer Metastasis Database - to evaluate the expression of transcripts encoding GABA shunt enzymes, and their correlation with clinical outcomes, across nearly three dozen cancers from The Cancer Genome Atlas (TCGA) Pan-Cancer Atlas initiative dataset. Of the three genes encoding enzymes of the GABA shunt, we have focused on ABAT, which encodes the GABA-catabolizing enzyme γ-aminobutyrate aminotransferase (ABAT; aka GABA-transaminase or GABA-T). ABAT is widely expressed across all 32 TCGA studies, ranging from moderate to high expression in skin cutaneous melanoma and low-grade glioma (average ± SEM = 6.5 ± 0.08; 13.3 ± 0.03 log2 RSEM), respectively. For the majority of these tissues, the expression of ABAT in tumor is lower than its expression in adjacent normal tissue. Importantly, ABAT expression is prognostic in 11 cancers: upregulation is favorably prognostic for ACC, BLCA, BRCA, and PAAD, while downregulation is unfavorably prognostic for KIRC, KIRP, LGG, LIHC, LUAD, MESO, and SKCM (Kaplan-Meier survival estimates, z-score = 0.5, p < 0.05). A subset of these cancers, as well as 4 other TCGA cancers, show a decrease in ABAT transcript expression with increasing tumor stage (p < 0.05). Furthermore, 7 cancers express lower levels of ABAT transcripts in the primary tumors of patients with metastases vs the primary tumors of patients without metastases. Additionally, metastases of 4 cancers express elevated levels of ABAT transcripts vs their respective primary tumors. These results imply a dynamic regulation of ABAT expression throughout cancer progression. As very few studies have directly addressed the role of ABAT in cancer progression, the identification of ABAT transcript expression as a broadly applicable multi-cancer prognostic marker may ultimately benefit a wide variety of patients. We would like to thank the Feist-Weiller Cancer Center and LSU Health, Shreveport for supporting this work. Citation Format: Erika L. Knott, Nancy J. Leidenheimer. Targeted bioinformatics analysis identifies high ABAT transcript expression as a multi-cancer marker of favorable patient prognosis [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 2408.

Abstract 2255: The impact of primary immunodeficiency informed molecular landscape on the biology and prognosis of refractory malignancies

Abstract Introduction: Cancer is one of the most common causes of death in patients with primary immunodeficiencies (PIDs). The immune surveillance concept postulates that PIDs are an important contributor to cancer growth and outcomes. Although several immunodeficiency syndromes are known to be associated with malignancies in children and adults, currently, the molecular mechanisms that link immune functions to cancers are poorly understood. Here we describe distinct molecular characteristics in adult cancers with respect to PID-associated genes that impact survival outcomes in affected patients. Methods: In this study, we integrated transcriptome data of 28 adult cancers from the public database The Cancer Genome Atlas (TCGA) with respective healthy tissues as control from the Genotype-Tissue Expression (GTEx) project. Unified datasets integrating GTEx healthy samples and TCGA cancer data were provided by the Recount2 protocol and differential expression analyses (DEA) were carried out using the integrated limma-voom and edgeR pipelines in the TCGAbiolinks R package. Gene ontology (GO) enrichment analyses were performed using the pathfindR R package. Mutational and clinical data of adult cancer patient samples in TCGA and the AACR Project GENIE were consolidated using cBioPortal and the maftools R package. PID-associated genes were curated from the latest registry by the European Society for Immunodeficiencies (ESID). Results: Of 307 PID-associated genes, 151 (32.0%) up-regulated and 88 (18.6%) down-regulated genes were identified across the 28 unified TCGA-GTEx datasets. From the identified differentially expressed gene sets, GO analysis revealed enrichment in immunological pathways related to complement and coagulation cascades (27/28 datasets), systemic lupus erythematosus (25/28), in addition to Fanconi anemia (22/28). The greatest frequency of mutations in PID-associated genes from the consolidated TCGA and GENIE datasets were KRAS (14.83%), KMT2D (10.03%), TERT (8.99%), PRKDC (7.59%) and PTEN (6.18%) across all cancer types, with a total of 259 affected PID-associated genes. Survival analyses using Cox proportional-hazards and Kaplan-Meier models will assess the clinical implications of the identified molecular alterations in affected patients. Conclusion: Our integrative approach aids in the elucidation of molecular mechanisms that bridge PIDs to tumour biology. These data may help identify important biomarkers and potential future targeted therapeutics. Citation Format: Son Tran, Luis Murguia-Favela, Aru Narendran. The impact of primary immunodeficiency informed molecular landscape on the biology and prognosis of refractory malignancies [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 2255.

