Heterogeneity In Drug Response Research Articles

Reducing State Conflicts between Network Motifs Synergistically Enhances Cancer Drug Effects and Overcomes Adaptive Resistance.

Inducing apoptosis in cancer cells is a primary goal in anti-cancer therapy, but curing cancer with a single drug is unattainable due to drug resistance. The complex molecular network in cancer cells causes heterogeneous responses to single-target drugs, thereby inducing an adaptive drug response. Here, we showed that targeted drug perturbations can trigger state conflicts between multi-stable motifs within a molecular regulatory network, resulting in heterogeneous drug responses. However, we revealed that properly regulating an interconnecting molecule between these motifs can synergistically minimize the heterogeneous responses and overcome drug resistance. We extracted the essential cellular response dynamics of the Boolean network driven by the target node perturbation and developed an algorithm to identify a synergistic combinatorial target that can reduce heterogeneous drug responses. We validated the proposed approach using exemplary network models and a gastric cancer model from a previous study by showing that the targets identified with our algorithm can better drive the networks to desired states than those with other control theories. Of note, our approach suggests a new synergistic pair of control targets that can increase cancer drug efficacy to overcome adaptive drug resistance.

Uncovering PROTAC Sensitivity and Efficacy by Multidimensional Proteome Profiling: A Case for STAT3.

Proteolysis-targeting chimera (PROTAC) is a powerful technology that can effectively trigger the degradation of target proteins. The intricate interplay among various factors leads to a heterogeneous drug response, bringing about significant challenges in comprehending drug mechanisms. Our study applied data-independent acquisition-based mass spectrometry to multidimensional proteome profiling of PROTAC (DIA-MPP) to uncover the efficacy and sensitivity of the PROTAC compound. We profiled the signal transducer and activator of transcription 3 (STAT3) PROTAC degrader in six leukemia and lymphoma cell lines under multiple conditions, demonstrating the pharmacodynamic properties and downstream biological responses. Through comparison between sensitive and insensitive cell lines, we revealed that STAT1 can be regarded as a biomarker for STAT3 PROTAC degrader, which was validated in cells, patient-derived organoids, and mouse models. These results set an example for a comprehensive description of the multidimensional PROTAC pharmacodynamic response and PROTAC drug sensitivity biomarker exploration.

Orally Administered Drugs and Their Complicated Relationship with Our Gastrointestinal Tract.

Orally administered compounds represent the great majority of all pharmaceutical compounds produced for human use and are the most popular among patients since they are practical and easy to self-administer. Following ingestion, orally administered drugs begin a "perilous" journey down the gastrointestinal tract and their bioavailability is modulated by numerous factors. The gastrointestinal (GI) tract anatomy can modulate drug bioavailability and accounts for interpatient drug response heterogeneity. Furthermore, host genetics is a contributor to drug bioavailability modulation. Importantly, a component of the GI tract that has been gaining notoriety with regard to drug treatment interactions is the gut microbiota, which shares a two-way interaction with pharmaceutical compounds in that they can be influenced by and are able to influence administered drugs. Overall, orally administered drugs are a patient-friendly treatment option. However, during their journey down the GI tract, there are numerous host factors that can modulate drug bioavailability in a patient-specific manner.

ASO Visual Abstract: Intra- and Interpatient Drug Response Heterogeneity Exist in Patients Undergoing Cytoreductive Surgery and Hyperthermic Intraperitoneal Chemotherapy for Non-gynecologic Cancers.

ASO Visual Abstract: Intra- and Interpatient Drug Response Heterogeneity Exist in Patients Undergoing Cytoreductive Surgery and Hyperthermic Intraperitoneal Chemotherapy for Non-gynecologic Cancers.

