Abstract
ESCC (Esophageal squamous cell carcinoma) is a heterogeneous cancer with diverse prognosis. Here, to explore the biological diversity of ESCC, we employed gene expression profiles from 360 ESCC tumors from East Asians to establish a comprehensive molecular classification and characterization of ESCC. Using the specific 185-gene signature generated by unsupervised consensus clustering of gene expression data, we defined four subtypes associated with distinct clinical metrics: tumors with high metastasis associated with EMT (epithelial to mesenchymal transition) and active MAP4K4/JNK signaling pathway; tumors with high chromosomal instability with up regulated MYC targes; well differentiated tumors with less aggressive and moderated tumors. The clinical relevance of these subtypes was stated by significant differences in prognosis. Importantly, 24% of all ESCCs (n = 360) were classified into the high metastasis subtype associated with poorly differentiation and unfavorable prognosis. We provided evidence that this subtype relates to tumor microenvironment. Collectively, these results might contribute to more precise personalized therapeutic strategies for each subtype of ESCC patients in the near future.