Using drug response data to identify molecular effectors, and molecular "omic" data to identify candidate drugs in cancer

Selenoprotein GPX1 is a prognostic and chemotherapy-related biomarker for brain lower grade glioma

OBJECTIVE

Glutathione peroxidase 1 (GPX1) is a major selenoprotein in most animal tissues, primarily expressed in the cytoplasm and mitochondria of cells and peroxidase structures of certain cells. GPX1 expression is highly correlated with carcinogenesis and disease progression. The goal of the study was to determine the association between GPX1 expression and tumor therapy, and to identify GPX1 prognostic value in various malignancies.

METHODS

The Cancer Genome Atlas (TCGA), Genotype-Tissue Expression (GTEx), and Human Protein Atlas (HPA) databases were used to detect the levels of GPX1 expression in human tumor tissues and normal tissues. Indeed, correlations between GPX1 and tumor purity, tumor mutation burden (TMB), microsatellite instability (MSI), and DNA mismatch repair genes (MMRs) were explored using the TCGA cohort. Functional and enrichment analyses were performed by the GeneMANIA database and Gene Set Enrichment Analysis (GSEA), respectively. Cox regression models and Kaplan - Meier curves were used to screen for independent risk factors and estimate brain lower-grade glioma (LGG) survival probability. The Chinese Glioma Genome Atlas (CGGA) database was used to determine whether GPX1 had a race-specific effect on overall survival (OS) in LGG. The cross-interaction between GPX1 and chemoradiotherapy on LGG OS was determined by Kaplan - Meier curves. Logistic regression models of multiplicative interactions were constructed. Furthermore, the relationship between GPX1 and LGG treatment regimens was also explored through the Genomics of Drug Sensitivity in Cancer (GDSC) database.

RESULTS

GPX1 was highly expressed in various tumors, GPX1 overexpression was significantly correlated with the poor prognosis of LGG. GPX1 was found to be an independent predictive factor for LGG in both univariate and multivariate Cox models. The nomogram showed a high predictive accuracy (C-index: 0.804, 95% CI: 0.74-0.86). In addition, GPX1 was significantly associated with TMB, MSI, and MMRs in diverse cancers. GPX1 was involved in IL6/JAK/STAT3, inflammatory response, and apoptosis signaling pathways. Besides, non-radiotherapy, chemotherapy, and low GPX1 expression were important factors affecting the better prognosis of LGG. GPX1 acted as a tumor promoter, which has taken the worst effect on LGG survival, but a multiplicative interaction of GPX1*chemoradiotherapy may improve the poor clinical outcome. GPX1 was negatively correlated with the half inhibition concentration (IC50) of temozolomide (TMZ) (Spearman = -0.44, P = 4.52 ×10^-26).

CONCLUSION

In LGG patients, high GPX1 expression was linked to a shorter OS. The interaction between GPX1 and chemoradiotherapy exhibits a beneficial clinical effect and chemotherapy was recommended for LGG patients, especially for those with high GPX1 expression. Besides, high GPX1 expression can predict TMZ sensitivity in LGG, providing potential evidence for chemotherapy. On the whole, this study presents a wealth of biological as well as clinical significance for the roles of GPX1 in human tumors, particularly in LGG.

Identification of phenocopies improves prediction of targeted therapy response over DNA mutations alone

DNA mutations in specific genes can confer preferential benefit from drugs targeting those genes. However, other molecular perturbations can “phenocopy” pathogenic mutations, but would not be identified using standard clinical sequencing, leading to missed opportunities for other patients to benefit from targeted treatments. We hypothesized that RNA phenocopy signatures of key cancer driver gene mutations could improve our ability to predict response to targeted therapies, despite not being directly trained on drug response. To test this, we built gene expression signatures in tissue samples for specific mutations and found that phenocopy signatures broadly increased accuracy of drug response predictions in-vitro compared to DNA mutation alone, and identified additional cancer cell lines that respond well with a positive/negative predictive value on par or better than DNA mutations. We further validated our results across four clinical cohorts. Our results suggest that routine RNA sequencing of tumors to identify phenocopies in addition to standard targeted DNA sequencing would improve our ability to accurately select patients for targeted therapies in the clinic.

