Proteins in Tumor-Derived Plasma Extracellular Vesicles Indicate Tumor Origin.

Cancer-derived extracellular vesicles (EVs) promote tumorigenesis, premetastatic niche formation, and metastasis via their protein cargo. However, the proteins packaged by patient tumors into EVs cannot be determined in vivo because of the presence of EVs derived from other tissues. We therefore developed a cross-species proteomic method to quantify the human tumor-derived proteome of plasma EVs produced by patient-derived xenografts of four cancer types. Proteomic profiling revealed individualized packaging of novel protein cargo, and machine learning accurately classified the type of the underlying tumor.

Diptoindonesin G is a middle domain HSP90 modulator for cancer treatment.

HSP90 inhibitors can target many oncoproteins simultaneously, but none have made it through clinical trials due to dose-limiting toxicity and induction of heat shock response, leading to clinical resistance. We identified diptoindonesin G (dip G) as an HSP90 modulator that can promote degradation of HSP90 clients by binding to the middle domain of HSP90 (Kd = 0.13 ± 0.02 µM) without inducing heat shock response. This is likely because dip G does not interfere with the HSP90-HSF1 interaction like N-terminal inhibitors, maintaining HSF1 in a transcriptionally silent state. We found that binding of dip G to HSP90 promotes degradation of HSP90 client protein estrogen receptor α (ER), a major oncogenic driver protein in most breast cancers. Mutations in the ER ligand-binding domain (LBD) are an established mechanism of endocrine resistance and decrease the binding affinity of mainstay endocrine therapies targeting ER, reducing their ability to promote ER degradation or transcriptionally silence ER. Because dip G binds to HSP90 and does not bind to the LBD of ER, unlike endocrine therapies, it is insensitive to ER LBD mutations that drive endocrine resistance. Additionally, we determined that dip G promoted degradation of WT and mutant ER with similar efficacy, downregulated ER- and mutant ER-regulated gene expression, and inhibited WT and mutant cell proliferation. Our data suggest that dip G is not only a molecular probe to study HSP90 biology and the HSP90 conformation cycle, but also a new therapeutic avenue for various cancers, particularly endocrine-resistant breast cancer harboring ER LBD mutations.

Co-clinical FDG-PET radiomic signature in predicting response to neoadjuvant chemotherapy in triple-negative breast cancer.

PURPOSE: We sought to exploit the heterogeneity afforded by patient-derived tumor xenografts (PDX) to first, optimize and identify robust radiomic features to predict response to therapy in subtype-matched triple negative breast cancer (TNBC) PDX, and second, to implement PDX-optimized image features in a TNBC co-clinical study to predict response to therapy using machine learning (ML) algorithms. METHODS: TNBC patients and subtype-matched PDX were recruited into a co-clinical FDG-PET imaging trial to predict response to therapy. One hundred thirty-one imaging features were extracted from PDX and human-segmented tumors. Robust image features were identified based on reproducibility, cross-correlation, and volume independence. A rank importance of predictors using ReliefF was used to identify predictive radiomic features in the preclinical PDX trial in conjunction with ML algorithms: classification and regression tree (CART), Naïve Bayes (NB), and support vector machines (SVM). The top four PDX-optimized image features, defined as radiomic signatures (RadSig), from each task were then used to predict or assess response to therapy. Performance of RadSig in predicting/assessing response was compared to SUVmean, SUVmax, and lean body mass-normalized SULpeak measures. RESULTS: Sixty-four out of 131 preclinical imaging features were identified as robust. NB-RadSig performed highest in predicting and assessing response to therapy in the preclinical PDX trial. In the clinical study, the performance of SVM-RadSig and NB-RadSig to predict and assess response was practically identical and superior to SUVmean, SUVmax, and SULpeak measures. CONCLUSIONS: We optimized robust FDG-PET radiomic signatures (RadSig) to predict and assess response to therapy in the context of a co-clinical imaging trial.

RON signalling promotes therapeutic resistance in ESR1 mutant breast cancer.

