Identification and validation of core genes in tumor-educated platelets for human gastrointestinal tumor diagnosis using network-based transcriptomic analysis.

Gastrointestinal (GI) tumors have increasing incidence worldwide with their underlying mechanisms still not being fully understood. The use of tumor-educated platelets (TEPs) in liquid biopsy is a newly-emerged blood-based cancer diagnostic method. Herein, we aimed to investigate the genomic changes of TEPs in GI tumor development and their potential functions using network-based meta-analysis combined with bioinformatic methods. We used a total of three eligible RNA-seq datasets, which were integrated using multiple meta-analysis methods on the NetworkAnalyst website, and identified 775 DEGs (differentially expressed genes; 51 up-regulated and 724 down-regulated genes) in GI tumor relative to healthy control (HC) samples. These TEP DEGs were mostly enriched in bone marrow-derived cell types and associated with gene ontology (GO) of "carcinoma" and could affect pathways of "Integrated Cancer Pathway" and "Generic transcription pathway" respectively for highly and lowly expressed DEGs. Combined network-based meta-analysis and protein-protein interaction (PPI) analysis identified cyclin dependent kinase 1 (CDK1) and heat shock protein family A (Hsp70) member 5 (HSPA5) to be the hub genes with the highest degree centrality (DC), being up-regulated and down-regulated in TEPs, respectively. GO and Kyoto Encyclopedia of Genes and Genomes (KEGG) results showed that the hub genes were primarily related to cell cycle and division, nucleobase-containing compound and carbohydrate transport, and endoplasmic reticulum unfolded protein response. Additionally, the nomogram model suggested that the two-gene signature owns extraordinary predictive power for GI tumor diagnosis. Further, the two-gene signature was demonstrated to have potential value for metastatic GI tumor diagnosis. The expression levels of CDK1 and HSPA5 in clinical platelet samples were verified to be consistent with the results from bioinformatic analysis. This study identified a two-gene signature encompassing CDK1 and HSPA5 that can be used as a biomarker for GI tumor diagnosis and maybe even cancer-associated thrombosis (CAT)-related prognosis.