Harnessing Connectivity in a Large-Scale Small-Molecule Sensitivity Dataset

SRDFM: Siamese Response Deep Factorization Machine to improve anti-cancer drug recommendation

Predicting the response of cancer patients to a particular treatment is a major goal of modern oncology and an important step toward personalized treatment. In the practical clinics, the clinicians prefer to obtain the most-suited drugs for a particular patient instead of knowing the exact values of drug sensitivity. Instead of predicting the exact value of drug response, we proposed a deep learning-based method, named Siamese Response Deep Factorization Machines (SRDFM) Network, for personalized anti-cancer drug recommendation, which directly ranks the drugs and provides the most effective drugs. A Siamese network (SN), a type of deep learning network that is composed of identical subnetworks that share the same architecture, parameters and weights, was used to measure the relative position (RP) between drugs for each cell line. Through minimizing the difference between the real RP and the predicted RP, an optimal SN model was established to provide the rank for all the candidate drugs. Specifically, the subnetwork in each side of the SN consists of a feature generation level and a predictor construction level. On the feature generation level, both drug property and gene expression, were adopted to build a concatenated feature vector, which even enables the recommendation for newly designed drugs with only chemical property known. Particularly, we developed a response unit here to generate weighted genetic feature vector to simulate the biological interaction mechanism between a specific drug and the genes. For the predictor construction level, we built this level integrating a factorization machine (FM) component with a deep neural network component. The FM can well handle the discrete chemical information and both low-order and high-order feature interactions could be sufficiently learned. Impressively, the SRDFM works well on both single-drug recommendation and synergic drug combination. Experiment result on both single-drug and synergetic drug data sets have shown the efficiency of the SRDFM. The Python implementation for the proposed SRDFM is available at at https://github.com/RanSuLab/SRDFM Contact: ran.su@tju.edu.cn, gbx@mju.edu.cn and weileyi@sdu.edu.cn.

A functional module states framework reveals transcriptional states for drug and target prediction

Cells are complex systems in which many functions are performed by different genetically defined and encoded functional modules. To systematically understand how these modules respond to drug or genetic perturbations, we develop a functional module states framework. Using this framework, we (1) define the drug-induced transcriptional state space for breast cancer cell lines using large public gene expression datasets and reveal that the transcriptional states are associated with drug concentration and drug targets, (2) identify potential targetable vulnerabilities through integrative analysis of transcriptional states after drug treatment and gene knockdown-associated cancer dependency, and (3) use functional module states to predict transcriptional state-dependent drug sensitivity and build prediction models for drug response. This approach demonstrates a similar prediction performance as approaches using high-dimensional gene expression values, with the added advantage of more clearly revealing biologically relevant transcriptional states and key regulators.

A community challenge for a pancancer drug mechanism of action inference from perturbational profile data

The Columbia Cancer Target Discovery and Development (CTD2) Center is developing PANACEA, a resource comprising dose-responses and RNA sequencing (RNA-seq) profiles of 25 cell lines perturbed with ∼400 clinical oncology drugs, to study a tumor-specific drug mechanism of action. Here, this resource serves as the basis for a DREAM Challenge assessing the accuracy and sensitivity of computational algorithms for de novo drug polypharmacology predictions. Dose-response and perturbational profiles for 32 kinase inhibitors are provided to 21 teams who are blind to the identity of the compounds. The teams are asked to predict high-affinity binding targets of each compound among ∼1,300 targets cataloged in DrugBank. The best performing methods leverage gene expression profile similarity analysis as well as deep-learning methodologies trained on individual datasets. This study lays the foundation for future integrative analyses of pharmacogenomic data, reconciliation of polypharmacology effects in different tumor contexts, and insights into network-based assessments of drug mechanisms of action.

•

Drug-perturbed RNA sequencing data can be used to identify drug targets

•

Technology-based drug-target definitions often subsume literature definitions

•

Literature and screening datasets provide complementary information on drug mechanisms

A cross-study analysis of drug response prediction in cancer cell lines

To enable personalized cancer treatment, machine learning models have been developed to predict drug response as a function of tumor and drug features. However, most algorithm development efforts have relied on cross-validation within a single study to assess model accuracy. While an essential first step, cross-validation within a biological data set typically provides an overly optimistic estimate of the prediction performance on independent test sets. To provide a more rigorous assessment of model generalizability between different studies, we use machine learning to analyze five publicly available cell line-based data sets: National Cancer Institute 60, ancer Therapeutics Response Portal (CTRP), Genomics of Drug Sensitivity in Cancer, Cancer Cell Line Encyclopedia and Genentech Cell Line Screening Initiative (gCSI). Based on observed experimental variability across studies, we explore estimates of prediction upper bounds. We report performance results of a variety of machine learning models, with a multitasking deep neural network achieving the best cross-study generalizability. By multiple measures, models trained on CTRP yield the most accurate predictions on the remaining testing data, and gCSI is the most predictable among the cell line data sets included in this study. With these experiments and further simulations on partial data, two lessons emerge: (1) differences in viability assays can limit model generalizability across studies and (2) drug diversity, more than tumor diversity, is crucial for raising model generalizability in preclinical screening.

An overview of machine learning methods for monotherapy drug response prediction

For an increasing number of preclinical samples, both detailed molecular profiles and their responses to various drugs are becoming available. Efforts to understand, and predict, drug responses in a data-driven manner have led to a proliferation of machine learning (ML) methods, with the longer term ambition of predicting clinical drug responses. Here, we provide a uniquely wide and deep systematic review of the rapidly evolving literature on monotherapy drug response prediction, with a systematic characterization and classification that comprises more than 70 ML methods in 13 subclasses, their input and output data types, modes of evaluation, and code and software availability. ML experts are provided with a fundamental understanding of the biological problem, and how ML methods are configured for it. Biologists and biomedical researchers are introduced to the basic principles of applicable ML methods, and their application to the problem of drug response prediction. We also provide systematic overviews of commonly used data sources used for training and evaluation methods.

Glioblastoma stem cells reprogram chromatin in vivo to generate selective therapeutic dependencies on DPY30 and phosphodiesterases

Glioblastomas are universally fatal cancers and contain self-renewing glioblastoma stem cells (GSCs) that initiate tumors. Traditional anticancer drug discovery based on in vitro cultures tends to identify targets with poor therapeutic indices and fails to accurately model the effects of the tumor microenvironment. Here, leveraging in vivo genetic screening, we identified the histone H3 lysine 4 trimethylation (H3K4me3) regulator DPY30 (Dpy-30 histone methyltransferase complex regulatory subunit) as an in vivo–specific glioblastoma dependency. On the basis of the hypothesis that in vivo epigenetic regulation may define critical GSC dependencies, we interrogated active chromatin landscapes of GSCs derived from intracranial patient-derived xenografts (PDXs) and cell culture through H3K4me3 chromatin immunoprecipitation and transcriptome analyses. Intracranial-specific genes marked by H3K4me3 included FOS, NFκB, and phosphodiesterase (PDE) family members. In intracranial PDX tumors, DPY30 regulated angiogenesis and hypoxia pathways in an H3K4me3-dependent manner but was dispensable in vitro in cultured GSCs. PDE4B was a key downstream effector of DPY30, and the PDE4 inhibitor rolipram preferentially targeted DPY30-expressing cells and impaired PDX tumor growth in mice without affecting tumor cells cultured in vitro. Collectively, the MLL/SET1 (mixed lineage leukemia/SET domain-containing 1, histone lysine methyltransferase) complex member DPY30 selectively regulates H3K4me3 modification on genes critical to support angiogenesis and tumor growth in vivo, suggesting the DPY30-PDE4B axis as a specific therapeutic target in glioblastoma.

