A nomogram model based on preoperative grey-scale US features and routine serum biomarkers to predict the outcome of infants with biliary atresia after Kasai portoenterostomy

Abstract

Objective: To establish a nomogram to predict the outcome of biliary atresia (BA) infants 3-months post- Kasai portoenterostomy (KPE).

Methods: BA Infants who underwent KPE from two hospitals were included in the training (n = 161) and validation cohorts (n = 64). A logistic regression equation (Equation A) for predicting the serum total bilirubin (TBIL) level 3-month post-KPE was established in the training cohort. Then, a nomogram was developed based on Equation A in the training cohort and validated in the validation cohort. Moreover, a new equation (Equation B) was generated based on the nomogram and the size of the enlarged hilar lymph nodes (LNs) in the validation cohort. The predictive performance of the nomogram was evaluated by the receiver operating characteristic (ROC) curve and by calculating the area under the ROC curve (AUC), sensitivity, specificity, and positive (PPV) and negative (NPV) prediction values.

Results: A nomogram based on gallbladder morphology and serum levels of TBIL and total protein (TP) was established with AUC (95%CI) of 0.673 (0.595, 0.745) and 0.647 (0.518, 0.763), sensitivity (95%CI) of 71.4% (62.1%,79.6%) and 81.8% (59.7%,94.8%), specificity (95%CI) of 63.3% (48.3%,76.6%) and 47.6% (32.0%,63.6%), PPV (95%CI) of 81.6% (72.5%,88.9%) and 45.0% (29.3%,61.5%), and NPV (95%CI) 49.2% (36.4%,62.1%) and 83.3% (62.6%,95.3%), respectively, in the training and validation cohorts. Furthermore, in the validation cohort, the AUC (95%CI) of Equation B was 0.798 (95%CI: 0.679, 0.888), which was significantly higher than that of the nomogram (P = 0.042).

Conclusion: A nomogram based on the pre-KPE gallbladder morphology, TBIL, and TP to predict the outcome of BA 3-months post-KPE is established. Moreover, the addition of the size of the enlarged hilar LNs into the nomogram further improves its predictive value.

Keywords: Kasai portoenterostomy; biliary taresia; hilar lymph nodes; nomogram; prognostic prediction.

Figure 1

(A) the gallbladder is not identifiable; (B) the length is less than 1.5 cm; (C) the gallbladder is with irregular outline; (D) the wall of gallbladder is not identifiable, and the gallbladder appears as a cyst in the gallbladder fossa; (E) the gallbladder has an irregular wall thickness; (F) the hyperechogenic mucosal lining of gallbladder is not smooth; (G) the triangular cord (TC) thickness is the thickness of the echogenic anterior wall of the anterior branch of the right portal vein; and (H) the hilar lymph nodes (LNs) are detected around porta hepatis. PV, portal vein; HA, hepatic artery. The arrows in (A–F) indicate the location of the gallbladder; The arrows in (G) indicate the location of TC thickness; The arrows in (H) indicate the location of hilar LNs.

Figure 2

(A) the nomogram in the training cohort; (B) the receiver operating characteristic (ROC) curve of the nomogram in the training cohort; and (C) the comparison of ROC curve between the nomogram and equation B in the validation cohort. TP, total protein; TBIL, total bilirubin.