RNA profiling technologies at single-cell resolutions, including single-cell and single-nuclei RNA sequencing (scRNA-Seq and snRNA-Seq, scnRNA-Seq for short), can help characterize the composition of tissues and reveal cells that influence key functions in both healthy and disease tissues. However, the use of these technologies is operationally challenging because of high costs and stringent sample-collection requirements. Computational deconvolution methods that infer the composition of bulk-profiled samples using scnRNA-Seq-characterized cell types can broaden scnRNA-Seq applications, but their effectiveness remains controversial. We produced the first systematic evaluation of deconvolution methods on datasets with either known or scnRNA-Seq-estimated compositions. Our analyses revealed biases that are common to scnRNA-Seq 10X Genomics assays and illustrated the importance of accurate and properly controlled data preprocessing and method selection and optimization. Moreover, our results suggested that concurrent RNA-Seq and scnRNA-Seq profiles can help improve the accuracy of both scnRNA-Seq preprocessing and the deconvolution methods that employ them. Indeed, our proposed method, Single-cell RNA Quantity Informed Deconvolution (SQUID), combined RNA-Seq transformation and dampened weighted least-squares deconvolution approaches to consistently outperform other methods in predicting the composition of cell mixtures and tissue samples. Furthermore, our analysis suggested that only SQUID could identify outcomes-predictive cancer cell subclones in pediatric acute myeloid leukemia and neuroblastoma datasets, suggesting that deconvolution accuracy improvements are vital to enabling its applications in the life sciences.

No abstract

Hepatoblastoma (HB) is the most common pediatric primary liver malignancy, and survival for high-risk disease approaches 50%. Mouse models of HB fail to recapitulate hallmarks of high-risk disease. The aim of this work was to generate murine models that show high-risk features including multifocal tumors, vascular invasion, metastasis, and circulating tumor cells (CTCs). HepT1 cells were injected into the livers or tail veins of mice, and tumor growth was monitored with magnetic resonance and bioluminescent imaging. Blood was analyzed with fluorescence activated cell sorting to identify CTCs. Intra- and extra-hepatic tumor samples were harvested for immunohistochemistry and RNA and DNA sequencing. Cell lines were grown from tumor samples and profiled with RNA sequencing. With intrahepatic injection of HepT1 cells, 100% of animals grew liver tumors and showed vascular invasion, metastasis, and CTCs. Mutation profiling revealed genetic alterations in seven cancer-related genes, while transcriptomic analyses showed changes in gene expression with cells that invade vessels. Tail vein injection of HepT1 cells resulted in multifocal, metastatic disease. These unique models will facilitate further meaningful studies of high-risk HB.

Introduction: Hepatoblastoma (HB) is the most common pediatric primary liver tumor and has the fastest rising incidence of all pediatric solid tumors. Patients with high-risk, treatment refractory, or relapse disease have a survival rate of less than 50%. The development of clinically relevant models of these aggressive tumors will facilitate studies to identify drugs that target these cells.Methods: Fresh, whole primary tumor samples were implanted into the livers of immunocompromised mice. Tumor growth was monitored with MRI and ELISA to measure serum human Alpha-fetoprotein (AFP), which is detectable in the blood of tumor-bearing animals. Tumors were validated with immunohistochemistry (IHC) for HB markers Glypican-3 (GPC3) and Beta-catenin; short tandem repeat (STR) DNA validation; next generation sequencing-based mutation profiling of 124 genes involved in pediatric solid tumors; RNA sequencing (RNA-seq), and single cell RNA-seq (scRNA-seq). Lung metastasis was also detected in models with serial sectioning and H&E staining. Cells derived from tumors were grown in vitro in adherent and spheroid conditions and used for high throughput drug screening of candidate agents. Tumors were serially passaged in animals for further in vivo drug testing of novel targeted agents.Results: Nine patient-derived xenograft (PDX) models were generated that represent low- and high-risk tumors, treatment refractory cases, and relapsed tumors. Passaging of these models showed consistent implantation rates at or above 80% with tumors detectable in 2 to 4 weeks. Eight of nine models secrete human serum AFP. All models mimic gene expression and histological patterns of their primary tumor counterparts as well as identical STR DNA profiles. The models also show gene expression consistent with an HB2/high-risk profile according to the Sumazin HB expression signature. Interestingly, two models represent unique sub-clones of a very aggressive HB relapse with different AFP secretion and transcriptomic expression. scRNA-seq of these two models indicated outgrowth of disparate disease sub-clones. The nine models also demonstrate a range of DNA mutations with three or four mutations per tumor; all variants present in the original clinical samples were conserved in the PDX models. Lung metastasis was evident in six of nine models. Two stable patient-derived cell lines (PDCLs) were developed from models, and these cell lines show expression of HB markers and secrete AFP with growth in culture. Drug screening of adherent and spheroid tumor cells support the efficacy of novel targeted agents and indicate a spectrum of sensitivity to cisplatin, a frontline standard chemotherapy agent. Importantly, the models replicate the chemotherapy responses of the corresponding patients. Additional in vitro and in vivo work showed the efficacy of a histone deacetylase inhibitor, panobinostat.Conclusions: These novel orthotopic PDX models of HB fully recapitulate the primary tumors and represent a platform for clinically relevant drug screening and testing. Citation Format: Sarah E Woodfield, Roma H Patel, Andres F Espinoza, Richard S Whitlock, Jessica Epps, Andrew Badachhape, Samuel R Larson, Rohit K Srivastava, Aayushi P Shah, Saiabhiroop R Govindu, Barry Zorman, Brandon J Mistretta, Kevin E Fisher, Ilavarasi Gandhi, Jacquelyn Reuther, Martin Urbicain, Aryana M Ibarra, Sakuni Rankothgedera, Kimberly R Holloway, Stephen F Sarabia, Andras Heczey, Ketan B Ghaghada, Kalyani R Patel, Dolores Lopez-Terrada, Angshumoy Roy, Preethi H Gunaratne, Pavel Sumazin, Sanjeev A Vasudevan. Patient-derived xenograft mouse models of hepatoblastoma for a personalized medicine pipeline [abstract]. In: Proceedings of the AACR Special Conference: Advances in the Pathogenesis and Molecular Therapies of Liver Cancer; 2022 May 5-8; Boston, MA. Philadelphia (PA): AACR; Clin Cancer Res 2022;28(17_Suppl):Abstract nr PO013.