Integrative clinical and molecular characterization of translocation renal cell carcinoma

Translocation renal cell carcinoma (tRCC) is a poorly characterized subtype of kidney cancer driven by MiT/TFE gene fusions. Here, we define the landmarks of tRCC through an integrative analysis of 152 patients with tRCC identified across genomic, clinical trial, and retrospective cohorts. Most tRCCs harbor few somatic alterations apart from MiT/TFE fusions and homozygous deletions at chromosome 9p21.3 (19.2% of cases). Transcriptionally, tRCCs display a heightened NRF2-driven antioxidant response that is associated with resistance to targeted therapies. Consistently, we find that outcomes for patients with tRCC treated with vascular endothelial growth factor receptor inhibitors (VEGFR-TKIs) are worse than those treated with immune checkpoint inhibitors (ICI). Using multiparametric immunofluorescence, we find that the tumors are infiltrated with CD8+ T cells, though the T cells harbor an exhaustion immunophenotype distinct from that of clear cell RCC. Our findings comprehensively define the clinical and molecular features of tRCC and may inspire new therapeutic hypotheses.

Microbial Metabolites of Flavanols in Urine are Associated with Enhanced Anti-Proliferative Activity in Bladder Cancer Cells In Vitro

Flavanols are metabolized by the gut microbiota to bioavailable metabolites, and the absorbed fraction is excreted primarily via urine. Uroepithelial cells are thus a potential site of activity due to exposure to high concentrations of these compounds. Chemoprevention by flavanols may be partly due to these metabolites. In Vitro work in this area relies on a limited pool of commercially available microbial metabolites, and little has been done in bladder cancer. The impact of physiologically relevant mixtures of flavanols and their metabolites remains unknown. Rats were fed various flavanols and urine samples, approximating the bioavailable metabolome, were collected. Urines were profiled by UPLC-MS/MS, and their anti-proliferative activities were assayed In Vitro in four bladder cancer models. Significant interindividual variability was observed for composition and proliferation. Microbial metabolite concentrations (valerolactones, phenylalkyl acids and hippuric acids) were positively associated with reduced bladder cancer proliferation In Vitro, while native flavanols were poorly correlated with activity. These results suggest that microbial metabolites may be responsible for chemoprevention in uroepithelial cells following flavanol consumption. This highlights the potential to use individual genetics and microbial metabotyping to design personalized dietary interventions for cancer prevention and/or adjuvant therapy to reduce bladder cancer incidence and improve outcomes.

Classification of pyroptosis patterns and construction of a novel prognostic model for prostate cancer based on bulk and single-cell RNA sequencing

BACKGROUND

Emerging evidence suggests an important role for pyroptosis in tumorigenesis and recurrence, but it remains to be elucidated in prostate cancer (PCa). Considering the low accuracy of common clinical predictors of PCa recurrence, we aimed to develop a novel pyroptosis-related signature to predict the prognosis of PCa patients based on integrative analyses of bulk and single-cell RNA sequencing (RNA-seq) profiling.

METHODS

The RNA-seq data of PCa patients was downloaded from several online databases. PCa patients were stratified into two Classes by unsupervised clustering. A novel signature was constructed by Cox and the Least Absolute Shrinkage and Selection Operator (LASSO) regression. The Kaplan-Meier curve was employed to evaluate the prognostic value of this signature and the single sample Gene Set Enrichment Analysis (ssGSEA) algorithm was used to analysis tumor-infiltrating immune cells. At single-cell level, we also classified the malignant cells into two Classes and constructed cell developmental trajectories and cell-cell interaction networks. Furthermore, RT-qPCR and immunofluorescence were used to validate the expression of core pyroptosis-related genes.

RESULTS

Twelve prognostic pyroptosis-related genes were identified and used to classify PCa patients into two prognostic Classes. We constructed a signature that identified PCa patients with different risks of recurrence and the risk score was proven to be an independent predictor of the recurrence free survival (RFS). Patients in the high-risk group had a significantly lower RFS (P<0.001). The expression of various immune cells differed between the two Classes. At the single-cell level, we classified the malignant cells into two Classes and described the heterogeneity. In addition, we observed that malignant cells may shift from Class1 to Class2 and thus have a worse prognosis.

CONCLUSION

We have constructed a robust pyroptosis-related signature to predict the RFS of PCa patients and described the heterogeneity of prostate cancer cells in terms of pyroptosis.

A Multiparametric Pharmacogenomic Strategy for Drug Repositioning predicts Therapeutic Efficacy for Glioblastoma Cell Lines

Background

Poor prognosis of glioblastoma patients and the extensive heterogeneity of glioblastoma at both the molecular and cellular level necessitates developing novel individualized treatment modalities via genomics-driven approaches.

Methods

This study leverages numerous pharmacogenomic and tissue databases to examine drug repositioning for glioblastoma. RNAseq of glioblastoma tumor samples from The Cancer Genome Atlas (TCGA, n=117) were compared to “normal” frontal lobe samples from Genotype-Tissue Expression Portal (GTEX, n=120) to find differentially expressed genes (DEGs). Using compound-gene expression data and drug activity data from the Library of Integrated Network-Based Cellular Signatures (LINCS, n=66,512 compounds) CCLE (71 glioma cell lines), and Chemical European Molecular Biology Laboratory (ChEMBL) platforms, we employed a summarized reversal gene expression metric (sRGES) to “reverse” the resultant disease signature for GBM and its subtypes. A multi-parametric strategy was employed to stratify compounds capable of blood brain barrier penetrance with a favorable pharmacokinetic profile (CNS-MPO).

Results

Significant correlations were identified between sRGES and drug efficacy in GBM cell lines in both ChEMBL(r=0.37,p&lt;.001) and Cancer Therapeutic Response Portal (CTRP) databases (r=0.35, p&lt;0.001). Our multiparametric algorithm identified two classes of drugs with highest sRGES and CNS-MPO: HDAC inhibitors (vorinostat and entinostat) and topoisomerase inhibitors suitable for drug repurposing.

Conclusions

Our studies suggest that reversal of glioblastoma disease signature correlates with drug potency for various GBM subtypes. This multiparametric approach may set the foundation for an early-phase personalized -omics clinical trial for glioblastoma by effectively identifying drugs that are capable of reversing the disease signature and have favorable pharmacokinetic and safety profiles.

Preclinical Drug Response Metric Based on Cellular Response Phenotype Provides Better Pharmacogenomic Variables with Phenotype Relevance

High-throughput screening of drug response in cultured cell lines is essential for studying therapeutic mechanisms and identifying molecular variants associated with sensitivity to drugs. Assessment of drug response is typically performed by constructing a dose-response curve of viability and summarizing it to a representative, such as IC₅₀. However, this is limited by its dependency on the assay duration and lack of reflections regarding actual cellular response phenotypes. To address these limitations, we consider how each response-phenotype contributes to the overall growth behavior and propose an alternative method of drug response screening that takes into account the cellular response phenotype. In conventional drug response screening methods, the ranking of sensitivity depends on either the metric used to construct the dose-response curve or the representative factor used to summarize the curve. This ambiguity in conventional assessment methods is due to the fact that assessment methods are not consistent with the underlying principles of population dynamics. Instead, the suggested phenotype metrics provide all phenotypic rates of change that shape overall growth behavior at a given dose and better response classification, including the phenotypic mechanism of overall growth inhibition. This alternative high-throughput drug-response screening would improve preclinical pharmacogenomic analysis and the understanding of a therapeutic mechanism of action.

MFmap: A semi-supervised generative model matching cell lines to tumours and cancer subtypes

Translating in vitro results from experiments with cancer cell lines to clinical applications requires the selection of appropriate cell line models. Here we present MFmap (model fidelity map), a machine learning model to simultaneously predict the cancer subtype of a cell line and its similarity to an individual tumour sample. The MFmap is a semi-supervised generative model, which compresses high dimensional gene expression, copy number variation and mutation data into cancer subtype informed low dimensional latent representations. The accuracy (test set F₁ score >90%) of the MFmap subtype prediction is validated in ten different cancer datasets. We use breast cancer and glioblastoma cohorts as examples to show how subtype specific drug sensitivity can be translated to individual tumour samples. The low dimensional latent representations extracted by MFmap explain known and novel subtype specific features and enable the analysis of cell-state transformations between different subtypes. From a methodological perspective, we report that MFmap is a semi-supervised method which simultaneously achieves good generative and predictive performance and thus opens opportunities in other areas of computational biology.