BACKGROUND: Oestrogen Receptor 1 (ESR1) mutations are frequently acquired in oestrogen receptor (ER)-positive metastatic breast cancer (MBC) patients who were treated with aromatase inhibitors (AI) in the metastatic setting. Acquired ESR1 mutations are associated with poor prognosis and there is a lack of effective therapies that selectively target these cancers. METHODS: We performed a proteomic kinome analysis in ESR1 Y537S mutant cells to identify hyperactivated kinases in ESR1 mutant cells. We validated Recepteur d'Origine Nantais (RON) and PI3K hyperactivity through phospho-immunoblot analysis, organoid growth assays, and in an in vivo patient-derived xenograft (PDX) metastatic model. RESULTS: We demonstrated that RON was hyperactivated in ESR1 mutant models, and in acquired palbociclib-resistant (PalbR) models. RON and insulin-like growth factor 1 receptor (IGF-1R) interacted as shown through pharmacological and genetic inhibition and were regulated by the mutant ER as demonstrated by reduced phospho-protein expression with endocrine therapies (ET). We show that ET in combination with a RON inhibitor (RONi) decreased ex vivo organoid growth of ESR1 mutant models, and as a monotherapy in PalbR models, demonstrating its therapeutic efficacy. Significantly, ET in combination with the RONi reduced metastasis of an ESR1 Y537S mutant PDX model. CONCLUSIONS: Our results demonstrate that RON/PI3K pathway inhibition may be an effective treatment strategy in ESR1 mutant and PalbR MBC patients. Clinically our data predict that ET resistance mechanisms can also contribute to CDK4/6 inhibitor resistance.

Regulated Phosphosignaling Associated with Breast Cancer Subtypes and Druggability.

Aberrant phospho-signaling is a hallmark of cancer. We investigated kinase-substrate regulation of 33,239 phosphorylation sites (phosphosites) in 77 breast tumors and 24 breast cancer xenografts. Our search discovered 2134 quantitatively correlated kinase-phosphosite pairs, enriching for and extending experimental or binding-motif predictions. Among the 91 kinases with auto-phosphorylation, elevated EGFR, ERBB2, PRKG1, and WNK1 phosphosignaling were enriched in basal, HER2-E, Luminal A, and Luminal B breast cancers, respectively, revealing subtype-specific regulation. CDKs, MAPKs, and ataxia-telangiectasia proteins were dominant, master regulators of substrate-phosphorylation, whose activities are not captured by genomic evidence. We unveiled phospho-signaling and druggable targets from 113 kinase-substrate pairs and cascades downstream of kinases, including AKT1, BRAF and EGFR. We further identified kinase-substrate-pairs associated with clinical or immune signatures and experimentally validated activated phosphosites of ERBB2, EIF4EBP1, and EGFR. Overall, kinase-substrate regulation revealed by the largest unbiased global phosphorylation data to date connects driver events to their signaling effects.

Folate Receptor α-Targeted 89Zr-M9346A Immuno-PET for Image-Guided Intervention with Mirvetuximab Soravtansine in Triple-Negative Breast Cancer.

Folate receptor α (FRα) is a well-studied tumor biomarker highly expressed in many epithelial tumors such as breast, ovarian, and lung cancers. Mirvetuximab soravtansine (IMGN853) is the antibody-drug conjugate of FRα-binding humanized monoclonal antibody M9346A and cytotoxic maytansinoid drug DM4. IMGN853 is currently being evaluated in multiple clinical trials, in which the immunohistochemical evaluation of an archival tumor or biopsy specimen is used for patient screening. However, limited tissue collection may lead to inaccurate diagnosis due to tumor heterogeneity. Herein, we developed a zirconium-89 (89Zr)-radiolabeled M9346A (89Zr-M9346A) as an immuno-positron emission tomography (immuno-PET) radiotracer to evaluate FRα expression in triple-negative breast cancer (TNBC) patients, providing a novel means to guide intervention with therapeutic IMGN853. In this study, we verified the binding specificity and immunoreactivity of 89Zr-M9346A by in vitro studies in FRαhigh cells (HeLa) and FRαlow cells (OVCAR-3). In vivo PET/computed tomography (PET/CT) imaging in HeLa xenografts and TNBC patient-derived xenograft (PDX) mouse models with various levels of FRα expression demonstrated its targeting specificity and sensitivity. Following PET imaging, the treatment efficiencies of IMGN853, pemetrexed, IMGN853 + pemetrexed, paclitaxel, and saline were assessed in FRαhigh and FRαlow TNBC PDX models. The correlation between 89Zr-M9346A tumor uptake and treatment response using IMGN853 in FRαhigh TNBC PDX model suggested the potential of 89Zr-M9346A PET as a noninvasive tool to prescreen patients based on the in vivo PET imaging for IMGN853-targeted treatment.

