Clinical data analysis reveals three subytpes of gastric cancer

Abstract

Gastric cancer is the fourth most common cancer and second leading cause of cancer-related death worldwide. Nowadays the accumulated large scale clinical data allows the clinicopathlogical review to identify the clinical factors, reveal their possible correlations, and mine the possible clinical patterns for gastric cancer. Here we analyze the clinical data of over 1500 gastric cancer patients histopathologically diagnosed and treated during 2006 to 2010. Specifically, we collect and preprocess the data by extracting 14 available clinical factors from three categories, i.e., the clinical background, immunohistochemistry data, and the caner's stage information. Then these factors are quantized and the significant factors and their correlations are calculated. Importantly, we define a distance between two patients by their clinical factors profile similarity and cluster all the patients into subgroups. We find that most of the patients fall into three major classes and we define them as three subtypes of gastric cancer. Each subtype is analyzed and characterized by its own significant factors and correlations. Our analysis may provide important insights for gastric cancer classification and diagnose.