Single-cell transcriptomics reveals cellular evolution underlying pleomorphic adenoma recurrence and malignant transformation

Pleomorphic adenoma (PA), the most common benign salivary gland tumor, presents unpredictable risks of recurrence and malignant transformation into carcinoma ex pleomorphic adenoma (CXPA), posing significant clinical challenges. To better delineate the tumor progression trajectory, we performed single-cell RNA sequencing on normal salivary glands, primary PA, recurrent PA (rPA), and CXPA, integrating cell trajectory reconstruction, differential gene expression analysis, and key gene network exploration to uncover molecular transitions and intercellular crosstalk underlying PA recurrence and malignant transformation. Our findings revealed GALNT13? myoepithelial cells as CXPA-specific malignant progenitors, marking early malignant conversion, while MIF? myoepithelial cells exhibited enhanced tissue-destructive capabilities. Fibroblasts played dual roles, enforcing fibrotic restraint in primary PA and promoting extracellular matrix degradation in CXPA. The tumor microenvironment demonstrated stage-specific adaptations, with CXPA favoring pro-inflammatory MIF–CD74/CD44 signaling, whereas rPA adopted immunosuppressive characteristics. Together, stromal reprogramming and immune-editing dynamics orchestrated microenvironmental adaptation, linking cellular heterogeneity to clinical aggressiveness. This study presents the first comprehensive molecular atlas of PA progression, highlighting the malignant specialization of myoepithelial subpopulations, fibroblast-mediated stromal remodeling, and immune-driven microenvironmental evolution, offering a valuable framework for precision stratification of malignant potential and future microenvironment-targeted therapeutic strategies. Overall design: Fresh tumor tissues from human normal salivary gland, primary pleomorphic adenoma (PA), recurrent PA (rPA), and carcinoma ex pleomorphic adenoma (CXPA) were collected during surgery. Single-cell suspensions were generated by enzymatic digestion and processed using the 10x Genomics Chromium Single Cell 3' platform. scRNA-seq libraries were constructed and sequenced for all samples. The data were analyzed to investigate cellular heterogeneity, infer tumor progression trajectories, and characterize changes in the tumor microenvironment during recurrence and malignant transformation.