Efficient computation of high-dimensional penalized generalized linear mixed models by latent factor modeling of the random effects.

Modern biomedical datasets are increasingly high-dimensional and exhibit complex correlation structures. Generalized linear mixed models (GLMMs) have long been employed to account for such dependencies. However, proper specification of the fixed and random effects in GLMMs is increasingly difficult in high dimensions, and computational complexity grows with increasing dimension of the random effects. We present a novel reformulation of the GLMM using a factor model decomposition of the random effects, enabling scalable computation of GLMMs in high dimensions by reducing the latent space from a large number of random effects to a smaller set of latent factors. We also extend our prior work to estimate model parameters using a modified Monte Carlo Expectation Conditional Minimization algorithm, allowing us to perform variable selection on both the fixed and random effects simultaneously. We show through simulation that through this factor model decomposition, our method can fit high-dimensional penalized GLMMs faster than comparable methods and more easily scale to larger dimensions not previously seen in existing approaches.

First-in-Class Humanized Antibody against Alternatively Spliced Tissue Factor Augments Anti-Metastatic Efficacy of Chemotherapy in a Preclinical Model of Pancreatic Ductal Adenocarcinoma.

Alternatively spliced tissue factor (asTF) promotes the progression of pancreatic ductal adenocarcinoma (PDAC) by activating ß1-integrins on PDAC cell surfaces. hRabMab1, a first-in-class humanized inhibitory anti-asTF antibody we recently developed, can suppress PDAC primary tumor growth as a single agent. Whether hRabMab1 has the potential to suppress metastases in PDAC is unknown. Following in vivo screening of three asTF-proficient human PDAC cell lines, we chose to make use of KRAS G12V-mutant human PDAC cell line PaCa-44, which yields aggressive primary orthotopic tumors with spontaneous spread to PDAC-relevant anatomical sites, along with concomitant severe leukocytosis. The experimental design featured orthotopic tumors formed by luciferase labeled PaCa-44 cells; administration of hRabMab1 alone or in combination with gemcitabine/paclitaxel (gem/PTX); and the assessment of the treatment outcomes on the primary tumor tissue as well as systemic spread. When administered alone, hRabMab1 exhibited poor penetration of tumor tissue; however, hRabMab1 was abundant in tumor tissue when co-administered with gem/PTX, which resulted in a significant decrease in tumor cell proliferation; leukocyte infiltration; and neovascularization. Gem/PTX alone reduced primary tumor volume, but not metastatic spread; only the combination of hRabMab1 and gem/PTX significantly reduced metastatic spread. RNA-seq analysis of primary tumors showed that the addition of hRabMab1 to gem/PTX enhanced the downregulation of tubulin binding and microtubule motor activity. In the liver, hRabMab1 reduced liver metastasis as a single agent. Only the combination of hRabMab1 and gem/PTX eliminated tumor cell-induced leukocytosis. We here demonstrate for the first time that hRabMab1 may help suppress metastasis in PDAC. hRabMab1's ability to improve the efficacy of chemotherapy is significant and warrants further investigation.

Myofibroblastic cancer-associated fibroblast subtype heterogeneity in pancreatic cancer.

BACKGROUND: Pancreatic ductal adenocarcinoma (PDAC) has a fibrotic stroma that has both tumor-promoting and tumor-restraining properties. Different types of cancer-associated fibroblasts (CAFs) have been described. Here, we investigated whether CAFs within the same subtype exhibit heterogeneous functions. METHODS: We evaluated the gene and protein expression differences in two myofibroblastic CAF (myCAF) lines using single-cell and bulk RNA-sequencing. We utilized proliferation and migration assays to determine the effect of different CAF lines on a tumor cell line. RESULTS: We found that myCAF lines express an activated stroma subtype gene signature, which is associated with a shorter survival in patients. Although both myCAF lines expressed α-smooth muscle actin (α-SMA), platelet-derived growth factor-α (PDGFR-α), fibroblast-activated protein (FAP), and vimentin, we observed heterogeneity between the two lines. Similarly, despite being consistent with myCAF gene expression overall, heterogeneity within specific genes was observed. We found that these differences extended to the functional level where the two myCAF lines had different effects on the same tumor cell line. The myCAF216 line, which had slightly increased inflammatory CAF-like gene expression and higher protein expression of α-SMA, PDGFR-α, and FAP was found to restrain migration of tumor cells. CONCLUSIONS: We found that two myCAF lines with globally similar expression characteristics had different effects on the same tumor cell line, one promoting and the other restraining migration. Our study highlights that there may be unappreciated heterogeneity within CAF subtypes. Further investigation and attention to specific genes or proteins that may drive this heterogeneity will be important.

