Abstract
We aimed to create a mitophagy-related risk model via data mining of gene expression profiles to predict prognosis in uveal melanoma (UM) and develop a novel method for improving the prediction of clinical outcomes. Together with clinical information, RNA-seq and microarray data were gathered from the Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) databases. ConsensusClusterPlus was used to detect mitophagy-related subgroups. The genes involved with mitophagy, and the UM prognosis were discovered using univariate Cox regression analysis. In an outside population, a mitophagy risk sign was constructed and verified using least absolute shrinkage and selection operator (LASSO) regression. Data from both survival studies and receiver operating characteristic (ROC) curve analyses were used to evaluate model performance, a bootstrap method was used test the model. Functional enrichment and immune infiltration were examined. A risk model was developed using six mitophagy-related genes (ATG12, CSNK2B, MTERF3, TOMM5, TOMM40, and TOMM70), and patients with UM were divided into low- and high-risk subgroups. Patients in the high-risk group had a lower chance of living longer than those in the low-risk group (p < 0.001). The ROC test indicated the accuracy of the signature. Moreover, prognostic nomograms and calibration plots, which included mitophagy signals, were produced with high predictive performance, and the risk model was strongly associated with the control of immune infiltration. Furthermore, functional enrichment analysis demonstrated that several mitophagy subtypes may be implicated in cancer, mitochondrial metabolism, and immunological control signaling pathways. The mitophagy-related risk model we developed may be used to anticipate the clinical outcomes of UM and highlight the involvement of mitophagy-related genes as prospective therapeutic options in UM. Furthermore, our study emphasizes the essential role of mitophagy in UM.