Cancer Therapeutics Response Portal Research Articles

Consistency of in vitro drug sensitivities within pharmacological classes

Multiple comparative analyses between the common drugs and cell lines of the Genomics of Drug Sensitivity in Cancer (GDSC) and the Cancer Therapeutics Response Portal (CTRP) have previously shown low consistency between the in vitro phenotypic measures of a drug in one study with the other. While several potential sources of inconsistency have been tested, the similar targets of tested compounds has yet to be tested as a contributing factor of discrepancy. This analysis includes two methods of reclassifying drugs into classes based on their targets to identify the truer set of consistent cell lines, showing an increased correlation between the two pharmacogenomic studies.

Integrative Transcriptomic Analysis Reveals Distinctive Molecular Traits and Novel Subtypes of Collecting Duct Carcinoma.

Simple SummaryThe treatment of collecting duct carcinoma (CDC) remains unsatisfactory since this highly aggressive kidney cancer has an unfavorable clinical behavior. Due to the rarity of CDC a complete biological characterization of this disease is still lacking. The aim of our study is to provide new insights into the molecular biology of CDC leveraging newly generated and publicly available gene expression profiles of CDC tumors. We identified unique gene expression programs and pathways that distinguished CDC from other renal malignancies. The CDC-specific expression signature predicted in vitro sensitivity to several small molecule inhibitors screened in cancer cell lines from multiple cancer types. Finally, we proved that CDC is a molecularly heterogeneous disease made up of at least two subtypes distinguished by cell signaling, metabolic and immune-related alterations. Altogether, these findings pave the way for future investigations with meaningful clinical implications aimed at improving the management of CDC patients.Collecting duct carcinoma (CDC) is a rare and highly aggressive kidney cancer subtype with poor prognosis and no standard treatments. To date, only a few studies have examined the transcriptomic portrait of CDC. Through integration of multiple datasets, we compared CDC to normal tissue, upper-tract urothelial carcinomas, and other renal cancers, including clear cell, papillary, and chromophobe histologies. Association between CDC gene expression signatures and in vitro drug sensitivity data was evaluated using the Cancer Therapeutic Response Portal, Genomics of Drug Sensitivity in Cancer datasets, and connectivity map. We identified a CDC-specific gene signature that predicted in vitro sensitivity to different targeted agents and was associated to worse outcome in clear cell renal cell carcinoma. We showed that CDC are transcriptionally related to the principal cells of the collecting ducts providing evidence that this tumor originates from this normal kidney cell type. Finally, we proved that CDC is a molecularly heterogeneous disease composed of at least two subtypes distinguished by cell signaling, metabolic and immune-related alterations. Our findings elucidate the molecular features of CDC providing novel biological and clinical insights. The identification of distinct CDC subtypes and their transcriptomic traits provides the rationale for patient stratification and alternative therapeutic approaches.

KESVR: An Ensemble Model for Drug Response Prediction in Precision Medicine Using Cancer Cell Lines Gene Expression.

Background: Cancer cell lines are frequently used in research as in-vitro tumor models. Genomic data and large-scale drug screening have accelerated the right drug selection for cancer patients. Accuracy in drug response prediction is crucial for success. Due to data-type diversity and big data volume, few methods can integrative and efficiently find the principal low-dimensional manifold of the high-dimensional cancer multi-omics data to predict drug response in precision medicine. Method: A novelty k-means Ensemble Support Vector Regression (kESVR) is developed to predict each drug response values for single patient based on cell-line gene expression data. The kESVR is a blend of supervised and unsupervised learning methods and is entirely data driven. It utilizes embedded clustering (Principal Component Analysis and k-means clustering) and local regression (Support Vector Regression) to predict drug response and obtain the global pattern while overcoming missing data and outliers’ noise. Results: We compared the efficiency and accuracy of kESVR to 4 standard machine learning regression models: (1) simple linear regression, (2) support vector regression (3) random forest (quantile regression forest) and (4) back propagation neural network. Our results, which based on drug response across 610 cancer cells from Cancer Cell Line Encyclopedia (CCLE) and Cancer Therapeutics Response Portal (CTRP v2), proved to have the highest accuracy (smallest mean squared error (MSE) measure). We next compared kESVR with existing 17 drug response prediction models based a varied range of methods such as regression, Bayesian inference, matrix factorization and deep learning. After ranking the 18 models based on their accuracy of prediction, kESVR ranks first (best performing) in majority (74%) of the time. As for the remaining (26%) cases, kESVR still ranked in the top five performing models. Conclusion: In this paper we introduce a novel model (kESVR) for drug response prediction using high dimensional cell-line gene expression data. This model outperforms current existing prediction models in terms of prediction accuracy and speed and overcomes overfitting. This can be used in future to develop a robust drug response prediction system for cancer patients using the cancer cell-lines guidance and multi-omics data.