Platinum-based drugs induce phenotypic alterations in nucleoli and Cajal bodies in prostate cancer cells

PurposePlatinum-based drugs are cytotoxic drugs commonly used in cancer treatment. They cause DNA damage, effects of which on chromatin and cellular responses are relatively well described. Yet, the nuclear stress responses related to RNA processing are incompletely known and may be relevant for the heterogeneity with which cancer cells respond to these drugs. Here, we determine the type and extent of nuclear stress responses of prostate cancer cells to clinically relevant platinum drugs.MethodsWe study nucleolar and Cajal body (CB) responses to cisplatin, carboplatin, and oxaliplatin with immunofluorescence-based methods in prostate cancer cells. We utilize organelle-specific markers NPM, Fibrillarin, Coilin, and SMN1, and study CB-regulatory proteins FUS and TDP-43 using siRNA-mediated downregulation.ResultsDifferent types of prostate cancer cells have different sensitivities to platinum drugs. With equally cytotoxic doses, cisplatin, and oxaliplatin induce prominent nucleolar and CB stress responses while the nuclear stress phenotypes to carboplatin are milder. We find that Coilin is a stress-specific marker for platinum drug response heterogeneity. We also find that CB-associated, stress-responsive RNA binding proteins FUS and TDP-43 control Coilin and CB biology in prostate cancer cells and, further, that TDP-43 is associated with stress-responsive CBs in prostate cancer cells.ConclusionOur findings provide insight into the heterologous responses of prostate cancer cells to different platinum drug treatments and indicate Coilin and TDP-43 as stress mediators in the varied outcomes. These results help understand cancer drug responses at a cellular level and have implications in tackling heterogeneity in cancer treatment outcomes.

Intra- and Interpatient Drug Response Heterogeneity Exist in Patients Undergoing Cytoreductive Surgery and Hyperthermic Intraperitoneal Chemotherapy for Nongynecologic Cancers.

Select patients with peritoneal metastases are treated with cytoreductive surgery and hyperthermic intraperitoneal chemotherapy (CRS/HIPEC). We assayed for intra- and interpatient drug response heterogeneity through testing of patient-derived tumor organoids (PDTOs). PDTOs were generated from CRS/HIPEC patients from December 2021 to September 2022 and subjected to an in vitro HIPEC drug screen. Drug response was assessed with a cell viability assay and cleaved caspase-3 staining. A total of 31 patients were consented for tissue collection. Viable tissue was harvested from 23, and PDTO generation was successful in 13 (56%). PDTOs were analyzed from six appendiceal, three colorectal, two small bowel, one gastric, and one adrenal tumor. Drug screen results were generated in as few as 7 days (62%), with an average time of 12 days. Most patients received mitomycin-C (MMC) intraoperatively (n = 9); however, in only three cases was this agent considered the optimal choice in vitro. Three sets of PDTOs were resistant (defined as > 50% PDTO viability) to all agents tested and two were pan-sensitive (defined as 3 or more agents with < 50% PDTO viability). In three patients, organoids were generated from multiple metastatic sites and intrapatient drug response heterogeneity was observed. Both intra- and interpatient drug response heterogeneity exist in patients undergoing CRS/HIPEC for nongynecologic abdominal cancers. Caution must be used when interpreting patient response to chemotherapeutic agents based on a single site of testing in those with metastatic disease.

Single cell proteomics analysis of drug response shows its potential as a drug discovery platform.

Single-cell analysis has clearly established itself in biology and biomedical fields as an invaluable tool that allows one to comprehensively understand the relationship between cells, including their types, states, transitions, trajectories, and spatial position. Scientific methods such as fluorescence labeling, nanoscale super-resolution microscopy, advances in single cell RNAseq and proteomics technologies, provide more detailed information about biological processes which were not evident with the analysis of bulk material. This new era of single-cell biology provides a better understanding of such complex biological systems as cancer, inflammation, immunity mechanism and aging processes, and opens the door into the field of drug response heterogeneity. The latest discoveries of cellular heterogeneity gives us a unique understanding of complex biological processes, such as disease mechanism, and will lead to new strategies for better and personalized treatment strategies. Recently, single-cell proteomics techniques that allow quantification of thousands of proteins from single mammalian cells have been introduced. Here we present an improved single-cell mass spectrometry-based proteomics platform called SCREEN (Single Cell pRotEomE aNalysis) for deep and high-throughput single-cell proteome coverage with high efficiency, less turnaround time and with an improved ability for protein quantitation across more cells than previously achieved. We applied this new platform to analyze the single-cell proteomic landscape under different drug treatment over time to uncover heterogeneity in cancer cell response, which for the first time, to our knowledge, has been achieved by mass spectrometry based analytical methods. We discuss challenges in single-cell proteomics, future improvements and general trends with the goal to encourage forthcoming technical developments.