Integrated single-dose kinome profiling data is predictive of cancer cell line sensitivity to kinase inhibitors.

Protein kinase activity forms the backbone of cellular information transfer, acting both individually and as part of a broader network, the kinome. Their central role in signaling leads to kinome dysfunction being a common driver of disease, and in particular cancer, where numerous kinases have been identified as having a causal or modulating role in tumor development and progression. As a result, the development of therapies targeting kinases has rapidly grown, with over 70 kinase inhibitors approved for use in the clinic and over double this number currently in clinical trials. Understanding the relationship between kinase inhibitor treatment and their effects on downstream cellular phenotype is thus of clear importance for understanding treatment mechanisms and streamlining compound screening in therapy development. In this work, we combine two large-scale kinome profiling data sets and use them to link inhibitor-kinome interactions with cell line treatment responses (AUC/IC50). We then built computational models on this data set that achieve a high degree of prediction accuracy (R2 of 0.7 and RMSE of 0.9) and were able to identify a set of well-characterized and understudied kinases that significantly affect cell responses. We further validated these models experimentally by testing predicted effects in breast cancer cell lines and extended the model scope by performing additional validation in patient-derived pancreatic cancer cell lines. Overall, these results demonstrate that broad quantification of kinome inhibition state is highly predictive of downstream cellular phenotypes.

Sub-Cluster Identification through Semi-Supervised Optimization of Rare-Cell Silhouettes (SCISSORS) in single-cell RNA-sequencing.

MOTIVATION: Single-cell RNA-sequencing (scRNA-seq) has enabled the molecular profiling of thousands to millions of cells simultaneously in biologically heterogenous samples. Currently, the common practice in scRNA-seq is to determine cell type labels through unsupervised clustering and the examination of cluster-specific genes. However, even small differences in analysis and parameter choosing can greatly alter clustering results and thus impose great influence on which cell types are identified. Existing methods largely focus on determining the optimal number of robust clusters, which can be problematic for identifying cells of extremely low abundance due to their subtle contributions toward overall patterns of gene expression. RESULTS: Here, we present a carefully designed framework, SCISSORS, which accurately profiles subclusters within broad cluster(s) for the identification of rare cell types in scRNA-seq data. SCISSORS employs silhouette scoring for the estimation of heterogeneity of clusters and reveals rare cells in heterogenous clusters by a multi-step semi-supervised reclustering process. Additionally, SCISSORS provides a method for the identification of marker genes of high specificity to the cell type. SCISSORS is wrapped around the popular Seurat R package and can be easily integrated into existing Seurat pipelines. AVAILABILITY AND IMPLEMENTATION: SCISSORS, including source code and vignettes, are freely available at https://github.com/jr-leary7/SCISSORS.

Identifying patient profiles of disparate care in resectable pancreas cancer using latent class analysis.

BACKGROUND AND OBJECTIVES: Disparities in pancreas cancer care are multifactorial, but factors are often examined in isolation. Research that integrates these factors in a single conceptual framework is lacking. We use latent class analysis (LCA) to evaluate the association between intersectionality and patterns of care and survival in patients with resectable pancreas cancer. METHODS: LCA was used to identify demographic profiles in resectable pancreas cancer (n = 140 344) diagnosed from 2004 to 2019 in the National Cancer Database (NCDB). LCA-derived patient profiles were used to identify differences in receipt of minimum expected treatment (definitive surgery), optimal treatment (definitive surgery and chemotherapy), time to treatment, and overall survival. RESULTS: Minimum expected treatment (hazard ratio [HR] 0.69, 95% confidence interval [CI]: 0.65, 0.75) and optimal treatment (HR 0.58, 95% CI: 0.55, 0.62) were associated with improved overall survival. Seven latent classes were identified based on age, race/ethnicity, and socioeconomic status (SES) attributes (zip code-linked education and income, insurance, geography). Compared to the referent group (≥65 years + White + med/high SES), the ≥65 years + Black profile had the longest time-to-treatment (24 days vs. 28 days) and lowest odds of receiving minimum (odds ratio [OR] 0.67, 95% CI: 0.64, 0.71) or optimal treatment (OR 0.76, 95% CI: 0.72, 0.81). The Hispanic patient profile had the lowest median overall survival-55.3 months versus 67.5 months. CONCLUSIONS: Accounting for intersectionality in the NCDB resectable pancreatic cancer patient cohort identifies subgroups at higher risk for inequities in care. LCA demonstrates that older Black patients and Hispanic patients are at particular risk for being underserved and should be prioritiz for directed interventions.