Artificial intelligence grading of breast cancer: a promising method to refine prognostic classification for management precision.

Artificial intelligence (AI)-based breast cancer grading may help to overcome perceived limitations of human assessment. Here, the potential value of AI grade was evaluated at the molecular level and in predicting patient outcome. A supervised convolutional neural network (CNN) model was trained on images of 612 breast cancers from The Cancer Genome Atlas (TCGA). The test set, obtained from the Cooperative Human Tissue Network (CHTN), comprised 1058 cancers with corresponding survival data. Upon reversal, a CNN was trained from images of 1537 CHTN cancers and tested on 397 TCGA cancers. In TCGA, mRNA models were trained using AI grade and Nottingham grade (NG) as labels. Performance of mRNA models in predicting patient outcome was evaluated using data from 1807 cancers from the Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) cohort. In selecting images for training, nucleolar prominence determined high- versus low-grade cancer cells. In CHTN, NG corresponded to significant survival stratification in stages 1, 2 and 3 cancers, while AI grade showed significance in stages 1 and 2 and borderline in stage 3 tumours. In METABRIC, the mRNA model trained from AI grade was not significantly different to the NG-based model. The gene which best described AI grade was TRIP13, a gene involved with mitotic spindle assembly. An AI grade trained from the morphologically distinctive feature of nucleolar prominence could transmit significant patient outcome information across three independent patient cohorts. AI grade shows promise in gene discovery and for second opinions.

Cytolytic activity score as a biomarker for antitumor immunity and clinical outcome in patients with gastric cancer.

BackgroundA simple measure of immune cytolytic activity (CYT) base on mRNA expression levels of two genes, GZMA and PRF1, was recently reported. Here, we aimed to evaluate the CYT score's potential as a measure of antitumor immunity and predictor of clinical outcome in gastric cancer (GC) patients.Materials and MethodsWe evaluated the correlations between tumor‐infiltrating immune cells and the CYT score in 238 GC samples from The Cancer Genome Atlas (TCGA). Next, we investigated CYT score associations with molecular subtypes, somatic mutation load, and immune checkpoint molecules in GC samples from TCGA and Asian Cancer Research Group (ACRG). Moreover, we evaluated the clinical significance of the CYT score calculated by reverse transcription (RT)‐quantitative PCR (qPCR) data in 123 GC samples and the association of the CYT score with the response to anti‐PD‐1 therapy in 7 GC samples from Kyushu University Hospital.ResultsThe CYT score positively correlated with the proportions of tumor‐infiltrating CD8+ T cells and macrophages and negatively correlated with the proportion of regulatory T cells in GC tissues. A high CYT score was associated with common immune checkpoint molecules, a high mutation, the Epstein–Barr virus subtype, and the microsatellite instability subtype in GC. Moreover, a low CYT score was a poor prognosis factor in patients with GC. Finally, the CYT score was higher in a responder to anti‐PD‐1 therapy compared to nonresponders.ConclusionThe CYT score reflects antitumor immunity and predicts clinical outcome in GC patients.

Pan-Cancer Genome-Wide DNA Methylation Analyses Revealed That Hypermethylation Influences 3D Architecture and Gene Expression Dysregulation in HOXA Locus During Carcinogenesis of Cancers.