Super.FELT: supervised feature extraction learning using triplet loss for drug response prediction with multi-omics data

BackgroundPredicting the drug response of a patient is important for precision oncology. In recent studies, multi-omics data have been used to improve the prediction accuracy of drug response. Although multi-omics data are good resources for drug response prediction, the large dimension of data tends to hinder performance improvement. In this study, we aimed to develop a new method, which can effectively reduce the large dimension of data, based on the supervised deep learning model for predicting drug response.ResultsWe proposed a novel method called Supervised Feature Extraction Learning using Triplet loss (Super.FELT) for drug response prediction. Super.FELT consists of three stages, namely, feature selection, feature encoding using a supervised method, and binary classification of drug response (sensitive or resistant). We used multi-omics data including mutation, copy number aberration, and gene expression, and these were obtained from cell lines [Genomics of Drug Sensitivity in Cancer (GDSC), Cancer Cell Line Encyclopedia (CCLE), and Cancer Therapeutics Response Portal (CTRP)], patient-derived tumor xenografts (PDX), and The Cancer Genome Atlas (TCGA). GDSC was used for training and cross-validation tests, and CCLE, CTRP, PDX, and TCGA were used for external validation. We performed ablation studies for the three stages and verified that the use of multi-omics data guarantees better performance of drug response prediction. Our results verified that Super.FELT outperformed the other methods at external validation on PDX and TCGA and was good at cross-validation on GDSC and external validation on CCLE and CTRP. In addition, through our experiments, we confirmed that using multi-omics data is useful for external non-cell line data.ConclusionBy separating the three stages, Super.FELT achieved better performance than the other methods. Through our results, we found that it is important to train encoders and a classifier independently, especially for external test on PDX and TCGA. Moreover, although gene expression is the most powerful data on cell line data, multi-omics promises better performance for external validation on non-cell line data than gene expression data. Source codes of Super.FELT are available at https://github.com/DMCB-GIST/Super.FELT.

Brd4-bound enhancers drive cell-intrinsic sex differences in glioblastoma

Sex can be an important determinant of cancer phenotype, and exploring sex-biased tumor biology holds promise for identifying novel therapeutic targets and new approaches to cancer treatment. In an established isogenic murine model of glioblastoma (GBM), we discovered correlated transcriptome-wide sex differences in gene expression, H3K27ac marks, large Brd4-bound enhancer usage, and Brd4 localization to Myc and p53 genomic binding sites. These sex-biased gene expression patterns were also evident in human glioblastoma stem cells (GSCs). These observations led us to hypothesize that Brd4-bound enhancers might underlie sex differences in stem cell function and tumorigenicity in GBM. We found that male and female GBM cells exhibited sex-specific responses to pharmacological or genetic inhibition of Brd4. Brd4 knockdown or pharmacologic inhibition decreased male GBM cell clonogenicity and in vivo tumorigenesis while increasing both in female GBM cells. These results were validated in male and female patient-derived GBM cell lines. Furthermore, analysis of the Cancer Therapeutic Response Portal of human GBM samples segregated by sex revealed that male GBM cells are significantly more sensitive to BET (bromodomain and extraterminal) inhibitors than are female cells. Thus, Brd4 activity is revealed to drive sex differences in stem cell and tumorigenic phenotypes, which can be abrogated by sex-specific responses to BET inhibition. This has important implications for the clinical evaluation and use of BET inhibitors.

Pan-Cancer Analysis of NOS3 Identifies Its Expression and Clinical Relevance in Gastric Cancer.

Background: NOS3 (endothelial NOS, eNOS) is a member of the nitric oxide synthase (NOS) enzyme family, mainly participating in nitric oxide (NO) generation. NOS3 has been reported to inhibit apoptosis and promote angiogenesis, proliferation, and invasiveness. However, the expression pattern of NOS3 and its diagnostic and prognostic potential has not been investigated in a pan-cancer perspective.Methods: Data from the Genotype-Tissue Expression (GTEx), the Cancer Genome Atlas (TCGA), the Cancer Cell Line Encyclopedia (CCLE), and the Cancer Therapeutics Response Portal (CTRP) were employed and NOS3 expression was comprehensively analyzed in normal tissues, cancer tissues, and cell lines. Immunohistochemical staining of tissue sections were used to validate the prognostic role of NOS3 in gastric cancer patients. GSVA and GSEA analyses were performed to investigate signaling pathways related to NOS3 expression.Results: In normal tissues, NOS3 was expressed highest in the spleen and lowest in the blood. NOS3 expression was increased in stomach adenocarcinoma (STAD) and significantly associated with poor prognosis of patients. Immunohistochemical staining validated that NOS3 was an independent prognostic factor of gastric cancer. Several canonical cancer-related pathways were found to be correlated with NOS3 expression in STAD. The expression of NOS3 was related to the response to QS-11 and brivinib in STAD.Conclusions: NOS3 was an independent prognostic factor for patients with STAD. Increased expression of NOS3 influenced occurrence and development of STAD through several canonical cancer-related pathways. Drug response analysis reported drugs to suppress NOS3. NOS3 might be a novel target for gastric cancer treatment.