Open-source curation of a pancreatic ductal adenocarcinoma gene expression analysis platform (pdacR) supports a two-subtype model.

Pancreatic ductal adenocarcinoma (PDAC) is an aggressive disease for which potent therapies have limited efficacy. Several studies have described the transcriptomic landscape of PDAC tumors to provide insight into potentially actionable gene expression signatures to improve patient outcomes. Despite centralization efforts from multiple organizations and increased transparency requirements from funding agencies and publishers, analysis of public PDAC data remains difficult. Bioinformatic pitfalls litter public transcriptomic data, such as subtle inclusion of low-purity and non-adenocarcinoma cases. These pitfalls can introduce non-specificity to gene signatures without appropriate data curation, which can negatively impact findings. To reduce barriers to analysis, we have created pdacR ( http://pdacR.bmi.stonybrook.edu , github.com/rmoffitt/pdacR), an open-source software package and web-tool with annotated datasets from landmark studies and an interface for user-friendly analysis in clustering, differential expression, survival, and dimensionality reduction. Using this tool, we present a multi-dataset analysis of PDAC transcriptomics that confirms the basal-like/classical model over alternatives.

Balance between immunoregulatory B cells and plasma cells drives pancreatic tumor immunity.

Plasma cell responses are associated with anti-tumor immunity and favorable response to immunotherapy. B cells can amplify anti-tumor immune responses through antibody production; yet B cells in patients and tumor-bearing mice often fail to support this effector function. We identify dysregulated transcriptional program in B cells that disrupts differentiation of naive B cells into anti-tumor plasma cells. The signaling network contributing to this dysfunction is driven by interleukin (IL) 35 stimulation of a STAT3-PAX5 complex that upregulates the transcriptional regulator BCL6 in naive B cells. Transient inhibition of BCL6 in tumor-educated naive B cells is sufficient to reverse the dysfunction in B cell differentiation, stimulating the intra-tumoral accumulation of plasma cells and effector T cells and rendering pancreatic tumors sensitive to anti-programmed cell death protein 1 (PD-1) blockade. Our findings argue that B cell effector dysfunction in cancer can be due to an active systemic suppression program that can be targeted to synergize with T cell-directed immunotherapy.

Truncated O-Glycan-Bearing MUC16 Enhances Pancreatic Cancer Cells Aggressiveness via α4ß1 Integrin Complexes and FAK Signaling.

Elevated levels of Mucin-16 (MUC16) in conjunction with a high expression of truncated O-glycans is implicated in playing crucial roles in the malignancy of pancreatic ductal adenocarcinoma (PDAC). However, the mechanisms by which such aberrant glycoforms present on MUC16 itself promote an increased disease burden in PDAC are yet to be elucidated. This study demonstrates that the CRISPR/Cas9-mediated genetic deletion of MUC16 in PDAC cells decreases tumor cell migration. We found that MUC16 enhances tumor malignancy by activating the integrin-linked kinase and focal adhesion kinase (ILK/FAK)-signaling axis. These findings are especially noteworthy in truncated O-glycan (Tn and STn antigen)-expressing PDAC cells. Activation of these oncogenic-signaling pathways resulted in part from interactions between MUC16 and integrin complexes (α4ß1), which showed a stronger association with aberrant glycoforms of MUC16. Using a monoclonal antibody to functionally hinder MUC16 significantly reduced the migratory cascades in our model. Together, these findings suggest that truncated O-glycan containing MUC16 exacerbates malignancy in PDAC by activating FAK signaling through specific interactions with α4 and ß1 integrin complexes on cancer cell membranes. Targeting these aberrant glycoforms of MUC16 can aid in the development of a novel platform to study and treat metastatic pancreatic cancer.

