The Cancer Genome Atlas Samples Research Articles

Abstract 3499: Non-small cell lung cancer (NSCLC) histology classification using DNA methylation data captured from liquid biopsies

Abstract Background: Non-small cell lung cancer (NSCLC) makes up ~85% of lung cancers, and its major subtypes are adenocarcinoma (AD) and squamous cell carcinoma (SC). The histology classification of NSCLC has significant therapeutic and prognostic implications. As liquid biopsies are rapid and non-invasive, computational methods to classify NSCLC subtypes using cell-free DNA (cfDNA) are desired. Methods: Using The Cancer Genome Atlas (TCGA) NSCLC methylation data, we developed a method to classify AD and SC subtypes from cfDNA samples. We mapped TCGA methylation data to Guardant Infinity’s promoter regions and transformed them to allow for the transfer of the TCGA trained model to Infinity cfDNA epigenomic data. An advantage of this approach is that TCGA training data originates from the primary tumor, while cfDNA samples have low tumor fraction (TF) and may also contain signals from metastatic tumors. We trained a penalized Logistic Regression model with LASSO. To better fit the model to Infinity cfDNA data, we performed region filtering. Results: Our model effectively classified AD and SC subtypes from TCGA samples (AUC 0.97) (10-fold cross validation; Table 1). When evaluating on cfDNA samples with epigenomic TF ≥ 1%, the final model shows reasonable performance (AUC 0.94) (Table 1); the performance decreases when reduced to TF > 0.05% (AUC 0.89) as expected (Table 1). When evaluating cfDNA from uncategorized lung samples, the predicted AD probabilities have significant association with KRAS mutations (p-value = 0.027). Conclusion: Our results show that the penalized Logistic Regression model trained with tissue TCGA methylation data through mapping, transformation, and region selection can classify NSCLC subtypes from cfDNA samples. This method may offer a fast, non-invasive tool to inform subtype-based clinical decisions, especially in the setting of poorly-differentiated carcinomas, and to identify subtype switches upon resistance to targeted therapies. Table 1: Model evaluation on tissue TCGA samples and cfDNA Infinity samples (different TF cutoffs) Sample type N samples AUC Accuracy Sensitivity Specificity Precision F1 score tissue TCGA 907 0.97 0.93 0.94 +/- 0.02 0.93 +/- 0.02 0.94 +/- 0.02 0.94 +/- 0.01 cfDNA Infinity 17 (TF ≥ 1%) 0.94 0.88 0.83 +/- 0.3 0.91 +/- 0.17 0.83 +/- 0.3 0.83 +/- 0.21 cfDNA Infinity 29 (TF > 0.05%) 0.89 0.83 0.88 +/- 0.23 0.81 +/- 0.17 0.64 +/- 0.28 0.74 +/- 0.2 Citation Format: Laura Tung, Denis Tolkunov, Catalin Barbacioru, Deepika Kale, Scott Shell, Jill Tsai, Leylah Drusbosky, Kaushal Parikh, Han-Yu Chuang. Non-small cell lung cancer (NSCLC) histology classification using DNA methylation data captured from liquid biopsies [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 3499.

Abstract 6548: Leveraging compact feature sets for TCGA-based molecular subtype classification on new samples

Abstract The NCI's The Cancer Genome Atlas (TCGA) project profiled over 10,000 tumor samples over the course of 10 years. As different tissue-specific working groups reviewed all of the available data, these patient samples were separated into distinct molecular subtypes, and these clusters were reported in various marker papers. While these assignments provided invaluable information about the common patterns of molecular characteristics in different types of cancer there was no consistent methodology for assigning new samples to these defined molecular subtypes.The NCI's Tumor Molecular Pathology group was formulated to create machine learning-based models that could be applied to non-TCGA samples and determine their TCGA mapped subtypes. Five modeling systems, JADBio, SKGrid by the Oregon Health and Science University, CloudForest by the Institute of Systems Biology, AKLIMATE by University of California Santa Cruz and subSCOPE by BC Cancer’s Genome Sciences Centre, were trained to recognize TCGA subtypes using multi-omic measurements from gene expression, DNA methylation, miRNA expression, copy number, and somatic mutation calls. While the TCGA samples were profiled using multi-omic technologies, single platform and/or compact feature set models also were assessed for their ability to assign these classifications. Each machine learning system created predictive models for 106 subtypes from 26 cancer types using as few features as possible, with a maximum of 100 features allowed for scored models. A set of 411,706 models was developed, composed of results of each of the learning methods across the various omic platforms. Top models, both multi-omic and single platform, were selected for each cancer type. On average, models were able to achieve an overall weighted F1 score of 0.895 with 42 features. While the top models for each cancer type had an overall weighted F1 mean performance of 0.936 with a mean of 29 features, in 20 of the 26 cancer types models using only gene expression provided the best performance. Analysis of features selected by the models showed some known onco-drivers were selected by many models, but many times different models would utilize features of different genes with similar levels of performance. Network-level analysis revealed that many genes of these selected features operated within the same pathways.Transferability of these models to external datasets was tested, taking TCGA breast cancer trained models and applying them to AURORA and METABRIC datasets. Interestingly, despite the data platform difference between TCGA (RNAseq) and METABRIC (microarray), model performance saw only minimal degradation of F1 values in transfer. This set of models and the training dataset will provide new opportunities for researchers and translational scientists to connect new tumors to the subtypes seen in the TCGA cohorts. Citation Format: Kyle Ellrott, Chris K. Wong, Christina Yau, Mauro A. Castro, Jordan Lee, Brian Karlberg, Jasleen K. Grewal, Vincenzo Lagani, Bahar Tercan, Verena Friedl, Toshinori Hinoue, Vladislav Uzunangelov, Lindsay Westlake, Xavier Loinaz, Ina Felau, Peggy Wang, Anab Kemal, Samantha J. Caesar-Johnson, Ilya Shmulevich, Alexander J. Lazar, Ioannis Tsamardinos, Katherine A. Hoadley, The Cancer Genome Atlas Analysis Network, Gordon A. Robertson, Theo A. Knijnenburg, Christopher C. Benz, Joshua M. Stuart, Jean C. Zenklusen, Andrew D. Cherniack, Peter W. Laird. Leveraging compact feature sets for TCGA-based molecular subtype classification on new samples [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 6548.

