A new classification method for MALDI imaging mass spectrometry data acquired on formalin-fixed paraffin-embedded...

Area of Science:

• Biomedical Mass Spectrometry
• Computational Pathology
• Cancer Biomarker Discovery

Background:

• Matrix-assisted laser desorption/ionization imaging mass spectrometry (MALDI IMS) holds promise for histopathological diagnosis, particularly for tumor typing and subtyping.
• Developing MALDI IMS applications requires extracting relevant spectral fingerprints and identifying biomarkers associated with these patterns.

Purpose of the Study:

• To introduce a novel data analysis method for automated classification model generation using characteristic spectral patterns (CSPs).
• To apply this method for discriminating primary lung and pancreatic cancers, and lung adenocarcinoma from squamous cell carcinoma.

Main Methods:

• Analysis of formalin-fixed paraffin-embedded (FFPE) tissue samples from 445 patients using MALDI IMS.
• Development and application of a novel method for extracting characteristic spectral patterns (CSPs) for automated spectral data classification.
• Cross-validation assessment of classification accuracy and comparison with conventional m/z value-based methods.
• LC-MS/MS peptide identification to validate spectral features within CSPs.

Main Results:

• Achieved 100% classification accuracy for discriminating primary lung and pancreatic cancer.
• Attained 82.8% accuracy for classifying lung adenocarcinoma versus squamous cell carcinoma.
• The CSP-based method outperformed conventional m/z value extraction for initial classification tasks.

Conclusions:

• The proposed CSP method enables automated generation of classification models for MALDI IMS spectral data.
• This approach significantly supports cancer typing and subtyping in histopathology.
• Identified spectral features within CSPs correspond to peptides relevant for accurate cancer classification.