Identifying biomarkers of breast cancer micrometastatic disease in bone marrow using a patient-derived xenograft mouse model.

BACKGROUND: Disseminated tumor cells (DTCs) found in the bone marrow (BM) of patients with breast cancer portend a poor prognosis and are thought to be intermediaries in the metastatic process. To assess the clinical relevance of a mouse model for identifying possible prognostic and predictive biomarkers of these cells, we have employed patient-derived xenografts (PDX) for propagating and molecularly profiling human DTCs. METHODS: Previously developed mouse xenografts from five breast cancer patients were further passaged by implantation into NOD/SCID mouse mammary fat pads. BM was collected from long bones at early, serial passages and analyzed for human-specific gene expression by qRT-PCR as a surrogate biomarker for the detection of DTCs. Microarray-based gene expression analyses were performed to compare expression profiles between primary xenografts, solid metastasis, and populations of BM DTCs. Differential patterns of gene expression were then compared to previously generated microarray data from primary human BM aspirates from patients with breast cancer and healthy volunteers. RESULTS: Human-specific gene expression of SNAI1, GSC, FOXC2, KRT19, and STAM2, presumably originating from DTCs, was detected in the BM of all xenograft mice that also developed metastatic tumors. Human-specific gene expression was undetectable in the BM of those xenograft lines with no evidence of distant metastases and in non-transplanted control mice. Comparative gene expression analysis of BM DTCs versus the primary tumor of one mouse line identified multiple gene transcripts associated with epithelial-mesenchymal transition, aggressive clinical phenotype, and metastatic disease development. Sixteen of the PDX BM associated genes also demonstrated a statistically significant difference in expression in the BM of healthy volunteers versus the BM of breast cancer patients with distant metastatic disease. CONCLUSION: Unique and reproducible patterns of differential gene expression can be identified that presumably originate from BM DTCs in mouse PDX lines. Several of these identified genes are also detected in the BM of patients with breast cancer who develop early metastases, which suggests that they may be clinically relevant biomarkers. The PDX model may also provide a clinically relevant system for analyzing and targeting these intermediaries of metastases.

QuantFusion: Novel Unified Methodology for Enhanced Coverage and Precision in Quantifying Global Proteomic Changes in Whole Tissues.

Single quantitative platforms such as label-based or label-free quantitation (LFQ) present compromises in accuracy, precision, protein sequence coverage, and speed of quantifiable proteomic measurements. To maximize the quantitative precision and the number of quantifiable proteins or the quantifiable coverage of tissue proteomes, we have developed a unified approach, termed QuantFusion, that combines the quantitative ratios of all peptides measured by both LFQ and label-based methodologies. Here, we demonstrate the use of QuantFusion in determining the proteins differentially expressed in a pair of patient-derived tumor xenografts (PDXs) representing two major breast cancer (BC) subtypes, basal and luminal. Label-based in-spectra quantitative peptides derived from amino acid-coded tagging (AACT, also known as SILAC) of a non-malignant mammary cell line were uniformly added to each xenograft with a constant predefined ratio, from which Ratio-of-Ratio estimates were obtained for the label-free peptides paired with AACT peptides in each PDX tumor. A mixed model statistical analysis was used to determine global differential protein expression by combining complementary quantifiable peptide ratios measured by LFQ and Ratio-of-Ratios, respectively. With minimum number of replicates required for obtaining the statistically significant ratios, QuantFusion uses the distinct mechanisms to "rescue" the missing data inherent to both LFQ and label-based quantitation. Combined quantifiable peptide data from both quantitative schemes increased the overall number of peptide level measurements and protein level estimates. In our analysis of the PDX tumor proteomes, QuantFusion increased the number of distinct peptide ratios by 65%, representing differentially expressed proteins between the BC subtypes. This quantifiable coverage improvement, in turn, not only increased the number of measurable protein fold-changes by 8% but also increased the average precision of quantitative estimates by 181% so that some BC subtypically expressed proteins were rescued by QuantFusion. Thus, incorporating data from multiple quantitative approaches while accounting for measurement variability at both the peptide and global protein levels make QuantFusion unique for obtaining increased coverage and quantitative precision for tissue proteomes.