Aberrant Expression and Subcellular Localization of ECT2 Drives Colorectal Cancer Progression and Growth.

ECT2 is an activator of RHO GTPases that is essential for cytokinesis. In addition, ECT2 was identified as an oncoprotein when expressed ectopically in NIH/3T3 fibroblasts. However, oncogenic activation of ECT2 resulted from N-terminal truncation, and such truncated ECT2 proteins have not been found in patients with cancer. In this study, we observed elevated expression of full-length ECT2 protein in preneoplastic colon adenomas, driven by increased ECT2 mRNA abundance and associated with APC tumor-suppressor loss. Elevated ECT2 levels were detected in the cytoplasm and nucleus of colorectal cancer tissue, suggesting cytoplasmic mislocalization as one mechanism of early oncogenic ECT2 activation. Importantly, elevated nuclear ECT2 correlated with poorly differentiated tumors, and a low cytoplasmic:nuclear ratio of ECT2 protein correlated with poor patient survival, suggesting that nuclear and cytoplasmic ECT2 play distinct roles in colorectal cancer. Depletion of ECT2 reduced anchorage-independent cancer cell growth and invasion independent of its function in cytokinesis, and loss of Ect2 extended survival in a Kras G12D Apc-null colon cancer mouse model. Expression of ECT2 variants with impaired nuclear localization or guanine nucleotide exchange catalytic activity failed to restore cancer cell growth or invasion, indicating that active, nuclear ECT2 is required to support tumor progression. Nuclear ECT2 promoted ribosomal DNA transcription and ribosome biogenesis in colorectal cancer. These results support a driver role for both cytoplasmic and nuclear ECT2 overexpression in colorectal cancer and emphasize the critical role of precise subcellular localization in dictating ECT2 function in neoplastic cells. SIGNIFICANCE: ECT2 overexpression and mislocalization support its role as a driver in colon cancer that is independent from its function in normal cell cytokinesis.

GM-CSF drives myelopoiesis, recruitment and polarisation of tumour-associated macrophages in cholangiocarcinoma and systemic blockade facilitates antitumour immunity.

OBJECTIVE: Intrahepatic cholangiocarcinoma (iCCA) is rising in incidence, and at present, there are limited effective systemic therapies. iCCA tumours are infiltrated by stromal cells, with high prevalence of suppressive myeloid populations including tumour-associated macrophages (TAMs) and myeloid-derived suppressor cells (MDSCs). Here, we show that tumour-derived granulocyte-macrophage colony-stimulating factor (GM-CSF) and the host bone marrow is central for monopoiesis and potentiation of TAMs, and abrogation of this signalling axis facilitates antitumour immunity in a novel model of iCCA. METHODS: Blood and tumours were analysed from iCCA patients and controls. Treatment and correlative studies were performed in mice with autochthonous and established orthotopic iCCA tumours treated with anti-GM-CSF monoclonal antibody. RESULTS: Systemic elevation in circulating myeloid cells correlates with poor prognosis in patients with iCCA, and patients who undergo resection have a worse overall survival if tumours are more infiltrated with CD68+ TAMs. Mice with spontaneous iCCA demonstrate significant elevation of monocytic myeloid cells in the tumour microenvironment and immune compartments, and tumours overexpress GM-CSF. Blockade of GM-CSF with a monoclonal antibody decreased tumour growth and spread. Mice bearing orthotopic tumours treated with anti-GM-CSF demonstrate repolarisation of immunosuppressive TAMs and MDSCs, facilitating T cell response and tumour regression. GM-CSF blockade dampened inflammatory gene networks in tumours and TAMs. Human tumours with decreased GM-CSF expression exhibit improved overall survival after resection. CONCLUSIONS: iCCA uses the GM-CSF-bone marrow axis to establish an immunosuppressive tumour microenvironment. Blockade of the GM-CSF axis promotes antitumour T cell immunity.

The Q61H mutation decouples KRAS from upstream regulation and renders cancer cells resistant to SHP2 inhibitors.

