Abstract
Background: Clear cell renal cell carcinoma (ccRCC) is a major type of kidney cancer, making up about 80% of cases, with advanced stages showing low survival rates. Current treatments face challenges like toxicity and drug resistance. Studies indicate lactate, through the Warburg effect, promotes an immune-suppressive tumor microenvironment (TME), prompting the development of the LAC-TME classifier using machine learning to predict outcomes and personalize treatment. Methods: The study used data from TCGA-KIRC set and E-MTAB-1980 set, analyzing gene expression, mutations, and clinical data. It employed differential expression analysis, immune infiltration assessment, and 101 machine learning algorithms to build the classifier, integrating lactate-related genes and TME features, with predictive capability verified. Results: The LAC-TME classifier, constructed by integrating 9 lactate-related differentially expressed genes and TME cells, demonstrated high predictive accuracy (C-index of 0.92 in the training set and 0.73 in the validation set). Patients were categorized into three groups: Lactatelow + TMElow (best prognosis), Lactatehigh + TMEhigh (poorest prognosis), and a mixed group. This classifier can predict 1- to 5-year survival rates, with an AUC of 0.88-0.92. Notably, the Lactatehigh + TMEhigh subgroup was associated with immunosuppression and poor response to immunotherapy. As the core lactate-related gene of the LAC-TME classifier, the knockdown of LGALS1 significantly inhibits the proliferation and migration of ccRCC cells, verifying the biological rationality of the classifier. Conclusion: The LAC-TME classifier, integrating metabolic and immune data, offers a new tool for ccRCC prognosis and treatment guidance. Further validation is needed to confirm its clinical potential, reflecting the ongoing need for robust medical research.