A Tissue Section-Based Mid-Infrared Spectroscopical Analysis of Salivary Gland Tumors Based on Enzymatic...

Salivary gland tumors (SGTs) are a rare and histologically heterogeneous group of neoplasms that are challenging to diagnose due to phenotypic heterogeneity and overlapping histomorphological markers. Accurate diagnosis is required for clinical management, particularly in unusual subtypes. The objective of this study was to ascertain whether attenuated total reflectance-Fourier transform infrared (ATR-FTIR) spectroscopy, in combination with enzymatic deglycosylation, would be useful in SGT classification by detecting glycosylation-related metabolic variations.

155 tissue sections, consisting of 80 SGTs and 75 controls, were analyzed. ATR-FTIR spectroscopy was used to record the mid-infrared (MIR) spectra (4000-400 cm^-1) of enzymatically untreated and deglycosylated samples. Spectral data were preprocessed and analyzed by principal component analysis (PCA) and orthogonal partial least squares-discriminant analysis (OPLS-DA). Enzymatic deglycosylation focused on sialic acid and fucose residues with α2-3,6,8 neuraminidase, α1-2,4,6 fucosidase O, and α1-3,4 fucosidase.

Tumor and control samples were discriminated with an OPLS-DA model, achieving an accuracy of 81.9% (78.7% for controls and 85.0% for tumors), especially in the glycosylation-relevant spectral range (850-1250 cm^-1). Classification between benign and malignant tumors was more challenging, with an accuracy of 70.0% (72.5% for benign and 67.5% for malignant cases). Enzymatic deglycosylation resulted in detectable changes in the MIR spectra, confirming the contribution of glycosylation to tumor-specific signatures. Benign vs. malignant tumor discrimination was still poor and was not much enhanced in the sense of incorporating glycosylation-specific regions.

ATR-FTIR spectroscopy coupled with enzymatic deglycosylation can distinguish tumor and control tissues based on glycan-associated spectral differences. Application of the technique to benign/malignant SGT discrimination is hampered by spectral overlap and tumor heterogeneity. Further research will be necessary to explore other clustering algorithms and larger and more homogeneous datasets for improved diagnostic accuracy.