Table 5_Development of a prognostic model for hepatocellular carcinoma based on microvascular invasion characteristic genes by spatial transcriptomics sequencing.xlsx

Microvascular invasion (MVI) is an independent risk factor for the recurrence and metastasis of hepatocellular carcinoma (HCC), associated with poor prognosis. Thus, MVI has significant clinical value for the treatment selection and prognosis assessment of patients with HCC. However, there is no reliable and precise method for assessing the postoperative prognosis of MVI patients. This study aimed to develop a new HCC prognosis prediction model based on MVI characteristic genes through spatial transcriptomics sequencing, distinguishing between high-risk and low-risk patients and evaluating patient prognosis. In this study, four MVI samples with different grades were selected for spatial transcriptomic sequencing to screen for MVI region-specific genes. On this basis, an HCC prognostic model was constructed using univariate Cox regression analysis, LASSO regression analysis, random survival forest, and stepwise multivariate Cox regression analysis methods. We constructed a 7-gene prognostic model based on MVI characteristic genes and demonstrated its applicability for predicting the prognosis of HCC patients in three external validation cohorts. Furthermore, our model showed superior predictive performance compared with three published HCC prediction prognostic models and could serve as an independent prognostic factor for HCC. Additionally, single nucleus RNA sequencing analysis and multiple immunofluorescence images revealed an increased proportion of macrophages in high-risk patient samples, suggesting that HCC tumor cells may promote HCC metastasis through MIF-CD74 cell interactions. To sum up, we have developed a 7-gene biomarker based on MVI that can predict the survival rate of HCC patients at different stages. This predictive model can be used to categorize into high- and low- risk groups, which is of great significance for the prognostic assessment and personalized treatment of HCC patients.

微血管侵犯（Microvascular invasion, MVI）是肝细胞癌（hepatocellular carcinoma, HCC）复发与转移的独立危险因素，与不良预后密切相关。因此，MVI对HCC患者的治疗选择与预后评估具有重要临床价值。然而，目前尚无可靠且精准的方法用于评估MVI阳性患者的术后预后情况。本研究旨在基于空间转录组测序（spatial transcriptomics sequencing）技术，依托MVI特征基因构建新型HCC预后预测模型，以区分高风险与低风险患者并评估患者预后。本研究选取4例不同分级的MVI样本进行空间转录组测序，筛选MVI区域特异性基因。在此基础上，采用单因素Cox回归分析（univariate Cox regression analysis）、LASSO回归分析（LASSO regression analysis）、随机生存森林（random survival forest）及逐步多因素Cox回归分析（stepwise multivariate Cox regression analysis）方法构建HCC预后模型。我们基于MVI特征基因构建了包含7个基因的预后模型，并在3个外部验证队列（external validation cohorts）中验证了该模型对HCC患者预后的预测效能。进一步研究显示，相较于已发表的3种HCC预后预测模型，本模型展现出更优的预测性能，且可作为HCC的独立预后因素。此外，单细胞核RNA测序（single nucleus RNA sequencing）分析与多重免疫荧光成像（multiple immunofluorescence images）结果表明，高风险患者样本中的巨噬细胞比例升高，提示HCC肿瘤细胞可能通过MIF-CD74细胞相互作用促进肝癌转移。综上，本研究开发了一种基于MVI的7基因生物标志物，可预测不同分期HCC患者的生存率。该预测模型可将患者划分为高、低风险组，对HCC患者的预后评估与个性化治疗具有重要意义。