DNA methylation dysregulation during carcinogenesis has been widely discussed in recent years. However, the pan-cancer DNA methylation biomarkers and corresponding biological mechanisms were seldom investigated. We identified differentially methylated sites and regions from 5,056 The Cancer Genome Atlas (TCGA) samples across 10 cancer types and then validated the findings using 48 manually annotated datasets consisting of 3,394 samples across nine cancer types from Gene Expression Omnibus (GEO). All samples’ DNA methylation profile was evaluated with Illumina 450K microarray to narrow down the batch effect. Nine regions were identified as commonly differentially methylated regions across cancers in TCGA and GEO cohorts. Among these regions, a DNA fragment consisting of ∼1,400 bp detected inside the HOXA locus instead of the boundary may relate to the co-expression attenuation of genes inside the locus during carcinogenesis. We further analyzed the 3D DNA interaction profile by the publicly accessible Hi-C database. Consistently, the HOXA locus in normal cell lines compromised isolated topological domains while merging to the domain nearby in cancer cell lines. In conclusion, the dysregulation of the HOXA locus provides a novel insight into pan-cancer carcinogenesis.

Predicting DNA methylation from genetic data lacking racial diversity using shared classified random effects

Public genomic repositories are notoriously lacking in racially and ethnically diverse samples. This limits the reaches of exploration and has in fact been one of the driving factors for the initiation of the All of Us project. Our particular focus here is to provide a model-based framework for accurately predicting DNA methylation from genetic data using racially sparse public repository data. Epigenetic alterations are of great interest in cancer research but public repository data is limited in the information it provides. However, genetic data is more plentiful. Our phenotype of interest is cervical cancer in The Cancer Genome Atlas (TCGA) repository. Being able to generate such predictions would nicely complement other work that has generated gene-level predictions of gene expression for normal samples.We develop a new prediction approach which uses shared random effects from a nested error mixed effects regression model. The sharing of random effects allows borrowing of strength across racial groups greatly improving predictive accuracy. Additionally, we show how to further borrow strength by combining data from different cancers in TCGA even though the focus of our predictions is DNA methylation in cervical cancer. We compare our methodology against other popular approaches including the elastic net shrinkage estimator and random forest prediction. Results are very encouraging with the shared classified random effects approach uniformly producing more accurate predictions – overall and for each racial group.

Cancer prognosis prediction using somatic point mutation and copy number variation data: a comparison of gene-level and pathway-based models

BackgroundGenomic profiling of solid human tumors by projects such as The Cancer Genome Atlas (TCGA) has provided important information regarding the somatic alterations that drive cancer progression and patient survival. Although researchers have successfully leveraged TCGA data to build prognostic models, most efforts have focused on specific cancer types and a targeted set of gene-level predictors. Less is known about the prognostic ability of pathway-level variables in a pan-cancer setting. To address these limitations, we systematically evaluated and compared the prognostic ability of somatic point mutation (SPM) and copy number variation (CNV) data, gene-level and pathway-level models for a diverse set of TCGA cancer types and predictive modeling approaches.ResultsWe evaluated gene-level and pathway-level penalized Cox proportional hazards models using SPM and CNV data for 29 different TCGA cohorts. We measured predictive accuracy as the concordance index for predicting survival outcomes. Our comprehensive analysis suggests that the use of pathway-level predictors did not offer superior predictive power relative to gene-level models for all cancer types but had the advantages of robustness and parsimony. We identified a set of cohorts for which somatic alterations could not predict prognosis, and a unique cohort LGG, for which SPM data was more predictive than CNV data and the predictive accuracy is good for all model types. We found that the pathway-level predictors provide superior interpretative value and that there is often a serious collinearity issue for the gene-level models while pathway-level models avoided this issue.ConclusionOur comprehensive analysis suggests that when using somatic alterations data for cancer prognosis prediction, pathway-level models are more interpretable, stable and parsimonious compared to gene-level models. Pathway-level models also avoid the issue of collinearity, which can be serious for gene-level somatic alterations. The prognostic power of somatic alterations is highly variable across different cancer types and we have identified a set of cohorts for which somatic alterations could not predict prognosis. In general, CNV data predicts prognosis better than SPM data with the exception of the LGG cohort.