Integrated Bottom-Up and Top-Down Proteomics of Patient-Derived Breast Tumor Xenografts.

Bottom-up proteomics relies on the use of proteases and is the method of choice for identifying thousands of protein groups in complex samples. Top-down proteomics has been shown to be robust for direct analysis of small proteins and offers a solution to the "peptide-to-protein" inference problem inherent with bottom-up approaches. Here, we describe the first large-scale integration of genomic, bottom-up and top-down proteomic data for the comparative analysis of patient-derived mouse xenograft models of basal and luminal B human breast cancer, WHIM2 and WHIM16, respectively. Using these well-characterized xenograft models established by the National Cancer Institute's Clinical Proteomic Tumor Analysis Consortium, we compared and contrasted the performance of bottom-up and top-down proteomics to detect cancer-specific aberrations at the peptide and proteoform levels and to measure differential expression of proteins and proteoforms. Bottom-up proteomic analysis of the tumor xenografts detected almost 10 times as many coding nucleotide polymorphisms and peptides resulting from novel splice junctions than top-down. For proteins in the range of 0-30 kDa, where quantitation was performed using both approaches, bottom-up proteomics quantified 3,519 protein groups from 49,185 peptides, while top-down proteomics quantified 982 proteoforms mapping to 358 proteins. Examples of both concordant and discordant quantitation were found in a ∼60:40 ratio, providing a unique opportunity for top-down to fill in missing information. The two techniques showed complementary performance, with bottom-up yielding eight times more identifications of 0-30 kDa proteins in xenograft proteomes, but failing to detect differences in certain posttranslational modifications (PTMs), such as phosphorylation pattern changes of alpha-endosulfine. This work illustrates the potency of a combined bottom-up and top-down proteomics approach to deepen our knowledge of cancer biology, especially when genomic data are available.

An Analysis of the Sensitivity of Proteogenomic Mapping of Somatic Mutations and Novel Splicing Events in Cancer.

Improvements in mass spectrometry (MS)-based peptide sequencing provide a new opportunity to determine whether polymorphisms, mutations, and splice variants identified in cancer cells are translated. Herein, we apply a proteogenomic data integration tool (QUILTS) to illustrate protein variant discovery using whole genome, whole transcriptome, and global proteome datasets generated from a pair of luminal and basal-like breast-cancer-patient-derived xenografts (PDX). The sensitivity of proteogenomic analysis for singe nucleotide variant (SNV) expression and novel splice junction (NSJ) detection was probed using multiple MS/MS sample process replicates defined here as an independent tandem MS experiment using identical sample material. Despite analysis of over 30 sample process replicates, only about 10% of SNVs (somatic and germline) detected by both DNA and RNA sequencing were observed as peptides. An even smaller proportion of peptides corresponding to NSJ observed by RNA sequencing were detected (<0.1%). Peptides mapping to DNA-detected SNVs without a detectable mRNA transcript were also observed, suggesting that transcriptome coverage was incomplete (∼80%). In contrast to germline variants, somatic variants were less likely to be detected at the peptide level in the basal-like tumor than in the luminal tumor, raising the possibility of differential translation or protein degradation effects. In conclusion, this large-scale proteogenomic integration allowed us to determine the degree to which mutations are translated and identify gaps in sequence coverage, thereby benchmarking current technology and progress toward whole cancer proteome and transcriptome analysis.