Identification of the Nerve-Cancer Cross-Talk-Related Prognostic Gene Model in Head and Neck Squamous Cell Carcinoma

The incidence of head and neck squamous cell carcinoma (HNSC) is increasing year by year. The nerve is an important component of the tumor microenvironment, which has a wide range of cross-talk with tumor cells and immune cells, especially in highly innervated organs, such as head and neck cancer and pancreatic cancer. However, the role of cancer-nerve cross-talk-related genes (NCCGs) in HNSC is unclear. In our study, we constructed a prognostic model based on genes with prognostic value in NCCGs. We used Pearson’s correlation to analyze the relationship between NCCGs and immune infiltration, microsatellite instability, tumor mutation burden, drug sensitivity, and clinical stage. We used single-cell sequencing data to analyze the expression of genes associated with stage in different cells and explored the possible pathways affected by these genes via gene set enrichment analysis. In the TCGA-HNSC cohort, a total of 23 genes were up- or downregulated compared with normal tissues. GO and KEGG pathway analysis suggested that NCCGs are mainly concentrated in membrane potential regulation, chemical synapse, axon formation, and neuroreceptor-ligand interaction. Ten genes were identified as prognosis genes by Kaplan-Meier plotter and used as candidate genes for LASSO regression. We constructed a seven-gene prognostic model (NTRK1, L1CAM, GRIN3A, CHRNA5, CHRNA6, CHRNB4, CHRND). The model could effectively predict the 1-, 3-, and 5-year survival rates in the TCGA-HNSC cohort, and the effectiveness of the model was verified by external test data. The genes included in the model were significantly correlated with immune infiltration, microsatellite instability, tumor mutation burden, drug sensitivity, and clinical stage. Single-cell sequencing data of HNSC showed that CHRNB4 was mainly expressed in tumor cells, and multiple metabolic pathways were enriched in high CHRNB4 expression tumor cells. In summary, we used comprehensive bioinformatics analysis to construct a prognostic gene model and revealed the potential of NCCGs as therapeutic targets and prognostic biomarkers in HNSC.

Uncovering drug repurposing candidates for head and neck cancers: insights from systematic pharmacogenomics data analysis

Effective treatment options for head and neck squamous cell carcinoma (HNSCC) are currently lacking. We exploited the drug response and genomic data of the 28 HNSCC cell lines, screened with 4,518 compounds, from the PRISM repurposing dataset to uncover repurposing drug candidates for HNSCC. A total of 886 active compounds, comprising of 418 targeted cancer, 404 non-oncology, and 64 chemotherapy compounds were identified for HNSCC. Top classes of mechanism of action amongst targeted cancer compounds included PI3K/AKT/MTOR, EGFR, and HDAC inhibitors. We have shortlisted 36 compounds with enriched killing activities for repurposing in HNSCC. The integrative analysis confirmed that the average expression of EGFR ligands (AREG, EREG, HBEGF, TGFA, and EPGN) is associated with osimertinib sensitivity. Novel putative biomarkers of response including those involved in immune signalling and cell cycle were found to be associated with sensitivity and resistance to MEK inhibitors respectively. We have also developed an RShiny webpage facilitating interactive visualization to fuel further hypothesis generation for drug repurposing in HNSCC. Our study provides a rich reference database of HNSCC drug sensitivity profiles, affording an opportunity to explore potential biomarkers of response in prioritized drug candidates. Our approach could also reveal insights for drug repurposing in other cancers.

Oligonucleotide conjugated antibody strategies for cyclic immunostaining

A number of highly multiplexed immunostaining and imaging methods have advanced spatial proteomics of cancer for improved treatment strategies. While a variety of methods have been developed, the most widely used methods are limited by harmful signal removal techniques, difficulties with reagent production and antigen sensitivity. Multiplexed immunostaining employing oligonucleotide (oligos)-barcoded antibodies is an alternative approach that is growing in popularity. However, challenges remain in consistent conjugation of oligos to antibodies with maintained antigenicity as well as non-destructive, robust and cost-effective signal removal methods. Herein, a variety of oligo conjugation and signal removal methods were evaluated in the development of a robust oligo conjugated antibody cyclic immunofluorescence (Ab-oligo cyCIF) methodology. Both non- and site-specific conjugation strategies were assessed to label antibodies, where site-specific conjugation resulted in higher retained binding affinity and antigen-specific staining. A variety of fluorescence signal removal methods were also evaluated, where incorporation of a photocleavable link (PCL) resulted in full fluorescence signal removal with minimal tissue disruption. In summary, this work resulted in an optimized Ab-oligo cyCIF platform capable of generating high dimensional images to characterize the spatial proteomics of the hallmarks of cancer.

KDM6B promotes activation of the oncogenic CDK4/6-pRB-E2F pathway by maintaining enhancer activity in MYCN-amplified neuroblastoma

The H3K27me2/me3 histone demethylase KDM6B is essential to neuroblastoma cell survival. However, the mechanism of KDM6B action remains poorly defined. We demonstrate that inhibition of KDM6B activity 1) reduces the chromatin accessibility of E2F target genes and MYCN, 2) selectively leads to an increase of H3K27me3 but a decrease of the enhancer mark H3K4me1 at the CTCF and BORIS binding sites, which may, consequently, disrupt the long-range chromatin interaction of MYCN and E2F target genes, and 3) phenocopies the transcriptome induced by the specific CDK4/6 inhibitor palbociclib. Overexpression of CDK4/6 or Rb1 knockout confers neuroblastoma cell resistance to both palbociclib and the KDM6 inhibitor GSK-J4. These data indicate that KDM6B promotes an oncogenic CDK4/6-pRB-E2F pathway in neuroblastoma cells via H3K27me3-dependent enhancer-promoter interactions, providing a rationale to target KDM6B for high-risk neuroblastoma.

The histone demethylase KDM6B is reported to be essential for neuroblastoma cell survival. Here the authors show that KDM6B regulates CDK4/6-pRB-E2F pathway through H3K27me3-dependent enhancer-promoter interactions in neuroblastoma.

Cholesterol Regulates the Tumor Adaptive Resistance to MAPK Pathway Inhibition

Although targeted MAPK pathway inhibition has achieved remarkable patient responses in many cancers, the development of resistance has remained a critical challenge. Adaptive tumor response underlies the drug resistance. Furthermore, such bypass mechanisms often lead to the activation of many pro-survival kinases, which complicates the rational design of combination therapies. Here, we performed global tyrosine phosphoproteomic (pTyr) analyses and demonstrated that targeted MAPK signaling inhibition in melanoma leads to a profound remodeling of the pTyr proteome. Intriguingly, altered cholesterol metabolism might drive, in a coordinated fashion, the activation of these kinases. Indeed, we found an accumulation of intracellular cholesterol in melanoma cells (with BRAF^V600E mutations) and non-small cell lung cancer cells (with KRAS^G12C mutations) treated with MAPK and KRAS^G12C inhibitors, respectively. Importantly, depletion of cholesterol not only prevents the feedback activation of pTyr signaling but also enhances the cytotoxic effects of MAPK pathway inhibitors, both in vitro and in vivo. Together, our findings suggest that cholesterol contributes to the tumor adaptive response upon targeted MAPK pathway inhibitors. These results also suggest that MAPK pathway inhibitors could be combined with cholesterol-lowering agents to achieve a more complete and durable response in tumors with hyperactive MAPK signaling.