A hypoxia-related signature for clinically predicting diagnosis, prognosis and immune microenvironment of hepatocellular carcinoma patients

BackgroundHypoxia plays an indispensable role in the development of hepatocellular carcinoma (HCC). However, there are few studies on the application of hypoxia molecules in the prognosis predicting of HCC. We aim to identify the hypoxia-related genes in HCC and construct reliable models for diagnosis, prognosis and recurrence of HCC patients as well as exploring the potential mechanism.MethodsDifferentially expressed genes (DEGs) analysis was performed using The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) database and four clusters were determined by a consistent clustering analysis. Three DEGs closely related to overall survival (OS) were identified using Cox regression and LASSO analysis. Then the hypoxia-related signature was developed and validated in TCGA and International Cancer Genome Consortium (ICGC) database. The Gene Set Enrichment Analysis (GSEA) was performed to explore signaling pathways regulated by the signature. CIBERSORT was used for estimating the fractions of immune cell types.ResultsA total of 397 hypoxia-related DEGs in HCC were detected and three genes (PDSS1, CDCA8 and SLC7A11) among them were selected to construct a prognosis, recurrence and diagnosis model. Then patients were divided into high- and low-risk groups. Our hypoxia-related signature was significantly associated with worse prognosis and higher recurrence rate. The diagnostic model also accurately distinguished HCC from normal samples and nodules. Furthermore, the hypoxia-related signature could positively regulate immune response. Meanwhile, the high-risk group had higher fractions of macrophages, B memory cells and follicle-helper T cells, and exhibited higher expression of immunocheckpoints such as PD1and PDL1.ConclusionsAltogether, our study showed that hypoxia-related signature is a potential biomarker for diagnosis, prognosis and recurrence of HCC, and it provided an immunological perspective for developing personalized therapies.

Identification of MicroRNA-Related Tumorigenesis Variants and Genes in the Cancer Genome Atlas (TCGA) Data.

MicroRNAs (miRNAs) are a class of small non-coding RNA that can down-regulate their targets by selectively binding to the 3′ untranslated region (3′UTR) of most messenger RNAs (mRNAs) in the human genome. Single nucleotide variants (SNVs) located in miRNA target sites (MTS) can disrupt the binding of targeting miRNAs. Anti-correlated miRNA–mRNA pairs between normal and tumor tissues obtained from The Cancer Genome Atlas (TCGA) can reveal important information behind these SNVs on MTS and their associated oncogenesis. In this study, using previously identified anti-correlated miRNA–mRNA pairs in 15 TCGA cancer types and publicly available variant annotation databases, namely dbNSFP (database for nonsynonymous SNPs’ functional predictions) and dbMTS (database of miRNA target site SNVs), we identified multiple functional variants and their gene products that could be associated with various types of cancers. We found two genes from dbMTS and 33 from dbNSFP that passed our stringent filtering criteria (e.g., pathogenicity). Specifically, from dbMTS, we identified 23 candidate genes, two of which (BMPR1A and XIAP) were associated with diseases that increased the risk of cancer in patients. From dbNSFP, we identified 65 variants located in 33 genes that were likely pathogenic and had a potential causative relationship with cancer. This study provides a novel way of utilizing TCGA data and integrating multiple publicly available databases to explore cancer genomics.

Abstract 6426: Novel mechanism of PIM2 inhibition leads to significant effect on solid tumors

