Machine learning-enabled spatial multi-omics uncovers lactate-driven targets and tumor microenvironmental...

Area of Science:

• Oncology
• Cancer Metabolism
• Molecular Biology

Background:

• Lactate accumulation is a hallmark of cancer metabolism.
• Its precise spatial and cell-type-specific roles in lung adenocarcinoma (LUAD) are not well understood.

Purpose of the Study:

• To investigate the spatial and cell-type-specific effects of lactate accumulation in LUAD.
• To identify lactate-associated molecular programs and their prognostic implications.

Main Methods:

• Integration of single-cell transcriptomics, spatial transcriptomics, spatial metabolomics, and immunofluorescence.
• Analysis of TCGA survival data and application of multiple machine-learning models (Random Forest, Elastic-Net, SVM, ANN, Decision Trees).

Main Results:

• High-lactate LUAD tumors showed increased epithelial and fibroblast cells, while low-lactate tumors enriched T/NK cells and macrophages.
• Spatial metabolomics revealed distinct lactate and pyruvate distributions across cell types; endothelial cells showed minimal lactate accumulation.
• Endothelial subclusters in high-lactate regions exhibited angiogenic and stress signatures, correlating with poor prognosis.
• Machine learning consistently identified endothelial and fibroblast programs as key drivers of high-lactate states and adverse outcomes.

Conclusions:

• Lactate accumulation significantly alters the LUAD microenvironment, promoting angiogenesis and immune suppression.
• Spatial profiling highlights lactate's role in prognostic stratification of LUAD patients.
• Lactate-centered pathways represent potential therapeutic targets for LUAD treatment.