Abstract 5078: Single-cell characterization of a carcinosarcoma patient-derived xenograft model

Carcinosarcoma (CS) is a rare aggressive biphasic tumor characterized by the presence of both carcinomatous (epithelial) and sarcomatous (mesenchymal) components. It can occur in various organs, including uterus and ovaries, and often exhibits rapid progression. The biphasic nature of CS makes it challenging to treat. In this study, we characterized a patient-derived xenograft (PDX) model developed from a 27 year-old female with CS of unknown primary origin in the NCI Patient-Derived Models Repository (https://pdmr.cancer.gov) to understand the transcriptomic features of these different components. PDX samples of the tumor were iteratively passaged in separate NSG mice, and samples from different lineages across passages P0-P2 were collected for analysis. Twenty-five separate carcinoma/sarcoma regions from 7 samples through laser capture microdissection were obtained for bulk RNA-seq. Fourteen samples were sequenced through single-cell RNA-seq using 10x Genomics Single Cell Chromium, including 2 samples in P0, 8 in P1 and 4 in P2. Mouse reads were excluded using XenoCell. Human-only reads were processed through standard QC filtering, integration and annotation with Cell Ranger, Seurat and other R packages. After quality-based filtering, we obtained 24,648 single cells from the integrated tumor samples. Through unsupervised clustering, cells separated into 16 Seurat-identified clusters, with consistent percentages of carcinoma and sarcoma components in P0, P1 and P2. Based on gene expression signature scores, each cluster was given a label descriptive of the cluster status. Five clusters had more epithelial/carcinoma characteristics (including one with a distinct ciliated signature), 2 clusters displayed a clear mesenchymal/sarcoma phenotype, 4 had low reads or gene counts, 2 had high cell cycle signature score and 3 clusters did not have a distinct carcinoma or sarcoma status. We identified differentially expressed genes in each annotated cluster relative to the other clusters and performed gene set enrichment analysis. Individual carcinoma clusters demonstrated upregulation of pathways related to apoptosis, hypoxia and inflammation (p-value < 0.05), while sarcoma clusters demonstrated upregulation of WNT signaling, EMT and angiogenesis (p-value < 0.05). Clusters with unclear carcinoma/sarcoma phenotype showed patterns of expression similar to carcinoma clusters, with the additional activation of oxidative phosphorylation and DNA repair pathways. Single cell data for a CS PDX model show that the biphasic nature of this histology is stable through passaging. Using single cell data, we identified the presence of clusters that are not fully committed to either the carcinoma or sarcoma phenotype. These clusters could represent transitional cell populations which may provide insight into biological drivers of cell plasticity in this tumor histology. Citation Format: Alida Palmisano, Yuri Kotliarov, Parimal Kumar, Ting-Chia Chang, Li Chen, Brandie Fullmer, Alyssa Chapman, Amanda Peach, Biswajit Das, Tomas Vilimas, Thomas Forbes, Shahanawaz Jiwani, Melinda G. Hollingshead, Yvonne A. Evrard, Sarah Shin, Hari Sankaran, Julia Krushkal, Alice Chen, Yingdong Zhao, Chris A. Karlovich, Mickey Williams, Lisa M. McShane, James H. Doroshow. Single-cell characterization of a carcinosarcoma patient-derived xenograft model [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 5078.