Identification of a novel prognostic signature based on vitamin metabolism clustering-related genes in lung adenocarcinoma.

Abstract

Background: Vitamins, and their metabolic processes play essential regulatory roles in controlling proliferation, differentiation, and growth in carcinogenesis. However, the role of vitamin metabolism in lung adenocarcinoma (LUAD) has rarely been reported. Here, we established a novel prognostic model based on vitamin metabolism-related genes in LUAD.

Methods: In this research, we aimed to identify vitamin metabolism associated with differentially expressed genes (DEGs) in LUAD utilizing The Cancer Genome Atlas (TCGA)-LUAD, GSE68465 and GSE72094 data. Unsupervised clustering classified patients into distinct subgroups. By utilizing least absolute shrinkage and selection operator (LASSO)-Cox regression analysis, vitamin metabolism-related genes could be used to construct prognostic model. Then the vitamin metabolism gene-related risk score (VRS) was calculated based on best cut-off splitting. Kaplan-Meier analysis, time-dependent receiver operating characteristic (ROC) analysis, univariate and multivariate Cox analyses, chemotherapeutic drugs sensitivity analysis, immune infiltration analysis and nomogram were conducted to verify our models' accuracy. Finally, CPS1 was identified as a relevant diagnostic marker using Random Forests algorithms, single-cell RNA sequencing data was used to confirm its expression.

Results: We investigated the relationship between vitamin metabolism patterns, overall survival (OS), and immune infiltration levels of patients with LUAD. A prognostic signature consisting of 11 genes was developed, which was able to classify patients into high and low VRS groups. Through gene enrichment analysis, cell cycle was mainly enriched. Compared to the low VRS group, the high VRS group exhibited poorer OS, as demonstrated by the Kaplan-Meier survival analysis. Furthermore, VRS was identified as an independent predictor of poor prognosis and poor OS, as indicated by both univariate and multivariate Cox regression analyses. Additionally, a nomogram was constructed to improve the accuracy of survival predictions in LUAD patients. We also found that the two groups of patients might respond differently to immune targets and anti-tumor drugs. CPS1 was identified as a relevant diagnostic marker and the expression was also as confirmed by single-cell RNA sequencing data.

Conclusions: Overall, our findings suggest that vitamin metabolism can influence the prognosis of LUAD patients, and our prognostic signature represents a potentially helpful resource for predicting patient outcomes and informing clinical decision-making.