Abstract 2757: Investigating regulatory drivers of prostate cancer metastasis in circulating tumor cells

Abstract Prostate cancer is the most common cancer in men and the second-leading cause of death. Most prostate cancer deaths are due to metastasis, particularly to the bone. Metastasis occurs when cells from the primary tumor travel through the blood as circulating tumor cells (CTCs) and settle in metastatic sites. CTCs are difficult to collect from patients, and understanding the relationships between CTCs and the tumor phenotypes of origin could reveal mechanisms of cancer metastasis. Hence, we utilized several genetically engineered mouse models (GEMMs) of metastatic prostate cancer to evaluate tumor phenotypes and CTCs. We compared the tumor phenotypes of several GEMMs of metastatic prostate cancer and identified a level of heterogeneity among tumors, particularly focusing on a GEMM that demonstrated severe metastasis, especially to the bone. Next, we analyzed CTCs from this GEMM and focused our analysis on transcriptional states that are regulated by proteins called master regulators (MRs). MRs control the expression of groups of genes called regulons, and pharmacologically inhibiting these MRs may lead to reversal of cancer phenotypes. We hypothesized that these MRs may be mechanistic drivers of CTC cell state in prostate cancer metastasis. We investigated MRs enriched in CTCs from mice and screened and validated Lap2, the lamina-associated polypeptide 2 protein, as a regulatory driver in mice with metastatic prostate cancer. Analysis of bulk RNA-seq profiles from primary and metastatic prostate samples in The Cancer Genome Atlas and Stand Up to Cancer cohorts, respectively, revealed that the regulon-based activity of Lap2 was significantly higher in the latter group. Moreover, analysis of single-cell RNA-seq profiles of CTCs isolated from prostate cancer patients confirmed aberrant Lap2 activity. Finally, the role of Lap2 was assessed in human tissue from an engineered multi-organ-on-a-chip model of metastasis. Silencing of Lap2 production in cells dissociated from mouse CTC-derived organoids inhibited extravasation to the bone in the model, suggesting a critical role for Lap2 in metastasis. Our findings illustrate the unique transcriptional profile of prostate cancer CTCs associated with their metastatic potential. We identified Lap2 as a novel driver of CTC dissemination and metastasis in mice and validated its activity in human prostate cancer, which can aid future therapies to prevent metastatic progression in patients. Citation Format: Aunika Zheng, Aleksandar Obradovic, Kacey Ronaldson-Bouchard, Gordana Vunjak-Novakovic, Andrea Califano, Peter A. Sims, Cory Abate-Shen, Arianna Giacobbe. Investigating regulatory drivers of prostate cancer metastasis in circulating tumor cells [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 2757.

Abstract 5552: Uncovering clinically significant tumor microenvironment interaction programs across diverse cancers

Abstract Background: Cell-cell interactions (CCI) within the tumor microenvironment (TME) play pivotal roles in various tumor behaviors, including cancer growth, metastasis, immune evasion, and resistance to therapy. Compared to front-line treatments, targeting specific CCIs marks a paradigm shift in cancer therapy, aiming for enhanced response rates with reduced side effects. Although current approaches targeting immune regulation pathways exhibit effective anti-tumor responses, substantial variability in response rates across patients and cancer types persists. Consequently, a critical imperative emerges to discover novel interaction programs (IPs) across diverse TMEs, demonstrating clear clinical impact across a broad spectrum of patients. Methods: Characterization of pan cancer IPs. We compiled a pan cancer single-cell RNA-seq (scRNA-seq) atlas, including 4 million cells from 890 tumors across 14 cancer types. In each sample, CCIs were inferred using a ligand-receptor (LR) analysis framework, generating ranked CCI scores based on consensus from various LR inference methods. To find IPs enriched across TMEs, we applied an unsupervised factorization approach. The outcome represents CCIs as factors, where sample loading indicates factors’ strength, and feature loading highlights the cells and LR pairs constituting each IP. Evaluating IPs in clinical RNA-seq cohorts. Given the more immediate translational potential of bulk cancer sequencing for patient stratification, we devised an approach to predict IPs' strengths in bulk RNA-seq. Random Forest regression models using pseudo-bulk expression of cancer genes were trained to predict the loading (strength) of factors (IPs) in each sample in the scRNA-seq pan-cancer atlas. These models were subsequently applied to cancer samples in The Cancer Genome Atlas (TCGA) cohort to assess the clinical impact of various IPs on patient mortality. Results: We assessed the impact of IPs on 5-year overall survival using Cox proportional hazard models. We identified multiple factors significantly associated with survival in at least 4 cancer types. Notably, a robust prognostic factor linked to the inclusion of anti-tumor natural killer cell interactions and exclusion of pro-tumor macrophage interactions was found in MESO, LGG, ACC, and UVM (HR 1.6-6.6, p < 0.05). The validation of IPs using spatial transcriptomics and evaluation of IP prediction's utility in targeted treatment cohorts are currently underway. Conclusion: We highlight a potent method for detecting clinically significant IPs involving cells in TME communicating through LR pairs. The ability to assess these scRNA-seq-derived IPs in clinical bulk RNA-seq cohorts signifies a valuable advancement. Our approach holds promise for expanding the repertoire of cell therapies, identifying new treatment targets, and improving our predictive capabilities for responses to existing therapies. Citation Format: Ido Nofech-Mozes, Vivian Wang, Philip Awadalla, Sagi Abelson. Uncovering clinically significant tumor microenvironment interaction programs across diverse cancers [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 5552.

