Prognostic Potential of a PSAT1‐Associated Gene Signature in Identifying High‐Risk Patients in Early‐Stage EGFR‐Mutant Lung Cancer

Abstract

De novocellular production of serine may contribute to tumor growth by providing precursors for macromolecular production and one‐carbon metabolism. Accordingly, multiple cancers exhibit increased expression of serine biosynthetic enzymes. For example, elevated PSAT1 (phosphoserine aminotransferase 1), which catalyzes the second step in converting 3‐phosphohydroxypyruvate to phosphoserine, is associated with poorer clinical outcomes. Several reports have implicated PSAT1 in the proliferation, migration, invasion, and chemo‐resistance of malignant cells. Our previous study found that PSAT1 translocated into the nucleus within EGFR‐activated NSCLC cells and was associated with poorer clinical outcomes of patients with EGFR‐mutant lung cancer, suggesting a relevant function for PSAT1 in EGFR‐driven tumorigenesis. In this study, we designed a bioinformatic approach to identify a PSAT1‐associated gene signature for EGFR‐mutant lung cancer (Fig. 1). Since PC9 cells are frequently utilized as an EGFR‐mutant lung cancer model, we speculated that a subset of PSAT1‐regulated genes independently contributes to EGFR‐driven lung tumorigenesis. For this, we transcriptionally profiled PSAT1 silenced PC9 cells. Analysis of the RNASeq identified 490 genes (279‐down, 211‐up) that were differentially expressed upon PSAT1 silencing (PSAT1‐mediated DEG). We next performed a comparative analysis between PSAT1‐mediated DEG and publicly available microarray datasets obtained from EGFR‐mutant patient tumors. Twenty‐five genes (13‐down, 12‐up) were differentially expressed in all EGFR‐mutant lung cancer datasets. These were classified as a PSAT1‐associated gene signature for EGFR‐mutant lung tumors and used to assess the relationship between this gene signature and both relapse‐free (RFS) and overall survival (OS) by the survival prediction tool from BRB‐ArrayTools. Thirteen and seventeen genes from the 25‐gene signature were found to predict the OS and RFS rates of EGFR‐mutant lung cancer, respectively. However, this signature failed to predict patient outcomes for either wild‐type or KRAS‐mutant lung tumors. Furthermore, the expression profile of RFS high‐risk patients with stage I EGFR‐mutant tumor phenocopies the high‐risk stage II tumor, supporting the prognostic power of PSAT1‐mediated gene signature for these EGFR‐mutant tumors. As early‐stage cancers dominate the gene signature obtained from human EGFR‐mutant lung tumors, analysis of genes involved in late‐stage tumor progression and metastasis is limited. Therefore, differentially expressed genes between parental PC9‐P and brain‐metastatic PC9‐BrM3 were extracted from the Nguyen et al.study (2009). Comparative analysis with RNA‐seq data found a potential PSAT1‐associated metastatic gene signature. In summary, our genomic approach yielded a PSAT1‐associated gene signature with prognostic value in EGFR‐mutant NSCLC patient outcomes.