Background: Liver transplant (LT) contributes to the cure of hepatocellular carcinoma (HCC) by extending the conventional surgical margin, while removing the concurrent tumor-favoring conditions of the diseased liver. Over the past two decades, HCC has emerged as a preferable indication for LT, accounting for 22-45% of the cases globally. However, despite the stringent selection criteria, recurrence still occurs in a substantial proportion of patients, leaving 10-13 months of median survival after tumor relapse. To address this challenge, a thorough exploration of molecular mechanism behind HCC recurrence after LT is required. Here, for the first time, we performed a comprehensive analysis to depict the evolving and unique ecosystem of HCC in the context of LT and immunosuppressant (IS) administration. Methods: We retrospectively reviewed our biobank and collected 91 paired (primary tumors (PT) from the explant after LT and relapsed tumors (RT) from hepatectomy) FFPE tumor samples from 15 patients for whole exome sequencing (WES). Also, from Jan 2021, we prospectively collected surgical tumor specimens and their adjacent non-tumor tissues from 15 primary and 6 post-LT intrahepatic recurrence patients for single-cell RNA sequencing. Moreover, 6 paired samples from 3 patients were used for spatial transcriptomic analysis. Results: By analysing WES data, we confirmed that RT samples represent recurrent, instead of de novo tumors. Also, the phylogenetic analysis demonstrated strong clonal relationship between PT and RT based on shared SNVs. As to mutation landscape differences of PT and RT, we identified a significantly enriched mutation panel from RT, including MUC16, FAT3, APOB, COL6A3 and POTEC. And the accuracy of this panel for post-LT HCC recurrence surveillance using plasma derived cfDNA sequencing has been preliminarily validated in a prospective cohort of 17 patients. From the single-cell profiling, distinct immune microenvironment between PT and RT is identified. As predicted, proliferating T cells are decreased in RT due to IS. Pseudotime analysis of T cells revealed that T cells from PT mainly distribute at the terminal end of cytotoxic stage whereas T cells from RT mostly represent a naïve state, further confirming the T cell blocking effect of IS. In terms of myeloid cells, we surprisingly found that FOLR2+ tumor associated macrophages, which is conventionally considered as a favorable prognosis factor, is increased in RT. Spatial transcriptomics and IHC also confirmed their active interplay with tumor cells at invasive margin. We postulate that such increase might be a result of the healthy nature of liver graft. Conclusions: Our comprehensive dissection of the HCC ecosystem provides deeper insights into the tumor relapse after LT. Targeting novel mutations and strengthen the antigen presenting process between increased TAM and disabled T cells can be the future direction in this field. Clinical trial information: NCT04506398.