| 摘要: |
| 目的：将超声造影和临床信息结合，开发并验证针对小肝癌（直径不超过3厘米且肿瘤个数不超过三个）微血管侵犯（MVI）的预测模型。
方法：本研究纳入2017年1月至2023年1月于我院行肝癌切除术的患者。使用病理结果作为检测MVI的参考标准，将患者分为MVI阳性组与阴性组，患者按照7:3的比例随机分配到建模队列与验证队列。采用卡方检验和两独立样本t检验或曼-惠特尼秩和检验在建模队列中比较MVI阳性组与阴性组各变量间的差异，随后利用多变量逻辑回归分析识别预测MVI的独立风险因素。基于这些风险因素，构建了临床因素模型、超声因素模型及融合因素模型，并比较三个模型的曲线下面积（AUC）、校准度以及临床预测模型决策曲线。
结果：共纳入423名患者，包括171例MVI阳性和252例MVI阴性患者。建模队列包含296例患者。单因素分析显示，MVI阳性组与阴性组在LI-RADS分级、甲胎蛋白（AFP）水平、造影后与造影前肿瘤面积增加比例、性别分布方面存在差异。多因素分析表明，更高的LI-RADS分级、更高的AFP水平、造影后肿瘤面积的增加是预测MVI的独立风险因素，OR分别为1.675、1.585、1.292。在建模队列中，临床因素模型、超声因素模型、融合因素模型的AUC分别为0.624、0.669、0.851，显示融合因素模型具有更高的区分度和更好的临床预测模型决策曲线。
结论：本研究基于超声指标与临床指标构建的MVI预测模型，相较于单独的临床指标或超声指标，展现出更高的预测效能，为临床制定个性化方案提供了更丰富的参考信息。 |
| 关键词: 超声造影 肝癌 微血管癌栓 预测 LI-RADS |
| DOI： |
| 投稿时间：2024-02-26修订日期：2024-09-10 |
| 基金项目: |
|
| Application Value of Contrast-enhanced Ultrasound LI-RADS Classification in Microvascular Invasion of Small Hepatocellular Carcinoma |
| Wang zhenbao,Liu guoan,Lai jiangqiong |
| (Department of Ultrasound Diagnosis, the 910th Hospital of Joint Logistic Support Force of PLA) |
| Abstract: |
| Objective: To establish and validate a predictive framework for microvascular invasion (MVI) in hepatocellular carcinoma (HCC) cases characterized by tumors ≤3 cm in diameter and limited to three lesions. This study leverages both contrast-enhanced ultrasound (CEUS) findings and clinical data for prediction.
Methods: This study enrolled HCC patients undergoing hepatectomy at our center between January 2017 and January 2023. Histopathological examination defined the presence of MVI, segregating subjects into MVI-positive and MVI-negative cohorts. Participants were randomly divided into training (70%) and validation (30%) sets. Initial analysis utilized chi-square, t-tests, or Mann-Whitney U tests for identifying variable disparities between MVI cohorts within the training set. Subsequent multivariable logistic regression discerned independent MVI predictors, facilitating the construction of clinical, ultrasound, and integrative predictive models. These models' performance was appraised through Area Under the Receiver Operating Characteristic Curve (AUC), calibration analysis, and decision curve evaluation.
Results: The study incorporated 423 individuals, with 171 diagnosed with MVI and 252 without. Analysis within the 296-patient training cohort highlighted significant variations in LI-RADS classification, alpha-fetoprotein (AFP) levels, changes in tumor size post-contrast, and gender across MVI statuses. Multivariate regression underscored elevated LI-RADS classification, increased AFP, and tumor size enlargement post-contrast as independent MVI risk contributors, yielding odds ratios of 1.675, 1.585, and 1.292, respectively. Comparative evaluation of the models revealed AUCs of 0.624 (clinical), 0.669 (ultrasound), and 0.851 (integrated), signifying the integrated model's superior predictive accuracy and clinical utility.
Conclusion: Our integrative model, combining CEUS and clinical metrics, outperforms single-factor models in predicting MVI in small HCC lesions. This model promises enhanced guidance for tailoring individualized therapeutic strategies. |
| Key words: Contrast-enhanced ultrasound (CEUS) Hepatocellular carcinoma (HCC) Microvascular invasion (MVI) Prediction LI-RADS |
