Background: Identifying patterns of pathological tumor resistant to treatment regimens and improving tumor’s chemotherapy sensitivity from longitudinal and multi-omics data integration are of paramount importance in order to provide the best chances of cure, especially, for colorectal cancer liver metastases. Methods: We elucidated genomic (ctDNA and whole exome sequencing), immunomic (DNA-based T cell receptor sequencing) and transcriptomic changes from liver metastases, peripheral blood and match primary tumors that accompany the evolution of colorectal cancer liver metastases. In total, 197 histopathologically distinct areas of liver metastases and 72 peripheral blood samples at multiple time points from 15 patients with colorectal cancer were analysis in this study. Results: In responding patients, mutation load from plasma were reduced from baseline (P< 0.001), but changed slightly in tumor tissues (P = 0.351). Transcriptomic analysis and immunohistochemistry revealed that increased infiltration of neutrophils and monocytes were associated with worse outcomes and insensitive response to chemotherapy (Neutrophils: P = 0.003; Monocytes: P = 0.032). Activation of fatty acid metabolism, JAK-STAT, PPAR, insulin and TGF-β signaling pathway were observed in progressive tumor. TCR diversity in response metastatic regions was significantly increased compared with non-responder (P = 0.008). Combination of bevacizumab with chemotherapy facilitated T cell infiltrating to the tumor microenvironment which might consequently benefit from checkpoint immunotherapy (P = 0.006). Conclusions: Our integrative and comparative genomic analysis provides a new paradigm for understanding the evolution and treatment resistance of colorectal cancer liver metastases, with implications for identifying ways to advance treatment regimen and monitoring treatment response of colorectal cancer liver metastases.