Multi-omics analysis and experiments uncover the link between cancer intrinsic drivers, stemness, and immunotherapy in ovarian cancer with validation in a pan-cancer census

Yilin Li¹Cen chen¹Xiaoyu Ji²Ningxiao Jiang³Fei Wang³Xiangqian Gao³Weiwei Chen³Qiang Tang⁴Yan Li⁴Shinan Zhang³Gaofeng Qin³Yingjiang Xu³Yanlin Wang³Lei Han^3*Jie Mei^1*

• ¹Sichuan Academy of Medical Sciences and Sichuan Provincial People's Hospital, Chengdu, Sichuan Province, China
• ²Huashan Hospital North, Fudan University, Shanghai, Shanghai Municipality, China
• ³Binzhou Medical University Hospital, Binzhou, China
• ⁴Meishan People's Hospital, Meishan, China

Background: Although immune checkpoint inhibitors (ICIs) represent a substantial breakthrough in cancer treatment, it is crucial to acknowledge that their efficacy is limited to a subset of patients. The heterogeneity and stemness of cancer render its response to immunotherapy variable, warranting the identification of robust biomarkers for evaluation.Methods: Publicly available Ovarian Cancer (OV) single-cell RNA (scRNA) sequence dataset was collected and analyzed to elucidate the intrinsic driver gene of OV cancer cells. Through genome-scale CRISPR screening of RNA sequencing data from Project Achilles, essential genes specific to OV were identified. A novel cancer stem cell index (CSCI) was developed and validated using multiple advanced algorithms and large-scale datasets, as well as corresponding clinical features, including 14 OV transcriptomic datasets, 7 pan-cancer ICI transcriptomic cohorts and one melanoma scRNA dataset derived from PD-1 treated patients.Results: Chromosomal 20q gain, 8q gain, and 5q loss have been identified as ovarian cancer-specific driving variations. By analyzing large-scale datasets of ovarian cancer transcriptomics, including scRNA and CRISPR cell line datasets, we have identified a gene set that influences tumor intrinsic drivers and stemness properties. We then developed the CSCI to predict the prognosis and response to immunotherapy in ovarian cancer patients using advanced machine learning algorithms. When applied to PD1/PD-L1 ICI transcriptomic cohorts, CSCI consistently and accurately predicts tumor progression and immunotherapy benefits, with a mean AUC greater than 0.8. Notably, compared to previously established signatures, CSCI demonstrates better predictive performance across multiple ovarian cancer datasets. Intriguingly, we discovered that amplification of CSE1L enhances the stemness of tumor-initiating cells, facilitates angiogenesis, and the formation of ovarian cancer, which can serve as a potential therapeutic target. Finally, experiments validated that CSE1L promotes progression, migration, and proliferation of ovarian cancer.Our study has uncovered a robust correlation between variations in cancer intrinsic drivers and stemness, as well as resistance to immunotherapy. This finding provides valuable insights for potential strategies to overcome immune resistance by targeting genes associated with stemness.