Identification of hub genes to determine drug-disease correlation in breast carcinomas.

The exact molecular mechanism underlying the heterogeneous drug response against breast carcinoma remains to be fully understood. It is urgently required to identify key genes that are intricately associated with varied clinical response of standard anti-cancer drugs, clinically used to treat breast cancer patients. In the present study, the utility of transcriptomic data of breast cancer patients in discerning the clinical drug response using machine learning-based approaches were evaluated. Here, a computational framework has been developed which can be used to identify key genes that can be linked with clinical drug response and progression of cancer, offering an immense opportunity to predict potential prognostic biomarkers and therapeutic targets. The framework concerned utilizes DeSeq2, Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), Cytoscape, and machine learning techniques to find these crucial genes. Total RNA extraction and qRT-PCR were performed to quantify relative expression of few hub genes selected from the networks. In our study, we have experimentally checked the expression of few key hub genes like APOA2, DLX5, APOC3, CAMK2B, and PAK6 that were predicted to play an immense role in breast cancer tumorigenesis and progression in response to anti-cancer drug Paclitaxel. However, further experimental validations will be required to get mechanistic insights of these genes in regulating the drug response and cancer progression which will likely to play pivotal role in cancer treatment and precision oncology.

A human mucosal melanoma organoid platform for modeling tumor heterogeneity and exploring immunotherapy combination options.

Mucosal melanoma (MM), an aggressive rare subtype of melanoma, is distinct from cutaneous melanoma and has poor prognoses. We addressed the lack of cell models for MM by establishing 30 organoids of human oral MM (OMM), which retained major histopathological and functional features of parental tumors. Organoid groups derived from chronologically or intratumorally distinct lesions within the same individual displayed heterogeneous genetics, expression profiles, and drug responses, indicating rapid tumor evolution and poor clinical response. Furthermore, transcriptome analysis revealed receptor tyrosine kinases (RTKs) signaling, particularly NGFR, a nerve growth factor receptor, was significantly up-regulated in OMMs and organoids from patients resistant to anti-programmed cell death protein 1 (anti-PD-1) therapy. Combining anti-PD-1 with anlotinib (a phase 2 multitarget RTK inhibitor for OMM) or NGFR knockdown enhanced the effective activity of immune cells in organoid-immune cell coculture systems. Together, our study suggested that OMM organoids serve as faithful models for exploring tumor evolution and immunotherapy combination strategies.

SpaRx: elucidate single-cell spatial heterogeneity of drug responses for personalized treatment.

Spatial cellular authors heterogeneity contributes to differential drug responses in a tumor lesion and potential therapeutic resistance. Recent emerging spatial technologies such as CosMx, MERSCOPE and Xenium delineate the spatial gene expression patterns at the single cell resolution. This provides unprecedented opportunities to identify spatially localized cellular resistance and to optimize the treatment for individual patients. In this work, we present a graph-based domain adaptation model, SpaRx, to reveal the heterogeneity of spatial cellular response to drugs. SpaRx transfers the knowledge from pharmacogenomics profiles to single-cell spatial transcriptomics data, through hybrid learning with dynamic adversarial adaption. Comprehensive benchmarking demonstrates the superior and robust performance of SpaRx at different dropout rates, noise levels and transcriptomics coverage. Further application of SpaRx to the state-of-the-art single-cell spatial transcriptomics data reveals that tumor cells in different locations of a tumor lesion present heterogenous sensitivity or resistance to drugs. Moreover, resistant tumor cells interact with themselves or the surrounding constituents to form an ecosystem for drug resistance. Collectively, SpaRx characterizes the spatial therapeutic variability, unveils the molecular mechanisms underpinning drug resistance and identifies personalized drug targets and effective drug combinations.

