背景：原发性肝癌作为全球高发恶性肿瘤，主要包括肝细胞癌（hepatocellular carcinoma，HCC）、肝内胆管癌（intrahepatic cholangiocarcinoma，iCCA）和混合型肝细胞癌-胆管细胞癌（combined hepatocellular-cholangiocarcinoma，CHC）三种病理亚型。其中，iCCA作为第二大原发性肝癌，近十年全球发病率与死亡率持续攀升，其高侵袭特征导致较高比例的患者疾病早期就伴随肝内转移，严重影响患者预后。与此同时，CHC作为兼具HCC与iCCA病理特征的特殊亚型，面临更严峻的临床困境：诊断层面因缺乏特异性生物标志物导致误诊率高，治疗层面因分子机制不明，尚未形成标准系统治疗方案。针对上述挑战，单细胞转录组测序技术凭借其高分辨率的技术优势，提供了全新视角。系统解析iCCA肝转移过程中肿瘤微环境（tumor microenvironment，TME）组分的动态演变特征和关键细胞亚群间的交互调控机制，以及CHC相较于HCC和iCCA的独特TME特征，将为iCCA与CHC的临床诊疗困境提供新的解决思路。

方法：本研究系统整合了多组学技术、功能实验及临床队列，分别针对iCCA及CHC开展研究。首先，本研究分别对iCCA肝转移研究队列（12例肝转移患者的24个原发或转移灶样本及4例无肝转移患者的4个肿瘤样本）和CHC发生发展研究队列（5例CHC患者的7个肿瘤样本、4例HCC患者的4个肿瘤样本以及4例iCCA患者的4个样本）进行了单细胞转录组测序。之后，通过对测序数据进行非负矩阵因式分解、细胞组成异质性分析、细胞互作网络分析、RNA速率、自组织映射和基因调控网络推断等分析，解析了iCCA肝转移过程中TME的动态特征以及CHC相较于HCC和iCCA的独特TME特征。此外，对于iCCA肝内转移机制研究，本研究进一步联合空间转录组测序数据、多重免疫组化及转录组测序数据验证了关键细胞亚群的空间定位与临床相关性，并通过体外和体内功能实验验证了数据分析关键结果的可靠性。针对CHC研究，本研究回顾性地纳入了10例经过系统治疗的CHC患者，评估了这些患者的临床获益情况，并对治疗前后的CHC患者（5例治疗前样本及1例治疗后样本）进行了单细胞转录组测序，通过分析揭示了CHC系统治疗前后TME发生的动态变化。

结果：在iCCA肝内转移机制研究中，鉴定出了具有上皮-间质转化特征的COL3A1+转移性肿瘤细胞亚群；揭示了COL3A1+转移性肿瘤细胞亚群可以通过阶段特异性机制驱动转移进程，在侵袭阶段通过TGFβ-JAG1-NOTCH3轴诱导内皮细胞间质化促进肿瘤侵袭，在定植阶段通过CXCL8-CXCR2、C5-C5AR1及TNF-TNFRSF1B配受体对介导中性粒细胞的募集及胞外诱捕网形成，保护肿瘤细胞免受细胞毒性免疫，促进肿瘤定植。在CHC的研究中，发现了CHC的肿瘤细胞具有HCC样和iCCA样双重基因程序特征；揭示了耗竭性CD8+T细胞可以通过CD274-PDCD1、CD80-CTLA4等免疫检查点相关配受体对与CXCL10+巨噬细胞发生串扰，形成CHC独特的抑制性TME；还发现了系统治疗可以通过重塑耗竭性CD8+T细胞的功能状态激活抗肿瘤免疫应答，使CHC患者从系统治疗中获益。

结论：本研究系统解析了iCCA肝转移过程中TME的时空动态演变特征，揭示了肝转移侵袭阶段与定植阶段的关键驱动机制。此外，本研究阐明了CHC相较于HCC和iCCA的独特肿瘤细胞分子特征及TME异质性，并揭示了系统治疗诱导的CHC的TME动态重塑机制。本研究构建了iCCA与CHC发生发展过程中单细胞分辨率的转录组图谱，为将来继续探究iCCA及CHC的发生发展机制提供了高分辨率数据支撑，为开发新的iCCA及CHC靶向干预策略提供了理论基础。

Background: Primary liver cancer, as a highly prevalent malignant tumor worldwide, mainly includes three pathological subtypes: hepatocellular carcinoma (HCC), intrahepatic cholangiocarcinoma (iCCA), and combined hepatocellular-cholangiocarcinoma (CHC).

