A decision tree-based combination of ezrin-interacting proteins to estimate the prognostic risk of patients with esophageal squamous cell carcinoma

Our previous studies have highlighted the importance of ezrin in esophageal squamous cell carcinoma (ESCC). Here our objective was to explore the clinical significance of ezrin-interacting proteins, which would provide a theoretical basis for understanding the function of ezrin and potential therapeutic targets for ESCC. We used affinity purification and mass spectrometry to identify PDIA3, CNPY2, and STMN1 as potential ezrin-interacting proteins. Confocal microscopy and coimmunoprecipitation analysis further confirmed the colocalization and interaction of ezrin with PDIA3, CNPY2, and STMN1. Tissue microarray data of ESCC samples (n = 263) showed that the 5-year overall survival (OS) and disease-free survival (DFS) were significantly lower for the CNPY2 (OS, P = .003; DFS, P = .011) and STMN1 (OS, P = .010; DFS, P = .002) high-expression groups compared with the low-expression groups. By contrast, overexpression of PDIA3 was significantly correlated with favorable survival (OS, P < .001; DFS, P = .001). Cox regression demonstrated the prognostic value of PDIA3, CNPY2, and STMN1 in ESCC. Furthermore, decision tree analysis revealed that the resulting classifier of both ezrin and its interacting proteins could be used to better predict OS and DFS of patients with ESCC. In conclusion, a signature of ezrin-interacting proteins accurately predicts ESCC patient survival or tumor recurrence.