The prediction of drug sensitivity by multi-omics fusion reveals the heterogeneity of drug response in pan-cancer

Cancer drug response prediction based on genomic information plays a crucial role in modern pharmacogenomics, enabling individualized therapy. Given the expensive and complexity of biological experiments, computational methods serve as effective tools in predicting cancer drug sensitivity. In this study, we proposed a novel method called Multi-Omics Integrated Collective Variational Autoencoders (MOICVAE), which leverages integrated omics knowledge, including genomic and transcriptomic data, to fill in missing cancer-drug associations and enhance drug sensitivity prediction. Our method employs an encoder-decoder network to learn latent feature representations from cell lines. These learned feature vectors are then fed into a collective variational autoencoder network to train an association matrix. We evaluated MOICVAE on the GDSC and CCLE benchmark datasets using 10-fold cross-validation and achieved impressive AUCs of 0.856 and 0.808, respectively, outperforming state-of-the-art methods. Furthermore, on the TCGA dataset, consisting of 25 drugs across 7 cancer types, MOICVAE exhibited an average AUC of 0.91 in predicting drug sensitivity. Additionally, significant differences were observed in survival, tumor inflammatory assessment, and tumor microenvironment between the predicted drug-sensitive and drug-resistant groups. These results are consistent with predictions made on the METABRIC dataset. Moreover, we discovered that fusing omics data based on mRNA and CNV (copy number variations) yielded superior results in drug sensitivity prediction. MOICVAE not only achieved higher accuracy in drug sensitivity prediction but also provided additional value for combining immunotherapy with chemotherapy, offering patients with more precise treatment options. The code and dataset for MOICVAE are freely available at https://github.com/wanggnoc/MOICVAE.

Viability Analysis and High-Content Live-Cell Imaging for Drug Testing in Prostate Cancer Xenograft-Derived Organoids

Tumor organoids have been pushed forward as advanced model systems for in vitro oncology drug testing, with the eventual goal to direct personalized cancer treatments. However, drug testing efforts suffer from a large variation in experimental conditions for organoid culturing and organoid treatment. Moreover, most drug tests are restricted to whole-well viability as the sole read-out, thereby losing important information about key biological aspects that might be impacted due to the use of administered drugs. These bulk read-outs also discard potential inter-organoid heterogeneity in drug responses. To tackle these issues, we developed a systematic approach for processing organoids from prostate cancer (PCa) patient-derived xenografts (PDXs) for viability-based drug testing and identified essential conditions and quality checks for consistent results. In addition, we generated an imaging-based drug testing procedure using high-content fluorescence microscopy in living PCa organoids to detect various modalities of cell death. Individual organoids and cell nuclei in organoids were segmented and quantified using a dye combination of Hoechst 33342, propidium iodide and Caspase 3/7 Green, allowing the identification of cytostatic and cytotoxic treatment effects. Our procedures provide important insights into the mechanistic actions of tested drugs. Moreover, these methods can be adapted for tumor organoids originating from other cancer types to increase organoid-based drug test validity, and ultimately, accelerate clinical implementation.