Bipartite graph-based approach for clustering of cell lines by gene expression-drug response associations.

In pharmacogenomic studies, the biological context of cell lines influences the predictive ability of drug-response models and the discovery of biomarkers. Thus, similar cell lines are often studied together based on prior knowledge of biological annotations. However, this selection approach is not scalable with the number of annotations, and the relationship between gene-drug association patterns and biological context may not be obvious. We present a procedure to compare cell lines based on their gene-drug association patterns. Starting with a grouping of cell lines from biological annotation, we model gene-drug association patterns for each group as a bipartite graph between genes and drugs. This is accomplished by applying sparse canonical correlation analysis (SCCA) to extract the gene-drug associations, and using the canonical vectors to construct the edge weights. Then, we introduce a nuclear norm-based dissimilarity measure to compare the bipartite graphs. Accompanying our procedure is a permutation test to evaluate the significance of similarity of cell line groups in terms of gene-drug associations. In the pharmacogenomic datasets CTRP2, GDSC2 and CCLE, hierarchical clustering of carcinoma groups based on this dissimilarity measure uniquely reveals clustering patterns driven by carcinoma subtype rather than primary site. Next, we show that the top associated drugs or genes from SCCA can be used to characterize the clustering patterns of haematopoietic and lymphoid malignancies. Finally, we confirm by simulation that when drug responses are linearly dependent on expression, our approach is the only one that can effectively infer the true hierarchy compared to existing approaches. Bipartite graph-based hierarchical clustering is implemented in R and can be obtained from CRAN: https://CRAN.R-project.org/package=hierBipartite. The source code is available at https://github.com/CalvinTChi/hierBipartite. The datasets were derived from sources in the public domain, which are the Cancer Cell Line Encyclopedia (https://portals.broadinstitute.org/ccle), the Cancer Therapeutics Response Portal (https://portals.broadinstitute.org/ctrp.v2.1/?page=#ctd2BodyHome), and the Genomics of Drug Sensitivity in Cancer (https://www.cancerrxgene.org/). These datasets can be downloaded using the PharmacoGx R package (https://bioconductor.org/packages/release/bioc/html/PharmacoGx.html). Supplementary data are available at Bioinformatics online.

Identification of Predictive ERBB Mutations by Leveraging Publicly Available Cell Line Databases

Although targeted therapies can be effective for a subgroup of patients, identification of individuals who benefit from the treatments is challenging. At the same time, the predictive significance of the majority of the thousands of mutations observed in the cancer tissues remains unknown. Here, we describe the identification of novel predictive biomarkers for ERBB-targeted tyrosine kinase inhibitors (TKIs) by leveraging the genetic and drug screening data available in the public cell line databases: Cancer Cell Line Encyclopedia, Genomics of Drug Sensitivity in Cancer, and Cancer Therapeutics Response Portal. We assessed the potential of 412 ERBB mutations in 296 cell lines to predict responses to 10 different ERBB-targeted TKIs. Seventy-six ERBB mutations were identified that were associated with ERBB TKI sensitivity comparable with non-small cell lung cancer cell lines harboring the well-established predictive EGFR L858R mutation or exon 19 deletions. Fourteen (18.4%) of these mutations were classified as oncogenic by the cBioPortal database, whereas 62 (81.6%) were regarded as novel potentially predictive mutations. Of the nine functionally validated novel mutations, EGFR Y1069C and ERBB2 E936K were transforming in Ba/F3 cells and demonstrated enhanced signaling activity. Mechanistically, the EGFR Y1069C mutation disrupted the binding of the ubiquitin ligase c-CBL to EGFR, whereas the ERBB2 E936K mutation selectively enhanced the activity of ERBB heterodimers. These findings indicate that integrating data from publicly available cell line databases can be used to identify novel, predictive nonhotspot mutations, potentially expanding the patient population benefiting from existing cancer therapies.

Identification of tumor mutation burden-related hub genes and the underlying mechanism in melanoma.

