Objective: To explore the application value of shear wave dispersion (SWD) and shear wave elastography (SWE) combined with serological indicators in the evaluation of liver fibrosis. Methods: A total of 219 patients with liver disorders who underwent liver biopsy were prospectively collected in Huashan Hospital, Fudan University from January 2021 to September 2022, including 130 males and 89 females, aged from 18 to 76 (42±12) years. All patients underwent SWD and SWE examinations before liver biopsy. Serological indicators including alanine aminotransferase(ALT), aspartate aminotransferase(AST), alkaline phosphatase(ALP)) and γ-glutamyl transpeptadase (GGT) were also collected. Based on pathological diagnosis of liver fibrosis stage (from S0 to S4), the distribution of dispersion slope and liver elastic modulus at different fibrosis stages were analyzed in all patients. All patients were divided 7: 3 into training set (156 cases) and validation set (63 cases) in chronological order. In training set, factors influencing liver fibrosis≥S2 stage and S4 stage were analysed using binary logistic regression. The predictive models were established for diagnosing liver fibrosis≥S2 stage and S4 stage by using R language, and the models were evaluated by the area under curve (AUC) and calibrated for validation. Results: The dispersion slope and elastic modulus increased with the severity of fibrosis, with statistically significant differences in different fibrosis stages (both P<0.001). In training set, dispersion slope, elastic modulus, ALT, AST, and GGT were influential factors in liver fibrosis≥S2 stage and S4 stage(both P<0.05), and prediction models were constructed based on these indicators. In training set, the AUCs of the predictive model, SWD and SWE for diagnosingliver fibrosis≥S2 stage were 0.743 (95%CI: 0.665-0.821), 0.709 (95%CI: 0.628-0.790) and 0.725 (95%CI: 0.647-0.804), respectively; for diagnosing liver fibrosis S4 stage, the AUCs were 0.988 (95%CI: 0.968-1.000), 0.908 (95%CI: 0.852-0.963) and 0.974 (95%CI: 0.945-1.000), respectively. In validation set, the AUC of the predictive model, SWD and SWE for diagnosing liver fibrosis≥S2 stage were 08.735 (95%CI: 0.612-0.859), 0.658 (95%CI:0.522-0.793) and 0.699 (95%CI:0.570-0.828), respectively; for diagnosing liver fibrosis S4 stage, the AUC were 0.976 (95%CI: 0.937-1.000), 0.872 (95%CI: 0.757-0.988) and 0.948 (95%CI: 0.889-1.000), respectively. The calibration curves of the prediction models were consistent in the training and validation sets. Conclusion: The predictive model of SWD and SWE combined with serological indicators is helpful in the diagnosis of stage of liver fibrosis non-invasively.
目的： 探讨剪切波频散成像（SWD）和弹性成像（SWE）联合血清学指标的预测模型在肝纤维化分期中的应用价值。 方法： 前瞻性收集2021年1月至2022年9月因肝功能异常就诊于复旦大学附属华山医院并进行肝穿刺活检患者219例，其中男130例，女89例，年龄18~76（42±12）岁。所有患者穿刺前均接受SWD和SWE检查。同时收集血清学指标［丙氨酸转氨酶（ALT）、天冬氨酸转氨酶（AST）、碱性磷酸酶（ALP）和γ-谷氨酰转肽酶（GGT）］。以病理肝纤维化分期（S0~S4期）为标准，分析总体人群的频散斜率和弹性模量在不同纤维化分期中的分布。将总体人群以入组时间按7∶3分为训练集（156例）和验证集（63例）。训练集中，使用二元logistic回归分析肝纤维化≥S2期和S4期的影响因素。用R语言构建诊断≥S2期和S4期预测模型，用曲线下面积（AUC）评价模型并进行校准验证。 结果： 频散斜率和弹性模量随纤维化严重程度而增加，在不同纤维化分期中差异具有统计学意义（均P<0.001）。训练集中，频散斜率、弹性模量、ALT、AST和GGT是肝纤维化≥S2期和S4期的影响因素（均P<0.05），基于这些指标构建预测模型。训练集中，预测模型、SWD和SWE诊断肝纤维化≥S2期的AUC分别为0.743（95%CI：0.665~0.821）、0.709（95%CI：0.628~0.790）和0.725（95%CI：0.647~0.804），诊断肝纤维化S4期的AUC分别为0.988（95%CI：0.968~1.000）、0.908（95%CI：0.852~0.963）和0.974（95%CI：0.945~1.000）；验证集中，预测模型、SWD和SWE诊断肝纤维化≥S2期的AUC分别为0.735（95%CI：0.612~0.859）、0.658（95%CI：0.522~0.793）和0.699（95%CI：0.570~0.828），诊断肝纤维化S4期的AUC分别为0.976（95%CI：0.937~1.000）、0.872（95%CI：0.757~0.988）和0.948（95%CI：0.889~1.000）。校准曲线显示训练集和验证集中预测模型一致性好。 结论： SWD和SWE联合血清学的预测模型有助于无创诊断肝纤维化程度。.