Cancer cells bearing distinct KRAS mutations exhibit variable sensitivity to SHP2 inhibitors (SHP2i). Here we show that cells harboring KRAS Q61H are uniquely resistant to SHP2i, and investigate the underlying mechanisms using biophysics, molecular dynamics, and cell-based approaches. Q61H mutation impairs intrinsic and GAP-mediated GTP hydrolysis, and impedes activation by SOS1, but does not alter tyrosyl phosphorylation. Wild-type and Q61H-mutant KRAS are both phosphorylated by Src on Tyr32 and Tyr64 and dephosphorylated by SHP2, however, SHP2i does not reduce ERK phosphorylation in KRAS Q61H cells. Phosphorylation of wild-type and Gly12-mutant KRAS, which are associated with sensitivity to SHP2i, confers resistance to regulation by GAP and GEF activities and impairs binding to RAF, whereas the near-complete GAP/GEF-resistance of KRAS Q61H remains unaltered, and high-affinity RAF interaction is retained. SHP2 can stimulate KRAS signaling by modulating GEF/GAP activities and dephosphorylating KRAS, processes that fail to regulate signaling of the Q61H mutant.

The PDAC Extracellular Matrix: A Review of the ECM Protein Composition, Tumor Cell Interaction, and Therapeutic Strategies.

Pancreatic ductal adenocarcinoma (PDAC) is notorious for a dense fibrotic stroma that is interlaced with a collagen-based extracellular matrix (ECM) that plays an important role in tumor biology. Traditionally thought to only provide a physical barrier from host responses and systemic chemotherapy, new studies have demonstrated that the ECM maintains biomechanical and biochemical properties of the tumor microenvironment (TME) and restrains tumor growth. Recent studies have shown that the ECM augments tumor stiffness, interstitial fluid pressure, cell-to-cell junctions, and microvascularity using a mix of biomechanical and biochemical signals to influence tumor fate for better or worse. In addition, PDAC tumors have been shown to use ECM-derived peptide fragments as a nutrient source in nutrient-poor conditions. While collagens are the most abundant proteins found in the ECM, several studies have identified growth factors, integrins, glycoproteins, and proteoglycans in the ECM. This review focuses on the dichotomous nature of the PDAC ECM, the types of collagens and other proteins found in the ECM, and therapeutic strategies targeting the PDAC ECM.

Pathology and Molecular Characteristics of Pancreatic Cancer.

Pancreatic ductal adenocarcinoma (PDAC) is the most common type of pancreatic cancer. However, it should be kept in mind that there are other pancreatic cancers that are classified by their cellular lineage: acinar cell carcinomas (acinar differentiation), neuroendocrine neoplasms (arising from the islets), solid-pseudopapillary neoplasms (showing no discernible cell lineage), and pancreatoblastomas (characterized by multiphenotypic differentiation, including acinar endocrine and ductal). This article focuses on the molecular and pathology alterations in PDAC.

High-Dimensional Precision Medicine From Patient-Derived Xenografts.

The complexity of human cancer often results in significant heterogeneity in response to treatment. Precision medicine offers the potential to improve patient outcomes by leveraging this heterogeneity. Individualized treatment rules (ITRs) formalize precision medicine as maps from the patient covariate space into the space of allowable treatments. The optimal ITR is that which maximizes the mean of a clinical outcome in a population of interest. Patient-derived xenograft (PDX) studies permit the evaluation of multiple treatments within a single tumor, and thus are ideally suited for estimating optimal ITRs. PDX data are characterized by correlated outcomes, a high-dimensional feature space, and a large number of treatments. Here we explore machine learning methods for estimating optimal ITRs from PDX data. We analyze data from a large PDX study to identify biomarkers that are informative for developing personalized treatment recommendations in multiple cancers. We estimate optimal ITRs using regression-based (Q-learning) and direct-search methods (outcome weighted learning). Finally, we implement a superlearner approach to combine multiple estimated ITRs and show that the resulting ITR performs better than any of the input ITRs, mitigating uncertainty regarding user choice. Our results indicate that PDX data are a valuable resource for developing individualized treatment strategies in oncology. Supplementary materials for this article are available online.

Gene Targeting in Disease Networks.