Among them, iCCA, as the second most common primary liver cancer, has seen a continuous increase in global morbidity and mortality over the past decade, and its highly invasive characteristics lead to intrahepatic metastasis at an early stage in a high percentage of patients, which seriously affects the prognosis of patients. Meanwhile, CHC, as a special subtype with both HCC and iCCA pathologic features, faces more severe clinical dilemmas: high misdiagnosis rate due to the lack of specific biomarkers at diagnosis level, and lack of standardized systemic therapeutic regimens due to the unknown molecular mechanism at therapeutic level. To address these challenges, single-cell transcriptome sequencing provides a new perspective by virtue of its high-resolution technology. The systematic analysis of the dynamic evolution of tumor microenvironment (TME) components and the interaction regulation mechanism between key cell subpopulations during liver metastasis of iCCA, as well as the unique TME characteristics of CHC compared with HCC and iCCA, will provide new ideas to solve the dilemmas of clinical diagnosis and treatment of iCCA and CHC.

Methods: This study systematically integrates multi-omics techniques, functional experiments and clinical cohorts to conduct iCCA and CHC studies, respectively. First, in this study, single-cell transcriptome sequencing was performed on the iCCA liver metastasis study cohort (24 samples of primary or metastatic foci from 12 iCCA patients with liver metastases and 4 tumor samples from 4 iCCA patients without liver metastases) and the CHC development and progression study cohort (7 tumor samples from 5 CHC patients, 4 tumor samples from 4 HCC patients, and 4 samples from 4 iCCA patients) respectively. After that, the dynamic features of TME during liver metastasis in iCCA and the heterogeneous features of TME in CHC compared with HCC and iCCA were resolved by non-negative matrix factorization, cellular compositional heterogeneity analysis, cellular interactions network analysis, RNA rate, self-organization mapping, and gene regulatory network inference of the sequencing data. In addition, for the iCCA intrahepatic metastasis mechanism study, this study further validated the spatial localization and clinical relevance of key cell subpopulations by combining spatial transcriptome sequencing data, multiplex immunohistochemistry, and transcriptome sequencing data, and verified the reliability of the key results of the data analyses by in vitro and in vivo functional experiments. For the CHC study, 10 systemically treated CHC patients were retrospectively enrolled in this study, and the clinical benefit of these patients was assessed. Single-cell transcriptome sequencing was performed on CHC patients before and after treatment (5 pre-treatment samples and 1 post-treatment sample), and the analysis revealed the dynamic changes in the TME occurring before and after systemic treatment of CHC.

Results: In the study of iCCA intrahepatic metastatic mechanism, a subpopulation of COL3A1+metastatic tumor cells with epithelial-mesenchymal transition characteristics was identified. It was revealed that the subpopulation of COL3A1+ metastatic tumor cells could drive metastatic progression through a stage-specific mechanism. In the invasive stage it can induce mesenchymalization of endothelial cells through the TGFβ-JAG1-NOTCH3 axis to promote tumor invasion, and in the colonization stage it can protect tumor cells from cytotoxic immunity and promote tumor by mediating neutrophil recruitment and extracellular trap network formation through the CXCL8-CXCR2, C5-C5AR1 and TNF-TNFRSF1B ligand receptor pairs to promote tumor colonization. In the study of CHC, it was found that tumor cells in CHC are characterized by dual gene programs, HCC-like and iCCA-like, and the exhausted CD8+T cells can crosstalk with CXCL10+macrophages through immune checkpoint-associated ligand-receptor pairs, such as CD274-PDCD1 and CD80-CTLA4, to form a unique suppressive TME in CHC. And this study also discovered that systemic therapy can activate anti-tumor immune responses by remodeling the functional state of exhausted CD8+ T cells, allowing CHC patients to benefit from systemic therapy.

Conclusion: This study systematically analyzed the spatiotemporal dynamic evolution of TME during liver metastasis of iCCA, and revealed the key driving mechanisms of the invasive and colonization phases of liver metastasis. In addition, this study elucidated the unique tumor cell molecular features and TME heterogeneity of CHC compared with HCC and iCCA, and revealed the dynamic remodeling mechanism of TME in systemic therapy-induced CHC. This study constructed a single-cell-resolution transcriptome map during the development of iCCA and CHC, which provides high-resolution data support for the further investigation of the mechanisms of iCCA and CHC, and provides a theoretical basis for the development of new targeted intervention strategies for iCCA and CHC.