Predicting cancer drug TARGETS - TreAtment Response Generalized Elastic-neT Signatures

We are now in an era of molecular medicine, where specific DNA alterations can be used to identify patients who will respond to specific drugs. However, there are only a handful of clinically used predictive biomarkers in oncology. Herein, we describe an approach utilizing in vitro DNA and RNA sequencing and drug response data to create TreAtment Response Generalized Elastic-neT Signatures (TARGETS). We trained TARGETS drug response models using Elastic-Net regression in the publicly available Genomics of Drug Sensitivity in Cancer (GDSC) database. Models were then validated on additional in-vitro data from the Cancer Cell Line Encyclopedia (CCLE), and on clinical samples from The Cancer Genome Atlas (TCGA) and Stand Up to Cancer/Prostate Cancer Foundation West Coast Prostate Cancer Dream Team (WCDT). First, we demonstrated that all TARGETS models successfully predicted treatment response in the separate in-vitro CCLE treatment response dataset. Next, we evaluated all FDA-approved biomarker-based cancer drug indications in TCGA and demonstrated that TARGETS predictions were concordant with established clinical indications. Finally, we performed independent clinical validation in the WCDT and found that the TARGETS AR signaling inhibitors (ARSI) signature successfully predicted clinical treatment response in metastatic castration-resistant prostate cancer with a statistically significant interaction between the TARGETS score and PSA response (p = 0.0252). TARGETS represents a pan-cancer, platform-independent approach to predict response to oncologic therapies and could be used as a tool to better select patients for existing therapies as well as identify new indications for testing in prospective clinical trials.

Identification of biological targets through the correlation between cell line chemosensitivity and protein expression pattern

Matching biological data sequences is one of the most interesting ways to discover new bioactive compounds. In particular, matching cell chemosensitivity with a protein expression profile can be a useful approach to predict the activity of compounds against definite biological targets. In this review, we discuss this correlation. First, we analyze case studies in which some known drugs, acting on known targets, show a good correlation between their antiproliferative activities and protein expression when a large panel of tumor cells is considered. Then, we highlight how the application of in silico methods based on the correlation between cell line chemosensitivity and gene/protein expression patterns might be a quick, cheap, and interesting approach to predict the biological activity of investigated molecules.

Predicting tumor response to drugs based on gene-expression biomarkers of sensitivity learned from cancer cell lines

Background

Human cancer cell line profiling and drug sensitivity studies provide valuable information about the therapeutic potential of drugs and their possible mechanisms of action. The goal of those studies is to translate the findings from in vitro studies of cancer cell lines into in vivo therapeutic relevance and, eventually, patients’ care. Tremendous progress has been made.

Results

In this work, we built predictive models for 453 drugs using data on gene expression and drug sensitivity (IC₅₀) from cancer cell lines. We identified many known drug-gene interactions and uncovered several potentially novel drug-gene associations. Importantly, we further applied these predictive models to ~ 17,000 bulk RNA-seq samples from The Cancer Genome Atlas (TCGA) and the Genotype-Tissue Expression (GTEx) database to predict drug sensitivity for both normal and tumor tissues. We created a web site for users to visualize and download our predicted data (https://manticore.niehs.nih.gov/cancerRxTissue). Using trametinib as an example, we showed that our approach can faithfully recapitulate the known tumor specificity of the drug.

Conclusions

We demonstrated that our approach can predict drugs that 1) are tumor-type specific; 2) elicit higher sensitivity from tumor compared to corresponding normal tissue; 3) elicit differential sensitivity across breast cancer subtypes. If validated, our prediction could have relevance for preclinical drug testing and in phase I clinical design.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12864-021-07581-7.