Profiling of whole transcriptomes has become a cornerstone of molecular biology and an invaluable tool for the characterization of clinical phenotypes and the identification of disease subtypes. Analyses of these data are becoming ever more sophisticated as we move beyond simple comparisons to consider networks of higher-order interactions and associations. Gene regulatory networks (GRNs) model the regulatory relationships of transcription factors and genes and have allowed the identification of differentially regulated processes in disease systems. In this perspective, we discuss gene targeting scores, which measure changes in inferred regulatory network interactions, and their use in identifying disease-relevant processes. In addition, we present an example analysis for pancreatic ductal adenocarcinoma (PDAC), demonstrating the power of gene targeting scores to identify differential processes between complex phenotypes, processes that would have been missed by only performing differential expression analysis. This example demonstrates that gene targeting scores are an invaluable addition to gene expression analysis in the characterization of diseases and other complex phenotypes.

Single-cell transcriptome analysis defines heterogeneity of the murine pancreatic ductal tree.

To study disease development, an inventory of an organ's cell types and understanding of physiologic function is paramount. Here, we performed single-cell RNA-sequencing to examine heterogeneity of murine pancreatic duct cells, pancreatobiliary cells, and intrapancreatic bile duct cells. We describe an epithelial-mesenchymal transitory axis in our three pancreatic duct subpopulations and identify osteopontin as a regulator of this fate decision as well as human duct cell dedifferentiation. Our results further identify functional heterogeneity within pancreatic duct subpopulations by elucidating a role for geminin in accumulation of DNA damage in the setting of chronic pancreatitis. Our findings implicate diverse functional roles for subpopulations of pancreatic duct cells in maintenance of duct cell identity and disease progression and establish a comprehensive road map of murine pancreatic duct cell, pancreatobiliary cell, and intrapancreatic bile duct cell homeostasis.

Isoforms of MUC16 activate oncogenic signaling through EGF receptors to enhance the progression of pancreatic cancer.

Aberrant expression of CA125/MUC16 is associated with pancreatic ductal adenocarcinoma (PDAC) progression and metastasis. However, knowledge of the contribution of MUC16 to pancreatic tumorigenesis is limited. Here, we show that MUC16 expression is associated with disease progression, basal-like and squamous tumor subtypes, increased tumor metastasis, and short-term survival of PDAC patients. MUC16 enhanced tumor malignancy through the activation of AKT and GSK3ß oncogenic signaling pathways. Activation of these oncogenic signaling pathways resulted in part from increased interactions between MUC16 and epidermal growth factor (EGF)-type receptors, which were enhanced for aberrant glycoforms of MUC16. Treatment of PDAC cells with monoclonal antibody (mAb) AR9.6 significantly reduced MUC16-induced oncogenic signaling. mAb AR9.6 binds to a unique conformational epitope on MUC16, which is influenced by O-glycosylation. Additionally, treatment of PDAC tumor-bearing mice with either mAb AR9.6 alone or in combination with gemcitabine significantly reduced tumor growth and metastasis. We conclude that the aberrant expression of MUC16 enhances PDAC progression to an aggressive phenotype by modulating oncogenic signaling through ErbB receptors. Anti-MUC16 mAb AR9.6 blocks oncogenic activities and tumor growth and could be a novel immunotherapeutic agent against MUC16-mediated PDAC tumor malignancy.

Modeling Between-Study Heterogeneity for Improved Replicability in Gene Signature Selection and Clinical Prediction.

In the genomic era, the identification of gene signatures associated with disease is of significant interest. Such signatures are often used to predict clinical outcomes in new patients and aid clinical decision-making. However, recent studies have shown that gene signatures are often not replicable. This occurrence has practical implications regarding the generalizability and clinical applicability of such signatures. To improve replicability, we introduce a novel approach to select gene signatures from multiple datasets whose effects are consistently non-zero and account for between-study heterogeneity. We build our model upon some rank-based quantities, facilitating integration over different genomic datasets. A high dimensional penalized Generalized Linear Mixed Model (pGLMM) is used to select gene signatures and address data heterogeneity. We compare our method to some commonly used strategies that select gene signatures ignoring between-study heterogeneity. We provide asymptotic results justifying the performance of our method and demonstrate its advantage in the presence of heterogeneity through thorough simulation studies. Lastly, we motivate our method through a case study subtyping pancreatic cancer patients from four gene expression studies.