Department of Radiology, Second Affiliated Hospital, Zhejiang University School of Medicine, Jiefang Rd 88#, Hangzhou, 310009 China.
Department of Radiology, Institute of Occupational Diseases, Zhejiang Academy of Medical Sciences, Hangzhou, China.
AJR Am J Roentgenol. 2020 Aug;215(2):390-397. doi: 10.2214/AJR.19.21921. Epub 2020 May 20.
The purpose of this study is to establish a diagnostic model for differentiating grade 3 (G3) pancreatic neuroendocrine tumors (PNETs) from pancreatic ductal adenocarcinomas (PDACs) and to analyze survival outcomes. Twenty patients with G3 PNETs and 58 patients with PDACs confirmed by surgery or biopsy were retrospectively included. Demographic and radiologic information was collected. Univariate analyses and binary logistic regression analyses were performed to identify independent factors and establish a diagnostic model. An ROC curve was created to determine diagnostic ability. Kaplan-Meier survival analysis was performed. Patients with G3 PNETs were more likely to present with normal carbohydrate antigen (CA) 19-9 levels, normal pancreatic ducts, and round tumors with well-defined margins and higher portal enhancement ratios than were patients with PDAC ( < 0.05). After multivariate analysis, a normal CA 19-9 level (odds ratio, 0.0125; 95% CI, 0.0008-0.2036), round tumor shape (odds ratio, 0.0143; 95% CI, 0.0004-0.5461), and pancreatic duct dilation of 4 mm or less (odds ratio, 17.9804; 95% CI, 1.0098-320.1711) were independent predictors of G3 PNETs. The AUC of the ROC curve was 0.916, and sensitivity and specificity were 90.0% and 81.0%, respectively. Furthermore, patients with G3 PNETs had better overall survival than patients with PDACs. Among patients in the G3 PNET subgroup, patients with liver or lymph node metastases had worse overall survival than patients without metastases. A diagnostic model was established to differentiate G3 PNETs from PDACs. A normal CA 19-9 level, round tumor shape, and pancreatic duct dilation of 4 mm or less were factors that were strongly predictive of G3 PNET.
本研究旨在建立一个用于区分 3 级（G3）胰腺神经内分泌肿瘤（PNETs）和胰腺导管腺癌（PDACs）的诊断模型，并分析生存结果。回顾性纳入 20 例经手术或活检证实的 G3 PNETs 患者和 58 例 PDACs 患者。收集了人口统计学和影像学信息。进行单因素分析和二元逻辑回归分析，以确定独立因素并建立诊断模型。绘制 ROC 曲线以确定诊断能力。进行 Kaplan-Meier 生存分析。与 PDAC 患者相比，G3 PNET 患者更可能表现为正常的癌抗原 19-9（CA19-9）水平、正常的胰管、圆形肿瘤且边界清晰、门静脉增强率较高（<0.05）。多因素分析后，正常 CA19-9 水平（比值比，0.0125；95%置信区间，0.0008-0.2036）、圆形肿瘤形状（比值比，0.0143；95%置信区间，0.0004-0.5461）和胰管扩张 4mm 或以下（比值比，17.9804；95%置信区间，1.0098-320.1711）是 G3 PNET 的独立预测因子。ROC 曲线的 AUC 为 0.916，灵敏度和特异性分别为 90.0%和 81.0%。此外，G3 PNET 患者的总生存率优于 PDAC 患者。在 G3 PNET 亚组中，有肝或淋巴结转移的患者总生存率低于无转移的患者。建立了一个诊断模型来区分 G3 PNET 和 PDAC。正常的 CA19-9 水平、圆形肿瘤形状和胰管扩张 4mm 或以下是强烈预测 G3 PNET 的因素。