Abstract Background: PIM2 is overexpressed in hematopoietic and solid cancers, and is involved in multiple signaling pathways. In hematologic malignancies, PIM2 has been shown to be a pro-survival factor and demonstrated to be a possible therapeutic target. However, the role of PIM2 as a therapeutic target in solid cancers has not been fully elucidated. Methods: PIM2 expression was compared between normal and cancer tissues in The Cancer Genome Atlas (TCGA) datasets. Pan-PIM inhibitor treatment, as well as PIM2 overexpression/knockdown technique was utilized to investigate the PIM2 role in solid cancers. Multiple cancer models were utilized for in vitro and in vivo experiments to investigate if PIM2 as a therapeutic target and examine mechanism of anti-tumor activity by JP11646. Results: PIM2 was overexpressed in several types of cancer tissues, including breast, esophageal, head-neck, kidney, liver and endometrial cancers in TCGA. PIM2 overexpression promoted tumor growth in in vivo breast cancer mouse model. PIM2 inhibition by siRNA as well as pan-PIM inhibitor JP11646 activated apoptosis as seen by increased cleaved PARP. JP11646 treatment inhibited cancer cell proliferation in a dose-dependent manner in vitro in various types of solid cancers, including head-neck, ovarian, breast, prostate, liver, pancreatic, colorectal and non-small cell lung cancers. Interestingly, treatment with pan-PIM inhibitor JP11646 resulted in selective downregulation of PIM2 protein expression, but not PIM1 or PIM3 in breast cancer cell lines, BT549 and MDA-MB-231. The treatment also resulted in downregulation of phosphorylated downstream targets, 4EBP1 and TSC2 in breast cancer cells. However, additional proteasome inhibitor treatment blocked PIM2 downregulation and apoptosis activation was not seen. The result suggests that proteasome complex is required for PIM2 degradation which is necessary to induce apoptosis. JP11646 treatment inhibited tumor growth in breast, liver, pancreas, and non-small cell lung cancer in vivo xenograft models and was compared to multiple standard of care treatments showing advantage in lung tumors. Conclusion: We found that PIM2 promoted cancer progression in solid tumors. We show novel mechanism of JP11646 targeting PIM2 for degradation and induction of apoptosis. We show that PIM2 may serve as a novel therapeutic target for multiple types of solid cancers with enhanced therapeutic effect through protein degradation rather than kinase inhibition alone. Citation Format: Eriko Katsuta, Kazuaki Takabe, Alex Adjei, Mateusz Opyrchal. Novel mechanism of PIM2 inhibition leads to significant effect on solid tumors [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 6426.

An Immunogram for an Individualized Assessment of the Antitumor Immune Response in Patients With Hepatocellular Carcinoma.

In clinical practice, the cancer-immunity cycle of an individual patient with hepatocellular carcinoma (HCC) must be described to support the clinical management of cancer. The present study explored the immunograms of patients with liver cancer based on liver RNA sequencing data to visually display the individualized cancer-immunity cycles. Two independent HCC cohorts [The Cancer Genome Atlas (TCGA) and Liver Cancer-RIKEN, Japan (LIRI-JP) HCC cohorts] with whole exome sequencing (WES) data, RNA sequencing data, and clinical data from TCGA and International Cancer Genome Consortium (ICGC) were enrolled in this study. This study constructed HCC immunograms of cancer immune cells to visually explore the anticancer immune responses of patients with HCC. The patterns of the HCC immunograms were categorized into two clusters: hot and cold HCC immunograms. Favorable overall survival (OS) and disease-free survival (DFS) were observed in the hot immunogram cluster in the TCGA cohort. The results for LIRI-JP cohort were similar to the TCGA cohort. The OS of patients with HCC presenting the hot immunogram was longer than patients with the cold immunogram in the LIRI-JP HCC cohort. Compared with cold immunograms, hot immunograms were characterized by higher levels of immune cell infiltration and stronger immune signatures, including cytolytic activity, IFN-γ signature, immunocostimulator, immunoinhibitor, chemokine, adhesion molecule, MHC I, MHC II, and non-class MHC levels. The main difference in molecular features between hot and cold immunograms was reflected in WNT-CTNNB1 alterations and copy number variant (CNV) and loss of heterozygosity (LOH) scores, which are the molecular features associated with resistance to immunotherapy and tumor escape. The immunogram patterns were distinct in terms of the different molecular features of HCC tumors. The HCC immunogram for the cancer-immune cycle was able to visualize the personalized antitumor immune response of patients with HCC, and the patterns of the HCC immunograms contributed to the clinical outcomes of patients, which may facilitate an individualized assessment of the antitumor immune response for optimal personalized immunotherapy.

