Pancreatic cancer is one of the most malignant diseases and has a poor prognosis. The screening and validation of biomarkers with predictive value for prognosis and treatment efficacy are important. To identify potential prognostic markers of pancreatic cancer patients, we conducted a study that included 99 pancreatic cancer patients. Three patients with PFS>18 months were enrolled in the treat group, and three patients with PFS<12 months were enrolled in the control group. Differentially expressed genes (DEGs) between these two groups were analyzed by whole-genome expression microarray. A total of 178 DEGs were identified, including 110 up-regulated and 68 down-regulated genes. Next, 24 candidate genes were selected for validation by qPCR based on fold change and previous studies. The results showed that the mRNA levels of four candidate genes, including ACSL5, SLC44A4, LOX, and TOX3, were correlated with PFS. Immunohistochemical staining was performed to validate the protein expression levels of these four markers. The results showed that patients with LOX high, ACSL5 low and TOX3 low expression had a significantly shorter PFS than those with LOX low, ACSL5 high and TOX3 high expression. Multivariable analysis revealed differentiation, tumor stage, LOX expression, and ACSL5 expression were independent prognostic factors for PFS. Then, we use the TCGA database to explore the underlying mechanism of LOX influence pancreatic cancer progression. Protein-protein interaction network of ACSL5 was established by STRING to uncover the potential regulation mechanism. Our findings reveal that LOX and ACSL5 are potential prognostic markers for the prognosis of pancreatic cancer patients.
Keywords: ACSL5; EMT; LOX; Pancreatic cancer; bioinformatic analysis; microarray; prognostic markers.