R2*值预估肝细胞癌病理分级的可行性

目的 探讨增强T2^*加权血管成像（ESWAN）序列的R2^*值预估肝细胞癌（HCC）病理分级的可行性。方法 回顾性分析69例（共71个病灶）接受1.5T MR（含ESWAN序列）检查并经手术病理证实为HCC患者的影像资料，按照病理学分级分为3组，其中低分化组25个，中分化组25个，高分化组21个病灶。由2名观察者分别测量各组HCC实质区的R2^*值，检验2名观察者测量数据的一致性，并分析R2^*值与HCC病理级别的相关性，对不同级别HCC的R2^*值进行组间两两比较；采用ROC曲线评估R2^*值预估低级别HCC的效能并确定界值。结果 2名观察者测量各组数据的一致性很好（ICC>0.75）。R2^*值与HCC病理分级呈弱相关（r_s=0.493，P<0.001），低、中和高分化组HCC的R2^*值分别为（30.57±9.95）Hz、（21.07±5.11）Hz和（19.08±4.83）Hz，低、中分化组间及低、高分化组间R2^*值差异有统计学意义（P均<0.001），中、高分化组间差异无统计学意义（P=0.350）。应用R2^*值预估低分化HCC的曲线下面积（AUC）为0.816，以R2^*值=21.96 Hz为界值，敏感度为80.0%，特异度为73.9%。结论 R2^*值可作为MR非强化方式预估低分化HCC的定量指标。

Feasibility of R2* value in predicting pathological grade of hepatocellular carcinoma

Objective To investigate the feasibility that R2^* value of enhanced T2 star-weighted angiography (ESWAN) sequence in evaluating the pathological grade of hepatocellular carcinoma (HCC). Methods Totally 69 patient (71 lesions) with HCC confirmed by surgery and pathology underwent 1.5T MR examination (included ESWAN sequence), and the imaging data were analyzed retrospectively. The tumors were pathological classified into three groups as follows: 25 poorly, 25 moderately and 21 well differentiated HCC. The R2^* values of HCC tumor parenchyma were measured by two observers, and the consistency of the data between the two observers were tested. The correlation between R2^* value and pathological grade of HCC was analyzed, and the R2^* values of different pathological grades were compared between each two groups. The ROC curve was used to evaluate the effect of R2^* value on the classification of different levels, and find out the cut-off value. Results The data consistency of two observers was good (ICC>0.75). The R2^* value was weak correlated with the pathological grade of HCC (r_s=0.493, P<0.001). The R2^* values of poorly, moderately and well differentiated HCC were (30.57±9.95)Hz, (21.07±5.11)Hz and (19.08±4.83)Hz, respectively. The R2^* values between poorly and moderately, poorly and well differentiated HCC had statistically significance (both P<0.001). The R2^* values between moderately and well differentiated HCC had no statistically significant difference (P=0.350). The area under curve (AUC) of R2^* value for diagnosing poorly differentiated HCC was 0.816, the cut-off value was 21.96 Hz, the sensitivity was 80.0%, the specificity was 73.9%. Conclusion R2^* value can be used as a MR non-enhancement quantitative index for determining the poorly differentiated HCC.