Identification and validation of a novel six-gene signature based on mucinous adenocarcinoma-related gene molecular typing in colorectal cancer

Background and objectivesColorectal mucinous adenocarcinoma (MAC) is a particular pathological type that has yet to be thoroughly studied. This study aims to investigate the characteristics of colorectal MAC-related genes in colorectal cancer (CRC), explore the role of MAC-related genes in accurately classifying CRC, and further construct a prognostic signature.MethodsCRC samples were collected from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO). MAC-related differentially expressed genes (DEGs) were analyzed in TCGA samples. Based on colorectal MAC-related genes, TCGA CRC samples were molecularly typed by the non-negative matrix factorization (NMF). According to the molecular subtype characteristics, the RiskScore signature was constructed through univariate Cox, the least absolute shrinkage and selection operator (LASSO), and multivariate Cox regression analyses. Clinical significance in CRC of the RiskScore signature was analyzed. A nomogram was further built based on the RiskScore signature.ResultsFrom the colorectal MAC-related genes, three distinct molecular subtypes were identified. A RiskScore signature composed of six CRC subtype-related genes (CALB1, MMP1, HOXC6, ZIC2, SFTA2, and HYAL1) was constructed. Patients with high-RiskScores had the worse prognoses. RiskScores led to differences in gene mutation characteristics, antitumor drug sensitivity, and tumor microenvironment of CRC. A nomogram based on the signature was developed to predict the one-, three-, and five-year survival of CRC patients.ConclusionMAC-related genes were able to classify CRC. A RiskScore signature based on the colorectal MAC-related molecular subtype was constructed, which had important clinical significance for guiding the accurate stratification of CRC patients.

Novel application of the ferroptosis-related genes risk model associated with disulfidptosis in hepatocellular carcinoma prognosis and immune infiltration.

Hepatocellular carcinoma (HCC) stands as the prevailing manifestation of primary liver cancer and continues to pose a formidable challenge to human well-being and longevity, owing to its elevated incidence and mortality rates. Nevertheless, the quest for reliable predictive biomarkers for HCC remains ongoing. Recent research has demonstrated a close correlation between ferroptosis and disulfidptosis, two cellular processes, and cancer prognosis, suggesting their potential as predictive factors for HCC. In this study, we employed a combination of bioinformatics algorithms and machine learning techniques, leveraging RNA sequencing data, mutation profiles, and clinical data from HCC samples in The Cancer Genome Atlas (TCGA), Gene Expression Omnibus (GEO), and the International Cancer Genome Consortium (ICGC) databases, to develop a risk prognosis model based on genes associated with ferroptosis and disulfidptosis. We conducted an unsupervised clustering analysis, calculating a risk score (RS) to predict the prognosis of HCC using these genes. Clustering analysis revealed two distinct HCC clusters, each characterized by significantly different prognostic and immune features. The median RS stratified HCC samples in the TCGA, GEO, and ICGC cohorts into high-and low-risk groups. Importantly, RS emerged as an independent prognostic factor in all three cohorts, with the high-risk group demonstrating poorer prognosis and a more active immunosuppressive microenvironment. Additionally, the high-risk group exhibited higher expression levels of tumor mutation burden (TMB), immune checkpoints (ICs), and human leukocyte antigen (HLA), suggesting a heightened responsiveness to immunotherapy. A cancer stem cell infiltration analysis revealed a higher similarity between tumor cells and stem cells in the high-risk group. Furthermore, drug sensitivity analysis highlighted significant differences in response to antitumor drugs between the two risk groups. In summary, our risk prognostic model, constructed based on ferroptosis-related genes associated with disulfidptosis, effectively predicts HCC prognosis. These findings hold potential implications for patient stratification and clinical decision-making, offering valuable theoretical insights in this field.

Development of a novel prognostic signature based on single-cell combined bulk RNA analysis in breast cancer.