Quality Assessments of Long-Term Quantitative Proteomic Analysis of Breast Cancer Xenograft Tissues.

Clinical proteomics requires large-scale analysis of human specimens to achieve statistical significance. We evaluated the long-term reproducibility of an iTRAQ (isobaric tags for relative and absolute quantification)-based quantitative proteomics strategy using one channel for reference across all samples in different iTRAQ sets. A total of 148 liquid chromatography tandem mass spectrometric (LC-MS/MS) analyses were completed, generating six 2D LC-MS/MS data sets for human-in-mouse breast cancer xenograft tissues representative of basal and luminal subtypes. Such large-scale studies require the implementation of robust metrics to assess the contributions of technical and biological variability in the qualitative and quantitative data. Accordingly, we derived a quantification confidence score based on the quality of each peptide-spectrum match to remove quantification outliers from each analysis. After combining confidence score filtering and statistical analysis, reproducible protein identification and quantitative results were achieved from LC-MS/MS data sets collected over a 7-month period. This study provides the first quality assessment on long-term stability and technical considerations for study design of a large-scale clinical proteomics project.

Patient-derived xenograft (PDX) models in basic and translational breast cancer research.

Patient-derived xenograft (PDX) models of a growing spectrum of cancers are rapidly supplanting long-established traditional cell lines as preferred models for conducting basic and translational preclinical research. In breast cancer, to complement the now curated collection of approximately 45 long-established human breast cancer cell lines, a newly formed consortium of academic laboratories, currently from Europe, Australia, and North America, herein summarizes data on over 500 stably transplantable PDX models representing all three clinical subtypes of breast cancer (ER+, HER2+, and "Triple-negative" (TNBC)). Many of these models are well-characterized with respect to genomic, transcriptomic, and proteomic features, metastatic behavior, and treatment response to a variety of standard-of-care and experimental therapeutics. These stably transplantable PDX lines are generally available for dissemination to laboratories conducting translational research, and contact information for each collection is provided. This review summarizes current experiences related to PDX generation across participating groups, efforts to develop data standards for annotation and dissemination of patient clinical information that does not compromise patient privacy, efforts to develop complementary data standards for annotation of PDX characteristics and biology, and progress toward "credentialing" of PDX models as surrogates to represent individual patients for use in preclinical and co-clinical translational research. In addition, this review highlights important unresolved questions, as well as current limitations, that have hampered more efficient generation of PDX lines and more rapid adoption of PDX use in translational breast cancer research.

Reproducibility of Differential Proteomic Technologies in CPTAC Fractionated Xenografts.

The NCI Clinical Proteomic Tumor Analysis Consortium (CPTAC) employed a pair of reference xenograft proteomes for initial platform validation and ongoing quality control of its data collection for The Cancer Genome Atlas (TCGA) tumors. These two xenografts, representing basal and luminal-B human breast cancer, were fractionated and analyzed on six mass spectrometers in a total of 46 replicates divided between iTRAQ and label-free technologies, spanning a total of 1095 LC-MS/MS experiments. These data represent a unique opportunity to evaluate the stability of proteomic differentiation by mass spectrometry over many months of time for individual instruments or across instruments running dissimilar workflows. We evaluated iTRAQ reporter ions, label-free spectral counts, and label-free extracted ion chromatograms as strategies for data interpretation (source code is available from http://homepages.uc.edu/~wang2x7/Research.htm ). From these assessments, we found that differential genes from a single replicate were confirmed by other replicates on the same instrument from 61 to 93% of the time. When comparing across different instruments and quantitative technologies, using multiple replicates, differential genes were reproduced by other data sets from 67 to 99% of the time. Projecting gene differences to biological pathways and networks increased the degree of similarity. These overlaps send an encouraging message about the maturity of technologies for proteomic differentiation.

Genome remodelling in a basal-like breast cancer metastasis and xenograft.

