Figure 1. Schematic of the search for a metastasis-associated signature in the hepatic microenvironment. Twenty noncancerous hepatic tissue samples from areas surrounding tumors, characterized as MIM (metastasis-inclined microenvironment) or MAM (metastasis-averse microenvironment) samples, were analyzed by cDNA microarray (Step 1). A metastasis signature composed of 17 immune-related genes, associated with T helper cell type 1 (Th1)– and Th2-like cytokines, was significantly and differentially expressed in samples with metastasis (Step 2). Following validation by quantitative real-time polymerase chain reaction (qRT-PCR) (Step 3), prediction analysis of microarrays (PAM), multi-dimensional scaling (MDS), Kaplan-Meier survival analysis, and Cox proportional hazards modeling demonstrated accurate classification of patients with metastasis and prediction of outcome based on the 17-gene signature (Step 4). The pro-inflammatory status of metastasis samples was confirmed by immunohistochemistry (IHC), and a significant increase in the abundance of macrophage colony stimulating factor type 1 (CSF1) in metastasis samples was shown by qRT-PCR and ELISA (Step 5). Peripheral blood mononuclear cells (PBMC), incubated with recombinant CSF1, recapitulated the significant Th1-Th2 cytokine shift observed in metastasis samples, indicating that CSF1 may play a role in promoting the metastatic phenotype (Step 5). In addition, fluorescence-activated cell sorting (FACS) suggested that CSF1 induced these cytokine shifts in T-cell populations (Step 5). Ab, antibody; T-Test, student’s t test; FDR, false discovery rate.