Three-color risk stratification for improving the diagnostic accuracy for biliary atresia

Objectives: It is challenging to early differentiate biliary atresia from other causes of cholestasis. We aimed to develop an algorithm with risk stratification to distinguish biliary atresia from infantile cholestasis.

Methods: In this study, we enrolled infants with cholestasis into 2 subgroups from January 2010 to April 2019. A prospective cohort (subgroup 2) of 187 patients (107 with biliary atresia and 80 without biliary atresia) underwent acoustic radiation force impulse elastography. Stepwise regression was used to identify significant predictors of biliary atresia. A sequential algorithm with risk stratification was constructed.

Results: Among 187 patients, shear wave speed > 1.35 m/s and presence of the triangular cord sign were considered high risk for biliary atresia (red), in which 73 of 78 patients (accuracy of 93.6%) with biliary atresia were identified. Afterwards, γ-GT, abnormal gallbladder, and clay stool were introduced into the algorithm and 55 intermediate-risk infants were identified (yellow) with a diagnostic accuracy of 60% for biliary atresia. Of the remaining 54 infants who were classified as low-risk patients (green), the accuracy for excluding biliary atresia was 98.1%. By applying a three-color risk stratification tool, 70.6% patients were identified as either high risk or low risk for biliary atresia (area under the curve, 0.983; sensitivity, 98.7%; specificity, 91.4%). We also estimated the risk of biliary atresia in different color groups, which was 94.7% (95%CI, 94.3-95.5%) in the red group and 7.2% (95%CI, 6.6-8.3%) in the green group.

Conclusions: Our simple noninvasive approach was able to identify biliary atresia with high accuracy.

Key points: • Five predictors, namely shear wave speed, triangle cord sign, γ-glutamyl transferase, abnormal gallbladder, and clay stool, were selected to identify biliary atresia in cholestasis. • Shear wave speed > 1.35 m/s and presence of the triangle cord sign were considered high-risk patients with a diagnostic accuracy of 93.6% for biliary atresia. • Risk for biliary atresia was high (red), intermediate (yellow), or low (green). In the red and green group, we achieved an extremely high diagnostic performance (area under the curve, 0.983; sensitivity, 98.7%; specificity, 91.4%).