Breast cancer (BC), a malignant tumor, is a significant cause of death and disability among women globally. Recent research indicates that copy number variation plays a crucial role in tumor development. In this study, we employed the Single-Cell Variational Aneuploidy Analysis (SCEVAN) algorithm to differentiate between malignant and non-malignant cells, aiming to identify genetic signatures with prognostic relevance for predicting patient survival. We analyzed gene expression profiles and associated clinical data from the Gene Expression Omnibus (GEO) and The Cancer Genome Atlas (TCGA) databases. Using the SCEVAN algorithm, we distinguished malignant from non-malignant cells and investigated cellular interactions within the tumor microenvironment (TME). We categorized TCGA samples based on differentially expressed genes (DEGs) between these cell types. Subsequent Kyoto Encyclopedia of Genes and Genomes pathway analysis was conducted. Additionally, we developed polygenic models for the DEGs using least absolute shrinkage and selection operator-penalized Cox regression analysis. To assess the prognostic accuracy of these characteristics, we generated Kaplan-Meier and receiver operating characteristic curves from training and validation datasets. We also monitored the expression variations of prognostic genes across the pseudotime of malignant cells. Patients were divided into high-risk and low-risk groups based on median risk scores to compare their TME and identify potential therapeutic agents. Lastly, polymerase chain reaction was used to validate seven pivotal genes. The SCEVAN algorithm identified distinct malignant and non-malignant cells in GSE180286. Cellchat analysis revealed significantly increased cellular communication, particularly between fibroblasts, endothelial cells and malignant cells. The DEGs were predominantly involved in immune-related pathways. TCGA samples were classified into clusters A and B based on these genes. Cluster A, enriched in immune pathways, was associated with poorer prognosis, whereas cluster B, predominantly involved in circadian rhythm pathways, showed better outcomes. We constructed a 14-gene prognostic signature, validated in a 1:1 internal TCGA cohort and external GEO datasets (GSE42568 and GSE146558). Kaplan-Meier analysis confirmed the prognostic signature's accuracy (p < 0.001). Receiver operating characteristic curve analysis demonstrated the predictive reliability of these prognostic features. Single-cell pseudotime analysis with monocle2 highlighted the distinct expression trends of these genes in malignant cells, underscoring the intratumoral heterogeneity. Furthermore, we explored the differences in TME between high- and low-risk groups and identified 16 significantly correlated drugs. Our findings suggest that the 14-gene prognostic signature could serve as a novel biomarker for forecasting the prognosis of BC patients. Additionally, the immune cells and pathways in different risk groups indicate that immunotherapy may be a crucial component of treatment strategies for BC patients.

Abstract B033: Mapping subclonal epigenetic evolution in colorectal cancer by spatial analysis of DNA hypermethylation

Abstract Aberrant patterns of DNA methylation (DNAme) have long been noted in colorectal cancer (CRC). Early studies found overall hypomethylation of the cancer genome, while subsequent work revealed focal hypermethylation at specific loci, especially CpG-rich promoters. Loci occupied by polycomb repressive complex 2 in stem cells are remarkably enriched in cancer-specific hypermethylation events. These Polycomb Target Genes (PTGs) include important transcriptional regulators of lineage specification and differentiation, suggesting that hypermethylation-mediated silencing of PTGs could play an active role in CRC initiation or progression. Despite these findings, we have limited knowledge of when, where, and how this hypermethylation arises. Preliminary evidence from bulk tumor samples in the Cancer Genome Atlas suggests that many of these PTG hypermethylation events may be almost completely clonal. Additionally, we have noted rare, densely methylated reads covering PTGs in cancer-adjacent, histologically normal tissue. These observations suggest that PTG hypermethylation may be a very early event in carcinogenesis and may play a permissive or even active role in that process. To better understand the timing and evolution of PTG hypermethylation events in CRC, we have developed a protocol that allows us to extract dozens of small (20µL), spatially annotated mini-bulk samples from live tumors. These mini-bulks may be frozen or dissociated to viable single-cell suspensions. Remaining tumor tissue is preserved in the form of spatially annotated FFPE blocks, which can be used for analyses such as high-plex immunofluorescence or spatial transcriptomics. Using this process, we have collected over 500 spatially annotated samples from 6 CRC cases, including 196 micro-bulk samples profiled using the Infinium MethylationEPIC array. Additionally, we have profiled DNAme in 147 single cells using a dramatically improved single-cell whole-genome bisulfite sequencing protocol developed in our lab. Analysis of EPIC array data reveals wide variation in DNAme heterogeneity between tumors. While in some tumors, the vast majority of DNA hypermethylation events are clonal, other tumors reveal significant intra-tumor heterogeneity, including two or more prominent DNAme subclones as well as a substantial number of ‘private’ hypermethylation events that occur in only one sample. Mapping DNAme information back onto our spatial data reveals that in at least one case, DNAme clones are spatially partitioned and are associated with distinct histology. Ongoing work includes inference of phylogenetic trees from DNA hypermethylation events using non-reversible models of methylation accumulation, allowing us to measure the relative timing of hypermethylation events, as well as integration of array, whole-genome, and single-cell methylation data to more completely characterize DNAme subclones, characterize cellular DNAme heterogeneity at variably methylated sites, and shed further light on the evolution of cancer-related DNA hypermethylation. Citation Format: Nathan J. Spix, Hsiao-yun Milliron, Manpreet Kalkat, Emily Eugster, David W. Chesla, David Sokol, Emily Jung, Paula Nolte, Kelly K. Krzyzanowski, Toshinori Hinoue, Hui Shen, Peter W. Laird. Mapping subclonal epigenetic evolution in colorectal cancer by spatial analysis of DNA hypermethylation [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Translating Cancer Evolution and Data Science: The Next Frontier; 2023 Dec 3-6; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(3 Suppl_2):Abstract nr B033.

Calreticulin, a potential coregulator of immune checkpoints and biomarker associated with tumor microenvironment and clinical prognostic significance in breast invasive carcinoma.