ParsVNN: parsimony visible neural networks for uncovering cancer-specific and drug-sensitive genes and pathways

Prediction of cancer-specific drug responses as well as identification of the corresponding drug-sensitive genes and pathways remains a major biological and clinical challenge. Deep learning models hold immense promise for better drug response predictions, but most of them cannot provide biological and clinical interpretability. Visible neural network (VNN) models have emerged to solve the problem by giving neurons biological meanings and directly casting biological networks into the models. However, the biological networks used in VNNs are often redundant and contain components that are irrelevant to the downstream predictions. Therefore, the VNNs using these redundant biological networks are overparameterized, which significantly limits VNNs' predictive and explanatory power. To overcome the problem, we treat the edges and nodes in biological networks used in VNNs as features and develop a sparse learning framework ParsVNN to learn parsimony VNNs with only edges and nodes that contribute the most to the prediction task. We applied ParsVNN to build cancer-specific VNN models to predict drug response for five different cancer types. We demonstrated that the parsimony VNNs built by ParsVNN are superior to other state-of-the-art methods in terms of prediction performance and identification of cancer driver genes. Furthermore, we found that the pathways selected by ParsVNN have great potential to predict clinical outcomes as well as recommend synergistic drug combinations.

MTH1 Inhibitor TH1579 Induces Oxidative DNA Damage and Mitotic Arrest in Acute Myeloid Leukemia

Acute myeloid leukemia (AML) is an aggressive hematologic malignancy, exhibiting high levels of reactive oxygen species (ROS). ROS levels have been suggested to drive leukemogenesis and is thus a potential novel target for treating AML. MTH1 prevents incorporation of oxidized nucleotides into the DNA to maintain genome integrity and is upregulated in many cancers. Here we demonstrate that hematologic cancers are highly sensitive to MTH1 inhibitor TH1579 (karonudib). A functional precision medicine ex vivo screen in primary AML bone marrow samples demonstrated a broad response profile of TH1579, independent of the genomic alteration of AML, resembling the response profile of the standard-of-care treatments cytarabine and doxorubicin. Furthermore, TH1579 killed primary human AML blast cells (CD45⁺) as well as chemotherapy resistance leukemic stem cells (CD45⁺Lin^-CD34⁺CD38^-), which are often responsible for AML progression. TH1579 killed AML cells by causing mitotic arrest, elevating intracellular ROS levels, and enhancing oxidative DNA damage. TH1579 showed a significant therapeutic window, was well tolerated in animals, and could be combined with standard-of-care treatments to further improve efficacy. TH1579 significantly improved survival in two different AML disease models in vivo. In conclusion, the preclinical data presented here support that TH1579 is a promising novel anticancer agent for AML, providing a rationale to investigate the clinical usefulness of TH1579 in AML in an ongoing clinical phase I trial. SIGNIFICANCE: The MTH1 inhibitor TH1579 is a potential novel AML treatment, targeting both blasts and the pivotal leukemic stem cells while sparing normal bone marrow cells.

Bioinformatics Analysis Identifies Precision Treatment with Paclitaxel for Hepatocellular Carcinoma Patients Harboring Mutant TP53 or Wild-Type CTNNB1 Gene

Hepatocellular carcinoma (HCC) is an aggressive and chemoresistant cancer type. The development of novel therapeutic strategies is still urgently needed. Personalized or precision medicine is a new trend in cancer therapy, which treats cancer patients with specific genetic alterations. In this study, a gene signature was identified from the transcriptome of HCC patients, which was correlated with the patients’ poorer prognoses. This gene signature is functionally related to mitotic cell cycle regulation, and its higher or lower expression is linked to the mutation in tumor protein p53 (TP53) or catenin beta 1 (CTNNB1), respectively. Gene–drug association analysis indicated that the taxanes, such as the clinically approved anticancer drug paclitaxel, are potential drugs targeting this mitotic gene signature. Accordingly, HCC cell lines harboring mutant TP53 or wild-type CTNNB1 genes are more sensitive to paclitaxel treatment. Therefore, our results imply that HCC patients with mutant TP53 or wild-type CTNNB1 genes may benefit from the paclitaxel therapy.

Differential BET Bromodomain Inhibition by Dihydropteridinone and Pyrimidodiazepinone Kinase Inhibitors

BRD4 and other members of the bromodomain and extraterminal (BET) family of proteins are promising epigenetic targets for the development of novel therapeutics. Among the reported BRD4 inhibitors are dihydropteridinones and benzopyrimidodiazepinones originally designed to target the kinases PLK1, ERK5, and LRRK2. While these kinase inhibitors were identified as BRD4 inhibitors, little is known about their binding potential and structural details of interaction with the other BET bromodomains. We comprehensively characterized a series of known and newly identified dual BRD4-kinase inhibitors against all eight individual BET bromodomains. A detailed analysis of 23 novel cocrystal structures of BET-kinase inhibitor complexes in combination with direct binding assays and cell signaling studies revealed significant differences in molecular shape complementarity and inhibitory potential. Collectively, the data offer new insights into the action of kinase inhibitors across BET bromodomains, which may aid the development of drugs to inhibit certain BET proteins and kinases differentially.

Integrative Characterization of the Role of IL27 In Melanoma Using Bioinformatics Analysis

Background

IL27 has been reported to play dual roles in cancer; however, its effects on the tumor microenvironment (TME), immunotherapy, and prognosis in melanoma remain largely unclear. This study was aimed to uncover the effects of IL27 on TME, immunotherapy and prognosis in patients with melanoma.

Methods

RNA-seq data, drug sensitivity data, and clinical data were obtained from TCGA, GEO, CCLE, and CTRP. Log-rank test was used to determine the survival value of IL27. Univariate and multivariate Cox regression analyses were employed to determine the independent predictors of survival outcomes. DAVID and GSEA were used to perform gene set functional annotations. ssGSEA was used to explore the association between IL27 and immune infiltrates. ConsensusClusterPlus was used to classify melanoma tissues into hot tumors or cold tumors.

Results

Clinically, IL27 was negatively correlated with Breslow depth (P = 0.00042) and positively associated with response to radiotherapy (P = 0.038). High IL27 expression showed an improved survival outcome (P = 0.00016), and could serve as an independent predictor of survival outcomes (hazard ratio: 0.32 - 0.88, P = 0.015). Functionally, elevated IL27 expression could induce an enhanced immune response and pyroptosis (R = 0.64, P = 1.2e-55), autophagy (R = 0.37, P = 7.1e-17) and apoptosis (R = 0.47, P = 1.1e-27) in patients with melanoma. Mechanistically, elevated IL27 expression was positively correlated with cytotoxic cytokines (including INFG and GZMB), enhanced immune infiltrates, and elevated CD8/Treg ratio (R = 0.14, P = 0.02), possibly driving CD8⁺ T cell infiltration by suppressing β-catenin signaling in the TME. Furthermore, IL27 was significantly associated with hot tumor state, multiple predictors of response to immunotherapy, and improved drug response in patients with melanoma.

Conclusions

IL27 was correlated with enriched CD8⁺ T cells, desirable therapeutic response and improved prognosis. It thus can be utilized as a promising modulator in the development of cytokine-based immunotherapy for melanoma.

Omics and Computational Modeling Approaches for the Effective Treatment of Drug-Resistant Cancer Cells

Chemotherapy is a mainstream cancer treatment, but has a constant challenge of drug resistance, which consequently leads to poor prognosis in cancer treatment. For better understanding and effective treatment of drug-resistant cancer cells, omics approaches have been widely conducted in various forms. A notable use of omics data beyond routine data mining is to use them for computational modeling that allows generating useful predictions, such as drug responses and prognostic biomarkers. In particular, an increasing volume of omics data has facilitated the development of machine learning models. In this mini review, we highlight recent studies on the use of multi-omics data for studying drug-resistant cancer cells. We put a particular focus on studies that use computational models to characterize drug-resistant cancer cells, and to predict biomarkers and/or drug responses. Computational models covered in this mini review include network-based models, machine learning models and genome-scale metabolic models. We also provide perspectives on future research opportunities for combating drug-resistant cancer cells.