Background: Tumor mutation burden (TMB) has emerged as an important predictive factor for drug resistance in cancers; however, the specific mechanism underlying TMB function in melanoma remains elusive.Methods: Data on somatic mutations, RNA sequencing (RNA-seq), miRNA sequencing (miRNA-seq), and clinical characteristics for 472 melanoma patients were extracted from the TCGA cohort. RNA-seq data of melanoma cell lines were obtained from the Cancer Cell Line Encyclopedia, and sensitivity of cell lines to therapeutic agents is available in the Cancer Therapeutics Response Portal. TMB was calculated based on somatic mutation data. Differentially expressed gene analysis, weighted gene co-expression network analysis, protein-protein interaction networks, Minimal Common Oncology Data Elements, and survival analysis were leveraged to determine TMB-related hub genes. Competing endogenous RNA (ceRNA) networks were constructed to explore the molecular mechanisms underlying hub gene function. The influence of key genes on drug sensitivity was analyzed to investigate their clinical significance.Results: Elevated TMB levels were significantly correlated with improved survival outcomes. In addition, six tumor-infiltrating immune cells, including naive B cells, regulatory T cells, memory resting CD4 T cells, memory B cells, activated mast cells, and resting NK cells, were significantly overexpressed in the low-TMB group relative to the high-TMB group. Furthermore, we identified FLNC, NEXN, and TNNT3 as TMB-related hub genes, and constructed their ceRNA networks, including five miRNAs (has-miR-590-3p, has-miR-374b-5p, has-miR-3127-5p, has-miR-1913, and has-miR-1291) and 31 lncRNAs (FAM66C, MIAT, NR2F2AS1, etc.). Finally, we observed that TMB-related genes were associated with distinct therapeutic responses to AKT/mTOR pathway inhibitors.Conclusions: We identified three TMB-associated key genes, established their ceRNA networks, and investigated their influence on therapeutic responses, which could provide insights into future precision medicine.

Comprehensive bioinformatics analysis of the TP53 signaling pathway in Wilms' tumor.

BackgroundDifferential expression of tumor protein 53 (TP53, or p53) has been observed in multiple cancers. However, the expression levels and prognostic role of TP53 signaling pathway genes in Wilms’ tumor (WT) have yet to be fully explored.MethodsThe expression levels of TP53 signaling pathway genes including TP53, mouse double minute 2 (MDM2), mouse double minute 4 (MDM4), cyclin-dependent kinase 2A (CDKN2A), cyclin-dependent kinase 2B (CDKN2B), and tumor suppressor p53-binding protein 1 (TP53BP1) in WT were analyzed using the Oncomine database. Aberration types, co-mutations, mutation locations, signaling pathways, and the prognostic role of TP53 in WT were investigated using cBioPortal. MicroRNA (miRNA) and transcription factor (TF) targets were identified with miRTarBase, miWalk, and ChIP-X Enrichment Analysis 3 (CheA3), respectively. A protein-protein network was constructed using GeneMANIA. The expression of TP53 signaling genes were confirmed in WT samples and normal kidney tissues using the Human Protein Atlas (HPA). Cancer Therapeutics Response Portal (CTRP) was used to analyze the small molecules potentially targeting TP53.ResultsTP53 was significantly expressed in the Cutcliffe Renal (P=0.010), but not in the Yusenko Renal (P=0.094). Meanwhile, MDM2 was significantly overexpressed in the Yusenko Renal (P=0.058), but not in the Cutcliffe Renal (P=0.058). The expression levels of MDM4 no significant difference between the tumor and normal tissue samples. The most common TP53 alteration was missense and the proportion of TP53 pathway-related mutations was 2.3%. Co-expressed genes included ZNF609 (zinc finger protein 609), WRAP53 (WD40-encoding RNA antisense to p53), CNOT2 (CC chemokine receptor 4-negative regulator of transcription 2), and CDH13 (cadherin 13). TP53 alterations indicated poor prognosis of WT (P=1.051e-4). The regulators of the TP53 pathway included miR-485-5p and TFs NR2F2 and KDM5B. The functions of TP53 signaling pathway were signal transduction in response to DNA damage and regulate the cell cycle. The small molecules targeting TP53 included PRIMA-1, RITA, SJ-172550, and SCH-529074.ConclusionsTP53 was found to be differentially expressed in WT tissues. TP53 mutations indicated poor outcomes of WT. Therefore, pifithrin-mu, PRIMA-1, RITA, SJ-172550, and SCH-529074 could be used in combination with traditional chemotherapy to treat WT.

Transcriptome Profiling of Acquired Gefitinib Resistant Lung Cancer Cells Reveals Dramatically Changed Transcription Programs and New Treatment Targets.

