Integrating bioinformatics and clinicopathological research of gastrointestinal stromal tumors: identification of aurora kinase A as a poor risk marker.

Abstract

For completely resected primary gastrointestinal stromal tumors (GISTs), mitotic rate, tumor size, and tumor location are important risk factors for recurrence. However, molecular markers for recurrence are still lacking. We reanalyzed GIST gene expression profile GSE8167 available from the Gene Expression Omnibus (GEO) and confirmed the prognostic influence of one selected gene, aurora kinase A (AURKA), in another cohort of 142 patients using immunohistochemistry (IHC). Thirty-two cases in GSE8167 were classified into two risk groups with distinct recurrence-free survival (RFS) and expression profiles using modified criteria of Miettinen et al. from the Armed Forces Institute of Pathology (AFIP-Miettinen). AURKA was among the 19 genes common to the top 50 features of the high-risk phenotype and a 67-gene signature called the complexity index in sarcomas. AURKA was significantly overexpressed in the high-risk group, and patients with high AURKA expression had significantly worse RFS than those with low expression. In the IHC-validated cohort, AURKA expression was significantly higher in nongastric tumors than in gastric tumors and was significantly correlated with AFIP-Miettinen risk group. Univariate analysis showed that RFS was significantly influenced by tumor size, mitotic count, AFIP-Miettinen risk group classification, and AURKA expression. However, only tumor size (P = 0.017), mitotic count (P = 0.007), and AURKA expression (P = 0.039) were identified as independent unfavorable prognostic factors for RFS in multivariate analysis. By integrating bioinformatics and clinicopathological studies, AURKA was identified as a marker for high-risk GIST.