Computational repurposing of therapeutic small molecules from cancer to pulmonary hypertension

Cancer therapies are being considered for treating rare noncancerous diseases like pulmonary hypertension (PH), but effective computational screening is lacking. Via transcriptomic differential dependency analyses leveraging parallels between cancer and PH, we mapped a landscape of cancer drug functions dependent upon rewiring of PH gene clusters. Bromodomain and extra-terminal motif (BET) protein inhibitors were predicted to rely upon several gene clusters inclusive of galectin-8 (LGALS8). Correspondingly, LGALS8 was found to mediate the BET inhibitor–dependent control of endothelial apoptosis, an essential role for PH in vivo. Separately, a piperlongumine analog’s actions were predicted to depend upon the iron-sulfur biogenesis gene ISCU. Correspondingly, the analog was found to inhibit ISCU glutathionylation, rescuing oxidative metabolism, decreasing endothelial apoptosis, and improving PH. Thus, we identified crucial drug-gene axes central to endothelial dysfunction and therapeutic priorities for PH. These results establish a wide-ranging, network dependency platform to redefine cancer drugs for use in noncancerous conditions.

Applications of Machine Learning to Predict Cisplatin Resistance in Lung Cancer

Purpose

Lung cancer, mainly lung adenocarcinoma, lung squamous cell carcinoma and small cell lung cancer, has the highest incidence and cancer-related mortality worldwide. Platinum-based chemotherapy plays an important role in the treatment of various lung cancer subtypes, but not all patients benefit from this treatment regimen; thus, it is worth identifying lung cancer patients who are resistant or sensitive to platinum-based therapy.

Methods

The drug response and sequencing data of 170 lung cancer cell lines were downloaded from the Genomics of Drug Sensitivity in Cancer (GDSC) database, and support vector machines (SVMs) and beam search were used to select an optimal gene panel that can predict the sensitivity of cell lines to cisplatin. Then, we used available cell line data to explore the potential mechanisms.

Results

In this work, the drug response and sequencing data of 170 lung cancer cell lines were downloaded from the GDSC database, and SVMs and beam search were used to screen a panel of genes related to lung cancer cell line resistance to cisplatin. A final panel of nine genes (PLXNC1, KIAA0649, SPTBN4, SLC14A2, F13A1, COL5A1, SCN2A, PLEC, and ALMS1) was identified, and achieved an area under the curve (AUC) of 0.873 ± 0.004. The natural logarithm of the half maximal inhibitory concentration (lnIC50) values of the mutant-type (panel-MT) group was significantly higher than that of the wild-type (panel-WT) group, regardless of the lung cancer subtype. The differentially expressed pathways between the two groups may explain this difference.

Conclusion

In this study, we found that a panel of nine genes can accurately predict sensitivity to cisplatin, which may provide individualized treatment recommendations to improve the prognosis of patients with lung cancer.

Development of targeted therapy of NRF2high esophageal squamous cell carcinoma

Esophageal squamous cell carcinoma (ESCC) is a deadly disease and one of the most aggressive cancers of the gastrointestinal tract. As a master transcription factor regulating the stress response, NRF2 is often mutated and becomes hyperactive, and thus causes chemo-radioresistance and poor survival in human ESCC. There is a great need to develop NRF2 inhibitors for targeted therapy of NRF2high ESCC. In this review, we mainly focus on three aspects, NRF2 inhibitors and their mechanisms of action, screening novel drug targets, and evaluation of NRF2 activity in the esophagus. A research strategy has been proposed to develop NRF2 inhibitors using human ESCC cells and mouse models.

Representation of molecules for drug response prediction

The rapid development of machine learning and deep learning algorithms in the recent decade has spurred an outburst of their applications in many research fields. In the chemistry domain, machine learning has been widely used to aid in drug screening, drug toxicity prediction, quantitative structure-activity relationship prediction, anti-cancer synergy score prediction, etc. This review is dedicated to the application of machine learning in drug response prediction. Specifically, we focus on molecular representations, which is a crucial element to the success of drug response prediction and other chemistry-related prediction tasks. We introduce three types of commonly used molecular representation methods, together with their implementation and application examples. This review will serve as a brief introduction of the broad field of molecular representations.

Inferred Ancestral Origin of Cancer Cell Lines Associates with Differential Drug Response

Disparities between risk, treatment outcomes and survival rates in cancer patients across the world may be attributed to socioeconomic factors. In addition, the role of ancestry is frequently discussed. In preclinical studies, high-throughput drug screens in cancer cell lines have empowered the identification of clinically relevant molecular biomarkers of drug sensitivity; however, the genetic ancestry from tissue donors has been largely neglected in this setting. In order to address this, here, we show that the inferred ancestry of cancer cell lines is conserved and may impact drug response in patients as a predictive covariate in high-throughput drug screens. We found that there are differential drug responses between European and East Asian ancestries, especially when treated with PI3K/mTOR inhibitors. Our finding emphasizes a new angle in precision medicine, as cancer intervention strategies should consider the germline landscape, thereby reducing the failure rate of clinical trials.

Establishment and characterization of NCC-LGFMS1-C1: a novel patient-derived cell line of low-grade fibromyxoid sarcoma

Low-grade fibromyxoid sarcoma (LGFMS) is a rare soft-tissue sarcoma genetically characterized by the presence of the FUS-CREB3L2 gene fusion. While LGFMS exhibits indolent features during its early stages, the rates of recurrence, metastasis, and death from the disease are high. Presently, the role of FUS-CREB3L2 gene fusions in the unique features of LGFMS is not clear, and there is no modality to improve the clinical outcomes of patients with LGFMS; thus, extensive studies on LGFMS are required. Patient-derived cancer cell lines are critical tools for cancer research. However, no cell line has been established for LGFMS. Here, we aimed to develop a novel cell line for LGFMS and successfully established it using surgically resected tumor tissues. The cells, named NCC-LGFMS1-C1, possessed the same fusion genes as their original tumor and visible copy number variations. The cells had a fibroblastic appearance, formed spheroids when they were seeded in a low-attachment dish, and exhibited constant growth and invasion. Additionally, we demonstrated the feasibility of high-throughput drug screening using these cells. In conclusion, the NCC-LGFMS1-C1 cell line is a useful tool for studying LGFMS.

A Novel Intercellular Communication-Associated Gene Signature for Prognostic Prediction and Clinical Value in Patients With Lung Adenocarcinoma

Background

Lung cancer remains the leading cause of cancer death globally, with lung adenocarcinoma (LUAD) being its most prevalent subtype. This study aimed to identify the key intercellular communication-associated genes (ICAGs) in LUAD.

Methods

Eight publicly available datasets were downloaded from the Gene Expression Omnibus (GEO) and The Cancer Genome Atlas (TCGA) databases. The prognosis-related ICAGs were identified and a risk score was developed by using survival analysis. Machine learning models were trained to predict LUAD recurrence based on the selected ICAGs and clinical information. Comprehensive analyses on ICAGs and tumor microenvironment were performed. A single-cell RNA-sequencing dataset was assessed to further elucidate aberrant changes in intercellular communication.

Results

Eight ICAGs with prognostic potential were identified in the present study, and a risk score was derived accordingly. The best machine-learning model to predict relapse was developed based on clinical information and the expression levels of these eight ICAGs. This model achieved a remarkable area under receiver operator characteristic curves of 0.841. Patients were divided into high- and low-risk groups according to their risk scores. DNA replication and cell cycle were significantly enriched by the differentially expressed genes between the high- and the low-risk groups. Infiltrating immune cells, immune functions were significantly related to ICAGs expressions and risk scores. Additionally, the changes of intercellular communication were modeled by analyzing the single-cell sequencing dataset.