Background: Targeted therapy for lung cancer with epidermal growth factor receptor (EGFR) mutations with tyrosine kinase inhibitors (TKIs) represents one of the major breakthroughs in lung cancer management. However, gradually developed resistance to these drugs prevents sustained clinical benefits and calls for resistant mechanism research and identification of new therapeutic targets. Acquired T790M mutation accounts for the majority of resistance cases, yet transcriptome changes in these cells are less characterized, and it is not known if new treatment targets exist by available drugs.Methods: Transcriptome profiling was performed for lung cancer cell line PC9 and its resistant line PC9GR after long-term exposure to gefitinib through RNA sequencing. Differentially expressed genes and changed pathways were identified along with existing drugs targeting these upregulated genes. Using 144 lung cancer cell lines with both gene expression and drug response data from the cancer cell line encyclopedia (CCLE) and Cancer Therapeutics Response Portal (CTRP), we screened 549 drugs whose response was correlated with these upregulated genes in PC9GR cells, and top drugs were evaluated for their response in both PC9 and PC9GR cells.Results: In addition to the acquired T790M mutation, the resistant PC9GR cells had very different transcription programs from the sensitive PC9 cells. Multiple pathways were changed with the top ones including TNFA signaling, androgen/estrogen response, P53 pathway, MTORC1 signaling, hypoxia, and epithelial mesenchymal transition. Thirty-two upregulated genes had available drugs that can potentially be effective in treating the resistant cells. From the response profiles of CCLE, we found 17 drugs whose responses were associated with at least four of these upregulated genes. Among the four drugs evaluated (dasatinib, KPT-185, trametinib, and pluripotin), all except trametinib demonstrated strong inhibitory effects on the resistant PC9GR cells, among which KPT185 was the most potent. KPT-185 suppressed growth, caused apoptosis, and inhibited migration of the PC9GR cells at similar (or better) rates as the sensitive PC9 cells in a dose-dependent manner.Conclusions: Acquired TKI-resistant lung cancer cells (PC9GR) have dramatically changed transcription and pathway regulation, which expose new treatment targets. Existing drugs may be repurposed to treat those patients with developed resistance to TKIs.

Abstract 4846: A quantitative proteo-genomic analysis for targeted therapies in the management of bladder cancer survival

Abstract Bladder cancer is the most common type of urothelial carcinoma (UC), and it represents about 5% of all new cancers in the United States. The diagnostics, treatment, and five-year survival rates of this cancer have seen no significant improvements for nearly last three decades. Due to the lack of development of new treatment options, bladder cancer poses a bigger threat than many other cancers. In this current study, we have analyzed over 400 patients' proteo-genomic sequencing and expression data from The Cancer Genome Atlas (TCGA) project using the cBioportal platform. We found two genes, TP53 and CDKN2A, which were significantly altered in the bladder cancer patient samples. The mutations and copy number alteration (CNA) of TP53 were present in more than 50% of patient samples, while CDKN2A was deleted in approximately 40% of patient samples. However, we found that patients with CDKN2A deletion significantly contributed to patient survival compared to TP53 mutational inactivation. Based on this finding, we further analyzed the survival associated genetic alterations in only those patient samples that have the CDK2NA deletion. Analyses indicated that bladder cancer patients with the CDKN2A gene deletion have differential expression of PIK3CA, RPS6KB1, CCND1, and TFRC and worse prognosis. Next, we analyzed protein level expression data using MD Anderson cell line project (MCLP portal) database and found that phospho-RPS6KB1 (p70-S6K_pT389) and CCND1 (Cyclin D1) were the least expressed proteins in bladder cancer cell lines compared to over fifteen different types of cancer cell lines. Additionally, data analyses of The Cancer Therapeutics Response Portal (CTRP) and MCLP database, we found two drugs- hyperforin and CD437 which have the highest Spearman's Rho correlation with CCND1 and RPS6KB1, respectively, with significantly low p-values. Importantly, these drugs sensitize the cell lines that have low CCND1 and RPS6KB1, respectively, and would likely be beneficial for patient survival. Next, we validated our findings in three bladder cancer cell lines, T24, 5637, and HT1376. We monitored the expression of the above proteins in these bladder cancer cell-lines and compared them with other cancer cell lines. Experimental data found a strong agreement with our high throughput cell-line and patient data analysis. Besides, both hyperforin and CD437 drugs exhibit a strong cytotoxic effect on all three bladder cell lines. Furthermore, hyperforin and CD437 were found to increase the expression of CDKN2A in addition to increasing CCND1 and RPS6KB1, respectively. In summary, we characterized four different genes PIK3CA, RPS6KB1, CCND1, and TFRC, which are associated with CDKN2A mediated survival in bladder cancer. Our findings also provide a novel therapeutic approach for bladder cancer patients. Funding: This work was supported by the DoD grant to Dr. Srivastava (DAMD17-03-1-0107) and by CHIRP funding (CHIRP ID# IAA-A-HL-14-007). Disclaimers: The contents of this publication are the sole responsibility of the author(s) and do not necessarily reflect the views, opinions or policies of The Uniformed Services University of the Health Sciences (USUHS), The Henry M. Jackson Foundation (HJF), the Department of Defense (DoD), the Departments of the Army, Navy, or Air Force. *Correspondence: meera.srivastava@usuhs.edu OR alakesh.bera.ctr@usuhs.edu. Citation Format: Sreejato Chatterjee, John Karaian, Madhan Subramanian, Eric Russ, Michael Eklund, Harvey B. Pollard, Meera Srivastava, Alakesh Bera. A quantitative proteo-genomic analysis for targeted therapies in the management of bladder cancer survival [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 4846.