As a promising immune checkpoint of immunogenic cell death (ICD) and multifunctional calcium-binding molecular chaperone, calreticulin (CALR) has been attracting increasing attention. CALR mainly locates in cellular endoplasmic reticulum and significantly affects cell proliferation, invasion, induction of apoptosis, and angiogenesis in breast invasive carcinoma (BRCA). CALR overexpression might be correlated with a worse outcome. Nonetheless, it remains obscure how CALR correlates with immune infiltration and survival prognosis of BRCA. In this study, we investigated CALR expression utilizing RNAseq data from the cancer genome atlas (TCGA) and genotype-tissue expression (GTEx) database. The prognostic value of CALR was analyzed using clinical survival data. Enrichment analysis was conducted using the R package "clusterProfiler." We downloaded the immune cell infiltration score of TCGA samples from published articles and online databases and performed a correlation analysis between immune cell infiltration levels and CALR expression. We further assessed the association between CALR and immunomodulators. Moreover, we also evaluated the expression of CALR in 100 formalin-fixed and paraffin-embedded breast cancer and adjacent normal breast tissue specimens. Our results found that CALR was highly expressed in BRCA, and CALR expression levels differed in pathological stages, T stages, and N stages. Besides, these results suggested that CALR overexpression may have adverse effects on the progression-free interval (PFI) and disease-free interval (DFI), which may be related to tumor proliferation, invasion, and metastasis, leading to tumor deterioration. Meanwhile, immune cell infiltration analysis revealed a correlation between the expression of CALR and the number of neutrophils and dendritic cells, suggesting that CALR was highly correlated with many immunomodulators in BRCA. Our results provide potential biomarkers of CALR in BRCA. CALR may interact synergistically with other immunomodulators to regulate the immune microenvironment, which could be utilized to develop new immunotherapy drugs.

Singleton mutations in large-scale cancer genome studies: uncovering the tail of cancer genome.

Singleton or low-frequency driver mutations are challenging to identify. We present a domain driver mutation estimator (DOME) to identify rare candidate driver mutations. DOME analyzes positions analogous to known statistical hotspots and resistant mutations in combination with their functional and biochemical residue context as determined by protein structures and somatic mutation propensity within conserved PFAM domains, integrating the CADD scoring scheme. Benchmarked against seven other tools, DOME exhibited superior or comparable accuracy compared to all evaluated tools in the prediction of functional cancer drivers, with the exception of one tool. DOME identified a unique set of 32917 high-confidence predicted driver mutations from the analysis of whole proteome missense variants within domain boundaries across 1331 genes, including 1192 noncancer gene census genes, emphasizing its unique place in cancer genome analysis. Additionally, analysis of 8799 TCGA(The Cancer Genome Atlas) and in-house tumor samples revealed 847 potential driver mutations, with mutations in tyrosine kinase members forming the dominant burden, underscoring its higher significance in cancer. Overall, DOME complements current approaches for identifying novel, low-frequency drivers and resistant mutations in personalized therapy.

Abstract A019: eIF4A and PDCD4 expression in TNBC underlies sensitivity to natural product translation inhibitor

Abstract Despite recent advances in immunotherapy and adjuvant-therapy, resistance to cancer therapeutics is common and requires novel approaches. Moreover, highly aggressive triple-negative breast cancers lack effective targeted treatments. Deregulation of translation factors is associated with the increased proliferation of TNBC cells, heightened CSC activity, and poor patient outcomes. Thus targeting translation regulatory factors upregulated in TNBC is a promising approach to developing novel treatments. Pateamine A (PatA), a natural marine macrodiolide product, inhibits RNA cap-dependent translation by binding and sequestering the eukaryotic translation initiation factor eIF4A. eIF4A promotes the translation of polypurine-rich mRNA sequences, which are enriched within breast cancer oncogenes. Here we report the anticancer efficacy of novel PatA analogs we have synthesized against breast cancer cell lines. The PatA analogs exhibited selective cytotoxicity to breast and colorectal cancers compared to non-cancerous cells measured by Cell Titer-Blue viability studies. Novel analogs show greater antiproliferative potency against models of triple-negative than against human mammary epithelial cells (HMLE). To further elucidate eIF4A as a molecular target, we conducted analyses of patient tissue samples in the Cancer Genome Atlas (TCGA). Our analyses reveal the overexpression of eIF4A in tumor tissues compared to normal tissues and gene correlation studies reveal the upregulation of key co-expressed tumor drivers including beta-catenin, fibronectin, and epithelial-to-mesenchymal transition (EMT) markers such as SNAIL and SLUG. Further analyses show a downregulation of tumor suppressor protein PDCD4, an endogenous inhibitor of eIF4A, in cancer tissues. We have generated eIF4A shRNA knockdown and PDCD4 shRNA knockdown cell lines to uncover the eIF4A-PDCD4 regulatory axis across breast adenocarcinoma and colorectal adenocarcinoma cell types. We show that the loss of eIF4A attenuates the mesenchymal and stem-like phenotype of breast cancer stem cells revealed by phenotypic assays. Our results suggest that targeting translation initiator eIF4A is an effective approach against highly proliferative and metastatic breast cancer types. The presented findings underscore the potential of eIF4A as a promising molecular target and supports translational control as a viable approach against aggressive cancers. Citation Format: Alagu Subramanian, Haoran Xue, Kenneth G Hull, Daniel Romo, Joseph Taube. eIF4A and PDCD4 expression in TNBC underlies sensitivity to natural product translation inhibitor [abstract]. In: Proceedings of the AACR-NCI-EORTC Virtual International Conference on Molecular Targets and Cancer Therapeutics; 2023 Oct 11-15; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2023;22(12 Suppl):Abstract nr A019.