Ex vivo drug response heterogeneity reveals personalized therapeutic strategies for patients with multiple myeloma

Multiple myeloma (MM) is a plasma cell malignancy defined by complex genetics and extensive patient heterogeneity. Despite a growing arsenal of approved therapies, MM remains incurable and in need of guidelines to identify effective personalized treatments. Here, we survey the ex vivo drug and immunotherapy sensitivities across 101 bone marrow samples from 70 patients with MM using multiplexed immunofluorescence, automated microscopy and deep-learning-based single-cell phenotyping. Combined with sample-matched genetics, proteotyping and cytokine profiling, we map the molecular regulatory network of drug sensitivity, implicating the DNA repair pathway and EYA3 expression in proteasome inhibitor sensitivity and major histocompatibility complex class II expression in the response to elotuzumab. Globally, ex vivo drug sensitivity associated with bone marrow microenvironmental signatures reflecting treatment stage, clonality and inflammation. Furthermore, ex vivo drug sensitivity significantly stratified clinical treatment responses, including to immunotherapy. Taken together, our study provides molecular and actionable insights into diverse treatment strategies for patients with MM.

516 Functional Impact of Intratumoral Heterogeneity in Therapeutic Vulnerabilities Across Enhancing and Non-Enhancing Tumor Components of Glioblastoma: A Paradigm-Shifting Personalized Medicine Approach

INTRODUCTION: Glioblastoma (GBM) remains a fatal disease without effective targeted therapeutics. This is largely attributed to molecular and spatial heterogeneity and absence of clinically applicable models to guide patient care. Moreover, as standard surgical resection is aimed at contrast-enhancing (CE) tumor, the molecular landscape and treatment response of the remaining non-contrast enhancing (nCE) component remains unknown despite its active role in tumor progression and local recurrence. METHODS: Freshly resected tumor tissue was subjected to the generation of 3D tumor models (brain tumor initiating-enriched cell lines [BTICs] and microcancers). To identify the best treatment strategy, a panel of 25 brain penetrant drugs (alone/ in combination) targeting GBM drivers (i.e. RTK/Ras/PI3K, RB, p53/apoptosis, DNA damage repair, transcription addiction) was developed and used to test therapeutic vulnerabilities. Comprehensive sequencing including genome and transcriptome profiling and pathway analysis from these different tumor regions is ongoing. RESULTS: Thirty-two patients have been prospectively enrolled, with generation of 38 microcancers and 17 BTICs, and drug screening completed in 26 patients. For a subset, four tumor regions (two CE and two nCE) were collected under neuronavigation. Among 25 drugs tested, 5 ± 3 (mean ± SD) treatment strategies showed strong (≥70%) growth inhibition at Cmax, with inter-tumor drug response heterogeneity ranging from 18.9% to 5.4% (p = 0.02). Intra-tumor drug response heterogeneity varied from high (5/25 drugs tested) to low (0/25, p &lt; 0.05) and was independent of enhancement status or spatial proximity. CONCLUSIONS: This is the first study to systematically assess treatment response from both CE and nCE patient-derived GBM models. By interrogating molecular landscapes and vulnerability profiles, we will be able to personalize and prioritize effective targeted interventions for GBM treatment.

Phenotypic deconvolution in heterogeneous cancer cell populations using drug-screening data

Tumor heterogeneity is an important driver of treatment failure in cancer since therapies often select for drug-tolerant or drug-resistant cellular subpopulations that drive tumor growth and recurrence. Profiling the drug-response heterogeneity of tumor samples using traditional genomic deconvolution methods has yielded limited results, due in part to the imperfect mapping between genomic variation and functional characteristics. Here, we leverage mechanistic population modeling to develop a statistical framework for profiling phenotypic heterogeneity from standard drug-screen data on bulk tumor samples. This method, called PhenoPop, reliably identifies tumor subpopulations exhibiting differential drug responses and estimates their drug sensitivities and frequencies within the bulk population. We apply PhenoPop to synthetically generated cell populations, mixed cell-line experiments, and multiple myeloma patient samples and demonstrate how it can provide individualized predictions of tumor growth under candidate therapies. This methodology can also be applied to deconvolution problems in a variety of biological settings beyond cancer drug response.

The complex nature of heterogeneity and its roles in breast cancer biology and therapeutic responsiveness.