Abstract 33: Drug sensitivity prediction modeling from genomics, transcriptomics and inferred protein activity

Abstract Background: Machine learning models that rely on single omics data for drug sensitivity prediction are challenging and frequently fail within precision medicine scenarios. Proteomics, for example, reflects the system biology and the regulatory network better than genomics. However, it is less likely available for many preclinical and in vivo data, mainly due to the cost of data generation. Currently, there are massive amounts of genomics data available, such as CNV, mutation, and RNA-seq signatures. However, these data do not characterize the post-translational modifications in proteins, limiting their utility for biomarker discovery. Objective: The main objective of this study is to overcome the lack of proteomics data gap. Herein, we propose a novel modeling approach to improve the prediction accuracy of drug sensitivity, that is, combining genomics and proteomics signatures. In addition to genomics and transcriptomics data, the model infers proteomic activity from gene expression using VIPER from in vitro data, and this is extendable to in vivo settings. Material and Methods: Using PharmacoGX package developed in our lab, we downloaded, curated, and annotated the genomic and pharmacologic data of the Cancer Cell line Encyclopedia (CCLE), as well as the Cancer Therapeutics Response Portal (CTRPV2) dataset that is a continuation of the CTRP project and the largest pharmacologic screen conducted to date, containing several hundreds of thousands of drug dose-response curves. In this study, we extracted from CCLE dataset the following signatures: RPPA, RNASeq, Mutation, and CNV, then inferred the VIPER protein using RNASeq. Then, we built a model that a) checks the different omics combinations, b) applies random forest with ten-fold cross-validation for sensitivity prediction using CTRPV2, and c) evaluates the significance of each model using the concordance index (CI) package developed in our lab. Results: The proposed model was tested in vitro using CCLE and CTRPv2 common cell lines. We tested the model with drugs having known biomarkers in pharmacogenomics literature. ERBB2 biomarker for lapatinib showed the best CI=0.95 by combining CNV and VIPER models while CI equals 0.77 and 0.9 for each of them, respectively. Moreover, MET biomarker for crizotinib showed best CI=0.89 by integrating RNASeq, Mutation, RPPA, and VIPER, while each model obtained CI=(0.45, 0.5, 0.85, 0.5), respectively. Conclusion: In conclusion, omics integration boosted the drug sensitivity prediction compared to single models. The application of the proposed model in vivo will improve drug development and increase the prediction quality of precision medicine. Citation Format: Hassan Mahmoud, Benjamin Haibe-Kains. Drug sensitivity prediction modeling from genomics, transcriptomics and inferred protein activity [abstract]. In: Proceedings of the AACR Special Conference on Advancing Precision Medicine Drug Development: Incorporation of Real-World Data and Other Novel Strategies; Jan 9-12, 2020; San Diego, CA. Philadelphia (PA): AACR; Clin Cancer Res 2020;26(12_Suppl_1):Abstract nr 33.

Computational Analysis of Transcriptomic and Proteomic Data for Deciphering Molecular Heterogeneity and Drug Responsiveness in Model Human Hepatocellular Carcinoma Cell Lines.

Hepatocellular carcinoma (HCC) is associated with high mortality due to its inherent heterogeneity, aggressiveness, and limited therapeutic regimes. Herein, we analyzed 21 human HCC cell lines (HCC lines) to explore intertumor molecular diversity and pertinent drug sensitivity. We used an integrative computational approach based on exploratory and single-sample gene-set enrichment analysis of transcriptome and proteome data from the Cancer Cell Line Encyclopedia, followed by correlation analysis of drug-screening data from the Cancer Therapeutics Response Portal with curated gene-set enrichment scores. Acquired results classified HCC lines into two groups, a poorly and a well-differentiated group, displaying lower/higher enrichment scores in a “Specifically Upregulated in Liver” gene-set, respectively. Hierarchical clustering based on a published epithelial–mesenchymal transition gene expression signature further supported this stratification. Between-group comparisons of gene and protein expression unveiled distinctive patterns, whereas downstream functional analysis significantly associated differentially expressed genes with crucial cancer-related biological processes/pathways and revealed concrete driver-gene signatures. Finally, correlation analysis highlighted a diverse effectiveness of specific drugs against poorly compared to well-differentiated HCC lines, possibly applicable in clinical research with patients with analogous characteristics. Overall, this study expanded the knowledge on the molecular profiles, differentiation status, and drug responsiveness of HCC lines, and proposes a cost-effective computational approach to precision anti-HCC therapies.

Significance of STAT3 in Immune Infiltration and Drug Response in Cancer.