Abstract C049: Genomic correlates with ancestry in lung squamous cell carcinoma

Abstract Outcome data from lung squamous cell carcinoma (SCC) show significantly lower overall and progression-free survival in Black patients. However, whether any genomic factors correlate with race and ancestry in this lung SCC, and whether this may contribute to disparities, has not been elucidated. We analyzed clinically annotated genomic data from 501 lung SCC tumor samples in the Cancer Genome Atlas (TCGA), with available ancestry admixture data. Here, we analyzed clinical data, gene expression profiles, mutations, and copy number changes to identify significantly differential events between tumors from patients of European (EUR) or African (AFR) ancestry. In tumors from AFR patients, we identified more frequent gain of chromosome arm 8q and/or the chr8q24 locus, where the oncogene MYC and its enhancers are located. 8q24 gain is associated both with global AFR ancestry and local AFR ancestry. Furthermore, expression of MYC targets is significantly increased in AFR samples relative to EUR samples, all of which suggest a role for the MYC pathway in differential cancer progression. MYC activity correlates with decreased immune infiltration and gene signaling, and patients with AFR ancestry are more likely to have decreased immune infiltrate and pathway activity. Our findings are consistent with observations in prostate cancer; this is the first study to find MYC enhancer amplification correlating with African ancestry in lung SCC. We are currently testing additional cohorts, including AACR GENIE, to validate these findings. Future studies will elucidate patient-specific treatment modalities. Citation Format: Sejal Jain, Alexandria Yao, Fatemeh Momen-Heravi, Diane Allen-Gipson, Xiaoyang Zhang, Brian S. Henick, Jian Carrot-Zhang, Alison M. Taylor. Genomic correlates with ancestry in lung squamous cell carcinoma [abstract]. In: Proceedings of the 16th AACR Conference on the Science of Cancer Health Disparities in Racial/Ethnic Minorities and the Medically Underserved; 2023 Sep 29-Oct 2;Orlando, FL. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2023;32(12 Suppl):Abstract nr C049.

Carbonic Anhydrase 4 Serves as A Novel Prognostic Biomarker and Therapeutic Target for Non-Small Cell Lung Cancer: A Study Based on TCGA Samples.

Carbonic anhydrase 4 (CA4) is a member of a large family of zinc metalloenzymes that catalyze the reversible hydration of carbon dioxide and was found to have low expression in non-small cell lung cancer (NSCLC). However, the specific role of CA4 in NSCLC and the underlying mechanisms remain unknown. The bioinformatic analysis on lung adenocarcinoma (LUAD) and lung squamous cell carcinoma (LUSC) datasets downloaded from The Cancer Genome Atlas (TCGA) database was performed. We found that CA4 expression was lower in tumors than that in normal tissues, which were verified by Real-time PCR. Lower CA4 levels were significantly associated with higher T stages in LUAD and LUSC cohorts. Multivariate analysis showed that CA4 is an independent prognostic factor for NSCLC. Furthermore, the expression of CA4 also correlated with immune infiltration and drug sensitivity. Ectopic expression of CA4 decreased NSCLC cell proliferation in vitro by CCK-8 assay. CA4 caused G0/G1 cell cycle arrest by cell experiments. Mechanistic studies found that CA affects the cell cycle and inhibits cell proliferation by downregulating the expression of CDK2. The present findings highlight the role of CA4 in NSCLC and identify CA4 as a potential novel diagnostic and prognostic biomarker for the treatment of NSCLC.

Bioinformatic analysis and verification of a lipid metabolism-related long noncoding RNA prognostic signature for head and neck squamous cell carcinoma

PurposeBoth lipid metabolism reprogramming and lncRNAs exert effects on tumor development. We aimed to predict the prognosis of head and neck squamous cell carcinoma (HNSCC) based on lipid metabolism-related (LR)-lncRNAs. MethodsLR-lncRNAs were determined from the RNA-ref profiles of HNSCC samples in The Cancer Genome Atlas (TCGA). The prognostic model was established by univariate Cox and Lasso regression analysis. Clinical relevance and predictive accuracy were investigated, and external validation was also performed in the Gene Expression Omnibus (GEO) cohort. Tumor immune infiltration and relevant functional analysis, including the association of autophagy with prognostic signatures, were conducted through single-sample gene set enrichment analysis (ssGSEA). The regulatory network of candidate LR-lncRNAs was investigated via coexpression, ceRNA and cis/trans acting interactions. Potential genes were selected through qRT–PCR analysis, and their effects on tumor biological activities and autophagic activity were explored after gene knockdown. ResultsA total of 222 LR-lncRNAs were identified. Among the 41 genes with prognostic significance, 17 lncRNAs were eligible for the risk model. Patients in the high-risk group had a poorer prognosis than those in the low-risk group, and the risk score was found to be positively associated with tumor microenvironment infiltration via multiple algorithms. Furthermore, improved prognosis was found in patients with high autophagic scores and low risk scores, and autophagy-related genes such as PINK1 and CCL2 showed significantly lower expression in the low-risk group. The expression of immune checkpoint genes such as CD28, CTLA4 and PDCD1 decreased dramatically in the high-risk group. The target genes of candidate lncRNAs were confirmed, such as ENO2 and PPAR-gamma. Furthermore, MIR4435-2HG was the most significantly overexpressed lncRNA in HNSCC cell lines and tumor samples, which could promote proliferation and migration and inhibit apoptosis. Additionally, MIR4435-2HG silencing activated autophagy by increasing LC3B expression. ConclusionThis study constructed an LR-lncRNA prognostic signature for HNSCC and indicated its relationships with tumor immunity and autophagy, which provides a promising future for LR-lncRNA-oriented prognostic tools and therapeutic targets.

AKR1B10 regulates M2 macrophage polarization to promote the malignant phenotype of gastric cancer.

