Construction of a mitochondria genes-based model for prognosis prediction, drug guidance and immune feature analysis in ovarian serous cystadenocarcinoma

Abstract Background Ovarian serous cystadenocarcinoma (OSC) is the most common pathological subtype of ovarian cancer (OC) associated with high mortality. Albeit dysregulated mitochondrial metabolism has been implicated with OC, the role of mitochondrial genes in OSC remains unclear. We sought to construct a model based on mitochondrial genes for prognosis prediction, drug guidance and immune feature analysis of OSC. Methods Differentially expressed genes (DEGs) and mitochondrial-related DEGs (MRGs) were identified through the Cancer Genome Atlas (TCGA)-OV dataset. Consensus clustering algorithm was applied to classify OSC patients into distinct MRGs subtypes. Prognosis-related MRGs were screened to construct the prognosis-related Risk score model, which was verified using GSE26193 dataset and immunohistochemistry (IHC) score model based on staining intensity and extent scores of MRGs. A visualized nomogram was developed to predict 1-, 3- and 5-year overall survival (OS) and drug response. The correlation of MRGs subtypes with risk subgroups and the association of Risk score model with immune response and infiltration were also investigated. Results 341 MRGs were identified from TCGA-OV, and OSC patients could be mainly divided into two MRGs subtypes. A novel prognostic Risk score model based on 7-MRGs, including ACOT13 , ACSS3 , COA6 , HINT2 , MRPL14 , NDUFC2 and NDUFV2 , was developed and validated via GSE26193 dataset and IHC score model. Patients in the low-risk group had a significantly longer OS. The nomogram exhibited good prognostic assessment accuracy in both training and validation datasets. Drug sensitivity analysis indicated that cisplatin, paclitaxel and docetaxel were more sensitive in the low-risk group; VEGFR inhibitor Axitinib, and BRAF inhibitors Vemurafenib and SB590885 showed better sensitivity in the high-risk group; moreover, patients in the low-risk group could have better anti-PD-1 immunotherapy response. Patients in “cluster1” MRGs subtype had lower risk scores and better immunotherapy response scores than the “cluster2” subgroup. More significant infiltrated tumor killing cells (CD8 + T cells) and higher M1 / M2 macrophage ratio were in “cluster1” patients. Conclusions Our novel 7 MRGs-based Risk score model has huge prospects to evaluate the prognosis and guide drug treatment. The favorable prognosis associated with the low-risk group is closely related to better immune response and more significant anti-tumor cellular infiltration.