Signal transducer and activator of transcription 3 (STAT3) is a transcription factor and regulates tumorigenesis. However, the functions of STAT3 in immune and drug response in cancer remain elusive. Hence, we aim to reveal the impact of STAT3 in immune infiltration and drug response comprehensively by bioinformatics analysis. The expression of STAT3 and its relationship with tumor stage were explored by Tumor Immune Estimation Resource (TIMER), Human Protein Altas (HPA), and UALCAN databases. The correlations between STAT3 and immune infiltration, gene markers of immune cells were analyzed by TIMER. Moreover, the association between STAT3 and drug response was evaluated by the Cancer Cell Line Encyclopedia (CCLE) and Cancer Therapeutics Response Portal (CTRP). The results suggested that the mRNA transcriptional level of STAT3 was lower in tumors than normal tissues and mostly unrelated to tumor stage. Besides, the protein expression of STAT3 decreased in colorectal and renal cancer compared with normal tissues. Importantly, STAT3 was correlated with immune infiltration and particularly regulated tumor-associated macrophage (TAM), M2 macrophage, T-helper 1 (Th1), follicular helper T (Treg), and exhausted T-cells. Remarkably, STAT3 was closely correlated with the response to specified inhibitors and natural compounds in cancer. Furthermore, the association between STAT3 and drug response was highly cell line type dependent. Significantly, the study provides thorough insight that STAT3 is associated with immunosuppression, as well as drug response in clinical treatment.

YM155-Adapted Cancer Cell Lines Reveal Drug-Induced Heterogeneity and Enable the Identification of Biomarker Candidates for the Acquired Resistance Setting.

Survivin is a drug target and its suppressant YM155 a drug candidate mainly investigated for high-risk neuroblastoma. Findings from one YM155-adapted subline of the neuroblastoma cell line UKF-NB-3 had suggested that increased ABCB1 (mediates YM155 efflux) levels, decreased SLC35F2 (mediates YM155 uptake) levels, decreased survivin levels, and TP53 mutations indicate YM155 resistance. Here, the investigation of 10 additional YM155-adapted UKF-NB-3 sublines only confirmed the roles of ABCB1 and SLC35F2. However, cellular ABCB1 and SLC35F2 levels did not indicate YM155 sensitivity in YM155-naïve cells, as indicated by drug response data derived from the Cancer Therapeutics Response Portal (CTRP) and the Genomics of Drug Sensitivity in Cancer (GDSC) databases. Moreover, the resistant sublines were characterized by a remarkable heterogeneity. Only seven sublines developed on-target resistance as indicated by resistance to RNAi-mediated survivin depletion. The sublines also varied in their response to other anti-cancer drugs. In conclusion, cancer cell populations of limited intrinsic heterogeneity can develop various resistance phenotypes in response to treatment. Therefore, individualized therapies will require monitoring of cancer cell evolution in response to treatment. Moreover, biomarkers can indicate resistance formation in the acquired resistance setting, even when they are not predictive in the intrinsic resistance setting.

Expression Levels of Therapeutic Targets as Indicators of Sensitivity to Targeted Therapeutics

Cancer precision medicine aims to predict the drug likely to yield the best response for a patient. Genomic sequencing of tumors is currently being used to better inform treatment options; however, this approach has had a limited clinical impact due to the paucity of actionable mutations. An alternative to mutation status is the use of gene expression signatures to predict response. Using data from two large-scale studies, The Genomics of Drug Sensitivity of Cancer (GDSC) and The Cancer Therapeutics Response Portal (CTRP), we investigated the relationship between the sensitivity of hundreds of cell lines to hundreds of drugs, and the relative expression levels of the targets these drugs are directed against. For approximately one third of the drugs considered (73/222 in GDSC and 131/360 in CTRP), sensitivity was significantly correlated with the expression of at least one of the known targets. Surprisingly, for 8% of the annotated targets, there was a significant anticorrelation between target expression and sensitivity. For several cases, this corresponded to drugs targeting multiple genes in the same family, with the expression of one target significantly correlated with sensitivity and another significantly anticorrelated suggesting a possible role in resistance. Furthermore, we identified nontarget genes that are significantly correlated or anticorrelated with drug sensitivity, and find literature linking several to sensitization and resistance. Our analyses provide novel and important insights into both potential mechanisms of resistance and relative efficacy of drugs against the same target.