Immunotherapy has brought new hope to gastric cancer (GC) patients. Exploring the immune infiltration pattern in GC and the key molecules is critical for optimizing the efficacy of immunotherapy. Aldo-keto reductase family 1 member B10 (AKR1B10) is an inflammatory regulator and is closely related to the prognosis of patients with GC. However, the function of AKR1B10 in GC remains unclear. In the present study, the CIBERSORT algorithm was used to analyze the immune infiltration pattern in 373 samples in the Cancer Genome Atlas (TCGA) database. Differentially expressed genes (DEGs) were seared by combing the TCGA database and the Gene Expression Omnibus (GEO) database, and the key molecule AKR1B10 was identified by weighted gene coexpression network analysis (WGCNA). The biological functions of AKR1B10 in stomach adenocarcinoma (STAD) were investigated in vitro. Macrophage polarization was the main immune infiltration pattern in GC, and the state of macrophage polarization was closely related to the pathological grading of GC and the clinical stage of patients. AKR1B10, MUC5AC, TFF2, GKN1, and PGC were significantly down-regulated in GC tissues. Low AKR1B10 expression induced M2 macrophage polarization and promoted the malignant phenotype of GC. M2 macrophage polarization is the main immune infiltration pattern in GC. Low AKR1B10 expression induces M2 macrophage polarization and promotes the malignant transformation of GC.

Identification of the HNSC88 Molecular Signature for Predicting Subtypes of Head and Neck Cancer.

Head and neck squamous cell carcinoma (HNSC) exhibits genetic heterogeneity in etiologies, tumor sites, and biological processes, which significantly impact therapeutic strategies and prognosis. While the influence of human papillomavirus on clinical outcomes is established, the molecular subtypes determining additional treatment options for HNSC remain unclear and inconsistent. This study aims to identify distinct HNSC molecular subtypes to enhance diagnosis and prognosis accuracy. In this study, we collected three HNSC microarrays (n = 306) from the Gene Expression Omnibus (GEO), and HNSC RNA-Seq data (n = 566) from The Cancer Genome Atlas (TCGA) to identify differentially expressed genes (DEGs) and validate our results. Two scoring methods, representative score (RS) and perturbative score (PS), were developed for DEGs to summarize their possible activation functions and influence in tumorigenesis. Based on the RS and PS scoring, we selected candidate genes to cluster TCGA samples for the identification of molecular subtypes in HNSC. We have identified 289 up-regulated DEGs and selected 88 genes (called HNSC88) using the RS and PS scoring methods. Based on HNSC88 and TCGA samples, we determined three HNSC subtypes, including one HPV-associated subtype, and two HPV-negative subtypes. One of the HPV-negative subtypes showed a relationship to smoking behavior, while the other exhibited high expression in tumor immune response. The Kaplan-Meier method was used to compare overall survival among the three subtypes. The HPV-associated subtype showed a better prognosis compared to the other two HPV-negative subtypes (log rank, p = 0.0092 and 0.0001; hazard ratio, 1.36 and 1.39). Additionally, within the HPV-negative group, the smoking-related subgroup exhibited worse prognosis compared to the subgroup with high expression in immune response (log rank, p = 0.039; hazard ratio, 1.53). The HNSC88 not only enables the identification of HPV-associated subtypes, but also proposes two potential HPV-negative subtypes with distinct prognoses and molecular signatures. This study provides valuable strategies for summarizing the roles and influences of genes in tumorigenesis for identifying molecular signatures and subtypes of HNSC.

Abstract A027: DNA methylation-based fetal niche and repetitive element profiling predict overall survival in clear cell renal cell carcinoma

Abstract Background: Fetal cell origin DNA methylation signature (FCO) estimates the proportion of cells in a mixture of cell types that are of fetal origin. Machine learning-based methods can predict DNA methylation of locus-specific repetitive elements (RE) by learning surrounding genetic and epigenetic information. In this study, we applied the FCO signature and RE methylation prediction to clear cell renal cell carcinoma (ccRCC) samples to study FCO and RE alterations in ccRCC carcinogenesis and survival. Methods: The FCO algorithm was applied to 47 ccRCC tumor samples and four adjacent normal samples from the Dartmouth Renal Tumor Biobank. 319 ccRCC tumor samples and 160 adjacent normal samples from The Cancer Genome Atlas (TCGA) were used for external validation. Linear regression models were used to compare the FCO signal between ccRCC tumor and normal samples adjusting for sex and age. Cox proportional hazard models were employed to study FCO with overall survival outcome in ccRCC samples, adjusting for sex, age, and tumor stage. The survival groups were dichotomized to FCO = 0% vs. FCO &amp;gt; 0%. The REMP package from Bioconductor was used to project LINE1 element methylation in 47 ccRCC samples. LINE1 median level methylation was compared by tumor grade. Cox proportional hazard models were employed to study LINE1 methylation level as a continuous variable with overall survival in ccRCC samples, adjusting for sex, age, and tumor stage. Results: The FCO signal is significantly lower in Dartmouth Renal Tumor Biobank archived ccRCC tumor samples compared to adjacent normal samples (Δ = 42.9%, p = 5.7e-11). By comparing the overall survival between FCO = 0% and FCO &amp;gt; 0% groups, we observed significantly worse survival outcomes in ccRCC patients with zero levels of FCO (HR: 2.46, 95% CI: 1.16, 5.18). Similar results were observed in TCGA samples. A significant decrease of FCO was found in ccRCC tumor samples compared to adjacent normal samples (Δ = 32.9%, p = 2.2e-308). A 1% increase in FCO is significantly associated with a better survival outcome (HR: 0.97, 95%CI 0.95, 0.99). For RE methylation prediction, we observed that the median LINE1 methylation level decreases with ccRCC tumor grade. A higher level of LINE1 median methylation is significantly associated with a better survival outcome (p = 0.01). Conclusion: Normal kidney samples possess a fetal niche reflected by DNA methylation, while ccRCC tumorigenesis depletes the fetal cells in normal kidneys. The depletion of FCO is associated with worse survival outcomes in ccRCC. The repetitive element LINE1 methylation level decreases with tumor grade and is related to ccRCC survival outcomes. These findings promise future investigations on the epigenetically targetable fetal niche and repetitive elements in ccRCC carcinogenesis and therapeutics. Citation Format: Ze Zhang, Lucas Salas. DNA methylation-based fetal niche and repetitive element profiling predict overall survival in clear cell renal cell carcinoma [abstract]. In: Proceedings of the AACR Special Conference: Advances in Kidney Cancer Research; 2023 Jun 24-27; Austin, Texas. Philadelphia (PA): AACR; Cancer Res 2023;83(16 Suppl):Abstract nr A027.