Conclusion

The present study identified eight key ICAGs in LUAD, which could contribute to patient stratification and act as novel therapeutic targets.

Functional genomics approaches to improve pre-clinical drug screening and biomarker discovery

Advances in sequencing technology have enabled the genomic and transcriptomic characterization of human malignancies with unprecedented detail. However, this wealth of information has been slow to translate into clinically meaningful outcomes. Different models to study human cancers have been established and extensively characterized. Using these models, functional genomic screens and pre-clinical drug screening platforms have identified genetic dependencies that can be exploited with drug therapy. These genetic dependencies can also be used as biomarkers to predict response to treatment. For many cancers, the identification of such biomarkers remains elusive. In this review, we discuss the development and characterization of models used to study human cancers, RNA interference and CRISPR screens to identify genetic dependencies, large-scale pharmacogenomics studies and drug screening approaches to improve pre-clinical drug screening and biomarker discovery.

Combining APR-246 and HDAC-Inhibitors: A Novel Targeted Treatment Option for Neuroblastoma

Simple Summary

Preclinical analyses identified APR-246 as a potent treatment option for neuroblastoma. However, a specific mode of action, sufficient biomarkers and promising combination partners are still missing. Here, we analyze the susceptibilities of different entities and relate them to gene expression profiles and previously described biomarkers. We propose a gene signature, consisting of 13 genes, as a novel predictive biomarker. Furthermore, we provide evidence that APR-246 directly targets metabolic weaknesses in neuroblastoma cell lines, thus hampering ROS detoxification. This makes APR-246 suitable to be combined with ROS-inducing HDAC inhibitors, a treatment combination that has not been described for neuroblastoma thus far.

Abstract

APR-246 (Eprenetapopt/PRIMA-1^Met) is a very potent anti-cancer drug in clinical trials and was initially developed as a p53 refolding agent. As an alternative mode of action, the elevation of reactive oxygen species (ROS) has been proposed. Through an in silico analysis, we investigated the responses of approximately 800 cancer cell lines (50 entities; Cancer Therapeutics Response Portal, CTRP) to APR-246 treatment. In particular, neuroblastoma, lymphoma and acute lymphocytic leukemia cells were highly responsive. With gene expression data from the Cancer Cell Line Encyclopedia (CCLE; n = 883) and patient samples (n = 1643) from the INFORM registry study, we confirmed that these entities express low levels of SLC7A11, a previously described predictive biomarker for APR-246 responsiveness. Combining the CTRP drug response data with the respective CCLE gene expression profiles, we defined a novel gene signature, predicting the effectiveness of APR-246 treatment with a sensitivity of 90% and a specificity of 94%. We confirmed the predicted APR-246 sensitivity in 8/10 cell lines and in ex vivo cultures of patient samples. Moreover, the combination of ROS detoxification-impeding APR-246 with approved HDAC-inhibitors, known to elevate ROS, substantially increased APR-246 sensitivity in cell cultures and in vivo in two zebrafish neuroblastoma xenograft models. These data provide evidence that APR-246, in combination with HDAC-inhibitors, displays a novel potent targeted treatment option for neuroblastoma patients.

Minimal information for chemosensitivity assays (MICHA): a next-generation pipeline to enable the FAIRification of drug screening experiments

Chemosensitivity assays are commonly used for preclinical drug discovery and clinical trial optimization. However, data from independent assays are often discordant, largely attributed to uncharacterized variation in the experimental materials and protocols. We report here the launching of Minimal Information for Chemosensitivity Assays (MICHA), accessed via https://micha-protocol.org. Distinguished from existing efforts that are often lacking support from data integration tools, MICHA can automatically extract publicly available information to facilitate the assay annotation including: 1) compounds, 2) samples, 3) reagents and 4) data processing methods. For example, MICHA provides an integrative web server and database to obtain compound annotation including chemical structures, targets and disease indications. In addition, the annotation of cell line samples, assay protocols and literature references can be greatly eased by retrieving manually curated catalogues. Once the annotation is complete, MICHA can export a report that conforms to the FAIR principle (Findable, Accessible, Interoperable and Reusable) of drug screening studies. To consolidate the utility of MICHA, we provide FAIRified protocols from five major cancer drug screening studies as well as six recently conducted COVID-19 studies. With the MICHA web server and database, we envisage a wider adoption of a community-driven effort to improve the open access of drug sensitivity assays.

High-throughput screening and genome-wide analyses of 44 anticancer drugs in the 1000 Genomes cell lines reveals an association of the NQO1 gene with the response of multiple anticancer drugs

Cancer patients exhibit a broad range of inter-individual variability in response and toxicity to widely used anticancer drugs, and genetic variation is a major contributor to this variability. To identify new genes that influence the response of 44 FDA-approved anticancer drug treatments widely used to treat various types of cancer, we conducted high-throughput screening and genome-wide association mapping using 680 lymphoblastoid cell lines from the 1000 Genomes Project. The drug treatments considered in this study represent nine drug classes widely used in the treatment of cancer in addition to the paclitaxel + epirubicin combination therapy commonly used for breast cancer patients. Our genome-wide association study (GWAS) found several significant and suggestive associations. We prioritized consistent associations for functional follow-up using gene-expression analyses. The NAD(P)H quinone dehydrogenase 1 (NQO1) gene was found to be associated with the dose-response of arsenic trioxide, erlotinib, trametinib, and a combination treatment of paclitaxel + epirubicin. NQO1 has previously been shown as a biomarker of epirubicin response, but our results reveal novel associations with these additional treatments. Baseline gene expression of NQO1 was positively correlated with response for 43 of the 44 treatments surveyed. By interrogating the functional mechanisms of this association, the results demonstrate differences in both baseline and drug-exposed induction.

In the burgeoning field of personalized medicine, genetic variation is recognized as a major contributor to patients’ differential responses to drugs. Lymphoblastoid cell lines (LCLs) are a consistent and convenient representation of cells used for in vitro research. Human genome sequencing with LCLs can identify new genes that influence individuals’ drug responses, including the dose-response relationship, which describes the relationship between physiological response and the amount of exposure to a substance. In this work, we conduct high-throughput screening and genome-wide association mapping using 680 LCLs from the 1000 Genomes Project to identify new genes that influence individual response to 44 widely used anticancer drugs. We found the NQO1 gene to be associated with the dose-response of several drugs, namely arsenic trioxide, erlotinib, trametinib, and the paclitaxel + epirubicin combination, and performed follow-up analyses to better understand its functional role in drug response. Our results indicate NQO1 expression is correlated with increased drug resistance and provide some evidence that SNP rs1800566 influences drug response by altering protein activity for these four treatments. With further research, NQO1 has potential use as a therapeutic target, for example, suppressing NQO1 expression to increase sensitivity to particular drugs.

Drug sensitivity prediction from cell line-based pharmacogenomics data: guidelines for developing machine learning models

The goal of precision oncology is to tailor treatment for patients individually using the genomic profile of their tumors. Pharmacogenomics datasets such as cancer cell lines are among the most valuable resources for drug sensitivity prediction, a crucial task of precision oncology. Machine learning methods have been employed to predict drug sensitivity based on the multiple omics data available for large panels of cancer cell lines. However, there are no comprehensive guidelines on how to properly train and validate such machine learning models for drug sensitivity prediction. In this paper, we introduce a set of guidelines for different aspects of training gene expression-based predictors using cell line datasets. These guidelines provide extensive analysis of the generalization of drug sensitivity predictors and challenge many current practices in the community including the choice of training dataset and measure of drug sensitivity. The application of these guidelines in future studies will enable the development of more robust preclinical biomarkers.

