CEUS LI-RADS对中国≤30mm肝细胞癌的诊断性能多中心分析
本研究纳入815例患者共856个≤30mm肝脏结节,分为回顾性推导队列和前瞻性验证队列,评估CEUS LI-RADS v2017并优化诊断标准。优化方案将早期廓清阈值从<60秒收紧至<50秒,并结合AFP>20 ng/mL重新分类LR-4/LR-M结节为LR-5++。结果显示LR-5++显著提高验证队列敏感度至71.43%和准确率至80.31%。
要点
- 标准LR-5特异度高但敏感度不足,≤30mm小肝癌易漏诊
- AFP>20 ng/mL联合CEUS征象可显著提高小肝癌检出敏感度
- 病毒相关性肝病或肝硬化患者中LR-5++提升敏感度最明显
核心结论
- 整合AFP>20 ng/mL的改良CEUS LI-RADS标准(LR-5++)可显著提高≤30mm肝细胞癌的诊断敏感度和准确率,弥补标准算法的敏感度不足
研究设计与样本
- 多中心研究,共纳入815例患者856个≤30mm肝脏结节,分为回顾性推导队列和前瞻性验证队列
- 采用CEUS LI-RADS v2017进行分类,并与优化标准进行比较,评价指标包括准确率、敏感度、特异度和阳性预测值
关键影像表现/征象
- 标准LR-5在验证队列中特异度为93.01%,敏感度仅为57.69%
- 早期廓清阈值从<60秒收紧至<50秒并将50至<60秒廓清的LR-M重新分类为LR-5+
- 整合AFP>20 ng/mL后LR-5++在验证队列中敏感度提升至71.43%,准确率提升至80.31%
- LR-5++特异度从93.01%轻微降至91.61%,差异无统计学意义
诊断标准与关键数值
- LR-5++标准:在CEUS LI-RADS基础上,将AFP>20 ng/mL的LR-4/LR-M结节重新分类为LR-5++
- LR-5+标准:早期廓清阈值从<60秒收紧至<50秒,将廓清起始50至<60秒的LR-M病灶重新分类为LR-5+
- 标准LR-5特异度为87.67-93.01%,敏感度为57.69-64.41%
- 验证队列中LR-5++敏感度从57.69%提升至71.43%(P<0.05),准确率从73.23%提升至80.31%(P<0.05)
临床应用与阅片要点
- 对于CEUS表现为LR-4或LR-M的≤30mm肝脏结节,若血清AFP>20 ng/mL应考虑升级为LR-5++诊断
- 病毒相关性肝病或肝硬化患者中使用LR-5++标准获益最大
- 报告时可注明采用改良CEUS LI-RADS标准以提高小肝癌诊断敏感度
局限性
- 部分数据来自回顾性推导队列,存在选择偏倚可能
- 研究人群为中国扩大高危人群,结果外推至其他人群需谨慎
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
英文原文(摘要)· 点击展开
RATIONALE AND OBJECTIVES: We aimed to validate and optimize the contrast-enhanced ultrasound (CEUS) Liver Imaging Reporting and Data System (LI-RADS) for diagnosing small hepatocellular carcinoma (sHCC; ≤ 30 mm) in a Chinese expanded high-risk population: a viral-related liver disease or cirrhosis (VC) group, and a non-VC group with non-cirrhotic, non-viral liver disease, including alcohol-related liver disease and metabolic dysfunction-associated steatotic liver disease.
MATERIALS AND METHODS: This multicenter study included 856 liver nodules (≤ 30 mm) from 815 patients, divided into retrospective derivation and prospective validation cohorts. CEUS LI-RADS v2017 was applied, then optimized by (1) tightening the early-washout cutoff from < 60 s to < 50 s, reclassifying LR-M lesions with washout onset from 50 to < 60 s as LR-5+; and (2) integrating serum alpha-fetoprotein (AFP) > 20 ng/mL, reclassifying LR-4/LR-M nodules as LR-5++. Standard and optimized criteria were compared using accuracy, sensitivity, specificity, and positive predictive value (PPV).
RESULTS: Across cohorts, standard LR-5 showed high specificity (87.67-93.01%) but suboptimal sensitivity (57.69-64.41%). In the validation cohort, LR-5++ significantly improved sensitivity from 57.69% to 71.43% (P < 0.05) and accuracy from 73.23% to 80.31% (P < 0.05), while specificity decreased slightly from 93.01% to 91.61% without reaching statistical significance. In subgroup analysis, LR-5++ improved sensitivity in both subgroups, most markedly in the VC group.
CONCLUSION: A study-derived modification of CEUS LI-RADS incorporating AFP >20 ng/mL (LR-5++) improved diagnostic performance for sHCC in this cohort, addressing the limitated sensitivity of the standard algorithm.
引用
- Liu, Tan, Tang, et al. Diagnostic Performance of CEUS LI-RADS for Hepatocellular Carcinoma ≤ 30 mm in Chinese Expanded High-Risk Patients: A Multicenter Analysis. Acad Radiol. 2026. PMID: 42816298. doi:10.1016/j.acra.2026.09.039
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。