Development of a Novel Orthotopic Gastric Cancer Mouse Model

BACKGROUND

Gastric cancer metastasis is a highly fatal disease with a five-year survival rate of less than 5%. One major obstacle in studying gastric cancer metastasis is the lack of faithful models available. The cancer xenograft mouse models are widely used to elucidate the mechanisms of cancer development and progression. Current procedures for creating cancer xenografts include both heterotopic (i.e., subcutaneous) and orthotopic transplantation methods. Compared to the heterotopic model, the orthotopic model has been shown to be the more clinically relevant design as it enables the development of cancer metastasis. Although there are several methods in use to develop the orthotopic gastric cancer model, there is not a model which uses various types of tumor materials, such as soft tissues, semi-liquid tissues, or culture derivatives, due to the technical challenges. Thus, developing the applicable orthotopic model which can utilize various tumor materials is essential.

RESULTS

To overcome the known limitations of the current orthotopic gastric cancer models, such as exposure of tumor fragments to the neighboring organs or only using firm tissues for the orthotopic implantation, we have developed a new method allowing for the complete insertion of soft tissue fragments or homogeneously minced tissues into the stomach submucosa layer of the immunodeficient NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NSG) mouse. With this completely-closed transplantation method, tumors with various types of tissue may be used to establish orthotopic gastric cancer models without the risks of exposure to nearby organs or cell leakage. This surgical procedure was highly reproducible in generating forty-eight mouse models with a surgery success rate of 96% and tumor formation of 93%. Among four orthotopic patient-derived xenograft (PDX) models that we generated in this study, we verified that the occurrence of organotropic metastasis in either the liver or peritoneal cavity was the same as that of the donor patients.

CONCLUSION

Here we describe a new protocol, step by step, for the establishment of orthotopic xenograft of gastric cancer. This novel technique will be able to increase the use of orthotopic models in broader applications for not only gastric cancer research but also any research related to the stomach microenvironment.

Insights into DNA-drug interactions in the era of omics

Despite the rise of sophisticated new targeting strategies in cancer chemotherapy, many classic DNA-binding drugs remain on the front line of the therapy against cancer. Based on examples primarily from the author's laboratory, this article reviews the capabilities of several DNA-binding drugs to alter gene expression. Research is ongoing about the molecular bases of the inhibition of gene expression and how alteration of the cellular transcriptome can commit cancer cells to die. The development of a variety of omic techniques allows us to gain insights into the effect of antitumor drugs. Genome-wide approaches provide unbiased genomic data that can facilitate a deeper understanding of the cellular response to DNA-binding drugs. Moreover, the results of large-scale genomic studies are gathered in publicly available databases that can be used in developing precision medicine in cancer treatment.

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

BACKGROUND

Long-term survival in numerous cancers often correlates with specific whole transcriptome profiles or the expression patterns of smaller numbers of transcripts. In some instances, these are better predictors of survival than are standard classification methods such as clinical stage or hormone receptor status in breast cancer. Here, we have used the method of "t-distributed stochastic neighbor embedding" (t-SNE) to show that, collectively, the expression patterns of small numbers of functionally-related transcripts from fifteen cancer pathways correlate with long-term survival in the vast majority of tumor types from The Cancer Genome Atlas (TCGA). We then ask whether the sequential application of t-SNE using the transcripts from a second pathway improves predictive capability or whether t-SNE can be used to refine the initial predictive power of whole transcriptome profiling.

METHODS

RNAseq data from 10,227 tumors in TCGA were previously analyzed using t-SNE-based clustering of 362 transcripts comprising 15 distinct cancer-related pathways. After showing that certain clusters were associated with differential survival, each relevant cluster was re-analyzed by t-SNE with a second pathway's transcripts. Alternatively, groups with differential survival identified by whole transcriptome profiling were subject to a second, t-SNE-based analysis.