Based on the Cancer Genome Atlas Database Development of a prognostic model of RNA binding protein in stomach adenocarcinoma

The purpose of this study was to identify potential RNA binding proteins associated with the survival of gastric adenocarcinoma, as well as the corresponding biological characteristics and signaling pathways of these RNA binding proteins. RNA sequencing and clinical data were obtained from the cancer genome map (N = 32, T = 375) and the comprehensive gene expression database (GSE84437, N = 433). The samples in The Cancer Genome Atlas were randomly divided into a development group and a test group. A total of 1495 RNA binding protein related genes were extracted. Using nonparametric tests to analyze the difference of RNA binding protein related genes, 296 differential RNA binding proteins were obtained, 166 were up-regulated and 130 were down regulated. Twenty prognosis-related RNA binding proteins were screened using Cox regression, including 14 high-risk genes (hazard ratio > 1.0) and 6 low-risk genes (hazard ratio < 1.0). Seven RNA binding protein related genes were screened from the final prognostic model and used to construct a new prognostic model. Using the development group and test group, the model was verified with survival analysis, receiver operating characteristics curves and prognosis analysis curves. A prediction nomogram was finally developed and showed good prediction performance.

Identification and validation of a novel prognostic model for gastric cancer based on m7G-related genes.

The role of N7-methyladenosine (m7G)-related genes in the progression and prognosis of gastric cancer (GC) remains unclear. This study aimed to explore prognostic biomarkers for GC based on m7G methylation regulators and to construct a prognostic risk model. RNA sequencing profiles with corresponding clinicopathological information associated with GC of which the histological type was stomach adenocarcinoma (STAD) were obtained from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO), respectively. A total of 29 m7G regulators were extracted from previous studies. According to the expression similarity of m7G regulators, the GC samples obtained from TCGA were further classified into 2 clusters demonstrating different overall survival (OS) rates and genetic heterogeneity, and the differentially expressed genes (DEGs) between these 2 clusters were defined as m7G-related genes. Univariate regression analysis and regression analysis were then used to obtain the prognostic m7G-related genes. The samples in TCGA and Genotype-Tissue Expression (GTEx) were used to verify the differential expression and prognostic value of these m7G-related genes contained in the prognostic model. Subsequently, the risk score was combined with other prognostic factors to develop a nomogram. The predictive ability of the nomogram was evaluated by the standard receiver operating characteristic (ROC) curve. Gene set enrichment analysis (GSEA) was used to identify activation pathways in both groups. Finally, the association between the prognostic model and the immune characteristics of GC were appraised. A prognostic model consisting of 11 m7G-related genes was constructed. GC patients in the high-risk group were shown to have a poor prognosis and this result was further demonstrated in each group. The risk model can be applied for patients with different clinical features. The results of GSEA showed that cell adhesion, cell junction, and focal adhesion were highly enriched in the high-risk group. In addition, we found that the expression of programmed cell death ligand 1 (PD-L1) was significantly elevated in the low-risk group, whereas programmed cell death ligand 2 (PD-L2) and tumor necrosis factor receptor superfamily member 4 (TNFRSF4) were overexpressed in the high-risk group. We successfully built and verified a m7G relevant prognostic model for predicting prognosis and providing a new train of thought for improving the treatment of GC.

A prognostic and immune infiltration analysis of ZEB2 in pan-cancer

[Objective] ZEB2 is known to be participated in several cancers, yet there is no comprehensive analysis about its function in pan-cancer. This study aimed to investigate the role ZEB2 plays in pan-cancer. [Methods] The data was downloaded from the University of California Santa Cruz (UCSC) Xena and the Cancer Genome Atlas (TCGA). The mRNA expression status of ZEB2 was studied in the TCGA_GTEx samples, TCGA samples and paired samples in TCGA, respectively, and protein expression status was obtained from the University of Alabama at Birmingham Cancer data analysis Portal (UALCAN). Kaplan-Meier analysis was applied to 33 kinds of tumors in TCGA, then among the cancers that ZEB2 can affect prognosis, clinical correlation analysis and univariate and multivariate Cox regression analysis were performed. Furthermore, to confirm the prognostic value of ZEB2 in cancers, nomogram models were constructed on lung adenocarcinoma (LUAD) and stomach adenocarcinoma (STAD). The relationship between ZEB2 mRNA expression and immune cell infiltrates was determined by tumor immune estimation resource (TIMER2.0). Protein-protein interaction (PPI) networks were constructed by the STRING. Functional enrichment analysis was also performed using the “ClusterProfiler” package. [Results] ZEB2 was expressed in most tumors, either upregulated or downregulated, and correlated with clinical characteristics in a part of tumors. The expression level of ZEB2 was also found to impact the prognosis of tumors. Additionally, ZEB2 expression was significantly related to immune cells infiltration level in tumor microenvironment. [Conclusion] we propose that ZEB2 has the potential to be a prognostic biomarker and a promising target for cancer immunotherapy.