胰腺神经内分泌肿瘤计算机断层扫描增强特征对术前病理分级的预测价值

Predictive value of enhanced computed tomography in preoperative pathological grading of pancreatic neuroendocrine neoplasm

目的探讨胰腺神经内分泌肿瘤(panNEN)计算机断层扫描(CT)增强特征预测术前病理分级的价值。方法回顾性分析2009年2月至2020年8月复旦大学附属中山医院经手术和病理确诊的136例panNEN患者的CT资料。根据2019年第5版世界卫生组织消化系统肿瘤分类标准将panNEN分为G1、G2、G3级神经内分泌瘤(NET)和神经内分泌癌(NEC),比较不同病理分级病灶实性部分CT值和相关参数的差异。采用受试者操作特征曲线下面积(AUC)评价CT值和相关参数对不同病理分级panNEN的鉴别诊断价值。统计学分析采用非参数Kruskal-Wallis检验和卡方检验。结果在136例panNEN患者中,共检出142个肿瘤病灶,其中G1、G2、G3级NET和NEC病灶数分别为58、73、7、4个。随着病理分级的升高,panNEN病灶实性部分的CT强化程度呈逐渐下降趋势。G2级NET患者动脉期和静脉期的CT值、CT值差值、相对密度指数均低于G1级NET患者[动脉期:105 HU(46~251 HU)比126 HU(57~195 HU)、62 HU(6~212 HU)比81 HU(23~166 HU)、1.1(0.4~2.5)比1.4(0.7~2.0)。静脉期:90 HU(49~159 HU)比107 HU(63~162 HU)、49 HU(9~134 HU)比62 HU(24~128 HU)、1.1(0.5~2.0)比1.4(0.9~2.0)],差异均有统计学意义(H_(动脉期)=2.880、2.607、3.482,H_(静脉期)=3.426、3.323、3.891,P均<0.05)。G3级NET患者动脉期CT值差值、病灶增强百分比、增强指数,以及静脉期CT值差值、增强指数均低于G2级NET患者[31 HU(27~52 HU)比62 HU(6~212 HU)、67%(59%~153%)比151%(15%~705%)、0.6 HU(0.5~0.9 HU)比1.3 HU(0.1~5.8 HU)、40 HU(36~52 HU)比49 HU(9~134 HU)、0.9(0.7~1.6)比1.5(0.2~5.4)],差异均有统计学意义(H=2.634、2.801、3.267、2.264、2.882,P均<0.05)。G1级和G2级NET动脉期和静脉期的相对密度指数具有最大AUC值,分别为0.679[95%可信区间(CI)0.587~0.771,P<0.01]、0.701(95%CI 0.609~0.793,P<0.01);G2级和G3级NET动脉期和静脉期的增强指数具有最大AUC值,分别为0.875(95%CI 0.793~0.956,P=0.001)、0.828(95%CI 0.700~0.956,P=0.004)。结论panNEN的CT增强扫描参数可用于预测术前病理分级。

Objective To investigate the value of imaging features of contrast-enhanced computed tomography(CT)of pancreatic neuroendocrine neoplasm(panNEN)in predicting preoperative pathological grade.Methods From February 2009 to August 2020,at Zhongshan Hospital affiliated to Fudan University,the CT data of 136 patients with panNEN diagnosed by surgery and pathology were retrospectively analyzed.According to World Health Organization classification of digestive tumors(5th edition,2019),panNEN was classified into G1,G2,G3 neuroendocrine tumor(NET)and neuroendocrine carcinoma(NEC).The differences in CT values and correlative parameters of solid part of lesions with different pathological grades were compared.The receiver operator characteristic curve was drawn and the area under the curve(AUC)was calculated to evaluate the CT value and related parameters in the differential diagnosis of panNEN of different pathological grades.The Kruskal-Wallis test and chi-square test were used for statistical analysis.Results A total of 142 tumor lesions were detected in 136 panNEN patients,of which the numbers of G1,G2,G3 NET and NEC were 58,73,7 and 4,respectively.Along with the increase of pathological grade,the degree of CT enhancement in the solid part of panNEN lesions gradually decreased.The CT value,CT value difference and relative density of arterial phase and venous phase of patients with G2 NET were all lower than those of patients with G1 NET(Arterial phase:105 HU,46 to 251 HU vs.126 HU,57 to 195 HU;62 HU,6 to 212 HU vs.81 HU,23 to 166 HU;1.1,0.4 to 2.5 vs.1.4,0.7 to 2.0.Venous phase:90 HU,49 to 159 HU vs.107 HU,63 to 162 HU;49 HU,9 to 134 HU vs.62 HU,24 to 128 HU;1.1,0.5 to 2.0 vs.1.4,0.9 to 2.0),and the differences were statistically significant(H_(arterial phase)=2.880,2.607 and 3.482,H_(venous phase)=3.426,3.323 and 3.891,all P<0.05).The CT value difference,the percentage of lesion enhancement,the index of lesion enhancement in arterial phase,and the CT value difference and the index of lesion enhancement in venous phase of patients with G3 NET were all lower than those of patients with G2 NET(31 HU,27 to 52 HU vs.62 HU,6 to 212 HU;67%,59%to 153%vs.151%,15%to 705%;0.6 HU,0.5 to 0.9 HU vs.1.3 HU,0.1 to 5.8 HU;40 HU,36 to 52 HU vs.49 HU,9 to 134 HU;0.9,0.7 to 1.6 vs.1.5,0.2 to 5.4),and the differences were statistically significant(H=2.634,2.801,3.267,2.264 and 2.882,all P<0.05).The relative density index of arterial phase and venous phase to identify G1 and G2 NET received maximum AUC value of 0.679(95%confidence interval(CI)0.587 to 0.771,P<0.01)and 0.701(95%CI 0.609 to 0.793,P<0.01),respectively.The index of lesion enhancement in arterial phase and venous phase to identify G2 and G3 NET received maximum AUC value of 0.875(95%CI 0.793 to 0.956,P=0.001)and 0.828(95%CI 0.700 to 0.956,P=0.004),respectively.Conclusion The enhanced CT parameters of panNEN can be used to predict the pathological grade before operation.