Heterogeneity is a complex feature of cells and tissues with many interacting components. Depending on the nature of the research context, interacting features of cellular, drug response, genetic, molecular, spatial, temporal, and vascular heterogeneity may be present. We describe the various forms of heterogeneity with examples of their interactions and how they play a role in affecting cellular phenotype and drug responses in breast cancer. While cellular heterogeneity may be the most widely described and invoked, many forms of heterogeneity are evident within the tumor microenvironment and affect responses to the endocrine and cytotoxic drugs widely used in standard clinical care. Drug response heterogeneity is a critical determinant of clinical response and curative potential and also is multifaceted when encountered. The interactive nature of some forms of heterogeneity is readily apparent. For example, the process of metastasis has the properties of both temporal and spatial heterogeneity within the host, whereas each individual metastatic deposit may exhibit cellular, genetic, molecular, and vascular heterogeneity. This review describes the many forms of heterogeneity, their integrated activities, and offers some insights into how heterogeneity may be understood and studied in the future.

Multiparametric Tumor Organoid Drug Screening Using Widefield Live-Cell Imaging for Bulk and Single-Organoid Analysis.

Patient-derived tumor organoids (PDTOs) hold great promise for preclinical and translational research and predicting the patient therapy response from ex vivo drug screenings. However, current adenosine triphosphate (ATP)-based drug screening assays do not capture the complexity of a drug response (cytostatic or cytotoxic) and intratumor heterogeneity that has been shown to be retained in PDTOs due to a bulk readout. Live-cell imaging is a powerful tool to overcome this issue and visualize drug responses more in-depth. However, image analysis software is often not adapted to the three-dimensionality of PDTOs, requires fluorescent viability dyes, or is not compatible with a 384-well microplate format. This paper describes a semi-automated methodology to seed, treat, and image PDTOs in a high-throughput, 384-well format using conventional, widefield, live-cell imaging systems. In addition, we developed viability marker-free image analysis software to quantify growth rate-based drug response metrics that improve reproducibility and correct growth rate variations between different PDTO lines. Using the normalized drug response metric, which scores drug response based on the growth rate normalized to a positive and negative control condition, and a fluorescent cell death dye, cytotoxic and cytostatic drug responses can be easily distinguished, profoundly improving the classification of responders and non-responders. In addition, drug-response heterogeneity can by quantified from single-organoid drug response analysis to identify potential, resistant clones. Ultimately, this method aims to improve the prediction of clinical therapy response by capturing a multiparametric drug response signature, which includes kinetic growth arrest and cell death quantification.

EXTH-28. INTRATUMORAL HETEROGENEITY OF THERAPEUTIC VULNERABILITIES ACROSS ENHANCING AND NON-ENHANCING TUMOR COMPONENTS OF GLIOBLASTOMA: A PARADIGM-SHIFTING PERSONALIZED MEDICINE APPROACH

Abstract Glioblastoma (GBM) is the most aggressive adult brain tumor with no effective therapy. Despite advances on its molecular characterization, targeted therapeutics remain out of reach. This lack-of-progress is largely attributed to molecular and spatial heterogeneity and absence of clinically applicable models to guide patient care. Moreover, as standard surgical resection is aimed at contrast-enhancing (CE) tumor, the molecular landscape and treatment response of the remaining non-contrast enhancing (nCE) component is largely unknown despite its active role in tumor progression and local recurrence. Here, we aimed to identify the functional impact of tumor heterogeneity and to explore therapeutic vulnerabilities of both CE and nCE components. Four tumor regions (two CE and two nCE) from each of five GBM patients were collected under neuronavigation. The resulting tissue was subjected to generation of 3D tumor models (a brain tumor initiating-enriched cell line and a spheroid-based mcancer model) with comparable success in both CE and nCE samples. A panel of 25 brain penetrant drugs (alone and in combination) targeting GBM drivers (i.e. RTK/Ras/PI3K, RB/cell cycle, p53/apoptosis, DNA damage repair, transcription addiction) was developed and used to test for therapeutic vulnerabilities. From each tumor region, 4.5 ± 3.7 (average ± SD) different targeting strategies were identified exhibiting at least 70% growth inhibition at Cmax. Drug response heterogeneity was relatively rare, varied among cases, and was independent of tumor enhancement status. To our knowledge, this is the first study to systemically assess treatment response from both CE and nCE patient-derived GBM models. Ongoing studies include comprehensive genome and transcriptome profiling and pathway analysis from these different tumor regions. We postulate that interrogation of molecular landscapes and vulnerability profiles of both CE and nCE tumor regions, will enable us to personalize and prioritize effective targeted interventions for GBM treatment.

