Elastic net-based identification of GAMT as potential diagnostic marker for early-stage gastric cancer

Abstract

Early-stage gastric cancer (GC) is asymptomatic. How to diagnose the early-stage GC is challenging. The sensitivity and specificity of diagnosing signatures for early-stage patients are still poor. Elastic-net-based analysis was used to identify potential diagnostic signatures of early-stage GC. The expression level of candidate gene was evaluated by immunohistochemistry staining. The potential function of candidate gene was verified by overexpressing in vitro. Consensus genes (including GAMT) were identified using the different strengths of the penalty. Surprisingly, GAMT was still identified even if some multicollinear variables were deleted directly. IHC staining showed that there are no GAMT-positive signals in the cell nuclei of all tumor tissues, while GAMT does express in nuclei of adjacent normal tissue. There are 16.33% positive cell nuclei in paracancerous tissues. In addition, the number of larger-area colonies of overexpression-GAMT group, empty-vector group, and AGS group is 70±29.21, 151.33±15.95, and 111.67±22.03, respectively. Number of larger colonies in group with overexpression of GAMT is significantly less than control groups. Elastic-net-penalty-based workflow is a effective tool to identify diagnostic biomarker for early-stage solid tumor. GAMT has strong potential to be the diagnostic biomarker for the early-stage GC.