Wang Xiao-Jie, Ke Jun-Li, Xu Jian-Xia, Zhou Jia-Ping, Lu Yuan-Fei, Zhou Qiao-Mei, Shi Dan, Yu Ri-Sheng
Department of Radiology, Second Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, China.
Department of Radiology, The Second Affiliated Hospital of Zhejiang Chinese Medical University, Hangzhou, China.
Front Oncol. 2021 Apr 1;11:627482. doi: 10.3389/fonc.2021.627482. eCollection 2021.
To investigate characteristic clinical and imaging features and establish a scoring system for preoperative prediction of malignancy in the bulging duodenal papilla.
A total of 147 patients with bulging duodenal papilla (Benign enlargement n = 67; malignant enlargement n = 80) from our hospital between 2010 and 2020 were retrospectively analyzed. We investigated meaningful clinical and CT imaging features and established the score model through logistic regression and weighted. The calibration test, the ROC, AUC, and cut-off points were performed in score model. The model was also divided into three score ranges for convenient clinical evaluation.
Three clinical and CT imaging features were finally included in the score model including direct bilirubin (DBil) increase >7 umol/L (3 points), pancreatic duct (PD) dilation >5 mm (2 points), and irregular shape (2 points). The AUCs of the primary predictive model and score model were 0.896 (95% CI, 0.835-0.940) and 0.896 (95% CI, 0.835-0.940), respectively. This scoring system presented with a sensitivity of 78.8% and a specificity of 88.1% when using 2.5 points as cutoff value. Three score ranges were also proposed for convenient clinical use as follows: 0-2 points; 3-4 points; 5-7 points. The number of patients with malignant duodenal papillary enlargement increased with the increasing scores.
We proposed a convenient scoring system to preoperative predict malignancy in the bulging duodenal papilla.
探讨十二指肠乳头隆起性病变的临床及影像学特征，建立术前预测其恶性程度的评分系统。
回顾性分析2010年至2020年我院收治的147例十二指肠乳头隆起性病变患者（良性肿大67例，恶性肿大80例）。研究有意义的临床及CT影像学特征，通过逻辑回归和加权建立评分模型。对评分模型进行校准测试、ROC曲线、AUC及截断点分析。该模型还分为三个评分范围以便于临床评估。
评分模型最终纳入三项临床及CT影像学特征，包括直接胆红素（DBil）升高>7 μmol/L（3分）、胰管（PD）扩张>5 mm（2分）及形态不规则（2分）。初步预测模型和评分模型的AUC分别为0.896（95%CI，0.835 - 0.940）和0.896（95%CI，0.835 - 0.940）。以2.5分为截断值时，该评分系统的灵敏度为78.8%，特异度为88.1%。还提出了三个便于临床使用的评分范围：0 - 2分；3 - 4分；5 - 7分。十二指肠乳头恶性肿大患者数量随评分增加而增多。
我们提出了一种简便的评分系统，用于术前预测十二指肠乳头隆起性病变的恶性程度。
