LI-RADS辅助征象替代阈值增长对HCC的诊断效能

目的 探讨2018版肝脏影像报告和数据系统（LI-RADS v2018）中肝细胞癌（HCC）的特异性辅助征象替代主要征象中的阈值增长后,其中的LR-5对HCC的诊断效能。方法 回顾性收集未经治疗并行钆塞酸二钠增强MRI（Gd-EOB-MRI）且经病理证实的HCC和其他肝内恶性肿瘤（OM）病人262例共262个病灶。由2名放射科医师依据LI-RADS v2018对病灶进行分析,采用t检验比较HCC和OM病灶中伴和不伴阈值增长的病灶大小;分析HCC和OM病灶间主要和辅助影像征象的差异,并确定HCC特异性辅助征象。分别计算标准LI-RADS v2018以及HCC特异性辅助征象替代阈值增长后LR-5对HCC的诊断效能,并采用McNemar检验比较其差异。结果 262个病灶中,HCC 187个（71.4%）,OM 75个（28.6%）。共47个HCC和29个OM病灶用于阈值增长评价,其中22个HCC和14个OM病灶出现阈值增长。HCC和OM病灶中,伴有阈值增长的病灶直径均小于不伴阈值增长者（均P<0.05）。主要征象中,OM较HCC更常见阈值增长;辅助征象中,HCC较OM更常见结中结和病灶内含脂（均P<0.05）。以结中结和病灶内含脂作为特异性辅助征象替代阈值增长,诊断HCC的敏感度、特异度、准确度分别为75.4%、88.6%、81.0%和74.9%、89.3%、81.0%,诊断效能与标准LI-RADS v2018（74.3%、88.6%、80.4%）的差异没有统计学意义（均P>0.05）。结论 当阈值增长被HCC特异性辅助征象替代后,并未影响LI-RADS v2018诊断HCC的效能,即阈值增长可以被结中结和病灶内含脂替代。

Diagnostic performance of LI-RADS for hepatocellular carcinoma after replacing the threshold growth by the auxiliary features

Objective To analyze the diagnostic performance of LR-5 of Liver Imaging Reporting and Data System version 2018 (LI-RADS v2018) for hepatocellular carcinoma (HCC) when threshold growth as a major feature of HCC was replaced by more HCC-specific auxiliary features. Methods The patients with untreated HCC and non-HCC other hepatic malignancy (OM) confirmed by pathology and underwent gadoxetate disodium contrast-enhanced magnetic resonance imaging (Gd-EOB MRI) were retrospectively collected. A total of 262 lesions in 262 patients were included in this study. The lesions were analyzed by two radiologists with LI-RADS v2018. In HCC and OM, the size of lesions with or without threshold growth were compared by t test. The differences in major and auxiliary features between HCC and OM were analyzed, and HCC-specific auxiliary features were determined. The diagnostic performance of LR-5 for HCC of LI-RADS v2018 and HCC-specific auxiliary features substitute for threshold growth were calculated separately, and the differences were compared using McNemar test. Results Among the 262 lesions, 187 (71.4%) were HCC and 75 (28.6%) were OM. A total of 47 HCC and 29 OM were used for threshold growth evaluation, of which 22 HCC; and 14 OM showed threshold growth. For both HCC and OM, the lesions with threshold growth were smaller than those without threshold growth (all P<0.05). In major features, threshold growth was more common in the OM than in the HCC, while in auxiliary features, nodule-in-nodule and fat-in-nodule were more common in the HCC than in the OM (all P<0.05). Using either specific auxiliary features nodule-in-nodule or fat-in-nodule instead of threshold growth, the sensitivity, specificity and accuracy of diagnosing HCC were 75.4%, 88.6% and 81.0% and 74.9%, 89.3% and 81.0%, respectively, and did not significantly differ from the diagnostic performance of LI-RADS v2018 (74.3%, 88.6% and 80.4%) (all P>0.05). Conclusion When the threshold growth is replaced by HCC-specific auxiliary features, the performance of LI-RADS v2018 in diagnosing HCC will not affected, and the threshold growth could be replaced by nodal-in-nodal and fat-in-nodule.