RESULTS

Sequential analyses employing either t-SNE➔t-SNE or whole transcriptome profiling➔t-SNE analyses were in many cases superior to either individual method at predicting long-term survival. We developed a dynamic and intuitive R Shiny web application to explore the t-SNE based transcriptome clustering and survival analysis across all TCGA cancers and all 15 cancer-related pathways in this analysis. This application provides a simple interface to select specific t-SNE clusters and analyze survival predictability using both individual or sequential approaches. The user can recreate the relationships described in this analysis and further explore many different cancer, pathway, and cluster combinations. Non-R users can access the application on the web at https://chpupsom19.shinyapps.io/Survival_Analysis_tsne_umap_TCGA. The application, R scripts performing survival analysis, and t-SNE clustering results of TCGA expression data can be accessed on GitHub enabling users to download and run the application locally with ease (https://github.com/RavulaPitt/Sequential-t-SNE/).

CONCLUSIONS

The long-term survival of patients correlated with expression patterns of 362 transcripts from 15 cancer-related pathways. In numerous cases, however, survival could be further improved when the cohorts were re-analyzed using iterative t-SNE clustering or when t-SNE clustering was applied to cohorts initially segregated by whole transcriptome-based hierarchical clustering.

The pro-tumorigenic host response to cancer therapies

Resistance to cancer therapy remains a major challenge in clinical oncology. Although the initial treatment phase is often successful, eventual resistance, characterized by tumour relapse or spread, is discouraging. The majority of studies devoted to investigating the basis of resistance have focused on tumour-related changes that contribute to therapy resistance and tumour aggressiveness. However, over the last decade, the diverse roles of various host cells in promoting therapy resistance have become more appreciated. A growing body of evidence demonstrates that cancer therapy can induce host-mediated local and systemic responses, many of which shift the delicate balance within the tumour microenvironment, ultimately facilitating or supporting tumour progression. In this Review, recent advances in understanding how the host response to different cancer therapies may promote therapy resistance are discussed, with a focus on therapy-induced immunological, angiogenic and metastatic effects. Also summarized is the potential of evaluating the host response to cancer therapy in an era of precision medicine in oncology.

Drivers of topoisomerase II poisoning mimic and complement cytotoxicity in AML cells

Recently approved cancer drugs remain out-of-reach to most patients due to prohibitive costs and only few produce clinically meaningful benefits. An untapped alternative is to enhance the efficacy and safety of existing cancer drugs. We hypothesized that the response to topoisomerase II poisons, a very successful group of cancer drugs, can be improved by considering treatment-associated transcript levels. To this end, we analyzed transcriptomes from Acute Myeloid Leukemia (AML) cell lines treated with the topoisomerase II poison etoposide. Using complementary criteria of co-regulation within networks and of essentiality for cell survival, we identified and functionally confirmed 11 druggable drivers of etoposide cytotoxicity. Drivers with pre-treatment expression predicting etoposide response (e.g., PARP9) generally synergized with etoposide. Drivers repressed by etoposide (e.g., PLK1) displayed standalone cytotoxicity. Drivers, whose modulation evoked etoposide-like gene expression changes (e.g., mTOR), were cytotoxic both alone and in combination with etoposide. In summary, both pre-treatment gene expression and treatment-driven changes contribute to the cell killing effect of etoposide. Such targets can be tweaked to enhance the efficacy of etoposide. This strategy can be used to identify combination partners or even replacements for other classical anticancer drugs, especially those interfering with DNA integrity and transcription.

A pan-cancer study of the transcriptional regulation of uricogenesis in human tumours: pathological and pharmacological correlates

Uric acid (UA) is the end product of the catabolism of purines, and its serum levels are commonly increased in cancer patients. We aimed to explore the transcriptional regulation of tumour uricogenesis in human tumours, and relate uricogenesis with tumour pathological and pharmacological findings. Using data from The Cancer Genome Atlas (TCGA), we analysed the expression levels of xanthine dehydrogenase (XDH) and adenine phosphoribosyltransferase (APRT), two key enzymes in UA production and the purine salvage pathway, respectively. We found large differences between tumour types and individual tumours in their expression of XDH and APRT. Variations in locus-specific DNA methylation and gene copy number correlated with the expression levels of XDH and APRT in human tumours respectively. We explored the consequences of this differential regulation of uricogenesis. Tumours with high levels of XDH mRNA were characterised by higher expression of several genes encoding pro-inflammatory and immune cytokines, and increased levels of tumour infiltration with immune cells. Finally, we studied cancer drug sensitivity using data from the National Cancer Institute-60 (NCI-60) database. A specific correlation was found between the expression levels of APRT and cell sensitivity to the chemotherapeutic agent 5-fluorouracil (5-FU). Our findings underline the existence of great differences in uricogenesis between different types of human tumours. The study of uricogenesis offers promising perspectives for the identification of clinically relevant molecular biomarkers and for tumour stratification in the therapeutic context.

