Lactate-related genes signature as a novel prognostic landscape in laryngeal squamous cell carcinoma: insights from 156 machine learning algorithms and in vitro validation

Lactate has been involved in the progress of the tumor as a signal molecule and modified protein. This study investigates lactate-related genes (LRGs) as prognostic indicators for laryngeal squamous cell carcinoma (LSCC). Differentially expressed genes (DEGs) were identified from the LSCC TCGA dataset, and key LSCC-related genes were further characterized using Weighted Gene Co-expression Network Analysis (WGCNA). Subsequently, lactate-related genes obtained from public databases were integrated with the DEGs and the identified LSCC-related genes, yielding 69 candidate genes. A comprehensive risk prediction model was then developed, incorporating 12 hub LRGs, through 10 machine learning algorithms. The robustness and predictive accuracy of the model were rigorously validated via survival analysis, ROC curve assessment. Single-cell analysis showed high lactate scores in fibroblasts and epithelial cells, and hub genes were mainly expressed in epithelial cells. Mendelian randomization and in vitro experiments explored the genetic and mechanistic links between lactate, confirming lactate's role in promoting LSCC progression, which was reversed by LDHB or amyloid precursor protein (APP) knockdown. The risk score developed in this study demonstrates excellent robustness and is capable of effectively predicting the prognosis of LSCC patients. Our study holds promise for advancing both prognostic prediction and therapeutic intervention in LSCC management.

Lactate; Laryngeal squamous cell carcinoma; Machine learning; Tumor microenvironment.