基于临床数据构建慢性乙型肝炎患者肝纤维化的新型预测评分：S-risk score

基于临床数据构建一种预测慢性乙型肝炎肝纤维化的无创诊断模型。

收集2021年1月至2023年7月宁波市医疗中心李惠利医院收治的165例CHB患者病例资料作回顾性分析，根据肝活检病理结果将患者分为无肝纤维化组（S0，n = 22）和肝纤维化组（≥S1，n = 143）。收集患者的血清学指标和临床数据，运用单因素和多因素 logitstic回归分析筛选出独立预测指标并建立模型，同时采用受试者工作特征曲线（ROC）评价模型的预测效能。

单因素分析结果显示，两组患者在白蛋白、谷草转氨酶、甘油三酯、总胆汁酸、胆碱酯酶、凝血酶原时间、 BMI、血清Ⅳ胶原和血清透明质酸等指标中存在差异（P＜0.05）。通过logistic多因素的回归分析构建肝纤维化模型S-risk score = −4.30+0.12×白蛋白+0.02×谷草转氨酶−0.05×碱性磷酸酶+0.29×甘油三酯+0.06×总胆汁酸−0.47×凝血酶原时间+0.20×BMI+0.03×血清Ⅳ胶原测定+0.02×血清透明质酸。该评分下的ROC曲线下的面积为0.866，其预测肝纤维化的准确性明显优于APRI和FIB-4两项评分模型。

我们构建的S-risk score模型对CHB患者肝纤维化有良好的预测能力，其预测准确性均高于APRI和FIB-4两项评分模型。

Construction of a new predictive score for liver fibrosis in patients with chronic hepatitis B based on clinical data: S-risk score

ObjectiveTo construct a non-invasive diagnostic model for predicting LF in CHB based on clinical data. MethodsThe data of 165 CHB patients admitted to Ningbo Medical Center Lihuili Hospital from January 2021 to July 2023 were collected and analyzed retrospectively. According to the pathological results of liver biopsy, the patients were divided into the group without LF (S0, n = 22) and the group with LF (≥S1, n = 143). Serological indicators and clinical data of patients were collected, and independent predictive indicators were screened out and models established using univariate and multivariate logistic regression analyses. The receiver operating characteristic curve (ROC) was used to evaluate the predictive accuracy of the comprehensive model. ResultsUnivariate analysis showed that there were differences between the two groups in albumin, aspartate aminotransferase, triglyceride, total bile acid, cholinesterase, prothrombin activity, BMI, serum collagen IV and serum hyaluronic acid. Through logistic multi-factor regression analysis, we constructed a LF model of S-risk score = −4.30 + 0.12 × albumin + 0.02 × aspartate aminotransferase − 0.05 × alkaline phosphatase + 0.29 × triglyceride + 0.06 × total bile acid − 0.47 × prothrombin activity + 0.20 × BMI + 0.03 × serum Ⅳ collagen determination + 0.02 × serum hyaluronic acid. The area under the ROC curve under this score was 0.866, and its accuracy in predicting LF is significantly better than the two score models of APRI and FIB-4. ConclusionThe S-risk score model we constructed has a good predictive ability for LF in CHB patients, and its prediction accuracy is higher than those of APRI and FIB-4 scoring models.