Binary pan-cancer classes with distinct vulnerabilities defined by pro- or anti-cancer YAP/TEAD activity

Cancer heterogeneity impacts therapeutic response, driving efforts to discover over-arching rules that supersede variability. Here, we define pan-cancer binary classes based on distinct expression of YAP and YAP-responsive adhesion regulators. Combining informatics with in vivo and in vitro gain- and loss-of-function studies across multiple murine and human tumor types, we show that opposite pro- or anti-cancer YAP activity functionally defines binary YAP^on or YAP^off cancer classes that express or silence YAP, respectively. YAP^off solid cancers are neural/neuroendocrine and frequently RB1^-/-, such as retinoblastoma, small cell lung cancer, and neuroendocrine prostate cancer. YAP silencing is intrinsic to the cell of origin, or acquired with lineage switching and drug resistance. The binary cancer groups exhibit distinct YAP-dependent adhesive behavior and pharmaceutical vulnerabilities, underscoring clinical relevance. Mechanistically, distinct YAP/TEAD enhancers in YAP^off or YAP^on cancers deploy anti-cancer integrin or pro-cancer proliferative programs, respectively. YAP is thus pivotal across cancer, but in opposite ways, with therapeutic implications.

A gentle introduction to understanding preclinical data for cancer pharmaco-omic modeling

A central goal of precision oncology is to administer an optimal drug treatment to each cancer patient. A common preclinical approach to tackle this problem has been to characterize the tumors of patients at the molecular and drug response levels, and employ the resulting datasets for predictive in silico modeling (mostly using machine learning). Understanding how and why the different variants of these datasets are generated is an important component of this process. This review focuses on providing such introduction aimed at scientists with little previous exposure to this research area.

Interpretable deep recommender system model for prediction of kinase inhibitor efficacy across cancer cell lines

Computational models for drug sensitivity prediction have the potential to significantly improve personalized cancer medicine. Drug sensitivity assays, combined with profiling of cancer cell lines and drugs become increasingly available for training such models. Multiple methods were proposed for predicting drug sensitivity from cancer cell line features, some in a multi-task fashion. So far, no such model leveraged drug inhibition profiles. Importantly, multi-task models require a tailored approach to model interpretability. In this work, we develop DEERS, a neural network recommender system for kinase inhibitor sensitivity prediction. The model utilizes molecular features of the cancer cell lines and kinase inhibition profiles of the drugs. DEERS incorporates two autoencoders to project cell line and drug features into 10-dimensional hidden representations and a feed-forward neural network to combine them into response prediction. We propose a novel interpretability approach, which in addition to the set of modeled features considers also the genes and processes outside of this set. Our approach outperforms simpler matrix factorization models, achieving R [Formula: see text] 0.82 correlation between true and predicted response for the unseen cell lines. The interpretability analysis identifies 67 biological processes that drive the cell line sensitivity to particular compounds. Detailed case studies are shown for PHA-793887, XMD14-99 and Dabrafenib.

Gene Fusions Create Partner and Collateral Dependencies Essential to Cancer Cell Survival

Gene fusions frequently result from rearrangements in cancer genomes. In many instances, gene fusions play an important role in oncogenesis; in other instances, they are thought to be passenger events. Although regulatory element rearrangements and copy number alterations resulting from these structural variants are known to lead to transcriptional dysregulation across cancers, the extent to which these events result in functional dependencies with an impact on cancer cell survival is variable. Here we used CRISPR-Cas9 dependency screens to evaluate the fitness impact of 3,277 fusions across 645 cell lines from the Cancer Dependency Map. We found that 35% of cell lines harbored either a fusion partner dependency or a collateral dependency on a gene within the same topologically associating domain as a fusion partner. Fusion-associated dependencies revealed numerous novel oncogenic drivers and clinically translatable alterations. Broadly, fusions can result in partner and collateral dependencies that have biological and clinical relevance across cancer types. SIGNIFICANCE: This study provides insights into how fusions contribute to fitness in different cancer contexts beyond partner-gene activation events, identifying partner and collateral dependencies that may have direct implications for clinical care.

SLC7A11 Is a Superior Determinant of APR-246 (Eprenetapopt) Response than TP53 Mutation Status

APR-246 (eprenetapopt) is in clinical development with a focus on hematologic malignancies and is promoted as a mutant-p53 reactivation therapy. Currently, the detection of at least one TP53 mutation is an inclusion criterion for patient selection into most APR-246 clinical trials. Preliminary results from our phase Ib/II clinical trial investigating APR-246 combined with doublet chemotherapy [cisplatin and 5-fluorouracil (5-FU)] in metastatic esophageal cancer, together with previous preclinical studies, indicate that TP53 mutation status alone may not be a sufficient biomarker for APR-246 response. This study aims to identify a robust biomarker for response to APR-246. Correlation analysis of the PRIMA-1 activity (lead compound to APR-246) with mutational status, gene expression, protein expression, and metabolite abundance across over 700 cancer cell lines (CCL) was performed. Functional validation and a boutique siRNA screen of over 850 redox-related genes were also conducted. TP53 mutation status was not consistently predictive of response to APR-246. The expression of SLC7A11, the cystine/glutamate transporter, was identified as a superior determinant of response to APR-246. Genetic regulators of SLC7A11, including ATF4, MDM2, wild-type p53, and c-Myc, were confirmed to also regulate cancer-cell sensitivity to APR-246. In conclusion, SLC7A11 expression is a broadly applicable determinant of sensitivity to APR-246 across cancer and should be utilized as the key predictive biomarker to stratify patients for future clinical investigation of APR-246.

Establishment and characterization of novel patient-derived cell lines from giant cell tumor of bone

Giant cell tumor of bone (GCTB) is a locally aggressive and rarely metastasizing tumor. GCTB is characterized by the presence of unique giant cells and a recurrent mutation in the histone tail of the histone variant H3.3, which is encoded by H3F3A on chromosome 1. GCTB accounts for ~ 5% of primary bone tumors. Although GCTB exhibits an indolent course, it has the potential to develop aggressive behaviors associated with local recurrence and distant metastasis. Currently, complete surgical resection is the only curative treatment, and novel therapeutic strategies are required. Patient-derived cancer cell lines are critical tools for basic and pre-clinical research. However, only a few GCTB cell lines have been reported, and none of them are available from public cell banks. Therefore, we aimed to establish novel GCTB cell lines in the present study. Using curetted tumor tissues of GCTB, we established two cell lines and named them NCC-GCTB2-C1 and NCC-GCTB3-C1. These cells harbored a typical mutation in histones and exhibited slow but constant growth, formed spheroids, and had invasive capabilities. We demonstrated the utility of these cell lines for high-throughput drug screening using 214 anticancer agents. We concluded that NCC-GCTB2-C1 and NCC-GCTB3-C1 cell lines were useful for the in vitro study of GCTB.

Bayesian multi-source regression and monocyte-associated gene expression predict BCL-2 inhibitor resistance in acute myeloid leukemia

The FDA recently approved eight targeted therapies for acute myeloid leukemia (AML), including the BCL-2 inhibitor venetoclax. Maximizing efficacy of these treatments requires refining patient selection. To this end, we analyzed two recent AML studies profiling the gene expression and ex vivo drug response of primary patient samples. We find that ex vivo samples often exhibit a general sensitivity to (any) drug exposure, independent of drug target. We observe that this "general response across drugs" (GRD) is associated with FLT3-ITD mutations, clinical response to standard induction chemotherapy, and overall survival. Further, incorporating GRD into expression-based regression models trained on one of the studies improved their performance in predicting ex vivo response in the second study, thus signifying its relevance to precision oncology efforts. We find that venetoclax response is independent of GRD but instead show that it is linked to expression of monocyte-associated genes by developing and applying a multi-source Bayesian regression approach. The method shares information across studies to robustly identify biomarkers of drug response and is broadly applicable in integrative analyses.

oncoPredict: an R package for predicting in vivo or cancer patient drug response and biomarkers from cell line screening data

Cell line drug screening datasets can be utilized for a range of different drug discovery applications from drug biomarker discovery to building translational models of drug response. Previously, we described three separate methodologies to (1) correct for general levels of drug sensitivity to enable drug-specific biomarker discovery, (2) predict clinical drug response in patients and (3) associate these predictions with clinical features to perform in vivo drug biomarker discovery. Here, we unite and update these methodologies into one R package (oncoPredict) to facilitate the development and adoption of these tools. This new OncoPredict R package can be applied to various in vitro and in vivo contexts for drug and biomarker discovery.