Challenging transcription by DNA-binding antitumor drugs

Cancer has been associated with altered gene expression. Therefore, transcription and its regulation by transcription factors are considered key points to be explored in the pursuit of more efficient antitumor agents. This paper reviews the effects of DNA-binding drugs on the interaction between transcription factors and DNA, and it discusses recent advances in the understanding of the mechanisms by which small compounds interfere with the activity of transcription factors and gene expression. Many DNA-binding drugs, some of them in clinical use, can compete with a variety of transcription factors for their preferred binding sites in gene promoters, or they can covalently modify DNA, thus preventing transcription factors from recognizing their binding sites. On the other hand, transcription factor activity can be impaired through modification of the protein factors or their complexes. Several "omic" tools have been developed to explore the genome-wide changes in gene expression induced by DNA-binding drugs, which reveal details of the mechanisms of action. Transcriptomic profiles obtained from drug-treated cells and of samples collected from patients upon treatment provide insights into the in vivo mechanisms of drug action related to the inhibition of gene transcription. The information available about the molecular structure and mechanisms of action of both transcription factors and DNA-binding drugs, together with the new opportunities provided by functional genomics, should encourage the development of new more-selective DNA-binding antitumor drugs to target a single gene with little effect on others.

Estimating Potency in High-Throughput Screening Experiments by Maximizing the Rate of Change in Weighted Shannon Entropy

High-throughput in vitro screening experiments can be used to generate concentration-response data for large chemical libraries. It is often desirable to estimate the concentration needed to achieve a particular effect, or potency, for each chemical tested in an assay. Potency estimates can be used to directly compare chemical profiles and prioritize compounds for confirmation studies, or employed as input data for prediction modeling and association mapping. The concentration for half-maximal activity derived from the Hill equation model (i.e., AC50) is the most common potency measure applied in pharmacological research and toxicity testing. However, the AC50 parameter is subject to large uncertainty for many concentration-response relationships. In this study we introduce a new measure of potency based on a weighted Shannon entropy measure termed the weighted entropy score (WES). Our potency estimator (Point of Departure, PODWES) is defined as the concentration producing the maximum rate of change in weighted entropy along a concentration-response profile. This approach provides a new tool for potency estimation that does not depend on the assumption of monotonicity or any other pre-specified concentration-response relationship. PODWES estimates potency with greater precision and less bias compared to the conventional AC50 assessed across a range of simulated conditions.

Preclinical mouse cancer models: a maze of opportunities and challenges

Significant advances have been made in developing novel therapeutics for cancer treatment, and targeted therapies have revolutionized the treatment of some cancers. Despite the promise, only about five percent of new cancer drugs are approved, and most fail due to lack of efficacy. The indication is that current preclinical methods are limited in predicting successful outcomes. Such failure exacts enormous cost, both financial and in the quality of human life. This Primer explores the current status, promise, and challenges of preclinical evaluation in advanced mouse cancer models and briefly addresses emerging models for early-stage preclinical development.

