Use of scRNA-seq to characterise the tumour microenvironment of high grade serous ovarian carincoma (HGSOC)

Abstract

High Grade Serous Ovarian Carcinoma (HGSOC) is the most common type of ovarian cancer. Patients with this disease typically experience relapse in their disease following surgical debulking and initially effective chemotherapy. HGSOC has been intensely studied at the genomic and transcriptomic levels in efforts to advance knowledge of the biological mechanisms that drive the behaviour of this malignancy, and so that new treatment strategies may curb the disease progression relapse. This body of work contributes an optimised protocol for generating robust 10X scRNA-seq libraries from fresh and preserved HGSOC tissue, aiming to dissect the cellular heterogeneity of HGSOC’s Tumour microenvironment (TME). Through unsupervised clustering analysis, it uncovers distinct cellular communities, elucidates transcriptomic signatures across HGSOC tumours, and augments bulk RNA-seq datasets via computational deconvolution, enhancing understanding of HGSOC's cellular complexity across an expanded clinical cohort.

The sequencing and analysis of these HGSOC patient tumours revealed 11 distinct cell types, including 2 that are novel in this tumour type; namely ciliated epithelial cells and metallothionein expressing T-cells. These 11 distinct cell types can be broadly categorised into 3 TME components (Tumour, Stroma and Immune) as in other previous tumour scRNA-seq studies. An additional analysis of these components examined the copy number variation (CNV) in the profiled cells and revealed HGSOC tumour cells to be mostly aneuploid while ciliated epithelial cells were diploid. A novel integrative subcluster analysis of HGSOC aneuploid tumour cells identified several apparently tumourigenic gene expression signatures. These include a KRT17+, protease inhibitory signature, an increased cellular metabolism signature, and an immune-reactive signature. Additionally, a ciliated cluster re-emerged within the HGSOC tumour cells, even though the diploid ciliated epithelial cells were not included in the integrative analysis.

Finally, the high granularity of HGSOC cellular composition revealed by scRNA-seq is utilised to perform deconvolution analyses to estimate cellular proportions and infer the TME of earlier bulk RNA-seq profiled HGSOC tumour samples. This investigation of earlier sequenced HGSOC samples revealed heterogeneity in the proportions of the TME compartments across the patient cohorts. Survival analysis using these inferred cellular proportions suggest that immune cell presence alone is not associated with survival, but metastatic fibroblast burden in tumour samples is significantly associated with worsen overall survival in HGSOC patients.

In conclusion, the laboratory protocol, the scRNA-seq datasets produced, and their analysis and application presented in this work expands the collective knowledge base of HGSOC. Specifically by characterising the cells of the HGSOC tumour microenvironment, and nuances of expression signatures of the malignant cells. The deconvolution approach showcases how scRNA-seq data can expand the clinical utility of earlier RNA-seq HGSOC datasets in a way that is scalable.