LAUSR

Although the postoperative recurrence rate for pancreatic neuroendocrine tumors (PNETs) is reported to be 13.5%‐30%, the paucity of valuable biomarkers to predict recurrence poses a problem for the early detection of relapse. Hence, this study aimed to identify new biomarkers to predict the recurrence of PNETs. We performed RNA sequencing (RNA‐Seq) on RNA isolated from frozen primary tumors sampled from all localized G1/G2 PNETs resected curatively from 1998 to 2015 in our institution. We calculated differentially expressed genes (DEGs) in tumor with and without recurrence (≥3 years) for the propensity‐matched cohort. Gene ontology analysis for the identified DEGs was also performed. Furthermore, we evaluated the expression levels of candidate genes as recurrence predictors via immunostaining. Comparison of transcriptional levels in tumors with and without recurrence identified 166 DEGs. Up‐ and downregulated genes with high significance in these tumors were mainly related to extracellular organization and cell adhesion, respectively. We observed the top three upregulated genes, C‐type lectin domain family 3 member A (CLEC3A), matrix metalloproteinase‐7 (MMP7), and lipocalin2 (LCN2) immunohistochemically and compared their levels in recurrent and nonrecurrent tumors. Significantly higher recurrence rate was shown in patients with positive expression of CLEC3A (P = 0.028), MMP7 (P = 0.003), and LCN2 (P = 0.040) than that with negative expression. We identified CLEC3A, MMP7, and LCN2 known to be associated with the phosphatidylinositol‐3‐kinase/Akt pathway, as potential novel markers to predict the postoperative recurrence of PNETs.