Integrated Proteomics and Machine Learning Approach Reveals PYCR1 as a Novel Biomarker to Predict Prognosis of Sinonasal Squamous Cell Carcinoma

Sinonasal squamous cell carcinoma (SNSCC) is a rare tumor with a high 5-year mortality rate. However, proteomic technologies have not yet been utilized to identify SNSCC-associated proteins, which could be used as biomarkers. In this study, we aimed to discover a biomarker to predict SNSCC patients using proteomic analysis integrated with machine learning models. Support vector machine (SVM), logistic regression (LR), random forest (RF), and gradient boost (GB) classifiers were developed to predict SNSCC based on proteomic profiles of SNSCC compared with nasal polyps (NP) as control. Seventeen feature proteins were found in all models, indicating possible biomarkers for SNSCC. Analysis of gene expression across multiple cancer types and their associations with cancer stage and patient survival in the TCGA-HNSC dataset identified a PYCR1 and MYO1B gene that could be a potential tumor-associated marker. The expression of PYCR1 was confirmed by RT-qPCR in SNSCC tissues, and its high expression was associated with poor overall survival, indicating PYCR1 as a potential tumor-associated biomarker to predict the prognosis of SNSCC.