MODL-04. ESTABLISHMENT OF A RELIABLE GLIOBLASTOMA TUMOR-ORGANOID PLATFORM ENABLES AUTOMATED HIGH-THROUGHPUT DRUG SCREENING ADVANCING PERSONALIZED MEDICINE

Abstract Patient-derived tumor organoids (TOs) generated from tumor tissues within a few days are promising new models since under optimal conditions they preserve genotype, phenotype, the tumor microenvironment and the cellular heterogeneity of the parental tumor. This exciting new technology could provide representative avatars of the patient tumor in large numbers and thus can be exploited for drug discovery and drug screening purposes more efficiently than conventional cell lines or xenografted mouse models. We established standardized TOs from fresh glioblastoma (GBM) tissue after enzymatic dissociation into single cell suspensions and seeding cells into anti-adhesive 384-well plates. Over 1000 standardized TOs were generated per patient with a success rate of over 90%. Morphologically, TOs formed after two days. Size was stable over 10 days, with a slow outgrowth in some cases. A steady increase of the ATP signal over time indicated proliferative activity in TOs. Immunofluorescent stainings for tumor cells (GFAP), immune cells (macrophages and T-cells), and extracellular matrix (Tensascin C) confirmed a strong resemblance of TOs to their corresponding parental tumor, which was further substantiated by RNA-seq and DNA-methylation. Next, automated robot-based high-throughput drug screening (aHTS) on TOs was performed using a drug library of 166 FDA-approved antineoplastic drugs. Cell viability was assessed by CellTiter-Glo3D. In total, aHTS on TOs from 11 GBM patients including two recurrent GBMs showed a strong heterogeneity of drug responses. The top ten best-performing drugs consisted of 35 compounds targeting 14 different modes of action. Proteasome and HDAC-inhibitors demonstrated the highest efficacy. Respective targets were validated by knockdown experiments in GBM cells. Altogether, our data corroborate the future use of standardized GBM TOs as patient avatars for personalized drug testing by utilizing an automated high-throughput drug screening platform to better account for inter-patient-individual drug responses.

Deep transfer learning of cancer drug responses by integrating bulk and single-cell RNA-seq data

Drug screening data from massive bulk gene expression databases can be analyzed to determine the optimal clinical application of cancer drugs. The growing amount of single-cell RNA sequencing (scRNA-seq) data also provides insights into improving therapeutic effectiveness by helping to study the heterogeneity of drug responses for cancer cell subpopulations. Developing computational approaches to predict and interpret cancer drug response in single-cell data collected from clinical samples can be very useful. We propose scDEAL, a deep transfer learning framework for cancer drug response prediction at the single-cell level by integrating large-scale bulk cell-line data. The highlight in scDEAL involves harmonizing drug-related bulk RNA-seq data with scRNA-seq data and transferring the model trained on bulk RNA-seq data to predict drug responses in scRNA-seq. Another feature of scDEAL is the integrated gradient feature interpretation to infer the signature genes of drug resistance mechanisms. We benchmark scDEAL on six scRNA-seq datasets and demonstrate its model interpretability via three case studies focusing on drug response label prediction, gene signature identification, and pseudotime analysis. We believe that scDEAL could help study cell reprogramming, drug selection, and repurposing for improving therapeutic efficacy.