Metabolic biomarkers significantly enhance the prediction of HBV-related ACLF occurrence and outcomes

作者：Zhang, Yan; Tan, Wenting; Wang, Xianbo; Zheng, Xin; Huang, Yan; Li, Beilin; Meng, Zhongji; Gao, Yanhang; Qian, Zhiping; Liu, Feng; Lu, Xiaobo; Shi, Yu; Shang, Jia; Yan, Huadong; Zheng, Yubao; Zhang, Weituo; Gu, Wenyi; Qiao, Liang; Deng, Guohong; Zhou, Yi; Hou, Yixin; Zhang, Qun; Xiong, Shue; Liu, Jing; Duan, Lihua; Chen, Ruochan; Chen, Jinjun; Jiang, Xiuhua; Luo, Sen; Chen, Yuanyuan; Jiang, Chang; Zhao, Jinming; Ji, Liujuan; Mei, Xue; Li, Jing; Li, Tao; Zheng, Rongjiong; Zhou, Xinyi

来源：Journal of Hepatology, 2023, 79(5): 1159-1171.

DOI：10.1016/j.jhep.2023.07.011

• diagnostic biomarkers
• clinical test development
• prediction model

Background & Aims: Acute-on-chronic liver failure (ACLF) is a clinical syndrome associated with high short-term mortality in patients with chronic liver disease. Chronic hepatitis B is the main cause of ACLF (HBV-ACLF) in China and other Asian countries. To improve disease management and survival for patients with ACLF, we aimed to discover novel biomarkers to enhance HBV-ACLF diagnosis and prognostication. Methods: We performed a metabolomics profiling of 1,024 plasma samples collected from patients with HBV-related chronic liver disease with acute exacerbation at hospital admission in a multi-year and multi-center prospective study (367 ACLF and 657 non-ACLF). The samples were randomly separated into equal halves as a discovery set and a validation set. We identified metabolites associated with 90-day mortality in the ACLF group and the progression to ACLF within 28 days in the non-ACLF group (pre-ACLF) using statistical analysis and machine learning. We developed diagnostic algorithms in the discovery set and used these to assess the findings in the validation set. Results: ACLF significantly altered the plasma metabolome, particularly in membrane lipid metabolism, steroid hormones, oxidative stress pathways, and energy metabolism. Numerous metabolites were significantly associated with 90-day mor-tality in the ACLF group and/or pre-ACLF in the non-ACLF group. We developed algorithms for the prediction of 90-day mortality in patients with ACLF (area under the curve 0.87 and 0.83 for the discovery set and validation set, respectively) and the diagnosis of pre-ACLF (area under the curve 0.94 and 0.88 for the discovery set and validation set, respectively). To translate our discoveries into practical clinical tests, we developed targeted assays using liquid chromatography -mass spectrometry. Conclusions: Based on novel metabolite biomarkers, we established tests for HBV-related ACLF with higher accuracy than existing methods. Clinical Trial Number: NCT02457637 and NCT03641872.

• 单位
  浙江大学; 南方医科大学; 华中科技大学; 南开大学; 吉林大学; 山东大学; 中山大学; 上海交通大学; 复旦大学