Wang Dong-Qing, Zeng Meng-Su, Jin Da-Yong, Lou Wen-Hui, Ji Yuan, Rao Sheng-Xiang, Shi Xun, Chen Cai-Zhong, Li Ren-Chen
Department of Diagnostic Radiology, Zhongshan Hospital, Fudan University, Shanghai 200032, China.
Zhonghua Zhong Liu Za Zhi. 2007 Nov;29(11):846-9.
To investigate the role of MRI in evaluating the peripancreatic vessel invasion and resectability of pancreatic carcinoma based on the comparison of MRI image with surgical exploration, and try to establish the criteria for assessment of the sensitivity, specificity and accuracy of resectability.
Forty-one pancreatic carcinoma patients confirmed by pathology received preoperative plain and contrast enhanced MRI scan, and 37 of them had additional coronal MRA scan. Peripancreatic vessel invasion was preoperatively assessed based on MRI features, and the vessel invasion degree from the uninvolved to the severely involved was divided into 6 grades represented by 1, 2a, 2b, 3a, 3b and 4, respectively. Compared with the findings during the surgery, the sensitivity and specificity of each vessel invasion grade were studied and the receiver operator characteristic curve (ROC) was drawn. The sensitivity, specificity, positive predictive value, negative predictive value and accuracy of resectability evaluation based on 2 and 2a degree were calculated respectively. The resectability of involved arteries and veins of grade 2 were also analyzed.
Of the 41 patients, 22 had resectable tumor, with 20 curative resection and 2 palliative. Compared with the findings during surgery, seven vessels including three arteries and four veins were not correctly interpreted by MRI. If grade 1,2a,2b,3a and 3b was used as the resectable standard,respectively, the sensitivity to predict the unresectbility was 78.3%, 84.8%, 67.4%, 56.5% and 47.8%, respectively. Receiver operator characteristic curve demonstrated that grade 2a was the optimal critical point. The sensitivity, specificity, positive predictive value, negative predictive value and accuracy of grade 2a in predicting the unresectbility were 84.8%, 98.5%, 92.9%, 96.6% and 95.9%.
Our data showed that grade 2a (tumor involvement < 2 cm long and < 1/2 circumference of the vessel) may be more sensitive and accurate in predicting the resectability, which may be considered as the line of demarcation between the respectable and unresectable cases in clinical practice.
通过将MRI图像与手术探查结果进行比较，探讨MRI在评估胰腺癌胰周血管侵犯及可切除性方面的作用，并尝试建立评估可切除性的敏感性、特异性及准确性的标准。
41例经病理证实的胰腺癌患者术前行MRI平扫及增强扫描，其中37例加做了冠状位MRA扫描。根据MRI表现术前评估胰周血管侵犯情况，将血管侵犯程度从无侵犯到严重侵犯分为6级，分别用1、2a、2b、3a、3b和4表示。与手术中的发现进行比较，研究各血管侵犯分级的敏感性和特异性，并绘制受试者操作特征曲线（ROC）。分别计算基于2级和2a级评估可切除性的敏感性、特异性、阳性预测值、阴性预测值及准确性。还分析了2级受累动静脉的可切除性。
41例患者中，22例肿瘤可切除，其中20例行根治性切除，2例行姑息性切除。与手术中的发现相比，MRI对包括3条动脉和4条静脉在内的7条血管的判断有误。若分别将1、2a、2b、3a和3b级作为可切除标准，预测不可切除性的敏感性分别为78.3%、84.8%、67.4%、56.5%和47.8%。受试者操作特征曲线显示2a级为最佳临界点。2a级预测不可切除性的敏感性、特异性、阳性预测值、阴性预测值及准确性分别为84.8%、98.5%、92.9%、96.6%和95.9%。
我们的数据表明，2a级（肿瘤累及血管长度<2 cm且<血管周长的1/2）在预测可切除性方面可能更敏感、准确，在临床实践中可作为可切除与不可切除病例的分界线。
