A metabolic map and artificial intelligence-aided identification of nasopharyngeal carcinoma via a single-cell Raman platform

Nasopharyngeal carcinoma (NPC) is a complex cancer influenced by various factors. This study explores the use of single-cell Raman spectroscopy as a potential diagnostic tool for investigating biomolecular changes associated with NPC carcinogenesis.

Seven NPC cell lines, one immortalised nasopharyngeal epithelial cell line, six nasopharyngeal mucosa tissues and seven NPC tissue samples were analysed by performing confocal Raman spectroscopic measurements and imaging. The single-cell Raman spectral dataset was used to quantify relevant biomolecules and build machine learning classification models. Metabolomic profiles were investigated using ultra-performance liquid chromatography-tandem mass spectrometer (UPLC-MS/MS).

By generating a metabolic map of seven NPC cell lines, we identified an interplay of altered metabolic processes involving nucleic acids, amino acids, lipids and sugars. The results from spatially resolved Raman maps and UPLC-MS/MS metabolomics were consistent, revealing an increase of unsaturated fatty acids in cancer cells, particularly in highly metastatic 5–8F and poorly differentiated CNE2 cells. The classification model achieved a nearly perfect classification when identifying NPC and non-NPC cells with an ROC-AUC of 0.99 and a value of 0.97 when identifying 13 tissue samples.

This study unveils a complex interplay of metabolic network and highlights the potential roles of unsaturated fatty acids in NPC progression and metastasis. This renders further research to provide deeper insights into NPC pathogenesis, identify new metabolic targets and improve the efficacy of targeted therapies in NPC. Artificial intelligence-aided analysis of single-cell Raman spectra has achieved high accuracies in the classification of both cancer cells and patient tissues, paving the way for a simple, less invasive and accurate diagnostic test.