118. Single-cell RNA sequencing and co-occurring cellular state analysis of high-grade serous ovarian cancer

Aly Abdelkareem
Nicholas Semenkovich

I’m a physician scientist and clinical fellow in Endocrinology, Metabolism, and Lipid Research at Washington University in St. Louis. I am currently a postdoctoral scholar with Dr. Aadel Chaudhuri working on cell-free DNA. I am developing machine learning methods to analyze cell free DNA and provide insights into chronic and metabolic disease (in addition to oncology). Prior to my fellowship, I completed an Internal Medicine residency at the Brigham and Women’s Hospital. I completed MD/PhD training at Washington University with Dr. Jeff Gordon exploring the impact of the gut microbiota on host epigenetic signaling, and my undergrad at MIT in Computer Science.


Nicholas Semenkovicha, Emilee Kotnika, Elena Lomonosovaa, Abul Usmania, Andrew Chenb, Faridi Qaiuma, David Mutcha, Aadel Chaudhuria, Katherine Fuha

aWashington University School of Medicine, St. Louis, MO, USA; bWashington University, St. Louis, MO, USA

High-grade serous carcinoma (HGSC) is the most lethal subtype of ovarian cancer, and a majority of patients are diagnosed at advanced stages. One standard-of-care is neoadjuvant chemotherapy followed by cytoreductive surgery, however up to 80% of HGSC patients develop recurrent disease. There exists a critical need to better understand the features of this tumor microenvironment that may highlight potential therapeutic targets, help stratify chemotherapy responders from non-responders, and uncover novel cell states that may serve as prognostic or predictive biomarkers.

  We obtained multiple biopsies from five patients with advanced-stage HGSC, both pre- and post-treatment, and analyzed these samples using single-cell RNAseq. We identified 20 distinct transcriptional clusters of cells, including a well-defined tumor subset enriched for EPCAM and KRT8. We annotated each cluster using known marker genes, and additionally validated these data through genome-wide copy number and developmental maturity analyses. We then performed transcriptome deconvolution to identify co-occurring transcriptional states using EcoTyper.

  We then performed bulk RNAseq on paired pre- and post-treatment tumor biopsies from 23 HGSC patients. Applying our fingerprints of ecotypes from the scRNA-seq data, we identified multiple distinct transcriptional states within the pre- and post-treatment HGSC tumor microenvironment, which were enriched for distinct co-occurring states comprised of multiple immunologic lineages, including CD4 T and NK cell states. We plan to validate these ecotypes using spatial transcriptomics and compare clinical outcomes across transcriptional states to determine the potential prognostic or predictive implications of our discoveries.