Background: Mucoepidermoid carcinoma (MEC) is the most common malignant salivary gland tumor; however, comprehensive molecular models that distinguish MEC from normal parotid tissue remain poorly defined.
Purpose: This study aimed to identify and internally evaluate a compact candidate expression signature distinguishing MEC from normal parotid tissue using a single public RNA-sequencing cohort.
Materials and Method: In this retrospective bioinformatics/in silico observational study, RNA-sequencing data from GSE282430, including 21 MEC and 6 normal parotid samples, were analyzed using differential expression analysis and Hallmark pathway enrichment. Recently published MEC-related gene signatures were reconstructed, harmonized, and intersected with the cohort. Logistic regression and least absolute shrinkage and selection operator (LASSO) regression were then used to evaluate tumor-versus-normal discrimination and derive a compact panel.
Results: A total of 385 genes were differentially expressed between MEC and normal tissue. A literature-derived union panel of 27 genes achieved an area under the receiver operating characteristic curve (AUC) of 1.00 for distinguishing MEC from normal parotid tissue. LASSO-based reduction yielded an exploratory three-gene signature comprising MLXIPL, POSTN, and SPP1, which showed an exploratory cross-validated AUC of 0.976. The selected genes represent complementary metabolic, stromal/extracellular matrix, and immune matrix processes, although functional interactions among them were not experimentally assessed.
Conclusion: Integration of four literature-derived gene sets identified an exploratory MLXIPL-POSTN-SPP1 candidate expression signature that distinguished MEC from normal parotid tissue within a small public RNA-sequencing cohort. These findings support further investigation of this candidate signature, but external cross-platform validation and experimental confirmation are required before diagnostic or clinical interpretation.