近日，瑞典卡罗林斯卡学院Rickard Strandberg团队使用新的CORE风险评分预测普通人群10年肝硬化风险。该项研究成果发表在2025年9月29日出版的《英国医学杂志》上。

为了开发并验证一种新的重大不良肝脏预后（MALO）的风险预测模型设置，研究组在瑞典、芬兰和英国进行了一项基于人群的队列研究。在480651名无肝病病史的个体中开发模型，并在初级保健或职业健康筛查中进行血液检测；在两个队列中验证了24191和449806名没有已知肝病史的个体。主要结局为代偿性和失代偿性肝硬化、肝细胞癌、肝移植和肝相关死亡率（统称为MALO）复合结局的10年风险。

研究组使用灵活的参数生存模型和几个容易获得的实验室生物标志物创建了一个新的风险模型。模型包括年龄、性别、天冬氨酸转氨酶、丙氨酸转氨酶和γ-谷氨酰转移酶。该模型的性能从鉴别（曲线下随时间变化的面积）、校准（校准曲线）和临床效用（决策曲线分析）三个方面进行评估。外部验证使用来自UK Biobank、FINRISK和Health 2000队列的数据，并与FIB-4评分进行比较。中位随访时间为28年，在此期间观察到7168例MALO事件。10年MALO事件风险为0.27%。新的风险评分，称为CORE（肝硬化结局风险估计值），曲线下10年的面积为88%(95%置信区间为87%至89%)，而FIB-4为79%（78%至80%）。CORE的校准在所有三个队列中都很好，根据决策曲线分析，CORE在所有风险阈值上都比FIB-4提供更高的净收益。

研究结果表明，CORE模型基于一种灵活的建模方法，并使用在初级保健中易于获得的生物标志物，在预测普通人群的肝脏相关结局时优于FIB-4，可能成为对普通人群中有肝脏疾病风险的患者进行分层的新手段。

附：英文原文

Title: Use of new CORE risk score to predict 10 year risk of liver cirrhosis in general population: population based cohort study

Author: Rickard Strandberg, Fredrik berg, Juho V Asteljoki, Panu K Luukkonen, Veikko Salomaa, Antti Jula, Annamari Lundqvist, Satu Mnnist, Markus Perola, Mats Talbck, Niklas Hammar, Hannes Hagstrm

Issue&Volume: 2025/09/29

Abstract:

Objective To develop and validate a novel risk prediction model for incident major adverse liver outcomes (MALO) in a primary care setting.

Design Population based cohort study.

Setting Sweden, with validation in Finland and the UK.

Participants Model development in 480651 individuals with no known history of liver disease and blood tests taken in primary care or at occupational healthcare screenings; validation in two cohorts with 24191 and 449806 individuals without known history of liver disease.

Main outcome measures 10 year risk of a composite outcome of compensated and decompensated cirrhosis, hepatocellular carcinoma, liver transplant, and liver related mortality, collectively referred to as MALO.

Results A new risk model was created using flexible parametric survival models and several easily available laboratory based biomarkers. The model includes age, sex, aspartate aminotransferase, alanine aminotransferase, and γ-glutamyl transferase. The model’s performance was assessed in terms of discrimination (time dependent area under the curve), calibration (calibration curves), and clinical utility (decision curve analysis). External validation was done using data from the UK Biobank and the FINRISK and Health 2000 cohorts and compared with the FIB-4 score. The median follow-up time was 28 years, and 7168 MALO events were observed in that time. The incident risk of MALO at 10 years was 0.27%. The new risk score, termed CORE (Cirrhosis Outcome Risk Estimator), achieved a 10 year area under the curve of 88% (95% confidence interval 87% to 89%) compared with 79% (78% to 80%) for FIB-4. The calibration of CORE was good in all three cohorts, and according to the decision curve analysis CORE provides a higher net benefit than FIB-4 for all risk thresholds.

Conclusions The CORE model, based on a flexible modelling approach and using biomarkers easily accessible in primary care, outperforms FIB-4 when predicting liver related outcomes in the general population and could be a novel means to stratify patients at risk for liver disease in the general population.

DOI: 10.1136/bmj-2024-083182

Source: https://www.bmj.com/content/390/bmj-2024-083182