Massively parallel DNA sequencing technologies provide an unprecedented ability to screen entire genomes for genetic changes associated with tumour progression. Here we describe the genomic analyses of four DNA samples from an African-American patient with basal-like breast cancer: peripheral blood, the primary tumour, a brain metastasis and a xenograft derived from the primary tumour. The metastasis contained two de novo mutations and a large deletion not present in the primary tumour, and was significantly enriched for 20 shared mutations. The xenograft retained all primary tumour mutations and displayed a mutation enrichment pattern that resembled the metastasis. Two overlapping large deletions, encompassing CTNNA1, were present in all three tumour samples. The differential mutation frequencies and structural variation patterns in metastasis and xenograft compared with the primary tumour indicate that secondary tumours may arise from a minority of cells within the primary tumour.

Ischemia in tumors induces early and sustained phosphorylation changes in stress kinase pathways but does not affect global protein levels.

Protein abundance and phosphorylation convey important information about pathway activity and molecular pathophysiology in diseases including cancer, providing biological insight, informing drug and diagnostic development, and guiding therapeutic intervention. Analyzed tissues are usually collected without tight regulation or documentation of ischemic time. To evaluate the impact of ischemia, we collected human ovarian tumor and breast cancer xenograft tissue without vascular interruption and performed quantitative proteomics and phosphoproteomics after defined ischemic intervals. Although the global expressed proteome and most of the >25,000 quantified phosphosites were unchanged after 60 min, rapid phosphorylation changes were observed in up to 24% of the phosphoproteome, representing activation of critical cancer pathways related to stress response, transcriptional regulation, and cell death. Both pan-tumor and tissue-specific changes were observed. The demonstrated impact of pre-analytical tissue ischemia on tumor biology mandates caution in interpreting stress-pathway activation in such samples and motivates reexamination of collection protocols for phosphoprotein analysis.

Comprehensive quantitative analysis of ovarian and breast cancer tumor peptidomes.

Aberrant degradation of proteins is associated with many pathological states, including cancers. Mass spectrometric analysis of tumor peptidomes, the intracellular and intercellular products of protein degradation, has the potential to provide biological insights on proteolytic processing in cancer. However, attempts to use the information on these smaller protein degradation products from tumors for biomarker discovery and cancer biology studies have been fairly limited to date, largely due to the lack of effective approaches for robust peptidomics identification and quantification and the prevalence of confounding factors and biases associated with sample handling and processing. Herein, we have developed an effective and robust analytical platform for comprehensive analyses of tissue peptidomes, which is suitable for high-throughput quantitative studies. The reproducibility and coverage of the platform, as well as the suitability of clinical ovarian tumor and patient-derived breast tumor xenograft samples with postexcision delay of up to 60 min before freezing for peptidomics analysis, have been demonstrated. Moreover, our data also show that the peptidomics profiles can effectively separate breast cancer subtypes, reflecting tumor-associated protease activities. Peptidomics complements results obtainable from conventional bottom-up proteomics and provides insights not readily obtainable from such approaches.

O-GlcNAcylation of MITF regulates its activity and CDK4/6 inhibitor resistance in breast cancer.

Cyclin-dependent kinases 4 and 6 (CDK4/6) play a pivotal role in cell cycle and cancer development. Targeting CDK4/6 has demonstrated promising effects against breast cancer. However, resistance to CDK4/6 inhibitors (CDK4/6i), such as palbociclib, remains a substantial challenge in clinical settings. Using high-throughput combinatorial drug screening and genomic sequencing, we find that the microphthalmia-associated transcription factor (MITF) is activated via O-GlcNAcylation by O-GlcNAc transferase (OGT) in palbociclib-resistant breast cancer cells and tumors. Mechanistically, O-GlcNAcylation of MITF at Serine 49 enhances its interaction with importin α/ß, thus promoting its translocation to nuclei, where it suppresses palbociclib-induced senescence. Inhibition of MITF or its O-GlcNAcylation re-sensitizes resistant cells to palbociclib. Moreover, clinical studies confirm the activation of MITF in tumors from patients who are palbociclib-resistant or undergoing palbociclib treatment. Collectively, our studies shed light on the mechanism regulating palbociclib resistance and present clinical evidence for developing therapeutic approaches to treat CDK4/6i-resistant breast cancer patients.

