Objective: To develop a nomogram model for preoperative diagnosis of proliferative hepatocellular carcinoma(HCC) based on gadolinium ethoxybenzyl diethylenetriamine pentaacetic acid (Gd-EOB-DTPA) enhanced magnetic resonance imaging (MRI), and to explore its clinical value. Methods: MRI and clinical pathological data of patients confirmed by pathology as proliferative HCC (178 cases) and non-proliferative HCC (378 cases) between September 2017 and November 2022 who underwent preoperative Gd-EOB-DTPA enhanced MRI scans were retrospectively collected. The MRI features and clinical pathological characteristics of proliferative and non-proliferative HCC were evaluated. Multivariable logistic regression analysis was utilized to identify independent predictive factors for proliferative HCC, the R software was used to construct the nomogram prediction model, and its diagnostic performance was evaluated through receiver operating characteristic (ROC) curve. The calibration curve and decision curve analysis (DCA) were drawn to evaluate the calibration performance and clinical application value of the nomogram model. The optimal cut-off value was selected by calculating the Youden index to distinguish high risk and low risk. Kaplan-Meier survival curve was used to analyze the survival prognosis of proliferative and non-proliferative HCC, and log-rank test was used for comparison. Results: There were significant differences in AFP level(χ2=17.244, P<0.001), morphology of tumor(χ2=13.669, P<0.001), intertumoral fat(χ2=10.495, P=0.001), arterial phase peritumoral enhancement(χ2=37.662, P<0.001), tumor capsule(χ2=23.961, P<0.001), substantial intratumoral necrosis(χ2=77.184, P<0.001), intratumoral hemorrhage(χ2=4.892, P=0.027), peritumoral hypointense in hepatobiliary phase(χ2=47.675, P<0.001), rim arterial phase hyperenhancement(χ2=115.976, P<0.001), intratumoral artery(χ2=15.528, P<0.001) and venous tumor thrombus(χ2=10.532, P=0.001) between proliferative and non-proliferative HCC groups. Multivariate Logistic regression analysis showed that AFP>200 ng/ml(OR=0.640, P=0.044), no intertumoral fat(OR=1.947, P=0.033), substantial intratumoral necrosis(OR=0.480, P=0.003), peritumoral hypointense in hepatobiliary phase(OR=0.432, P=0.001), and rim arterial phase hyperenhancement(OR=0.180, P<0.001) were independent predictors of preoperative diagnosis of proliferative HCC. Based on the independent predictors, a nomogram model for preoperative prediction of proliferative HCC was established. The area under the ROC curve of the model for predicting proliferative HCC was 0.772 (95%CI: 0.735～0.807), the sensitivity was 69.1%, and the specificity was 75.4%. The calibration curve and DCA curve showed that the calibration performance and clinical applicability of the nomogram model were good. Kaplan-Meier curve showed that the survival rate of patients with proliferative HCC after hepatectomy was significantly lower than that of non-proliferative HCC (P<0.001), and the high-risk group was significantly lower than the low-risk group (P<0.001). Conclusions: The nomogram prediction model based on Gd-EOB-DTPA enhanced MRI imaging features combined with AFP >200 ng/ml can accurately diagnose proliferative HCC before operation and predict prognosis.
目的： 研发基于钆塞酸二钠（Gd-EOB-DTPA）增强磁共振成像（MRI）术前诊断增殖型肝细胞癌（HCC）的列线图模型，并探讨其临床价值。方法： 回顾性收集2017年9月至2022年11月经病理证实为增殖型（178例）及非增殖型HCC（378例）患者的术前Gd-EOB-DTPA增强MRI影像学资料及其临床病理资料。评估增殖型与非增殖型HCC的MRI影像学特征及其临床病理特征。采用多因素logistic回归分析确定增殖型HCC的独立预测因素，采用R软件构建列线图预测模型，通过受试者操作特征曲线（ROC）评价其诊断效能，绘制校准曲线、决策曲线（DCA）以评估列线图模型的校准性能和临床应用价值。通过约登指数选择最佳阈值以区分高风险和低风险，使用Kaplan-Meier生存曲线分析增殖型和非增殖型HCC的生存预后，并通过log-rank检验进行比较。计量资料采用独立样本t检验或Mann-Whitney U检验。计数资料比较采用χ2检验。结果： 增殖型与非增殖型HCC患者在甲胎蛋白（AFP）水平（χ2=17.244，P<0.001）、肿瘤形态（χ2=13.669，P<0.001）、瘤内脂肪变性（χ2=10.495，P=0.001）、动脉期瘤周异常强化（χ2=37.662，P<0.001）、肿瘤包膜（χ2=23.961，P<0.001）、瘤内坏死（χ2=77.184，P<0.001）、瘤内出血（χ2=4.892，P=0.027）、肝胆期瘤周低信号（χ2=47.675，P<0.001）、动脉期环形高强化（χ2=115.976，P<0.001）、瘤内动脉（χ2=15.528，P<0.001）、静脉内癌栓（χ2=10.532，P=0.001）方面的差异均有统计学意义。多因素logistic回归分析显示AFP＞200 μg/L（OR=0.640，P=0.044）、瘤内无脂肪变性（OR=1.947，P=0.033）、瘤内坏死（OR=0.480，P=0.003）、肝胆期瘤周低信号（OR=0.432，P=0.001）、动脉期环形高强化（OR=0.180，P<0.001）是术前诊断增殖型HCC的独立预测因素，基于独立预测因素建立术前预测增殖型HCC列线图模型；该模型预测增殖型HCC的ROC曲线下面积为0.772（95%CI：0.735～0.807），灵敏度为69.1%、特异度为75.4%。校准曲线、DCA曲线显示列线图模型的校准性能和临床适用性均较好。Kaplan-Meier曲线显示，增殖型HCC患者的肝切除术后预后生存率显著低于非增殖型（P<0.001），高风险组显著低于低风险组（P<0.001）。结论： 基于Gd-EOB-DTPA增强MRI影像学特征联合AFP＞200 μg/L构建的列线图模型可较准确地术前诊断增殖型HCC和预测其预后。.