Tumor cell sensitivity to vemurafenib can be predicted from protein expression in a BRAF-V600E basket trial setting

BackgroundGenetics-based basket trials have emerged to test targeted therapeutics across multiple cancer types. However, while vemurafenib is FDA-approved for BRAF-V600E melanomas, the non-melanoma basket trial was unsuccessful, suggesting mutation status is insufficient to predict response. We hypothesized that proteomic data would complement mutation status to identify vemurafenib-sensitive tumors and effective co-treatments for BRAF-V600E tumors with inherent resistance.MethodsReverse Phase Proteomic Array (RPPA, MD Anderson Cell Lines Project), RNAseq (Cancer Cell Line Encyclopedia) and vemurafenib sensitivity (Cancer Therapeutic Response Portal) data for BRAF-V600E cancer cell lines were curated. Linear and nonlinear regression models using RPPA protein or RNAseq were evaluated and compared based on their ability to predict BRAF-V600E cell line sensitivity (area under the dose response curve). Accuracies of all models were evaluated using hold-out testing. CausalPath software was used to identify protein-protein interaction networks that could explain differential protein expression in resistant cells. Human examination of features employed by the model, the identified protein interaction networks, and model simulation suggested anti-ErbB co-therapy would counter intrinsic resistance to vemurafenib. To validate this potential co-therapy, cell lines were treated with vemurafenib and dacomitinib (a pan-ErbB inhibitor) and the number of viable cells was measured.ResultsOrthogonal partial least squares (O-PLS) predicted vemurafenib sensitivity with greater accuracy in both melanoma and non-melanoma BRAF-V600E cell lines than other leading machine learning methods, specifically Random Forests, Support Vector Regression (linear and quadratic kernels) and LASSO-penalized regression. Additionally, use of transcriptomic in place of proteomic data weakened model performance. Model analysis revealed that resistant lines had elevated expression and activation of ErbB receptors, suggesting ErbB inhibition could improve vemurafenib response. As predicted, experimental evaluation of vemurafenib plus dacomitinb demonstrated improved efficacy relative to monotherapies.Conclusions: Combined, our results support that inclusion of proteomics can predict drug response and identify co-therapies in a basket setting.

Abstract 2488: CellMinerCDB and CellMiner web-applications for genomics and pharmacogenomics analyses of cancer cell lines

Abstract The CellMiner (http://discover.nci.nih.gov/cellminer)and CellMinerCDB (https://discover.nci.nih.gov/cellminercdb/) web-applications provide complimentary functionality and datatsets that may be used for comparisons of genomic, molecular and pharmacological data within the NCI-60 cancerous cell lines, Cancer Cell Line Encyclopedia (CCLE), Genomics of Drug Sensitivity in Cancer (GDSC), Cancer Therapeutics Response Portal (CTRP), NCI/DTP small cell lung cancer (SCLC), and NCI Almanac cell line sets. CellMiner contains data for the NCI-60, including the most extensive sets of molecular and drug activity data (generated by the NCI Developmental Therapeutics Program https://dtp.cancer.gov),found for any of the databases. CellMinerCDB contains all the above mentioned cell line sets, including the substantially increased cell line numbers and tissue of origin types found in the CCLE, GDSC, and CTRP. The two web-applications have separate but complimentary functionalities. Each cell line set has some variable number of data types, some of which measure the same parameters, and some that do not. The number and make up of cell lines also varies, from 60 for the NCI-60, 69 for the SCLC, and ~1000 for the CCLE, GDSC, and CTRP. As there are partial overlaps of cell lines between many of these cell line sets, one may fill in some data type gaps by merging data from two sources, as well as do quality control by comparisons of the same data from multiple institutions. This rich set of data and functions facilitates the exploration of the relationships between and among molecular alterations and pharmacological responses in cancer cell lines from the omic perspective. Citation Format: William C. Reinhold, Sudir Varma, Yang-Hsin Wang, Fathi Elloumi, Yves Pommier. CellMinerCDB and CellMiner web-applications for genomics and pharmacogenomics analyses of cancer cell lines [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 2488.

Abstract SY42-01: Chemical biology-based approach to understanding and overcoming resistance of cancers to therapies

Abstract The ability of cancer cells to resist apoptotic death induced by radiation, chemotherapy, targeted therapies and immunotherapy has been a primary impediment to achieving long-lasting clinical responses. Cancer resistance is often studied through the lens of genetics, and thus on a drug-by-drug basis. I will discuss an alternative approach - studying resistance through the lens of cellular plasticity. As part of the NCI’s Cancer Target Discovery & Development Network, we use the Cancer Therapeutics Response Portal (CTRP), which enables small molecule sensitivity-cancer cell feature relationships to be explored on a broad basis. One insight gained from this resource suggests that a similar form of plasticity appears recurrently in multiple cancer contexts, and in response to multiple modalities (chemotherapy; targeted therapy; immunotherapy) thus opening the possibility of a pan-cancer explanation of resistance. Here, cell plasticity provides access to an apoptosis-deficient state and thus resistance, and then subsequent cancer mutations, rather than being the basis of resistance, are needed to optimize the fitness of the resistant state for growth. We found that this myofibroblastic-like cell state has profound changes in lipid biology that result in a novel vulnerability to ferroptotic cell death and a dependency on the activity of GPX4, a potentially druggable lipid hydroperoxidas. We hope this research will provide a blueprint for the cure of cancers in the future. Citation Format: Stuart L. Schreiber. Chemical biology-based approach to understanding and overcoming resistance of cancers to therapies [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr SY42-01.