Construction and Validation of a Novel Immunosignature for Overall Survival in Uveal Melanoma

Abstract

ObjectivesUveal melanoma (UM) is the most common primary intraocular malignancy in adults, and immune infiltration plays a crucial role in the prognosis of UM. This study aimed to generate an immunological marker-based predictive signature for the overall survival (OS) of UM patients.MethodsSingle-sample gene-set enrichment analysis (ssGSEA) was used to profile immune cell infiltration in 79 patients with UM from The Cancer Genome Atlas (TCGA) database. Univariate and multivariate least absolute shrinkage and selection operator (LASSO) Cox regressions were used to determine the prognostic factors for UM and construct the predictive immunosignature. Receiver operating characteristic (ROC) curves, decision curve analysis (DCA), and calibration curves were performed to evaluate the clinical ability and accuracy of the model. In addition, the predictive accuracy was compared between the immunosignature and the Tumor, Node, Metastasis (TNM) staging system of American Joint Committee on Cancer (AJCC). We further analyzed the differences in clinical characteristics, immune infiltrates, immune checkpoints, and therapy sensitivity between high- and low-risk groups characterized by the prognostic model.ResultsHigher levels of immune cell infiltration in UM were related to a lower survival rate. Matrix metallopeptidase 12 (MMP12), TCDD inducible poly (ADP-ribose) polymerase (TIPARP), and leucine rich repeat neuronal 3 (LRRN3) were identified as prognostic signatures, and an immunological marker-based prognostic signature was constructed with good clinical ability and accuracy. The immunosignature was developed with a concordance index (C-index) of 0.881, which is significantly better than that of the TNM staging system (p < 0.001). We further identified 1,762 genes with upregulated expression and 798 genes with downregulated expression in the high-risk group, and the differences between the high- and low-risk groups were mainly in immune-related processes. In addition, the expression of most of the immune checkpoint-relevant and immune activity-relevant genes was significantly higher in the high-risk group, which was more sensitive to therapy.ConclusionWe developed a novel immunosignature constructed by MMP12, TIPARP, and LRRN3 that could effectively predict the OS of UM.

Integrative analysis identifies key genes related to metastasis and a robust gene-based prognostic signature in uveal melanoma

“…In the past five years, several studies tried to identify essential genes involved in the development of UM [ 29 , 31 , 32 , 34 , 35 , 52 – 54 ]. These studies identified different aspects of UM such as clinical and pathological status along with the TME.…”

“…Other studies have reported variable gene signatures for predicting the OS of UM patients. The AUC of the ROC of the gene signature constructed by Xue et al was 0.8 [ 29 ], Zheng et al was 0.9 [ 30 ], Gu et al was 0.869 [ 31 ], Luo et al were 0.766 and 0.732 [ 32 ], and Li et al was 0.82 [ 33 ]. Although the AUC value of the ROC curves identified in our study was not the highest among them, was not even the lowest, which indicated that our gene signature was potentially significant.…”

Machine Learning-based Classifier to Decipher Immune Landscape of Uveal Melanoma and Predict Patient Outcomes

“…The current TNM staging system faces challenges in effectively assessing patient prognosis and performing risk strati cation. In recent years, the development of molecular subtype-based staging systems has improved the accuracy and speci city of prognostic predictions to some extent 3 . However, the practical application of these characteristics in clinical practice has been limited by inappropriate machine learning algorithms, underutilization of cohort information, lack of con dence in validation cohorts, and a lack of external datasets.…”