A Candidate Inflammation-Microenvironment Progression Signature for Colorectal Cancer Liver Metastasis: Bulk Transcriptomic Discovery, TCGA-COAD Prognosis, and Spatial Localization

Colorectal cancer liver metastasis (CRLM) is biologically heterogeneous, and transcriptomic signals from metastatic tissue can be confounded by host liver background. We integrated GSE50760 bulk transcriptomes, The Cancer Genome Atlas colon adenocarcinoma (TCGA-COAD) survival data, GSE17536 external survival data, and GSE225857 spatial transcriptomes to derive and assess a candidate inflammation-microenvironment progression signature. Differential expression between liver metastases and primary colorectal tumors was analyzed with limma using Benjamini–Hochberg false discovery rate correction. Genes with CRLM differential expression, concordant primary-tumor progression direction, and no typical hepatic-background annotation were retained. The resulting 39-gene score was calculated as the mean up-module expression minus the mean down-module expression. Liver metastases showed 418 upregulated and 88 downregulated genes. In TCGA-COAD, the score was associated with worse overall survival and remained associated after adjustment for age, sex, and American Joint Committee on Cancer stage (hazard ratio [HR] per 1 standard deviation [SD] 1.25; p = 0.038). In GSE17536, the score was associated with disease-free/recurrence-free survival (HR per 1 SD 1.72; p = 0.0021). Spatial analysis showed higher lesion-level scores in liver metastases than primary lesions, with L1/L2 heterogeneity. This candidate signature requires independent and prospective validation before clinical use.