Multi-Scale Spatial Analysis of the Tumor Microenvironment Reveals Features of Cabozantinib and Nivolumab Efficacy in Hepatocellular Carcinoma.
Multi-Scale Spatial Analysis of the Tumor Microenvironment Reveals Features of Cabozantinib and Nivolumab Efficacy in Hepatocellular Carcinoma.
复制标题
DOI：
10.3389/fimmu.2022.892250
复制
发表时间：
2022
影响因子：
7.3
通讯作者：
中科院分区：
文献类型：
--
作者：
同时抑制血管内皮生长因子 (VEGF) 和程序性细胞死亡蛋白 1 (PD-1) 或其配体 PD-L1 是晚期肝细胞癌 (HCC) 患者的标准治疗方法，但只有少数患者有反应，而且反应通常是短暂的。了解治疗对肿瘤微环境 (TME) 的影响可以深入了解治疗耐药机制。约翰霍普金斯大学西德尼金梅尔综合癌症中心对 14 名 HCC 患者进行了卡博替尼和纳武单抗联合治疗。其中，12 名患者（5 名有反应者 + 7 名无反应者）成功进行了切缘阴性切除，并进行了包含 37 个代表性肿瘤区域核心的组织微阵列 (TMA) 构建。使用 TMA，我们对一组 27 细胞谱系和功能标记物进行了成像质谱流式分析 (IMC)。然后对所有多重图像进行分割以生成单细胞数据集，该数据集能够（1）肿瘤免疫区室分析和（2）基于图形嵌入方法的细胞群落分析。这些层次结构的结果被合并到与响应相关的生物过程模式中。对 37 个多重图像进行图像处理，区分出 59,453 个细胞，然后将其聚类为 17 种细胞类型。区室分析显示，在NR的免疫肿瘤边界处，肿瘤细胞上的PD-L1水平显着高于偏远区域；然而，颗粒酶 B 的表达却显示出相反的模式。我们还发现，CD8+ T 细胞与精氨酸酶 1hi (Arg1hi) 巨噬细胞（而不是 CD4+ T 细胞）非常接近，是无应答者 TME 的显着特征。此外，细胞群落分析提取了 8 种类型的细胞间相互作用网络，称为细胞群落 (CC)。我们观察到，在无应答者中，富含巨噬细胞的 CC (MCC) 和富含淋巴细胞的 CC (LCC) 与肿瘤 CC 强烈通讯，而在应答者中，这种通讯因 MCC 和 LCC 之间的接触而受到破坏。这些结果证明了多重图像分析的新应用的可行性，该分析广泛适用于免疫肿瘤学中病理标本的定量分析，并进一步证明 CD163-Arg1hi 巨噬细胞可能是 HCC 的治疗靶点。研究结果还为开发旨在预测临床试验结果的机械定量系统药理学模型提供了关键信息。
Concomitant inhibition of vascular endothelial growth factor (VEGF) and programmed cell death protein 1 (PD-1) or its ligand PD-L1 is a standard of care for patients with advanced hepatocellular carcinoma (HCC), but only a minority of patients respond, and responses are usually transient. Understanding the effects of therapies on the tumor microenvironment (TME) can provide insights into mechanisms of therapeutic resistance. 14 patients with HCC were treated with the combination of cabozantinib and nivolumab through the Johns Hopkins Sidney Kimmel Comprehensive Cancer Center. Among them, 12 patients (5 responders + 7 non-responders) underwent successful margin negative resection and are subjects to tissue microarray (TMA) construction containing 37 representative tumor region cores. Using the TMAs, we performed imaging mass cytometry (IMC) with a panel of 27-cell lineage and functional markers. All multiplexed images were then segmented to generate a single-cell dataset that enables (1) tumor-immune compartment analysis and (2) cell community analysis based on graph-embedding methodology. Results from these hierarchies are merged into response-associated biological process patterns. Image processing on 37 multiplexed-images discriminated 59,453 cells and was then clustered into 17 cell types. Compartment analysis showed that at immune-tumor boundaries from NR, PD-L1 level on tumor cells is significantly higher than remote regions; however, Granzyme B expression shows the opposite pattern. We also identify that the close proximity of CD8+ T cells to arginase 1hi (Arg1hi) macrophages, rather than CD4+ T cells, is a salient feature of the TME in non-responders. Furthermore, cell community analysis extracted 8 types of cell-cell interaction networks termed cellular communities (CCs). We observed that in non-responders, macrophage-enriched CC (MCC) and lymphocyte-enriched CC (LCC) strongly communicate with tumor CC, whereas in responders, such communications were undermined by the engagement between MCC and LCC. These results demonstrate the feasibility of a novel application of multiplexed image analysis that is broadly applicable to quantitative analysis of pathology specimens in immuno-oncology and provides further evidence that CD163-Arg1hi macrophages may be a therapeutic target in HCC. The results also provide critical information for the development of mechanistic quantitative systems pharmacology models aimed at predicting outcomes of clinical trials.