Current dichotomy between traditional molecular biological and omic research in cancer biology and pharmacology

There is currently a split within the cancer research community between traditional molecular biological hypothesis-driven and the more recent “omic” forms or research. While the molecular biological approach employs the tried and true single alteration-single response formulations of experimentation, the omic employs broad-based assay or sample collection approaches that generate large volumes of data. How to integrate the benefits of these two approaches in an efficient and productive fashion remains an outstanding issue. Ideally, one would merge the understandability, exactness, simplicity, and testability of the molecular biological approach, with the larger amounts of data, simultaneous consideration of multiple alterations, consideration of genes both of known interest along with the novel, cross-sample comparisons among cell lines and patient samples, and consideration of directed questions while simultaneously gaining exposure to the novel provided by the omic approach. While at the current time integration of the two disciplines remains problematic, attempts to do so are ongoing, and will be necessary for the understanding of the large cell line screens including the Developmental Therapeutics Program’s NCI-60, the Broad Institute’s Cancer Cell Line Encyclopedia, and the Wellcome Trust Sanger Institute’s Cancer Genome Project, as well as the the Cancer Genome Atlas clinical samples project. Going forward there is significant benefit to be had from the integration of the molecular biological and the omic forms or research, with the desired goal being improved translational understanding and application.

rcellminer: exploring molecular profiles and drug response of the NCI-60 cell lines in R

PURPOSE

The rcellminer R package provides a wide range of functionality to help R users access and explore molecular profiling and drug response data for the NCI-60. The package enables flexible programmatic access to CellMiner's unparalleled breadth of NCI-60 data, including gene and protein expression, copy number, whole exome mutations, as well as activity data for ∼21K compounds, with information on their structure, mechanism of action and repeat screens. Functions are available to easily visualize compound structures, activity patterns and molecular feature profiles. Additionally, embedded R Shiny applications allow interactive data exploration.

AVAILABILITY AND IMPLEMENTATION

rcellminer is compatible with R 3.2 and above on Windows, Mac OS X and Linux. The package, documentation, tutorials and Shiny-based applications are available through Bioconductor (http://www.bioconductor.org/packages/rcellminer); ongoing updates will occur according to the Bioconductor release schedule with new CellMiner data. The package is free and open-source (LGPL 3).

CONTACT

lunaa@cbio.mskcc.org or vinodh.rajapakse@nih.gov.

Functional genomics to uncover drug mechanism of action

The upswing in US Food and Drug Administration and European Medicines Agency drug approvals in 2014 may have marked an end to the dry spell that has troubled the pharmaceutical industry over the past decade. Regardless, the attrition rate of drugs in late clinical phases remains high, and a lack of target validation has been highlighted as an explanation. This has led to a resurgence in appreciation of phenotypic drug screens, as these may be more likely to yield compounds with relevant modes of action. However, cell-based screening approaches do not directly reveal cellular targets, and hence target deconvolution and a detailed understanding of drug action are needed for efficient lead optimization and biomarker development. Here, recently developed functional genomics technologies that address this need are reviewed. The approaches pioneered in model organisms, particularly in yeast, and more recently adapted to mammalian systems are discussed. Finally, areas of particular interest and directions for future tool development are highlighted.

Durability of Kinase-Directed Therapies--A Network Perspective on Response and Resistance

Protein kinase-directed cancer therapies yield impressive initial clinical responses, but the benefits are typically transient. Enhancing the durability of clinical response is dependent upon patient selection, using drugs with more effective pharmacology, anticipating mechanisms of drug resistance, and applying concerted drug combinations. Achieving these tenets requires an understanding of the targeted kinase's role in signaling networks, how the network responds to drug perturbation, and patient-to-patient network variations. Protein kinases create sophisticated, malleable signaling networks with fidelity coded into the processes that regulate their presence and function. Robust and reliable signaling is facilitated through network processes (e.g., feedback regulation, and compensatory signaling). The routine use of kinase-directed therapies and advancements in both genomic analysis and tumor cell biology are illuminating the complexity of tumor network biology and its capacity to respond to perturbations. Drug efficacy is attenuated by alterations of the drug target (e.g., steric interference, compensatory activity, and conformational changes), compensatory signaling (bypass mechanisms and phenotype switching), and engagement of other oncogenic capabilities (polygenic disease). Factors influencing anticancer drug response and resistance are examined to define the behavior of kinases in network signaling, mechanisms of drug resistance, drug combinations necessary for durable clinical responses, and strategies to identify mechanisms of drug resistance.