SsGSEA score-based Ras dependency indexes derived from gene expression data reveal potential Ras addiction mechanisms with possible clinical implications

For nearly a decade, the difficulties associated with both the determination and reproducibility of Ras-dependency indexes (RDIs) have limited their application and further delineation of the biology underlying Ras dependency. In this report, we describe the application of a computational single sample gene set enrichment analysis (ssGSEA) method to derive RDIs with gene expression data. The computationally derived RDIs across the Cancer Cell Line Encyclopedia (CCLE) cell lines show excellent agreement with the experimentally derived values and high correlation with a previous in-house siRNA effector node (siREN) study and external studies. Using EMT signature-derived RDIs and data from cell lines representing the extremes in RAS dependency, we identified enriched pathways distinguishing these classes, including the Fas signaling pathway and a putative Ras-independent pathway first identified in NK cells. Importantly, extension of the method to patient samples from The Cancer Genome Atlas (TCGA) showed the same consensus differential expression patterns for these two pathways across multiple tissue types. Last, the computational RDIs displayed a significant association with TCGA cancer patients’ survival outcomes. Together, these lines of evidence confirm that our computationally derived RDIs faithfully represent a measure of Ras dependency in both cancer cell lines and patient samples. The application of such computational RDIs can provide insights into Ras biology and potential clinical applications.

144 Increased normal tissue telomere length is associated with decreased survival in melanoma patients

Normal tissue telomere length (nTL) varies within the population, affected by inherited genetics and environmental risk factors. Previous work has investigated whether telomere length interplays with overall cancer survival. We aim to determine whether nTL was predictive of melanoma patient survival. Both normal and tumor tissue telomere lengths from The Cancer Genome Atlas (TCGA) skin melanoma dataset, along with other cancers, were obtained from supplementary data of the Barthel et al. study (Nat Genet 49, 349–357 (2017)), which were estimated from whole genome/whole exome sequencing data. Univariate and multivariate Cox proportional hazard regression models were applied to identify variables predicting survival, adjusting for age, gender and stage. nTL from individuals with melanoma did not demonstrate a normal distribution. The majority of patients (56%) had short nTL, defined as lower than one-third of the range, while only 8% of patients demonstrated nTL in the intermediate range. Analyzed as a continuous variable, increased nTL was negatively correlated with both overall (OS) and progression free survival (PFS) for melanoma patients (p=0.005, p=0.042), even after adjusting for age, gender and stage. In contrast, tumor telomere length did not demonstrate an association with survival. When stratified into groups, those with longer than median nTL showed worse OS and PFS compared to those with shorter nTL (p<0.001, p=0.007). We then expanded this analysis to 30 other cancers in TCGA. When stratified into groups, longer nTL correlated with worse OS within bladder and colorectal cancers. However, unlike melanoma, no other cancer demonstrated a significant correlation between nTL and OS when nTL was treated as a continuous variable. In conclusion, normal tissue telomere length may represent a prognostic marker for melanoma survival.

Comprehensive analysis of advanced-stage solid tumors from TCGA reveal widespread variation of genomics evidence levels across cancer types.

e13547 Background: Improved scalability and affordability of next generation sequencing (NGS) has pivoted cancer care toward genomics-driven treatment decisions. Particularly in advanced-stage or refractory cancer, clinical insights gleaned from NGS have become an integral option as these patients have typically exhausted all lines of available therapy. As precision oncology evolves, NGS is expected to have a differential impact based on the cancer type. In this study, a comprehensive NGS panel was used to determine the strength of clinical evidence in various advanced stage tumor samples from The Cancer Genome Atlas (TCGA). Methods: A hybrid capture panel, Oncofoco, was developed to evaluate SNVs, INDELs, CNVs and TMB in 366 genes. The panel’s utility was validated by interrogating a broader cohort of 2847 TCGA samples (advanced tumors with T3 or T4; or N &gt; = 1; or M &gt; = 1). Watsonä for Genomics, an artificial intelligence offering, was used for variant interpretation and annotation of the 366 genes. A clinical evidence classification system that evaluated the strength of biomarker/drug response associations was used for annotation with level 1/R1 strongest and level 4 weakest from clinical literature, FDA drug labels and guidelines (PMID:28890946). Results: The highest level of evidence for the top nine frequently occurring advanced stage cancers in TCGA is shown in Table. Conclusions: Thyroid cancer and cutaneous melanoma have emerged as the cancer types with the most level 1 evidence (FDA approved drugs) owing to BRAF V600E mutations. Kidney and prostate cancers show no cases with level 1 evidence and also had the largest fraction of unactionable tumors. Over half of colorectal cancer cases had level R1 resistance evidence attributed to KRAS and NRAS mutations. The clinical utility of NGS in late-stage refractory cancer varies widely by tumor type. The presence of level 3 and level 4 evidence in all cancer types bodes well for the development of new targeted drugs. [Table: see text]

Construction of a 13-Gene Signature as a Novel Prognostic Marker for Patients with Clear Cell Renal Cell Carcinoma and the Role of XCR1 in Cell Proliferation.