Evaluation of Copanlisib in Combination with Eribulin in Triple-negative Breast Cancer Patient-derived Xenograft Models.

The PI3K pathway regulates essential cellular functions and promotes chemotherapy resistance. Activation of PI3K pathway signaling is commonly observed in triple-negative breast cancer (TNBC). However previous studies that combined PI3K pathway inhibitors with taxane regimens have yielded inconsistent results. We therefore set out to examine whether the combination of copanlisib, a clinical grade pan-PI3K inhibitor, and eribulin, an antimitotic chemotherapy approved for taxane-resistant metastatic breast cancer, improves the antitumor effect in TNBC. A panel of eight TNBC patient-derived xenograft (PDX) models was tested for tumor growth response to copanlisib and eribulin, alone or in combination. Treatment-induced signaling changes were examined by reverse phase protein array, immunohistochemistry (IHC) and 18F-fluorodeoxyglucose PET (18F-FDG PET). Compared with each drug alone, the combination of eribulin and copanlisib led to enhanced tumor growth inhibition, which was observed in both eribulin-sensitive and -resistant TNBC PDX models, regardless of PI3K pathway alterations or PTEN status. Copanlisib reduced PI3K signaling and enhanced eribulin-induced mitotic arrest. The combination enhanced induction of apoptosis compared with each drug alone. Interestingly, eribulin upregulated PI3K pathway signaling in PDX tumors, as demonstrated by increased tracer uptake by 18F-FDG PET scan and AKT phosphorylation by IHC. These changes were inhibited by the addition of copanlisib. These data support further clinical development for the combination of copanlisib and eribulin and led to a phase I/II trial of copanlisib and eribulin in patients with metastatic TNBC. SIGNIFICANCE: In this research, we demonstrated that the pan-PI3K inhibitor copanlisib enhanced the cytotoxicity of eribulin in a panel of TNBC PDX models. The improved tumor growth inhibition was irrespective of PI3K pathway alteration and was corroborated by the enhanced mitotic arrest and apoptotic induction observed in PDX tumors after combination therapy compared with each drug alone. These data provide the preclinical rationale for the clinical testing in TNBC.

A Pan-Cancer Patient-Derived Xenograft Histology Image Repository with Genomic and Pathologic Annotations Enables Deep Learning Analysis.

Patient-derived xenografts (PDX) model human intra- and intertumoral heterogeneity in the context of the intact tissue of immunocompromised mice. Histologic imaging via hematoxylin and eosin (H&E) staining is routinely performed on PDX samples, which could be harnessed for computational analysis. Prior studies of large clinical H&E image repositories have shown that deep learning analysis can identify intercellular and morphologic signals correlated with disease phenotype and therapeutic response. In this study, we developed an extensive, pan-cancer repository of >1,000 PDX and paired parental tumor H&E images. These images, curated from the PDX Development and Trial Centers Research Network Consortium, had a range of associated genomic and transcriptomic data, clinical metadata, pathologic assessments of cell composition, and, in several cases, detailed pathologic annotations of neoplastic, stromal, and necrotic regions. The amenability of these images to deep learning was highlighted through three applications: (i) development of a classifier for neoplastic, stromal, and necrotic regions; (ii) development of a predictor of xenograft-transplant lymphoproliferative disorder; and (iii) application of a published predictor of microsatellite instability. Together, this PDX Development and Trial Centers Research Network image repository provides a valuable resource for controlled digital pathology analysis, both for the evaluation of technical issues and for the development of computational image-based methods that make clinical predictions based on PDX treatment studies. Significance: A pan-cancer repository of >1,000 patient-derived xenograft hematoxylin and eosin-stained images will facilitate cancer biology investigations through histopathologic analysis and contributes important model system data that expand existing human histology repositories.