CT特征鉴别非富血供胰腺神经内分泌肿瘤与胰腺导管腺癌的初步研究

CT features differentiate nonhypervascular pancreatic neuroendocrine neoplasm and pancreatic ductal adenocarcinoma: preliminary study

目的探索能用于鉴别非富血供胰腺神经内分泌肿瘤(pNEN)与胰腺导管腺癌(PDAC)的CT特征。方法回顾性纳入四川大学华西医院2010年5月至2017年5月期间经过病理确诊为非富血供pNEN和PDAC的患者,分析二者的CT特征,通过多元逻辑回归筛选CT特征并计算其诊断效能。结果 40例非富血供pNEN(无功能性33例,功能性7例)及80例PDAC患者纳入本研究。非富血供pNEN与PDAC间差异有统计学意义的特征包括肿瘤位置、肿瘤长径、肿瘤边界、肿瘤实质均匀、肿瘤内钙化及肿瘤内血管影(P<0.05)。将各项有统计学意义的特征经过多元逻辑回归分析后提示肿瘤边界[OR为14.63,95%CI为(2.82,75.99)]、肿瘤内钙化[OR为4.00,95%CI为(1.03,15.59)]及肿瘤位置[OR为3.09,95%CI为(1.19,7.99)]能够独立鉴别出非富血供pNEN。再根据有统计学意义的特征最终得出非富血供pNEN与PDAC鉴别诊断的多元逻辑回归模型,其诊断敏感度为70.00%,95%CI为(53.5, 83.4);特异度为83.54%,95%CI为(73.5,90.9);受试者工作特征曲线下面积为0.824,95%CI为(0.743,0.887)。结论 CT特征多元逻辑回归模型可应用于非富血供pNEN与PDAC鉴别诊断,其中肿瘤边界及肿瘤内钙化特征在非富血供pNEN与PDAC鉴别诊断中有一定的应用价值。

Objective To explore CT features that can be used to identify nonhypervascular pancreatic neuroendocrine neoplasm （pNEN） and pancreatic ductal adenocarcinoma （PDAC）. Methods The patients with pathologically confirmed the pNEN and PDAC were retrospectively included from May 2010 to May 2017. The CT features were analyzed. The CT features were extracted by the multivariate logistic regression, and their diagnostic performances were calculated. Results Forty patients with the nonhypervascular pNEN （33 unfunctional, 7 functional） and 80 patients with the PDAC were included in this study. The features of significant differences between the nonhypervascular pNEN and the PDAC included： the location, long diameter, margin, uniform lesions, calcification, and vascular shadows of the lesion （P〈0.05）. The margin[OR=14.63, 95% CI （2.82, 75.99）], calcification [ OR=4.00, 95% CI （1.03, 15.59）], and location [ OR=3.09, 95% CI（1.19, 7.99）] of the lesion could independently identify the nonhypervascular pNEN. The multivariate logistic regression model of the differential diagnosis of the nonhypervascular pNEN and PDAC was obtained through the CT features of significant differences. The diagnostic sensitivity was 70.00%, 95% CI （53.5, 83.4）; specificity was 83.54%, 95% CI （73.5, 90.9）; and area under the receiver operating curve was 0.824, 95% CI （0.743, 0.887）. Conclusions Multivariate logistic regression model of CT features is helpful for differential diagnosis of nonhypervascular pNEN and PDAC. Features of margin and calcification of lesion are more valuable in differential diagnosis of nonhypervascular pNEN and PDAC.