Integrated multi-omics profiling to dissect the spatiotemporal evolution of metastatic hepatocellular carcinoma.

556 Background: Comprehensive molecular analyses of metastatic hepatocellular carcinoma (HCC) are lacking. Methods: Here, we generated genomic, transcriptomic, and T cell repertoire data coupled with digital pathology and digital spatial profiling of 257 primary, 28 relapsed, and 176 metastatic regions from 182 HCC patients. Results: Primary tumors rich in hypoxia signatures tended to give rise to polyclonal dissemination, which was associated with a poor clinical outcome. Genomic divergence between primary and metastatic HCC was high, and early dissemination was prevalent. The remarkable neoantigen intratumor heterogeneity observed in metastases was associated with decreased T cell reactivity, which may have resulted from disruptions to neoantigen presentation. We identified macroevolutionary somatic copy number alterations as highly selected events driving metastasis. Furthermore, we found that subclonal Wnt pathway mutations were not selected during spreading, whereas subclones without Wnt mutations showed a strong selective advantage for metastasis and were characterized by a reactive microenvironment rich in activated fibroblasts favoring an invasive tumor cell phenotype. Finally, metastases without Wnt mutations exhibited higher enrichment of immunosuppressive B cells that mediated terminal exhaustion of CD8+ T cells via HLA-E – NKG2A checkpoint axis. Conclusions: Collectively, our results provide a multi-dimensional dissection of the complex evolutionary process of metastasis.