Liver Disease Center, The Affiliated Hospital of Qingdao University, Qingdao, China.
Cancer Med. 2023 Jan;12(1):898-912. doi: 10.1002/cam4.4858. Epub 2022 Jun 1.
Metabolic reprogramming is one of the hallmarks of cancer, but metabolic pathway activity-related subtypes of hepatocellular carcinoma (HCC) have not been identified.
Based on the quantification results of 41 metabolic pathway activities by gene set variation analysis, the training cohort (n = 609, merged by TCGA and GSE14520) was clustered into three subtypes (C1, C2, and C3) with the nonnegative matrix factorization method. Totally 1371 differentially expressed genes among C1, C2, and C3 were identified, and an 8-gene risk score was established by univariable Cox regression analysis, least absolute shrinkage and selection operator method, and multivariable Cox regression analysis.
C1 had the strongest metabolic activity, good prognosis, the highest CTNNB1 mutation rate, with massive infiltration of eosinophils and natural killer cells. C2 had the weakest metabolic activity, poor prognosis, was younger, was inclined to vascular invasion and advanced stage, had the highest TP53 mutation rate, exhibited a higher expression level of immune checkpoints, accompanied by massive infiltration of regulatory T cells. C3 had moderate metabolic activity and prognosis, the highest LRP1B mutation rate, and a higher infiltration level of neutrophils and macrophages. Internal cohorts (TCGA, n = 370; GSE14520, n = 239), external cohorts (ICGC, n = 231; GSE116174, n = 64), and clinical subgroup validation showed that the risk score was applicable for patients with diverse clinical features and was effective in predicting the prognosis and malignant progression of patients with HCC. Compared with the low-risk group, the high-risk group had a poor prognosis, enhanced cancer stem cell characteristics, activated DNA damage repair, weakened metabolic activity, cytolytic activity, and interferon response.
We identified HCC subtypes from the perspective of metabolism-related pathway activity and proposed a robust prognostic signature for HCC.
代谢重编程是癌症的标志之一,但尚未确定与肝癌(HCC)代谢途径活性相关的亚型。
基于基因集变异分析定量的 41 种代谢途径活性的结果,采用非负矩阵分解方法对包含 TCGA 和 GSE14520 的训练队列(n=609)进行聚类,将其分为 3 个亚型(C1、C2 和 C3)。鉴定 C1、C2 和 C3 之间的 1371 个差异表达基因,采用单变量 Cox 回归分析、最小绝对值收缩和选择算子法以及多变量 Cox 回归分析建立 8 基因风险评分。
C1 的代谢活性最强,预后良好,CTNNB1 突变率最高,大量嗜酸性粒细胞和自然杀伤细胞浸润。C2 的代谢活性最弱,预后不良,较年轻,倾向于血管侵犯和晚期,TP53 突变率最高,免疫检查点表达水平较高,伴有大量调节性 T 细胞浸润。C3 的代谢活性和预后适中,LRP1B 突变率最高,中性粒细胞和巨噬细胞浸润水平较高。内部队列(TCGA,n=370;GSE14520,n=239)、外部队列(ICGC,n=231;GSE116174,n=64)和临床亚组验证表明,风险评分适用于具有不同临床特征的患者,可有效预测 HCC 患者的预后和恶性进展。与低危组相比,高危组预后不良,增强了癌症干细胞特征,激活了 DNA 损伤修复,减弱了代谢活性、细胞溶解活性和干扰素反应。
我们从代谢相关途径活性的角度鉴定了 HCC 亚型,并提出了一种用于 HCC 的稳健预后特征。