ObjectiveThe tumor microenvironment plays a key role in regulating tumor progression. This research aimed to develop the biomarker related to tumor microenvironment in clear cell renal cell carcinoma (ccRCC).MethodsThe ESTIMATE algorithm was used to evaluate the immune score of ccRCC cases from The Cancer Genome Atlas (TCGA). Differentially expressed genes between high and low immune scores were identified and a 13-gene signature was constructed by the LASSO Cox regression model to predict overall survival (OS) for ccRCC cases in TCGA or International Cancer Genome Consortium (ICGC) project. The immune cell fractions were calculated by the TIMER algorithm. Cell viability and gene expression were determined by CCK-8 and qRT-PCR, respectively.ResultsThe OS of patients with high immune scores was worse than that of patients with low immune scores. The OS between ccRCC patients from TCGA or ICGC cohort in high- and low-risk groups stratified by the gene signature was significantly different. Subgroup analysis also showed a robust prognostic ability of the gene signature. Multivariate Cox regression analysis demonstrated that this gene signature was an independent prognostic factor. The nomogram that integrated the gene signature and three clinicopathological risk factors had a favorably predictive ability in predicting 3, 5 and 10 year survival. Moreover, the high-risk group had a significantly higher abundance of B cell, T cell, CD4, neutrophil and DC infiltration. Among 13 genes, X-C motif chemokine receptor1 (XCR1) was upregulated in ccRCC cells and exerted an inhibitory effect on cell proliferation.ConclusionThis study constructs a 13-gene signature as a novel prognostic marker to predict the survival of ccRCC patients and XCR1 may serve as a therapeutic target.

Sequential analysis of transcript expression patterns improves survival prediction in multiple cancers

BackgroundLong-term survival in numerous cancers often correlates with specific whole transcriptome profiles or the expression patterns of smaller numbers of transcripts. In some instances, these are better predictors of survival than are standard classification methods such as clinical stage or hormone receptor status in breast cancer. Here, we have used the method of “t-distributed stochastic neighbor embedding” (t-SNE) to show that, collectively, the expression patterns of small numbers of functionally-related transcripts from fifteen cancer pathways correlate with long-term survival in the vast majority of tumor types from The Cancer Genome Atlas (TCGA). We then ask whether the sequential application of t-SNE using the transcripts from a second pathway improves predictive capability or whether t-SNE can be used to refine the initial predictive power of whole transcriptome profiling.MethodsRNAseq data from 10,227 tumors in TCGA were previously analyzed using t-SNE-based clustering of 362 transcripts comprising 15 distinct cancer-related pathways. After showing that certain clusters were associated with differential survival, each relevant cluster was re-analyzed by t-SNE with a second pathway’s transcripts. Alternatively, groups with differential survival identified by whole transcriptome profiling were subject to a second, t-SNE-based analysis.ResultsSequential analyses employing either t-SNE➔t-SNE or whole transcriptome profiling➔t-SNE analyses were in many cases superior to either individual method at predicting long-term survival. We developed a dynamic and intuitive R Shiny web application to explore the t-SNE based transcriptome clustering and survival analysis across all TCGA cancers and all 15 cancer-related pathways in this analysis. This application provides a simple interface to select specific t-SNE clusters and analyze survival predictability using both individual or sequential approaches. The user can recreate the relationships described in this analysis and further explore many different cancer, pathway, and cluster combinations. Non-R users can access the application on the web at https://chpupsom19.shinyapps.io/Survival_Analysis_tsne_umap_TCGA. The application, R scripts performing survival analysis, and t-SNE clustering results of TCGA expression data can be accessed on GitHub enabling users to download and run the application locally with ease (https://github.com/RavulaPitt/Sequential-t-SNE/).ConclusionsThe long-term survival of patients correlated with expression patterns of 362 transcripts from 15 cancer-related pathways. In numerous cases, however, survival could be further improved when the cohorts were re-analyzed using iterative t-SNE clustering or when t-SNE clustering was applied to cohorts initially segregated by whole transcriptome-based hierarchical clustering.

