摘要目的:探讨胆道闭锁(biliary atresia，BA)患儿的术前实验室检查结果与肝纤维化程度的相关性，拟建立更好的评估肝纤维化的无创预测模型。方法:回顾性分析2021年4月至2023年4月在浙江大学医学院附属儿童医院新生儿外科就诊的100例BA患儿的临床资料。采用Mann-Whitney U检验或Kruskal-Wallis H检验比较不同Ohkuma's肝纤维化分级和不同胆管增生分级患儿间各术前实验室指标的差异。根据多元logistic回归分析方法建立整合模型，各实验室指标同时绘制受试者操作特征(receiver operating characteristic，ROC)曲线，比较其对肝纤维化程度的预测效能。 结果:BA患儿Kasai手术年龄、血清基质金属蛋白酶-7(matrix metalloproteinase-7，MMP-7)、丙氨酸转氨酶(alanine aminotransferase，ALT)、天冬氨酸转氨酶(aspartate aminotransferase，AST)、直接胆红素(direct bilirubin，DBil)、间接胆红素(indirect bilirubin，IBil)、前白蛋白(prealbumin，PAB)、天冬氨酸转氨酶/血小板比值指数(aspartate aminotransferase to platelet ratio index，APRI)在不同肝纤维化分级及不同胆管增生分级患儿间差异有统计学意义(均 P<0.05)。肝纤维化分级越高，胆管增生分级普遍越高( P<0.001)。整合模型的ROC曲线下面积为0.876(95% CI：0.806，0.945)，敏感性为68.3%，特异性为93.2%，准确性为83.0%，阳性预测值为87.5%，阴性预测值为80.9%，高于各实验室指标的单独预测效能。 结论:Kasai手术年龄、MMP-7、APRI与BA患儿肝纤维化分级、胆管增生分级相关。多实验室指标的创新整合模型对于无创预测BA肝纤维化程度具有良好的诊断效能和准确度。

abstractsObjective:To explore the correlation between the results of preoperative laboratory tests and the degree of liver fibrosis (LF) to formulate a better noninvasive prediction model for evaluating LF in children with biliary atresia (BA).Methods:From April 2021 to April 2023, the relevant clinical data were retrospectively reviewed for 100 BA children.Mann-Whitney U or Kruskal-Wallis H test was utilized for comparing the differences of preoperative laboratory parameters between children with varying Ohkuma's grades of LF and bile duct hyperplasia (BDH). According to multivariate Logistic regression analysis, an integrated model was established.And receiver operating characteristic (ROC) curve was plotted for individual laboratory parameters to compare its predictive efficiency for the degree of LF.Results:Age at Kasai operation, serum matrix metalloproteinase-7 (MMP-7), alanine aminotransferase (ALT), aspartate aminotransferase (AST), direct bilirubin (DBil), indirect bilirubin (IBil), prealbumin (PAB) and aspartate aminotransferase to platelet ratio index (APRI) varied markedly between children with varying grades of LF and BDH (all P<0.05). The higher the grade of LF, the higher the grade of BDH ( P<0.001). The area under the ROC curve of the integrated model was 0.876(95% CI: 0.806, 0.945) with a sensitivity of 68.3%, a specificity of 93.2%, an accuracy of 83.0%, a positive predictive value of 87.5% and a negative predictive value of 80.9%, which were higher than the prediction efficiency of each laboratory parameter alone. Conclusions:Age of Kasai operation and level of MMP-7/APRI are correlated with the grades of LF and BDH.The innovative integrated model of multiple laboratory parameters offers better diagnostic efficiency and accuracy for non-invasive prediction of the degree of LF in BA.