LncTx: A network-based method to repurpose drugs acting on the survival-related lncRNAs in lung cancer

Despite the fact that an increased amount of survival-related lncRNAs have been found in cancer, few drugs that target lncRNAs are approved for treatment. Here, we developed a network-based algorithm, LncTx, to repurpose the medications that potentially act on survival-related lncRNAs in lung cancer. We used eight survival-related lncRNAs derived from our previous study to test the efficacy of this method. LncTx calculates the shortest path length (proximity) between the drug targets and the lncRNA-correlated proteins in the protein-protein interaction network (interactome). LncTx contains seven different proximity measures, which are calculated in the unweighted or weighted interactome. First, to test the performance of LncTx in predicting correct indication of drugs, we benchmarked the proximity measures based on the accuracy of differentiating anticancer drugs from non-anticancer drugs. The closest proximity weighted by clustering coefficient (closestCC) has the best performance (AUC around 0.8) compared to other proximity measures across all survival-related lncRNAs. The majority of the other six proximity measures have decent performance as well, with AUC greater than 0.7. Second, to evaluate whether LncTx can repurpose the drugs effectively acting on the lncRNAs, we clustered the drugs according to their proximities by hierarchical clustering. The drugs with smaller proximity (proximal drugs) were proved to be more effective than the drugs with larger proximity (distal drugs). In conclusion, LncTx enables us to accurately identify anticancer drugs and can potentially be an index to repurpose effective agents acting on survival-related lncRNAs in lung cancer.

Machine learning in the prediction of cancer therapy

Resistance to therapy remains a major cause of cancer treatment failures, resulting in many cancer-related deaths. Resistance can occur at any time during the treatment, even at the beginning. The current treatment plan is dependent mainly on cancer subtypes and the presence of genetic mutations. Evidently, the presence of a genetic mutation does not always predict the therapeutic response and can vary for different cancer subtypes. Therefore, there is an unmet need for predictive models to match a cancer patient with a specific drug or drug combination. Recent advancements in predictive models using artificial intelligence have shown great promise in preclinical settings. However, despite massive improvements in computational power, building clinically useable models remains challenging due to a lack of clinically meaningful pharmacogenomic data. In this review, we provide an overview of recent advancements in therapeutic response prediction using machine learning, which is the most widely used branch of artificial intelligence. We describe the basics of machine learning algorithms, illustrate their use, and highlight the current challenges in therapy response prediction for clinical practice.

Identification of genes and pathways related to breast cancer metastasis in an integrated cohort

BACKGROUND

Breast cancer is the most common malignant disease in women. Metastasis is the most common cause of death from this cancer. Screening genes related to breast cancer metastasis may help elucidate the mechanisms governing metastasis and identify molecular targets for antimetastatic therapy. The development of advanced algorithms enables us to perform cross-study analysis to improve the robustness of the results.

MATERIALS AND METHODS

Ten data sets meeting our criteria for differential expression analyses were obtained from the Gene Expression Omnibus (GEO) database. Among these data sets, five based on the same platform were formed into a large cohort using the XPN algorithm. Differentially expressed genes (DEGs) associated with breast cancer metastasis were identified using the differential expression via distance synthesis (DEDS) algorithm. A cross-platform method was employed to verify these DEGs in all ten selected data sets. The top 50 validated DEGs are represented with heat maps. Based on the validated DEGs, Gene Ontology (GO) functional and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were performed. Protein interaction (PPI) networks were constructed to further illustrate the direct and indirect associations among the DEGs. Survival analysis was performed to explore whether these genes can affect breast cancer patient prognosis.

RESULTS

A total of 817 DEGs were identified using the DEDS algorithm. Of these DEGs, 450 genes were validated by the second algorithm. Enriched KEGG pathway terms demonstrated that these 450 DEGs may be involved in the cell cycle and oocyte meiosis in addition to their functions in ECM-receptor interaction and protein digestion and absorption. PPI network analysis for the proteins encoded by the DEGs indicated that these genes may be primarily involved in the cell cycle and extracellular matrix. In particular, several genes played roles in multiple signalling pathways and were related to patient survival. These genes were also observed to be targetable in the CTD2 database.

CONCLUSIONS

Our study analysed multiple cross-platform data sets using two different algorithms, helping elucidate the molecular mechanisms and identify several potential therapeutic targets of metastatic breast cancer. In addition, several genes exhibited promise for applications in targeted therapy against metastasis in future research.

Combined morphological and proteome profiling reveals target-independent impairment of cholesterol homeostasis

Unbiased profiling approaches are powerful tools for small-molecule target or mode-of-action deconvolution as they generate a holistic view of the bioactivity space. This is particularly important for non-protein targets that are difficult to identify with commonly applied target identification methods. Thereby, unbiased profiling can enable identification of novel bioactivity even for annotated compounds. We report the identification of a large bioactivity cluster comprised of numerous well-characterized drugs with different primary targets using a combination of the morphological Cell Painting Assay and proteome profiling. Cluster members alter cholesterol homeostasis and localization due to their physicochemical properties that lead to protonation and accumulation in lysosomes, an increase in lysosomal pH, and a disturbed cholesterol homeostasis. The identified cluster enables identification of modulators of cholesterol homeostasis and links regulation of genes or proteins involved in cholesterol synthesis or trafficking to physicochemical properties rather than to nominal targets.

Integrative Characterization of Immune-relevant Genes in Hepatocellular Carcinoma

BACKGROUND AND AIMS

Tumor microenvironment plays an essential role in cancer development and progression. Cancer immunotherapy has become a promising approach for the treatment of hepatocellular carcinoma (HCC). We aimed to analyze the HCC immune microenvironment characteristics to identify immune-related genetic changes.

METHODS

Key immune-relevant genes (KIRGs) were obtained through integrating the differentially expressed genes of The Cancer Genome Atlas, immune genes from the Immunology Database and Analysis Portal, and immune differentially expressed genes determined by single-sample gene set enrichment analysis scores. Cox regression analysis was performed to mine therapeutic target genes. A regulatory network based on KIRGs, transcription factors, and immune-related long non-coding RNAs (IRLncRNAs) was also generated. The outcomes of risk score model were validated in a testing cohort and in clinical samples using tissue immunohistochemistry staining. Correlation analysis between risk score and immune checkpoint genes and immune cell infiltration were investigated.

RESULTS

In total, we identified 21 KIRGs, including programmed cell death-1 (PD-1) and cytotoxic T-lymphocyte associated protein 4 (CTLA4), and found IKBKE, IL2RG, EDNRA, and IGHA1 may be equally important to PD-1 or CTLA4. Meanwhile, KIRGs, various transcription factors, and IRLncRNAs were integrated to reveal that the NRF1-AC127024.5-IKBKE axis might be involved in tumor immunity regulation. Furthermore, the immune-related risk score model was established according to KIRGs and key IRLncRNAs, and verified more obvious discriminating power in the testing cohort. Correlation analysis indicated TNFSF4 , LGALS9 , KIAA1429 , IDO2, and CD276 were closely related to the risk score, and CD4 T cells, macrophages, and neutrophils were the primary immune infiltration cell types.

CONCLUSIONS

Our results highlight the importance of immune genes in the HCC microenvironment and further unravel the underlying molecular mechanisms in the development of HCC.