Abstract P6-07-04: Mechanosensing calcium channel PIEZO2 accelerates epithelial-mesenchymal transition in triple-negative breast cancer

Abstract Introduction: Breast cancer is the most common female malignancy. Despite improvements in therapeutic options, metastatic disease remains incurable with limited treatments especially in triple-negative breast cancer (TNBC). Clinical outcomes in patients with TNBC continue to be worse as compared to other breast cancer subtypes. Therefore, novel therapeutic targets for TNBC are in urgent need. PIEZO2, which is a transmembrane calcium channel, is known to function as a channel-type mechanosensor in various aspects of mechanotransduction. Recently there have been reports that PIEZO2 may be involved in carcinogenesis and cancer progression. However, the role of PIEZO2 in breast cancer has not been elucidated. We hypothesized that PIEZO2 plays significant roles in progression of TNBC. Methods: To investigate the role of PIEZO2 in human breast cancer patients, The Cancer Genome Atlas (TCGA) breast cancer cohort, in which there are mRNA expression data from RNA-seq and clinical data, was utilized. PIEZO2 expression level was compared among breast cancer subtypes. Patients’ survival was compared based on PIEZO2 expression in each subtype. We used gene set enrichment analysis (GSEA) algorithm to evaluate pathways associated with PIEZO2 expression. To further elucidate the mechanisms of PIEZO2 roles, TNBC cell line, MDA-MB-231 was utilized for in vitro experiments. PIEZO2 expression was downregulated by small interfering RNA (siRNA). Cell viability was quantified by MTT assay. Protein expression was compared by western blot. Results: Breast cancer showed the highest PIEZO2 expression level, among various types of cancers in TCGA. In the breast cancer cohort, PIEZO2 expression level was lower in the TNBC subtype compared to other subtypes (p&amp;lt;0.001). Thus we hypothesized that PIEZO2 has different roles in each subtype. Survival analyses revealed that patients with high expression of PIEZO2 tumors showed significantly worse overall survival (p=0.016) in the TNBC cohort. However, there was no significant difference in the hormone receptor (HR)-positive (p=0.555) and the HER2-positive (p=0.380) cohorts. To explore underlying mechanisms how PIEZO2 high expressing tumors associate with worse prognosis in TNBC, GSEA was conducted. PIEZO2 high expressing tumors enriched epithelial-mesenchymal transition (EMT; NES=1.90, p=0.012) and Angiogenesis (NES=1.91, p&amp;lt;0.001) gene sets, as well as TAZ/YAP gene set (NES=1.57, p=0.028). PIEZO2 knockdown by siRNA significantly suppressed cell viability. This finding was validated by PIEZO inhibitor GSMTx-4 treatment. Further, expression levels of EMT related proteins, such as Vimentin and Snail, were downregulated in the PIEZO2 knockdown MDA-MB-231 cells compared to control cells. Conclusion: We describe a novel signaling pathway mediated by PIEZO2 leading to worse clinical outcomes in patients with TNBC by activation of EMT. Further work is ongoing to validate PIEZO2 as a novel therapeutic target for TNBC. Citation Format: Eriko Katsuta, Marija Vujcic, Shipra Gandhi, Arthur Beyder, Kazuaki Takabe, Mateusz Opyrchal. Mechanosensing calcium channel PIEZO2 accelerates epithelial-mesenchymal transition in triple-negative breast cancer [abstract]. In: Proceedings of the 2019 San Antonio Breast Cancer Symposium; 2019 Dec 10-14; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2020;80(4 Suppl):Abstract nr P6-07-04.