PhD Scientific Days 2025

Pathological and Oncological Sciences III.

Breast Cancer Characterization via LA-REIMS: Exploring a Novel Mass Spectrometry-Based Diagnostic Approach

Name of the presenter

Karancsi Zsófia

Institute/workplace of the presenter

Department of Pathology Forensic and Insurance Medicine

Authors

Zsófia Karancsi¹, András Dénes Márton², Gabriel Stefán Horkovics-Kovács³, Bálint András Deák¹, Anna Mária Tőkés¹, Tamás Karancsi⁴

1: Semmelweis University, Department of Pathology Forensic and Insurance Medicine
2: University of Technology and Economics, Department of Chemical and Environmental Process Engineering
3: University of Regensburg, Department for Multimodal Imaging of Intracellular Communication
4: Ambimass Kft.

Text of the abstract

Introduction
This study evaluates the feasibility of Laser-Assisted Rapid Evaporative Ionization Mass Spectrometry (LA-REIMS) for distinguishing breast cancer subtypes and normal tissue based on lipidomic signatures. While mass spectrometry (MS) with chromatography is the gold standard for lipid metabolism analysis, LA-REIMS enables real-time lipidomic profiling, possibly supporting intraoperative decision-making. By integrating spatial data, LA-REIMS findings can be directly compared with H&E-stained slides, linking molecular signatures to histopathology.

Patients and Methods
Fresh-frozen tissue from 41 breast cancer cases, 34 normal tissues, and 5 benign tumors was analyzed. All patients underwent primary surgery at Semmelweis University, with pregnancy-associated and recurrent cases excluded. Histological subtypes included 68.2% no special type (NST), 12.1% invasive lobular carcinoma (ILC), and 19.5% others. Molecular subtypes comprised 10 Luminal A, 15 Luminal B1, 2 Luminal B2, 4 HER2, and 9 TNBC. Tissue slices (10–20 µm) were analyzed with LA-REIMS at 75 µm resolution, and data were processed using primary and linear discriminant analysis and cross-validation.

Results
Preliminary analysis of 20 cancer cases (25,064 scans) and 16 normal samples (7,312 scans) demonstrated 87.88% accuracy excluding outliers and 87.57% including outliers in distinguishing tumors from normal tissue.
To differentiate ER-positive (12 cases) and ER-negative (8 cases) tumors, we analyzed 13,663 scans (ER-positive) and 11,401 scans (ER-negative), achieving 75.76% accuracy excluding outliers and 73.93% including outliers.
Finally, we differentiated grade 3 tumors (14 cases, 5,976 spectra) from grade 1 and 2 cases (6 cases, 23,563 spectra) with 76.60% accuracy excluding outliers and 68.82% including outliers.

Discussion
Our preliminary findings show promising results, suggesting that LA-REIMS could aid real-time tumor classification and molecular subtyping. We aim to expand our dataset, improving accuracy across molecular subtypes and identifying key differentiation markers. Final results may contribute to biomarker discovery, enhancing personalized treatment, risk stratification, and intraoperative decision-making in breast cancer management.