脂肪酸相关基因特征可对皮肤黑色素瘤预后不良患者进行分层并表征其 TIME

The aim of this study is to build a prognostic model for cutaneous melanoma (CM) using fatty acid-related genes and evaluate its capacity for predicting prognosis, identifying the tumor immune microenvironment (TIME) composition, and assessing drug sensitivity.

Through the analysis of transcriptional data from TCGA-SKCM and GTEx datasets, we screened for differentially expressed fatty acids-related genes (DEFAGs). Additionally, we employed clinical data from TCGA-SKCM and GSE65904 to identify genes associated with prognosis. Subsequently, utilizing all the identified prognosis-related fatty acid genes, we performed unsupervised clustering analysis using the ConsensusClusterPlus R package. We further validated the significant differences between subtypes through survival analysis and pathway analysis. To predict prognosis, we developed a LASSO-Cox prognostic signature. This signature's predictive ability was rigorously examined through multivariant Cox regression, survival analysis, and ROC curve analysis. Following this, we constructed a nomogram based on the aforementioned signature and evaluated its accuracy and clinical utility using calibration curves, cumulative hazard rates, and decision curve analysis. Using this signature, we stratified all cases into high- and low-risk groups and compared the differences in immune characteristics and drug treatment responsiveness between these two subgroups. Additionally, in this study, we provided preliminary confirmation of the pivotal role of CD1D in the TIME of CM. We analyzed its expression across various immune cell types and its correlation with intercellular communication using single-cell data from the GSE139249 dataset.

In this study, a total of 84 DEFAGs were identified, among which 18 were associated with prognosis. Utilizing these 18 prognosis-related genes, all cases were categorized into three subtypes. Significant differences were observed between subtypes in terms of survival outcomes, the expression of the 18 DEFAGs, immune cell proportions, and enriched pathways. A LASSO-Cox regression analysis was performed on these 18 genes, leading to the development of a signature comprising 6 DEFAGs. Risk scores were calculated for all cases, dividing them into high-risk and low-risk groups. High-risk patients exhibited significantly poorer prognosis than low-risk patients, both in the training group (p < 0.001) and the test group (p = 0.002). Multivariate Cox regression analysis indicated that this signature could independently predict outcomes [HR = 2.03 (1.69-2.45), p < 0.001]. The area under the ROC curve for the training and test groups was 0.715 and 0.661, respectively. Combining risk scores with clinical factors including metastatic status and patient age, a nomogram was constructed, which demonstrated significant predictive power for 3 and 5 years patient outcomes. Furthermore, the high and low-risk subgroups displayed differences in the composition of various immune cells, including M1 macrophages, M0 macrophages, and CD8 + T cells. The low-risk subgroup exhibited higher StromalScore, ImmuneScore, and ESTIMATEScore (p < 0.001) and demonstrated better responsiveness to immune therapy for patients with PD1-positive and CTLA4-negative or positive expressions (p < 0.001). The signature gene CD1D was found to be mainly expressed in monocytes/macrophages and dendritic cells within the TIME. Through intercellular communication analysis, it was observed that cases with high CD1D expression exhibited significantly enhanced signal transductions from other immune cells to monocytes/macrophages, particularly the (HLA-A/B/C/E/F)-CD8A signaling from natural killer (NK) cells to monocytes/macrophages (p < 0.01).

The prognostic signature constructed in this study, based on six fatty acid-related genes, exhibits strong capabilities in predicting patient outcomes, identifying the TIME, and assessing drug sensitivity. This signature can aid in p