In our study, we evaluated the performance of the simple biliary atresia scoring system (SBASS), to differentiate between BA and other causes of neonatal cholestatic disorders. We found that this scoring system has a high sensitivity of 96.2%, specificity of 61.9%, positive predictive value of 86.2%, negative predictive value of 86.7% and accuracy of 86.3% using a cut-off of 3 and above, in diagnosis of BA. The AUC of this scoring system was 0.901 (95% CI,0.823–0.978), indicating good discrimination of BA.
Our study showed that GGT > 200U/L has a high sensitivity (94.2%) and a high positive predictive value (87.5%). Several studies have also reported higher GGT levels in BA compared to other cholestatic disorders, with a GGT level > 300 U/L having 85% accuracy in diagnosing BA [2, 6, 9]. Additionally, our study showed that, the ratio of direct hyperbilirubinemia to total bilirubin > 70% has sensitivity of 84.6%, specificity of 23% and positive predictive value of 73%. This is in line with a study reported by Harpavat et al. that found elevated serum direct conjugated bilirubin in 34 BA patients within the first 96 hours of age[10]. In 2016, Hapavat et al. initiated a prospective study using two-step screening with direct bilirubin measurement at 60 hours of life and at 2 weeks of age; the sensitivity, specificity, and positive predictive value were 100%, 99.9%, and 18.2% respectively [11]. They also postulated that a direct conjugated bilirubin to total bilirubin ratio < 15% has a specificity of 99.3%.
Numerous ultrasonographic features have been described as useful pointers to the diagnosis of biliary atresia. In our study, we found that the triangular cord sign is an important ultrasonographic feature in diagnosing BA, which is consistent with findings from other studies. In a meta-analysis of 17 studies from 1996 to 2015, Yoon et al. reported a sensitivity of 85% and specificity of 97% for the diagnosis of BA using the triangular cord sign [12]. Additionally, in our study, a gall bladder length < 1.6cm showed sensitivity of 63.5% and specificity of 76.2% for diagnosing BA. Yoon et al reported that the most common cut off sizes for gall bladder length (< 1.5cm or < 1.9cm) in 8 studies had a sensitivity range of 50%-100% and specificity range of 68%-99% [12]. Humphrey and Stinger reported that the features with the greatest individual sensitivity and specificity respectively in diagnosing BA were triangular cord sign (73% and 100%), abnormal gall bladder wall (91% and 95%), abnormal gall bladder shape (70% and 100%), and absence of the common bile duct (93% and 92%) [13]. Nevertheless, despite the extensive studies on radiological features indicating BA, there is no single feature with sufficient sensitivity and specificity to reliably differentiate BA from other causes of neonatal cholestasis [1].
An effective bedside diagnostic score for biliary atresia (BA) in cholestatic infants is still lacking, and none of the few proposed in the past has achieved widespread application in clinical practice [7]. El Guindi and colleagues used a twelve-point diagnostic score which consists of clinical, ultrasound findings, laboratory and histopathological results [2]. However, the scoring system is not reproducible as reported by Sciveres, and taking a liver biopsy in all children with direct hyperbilirubinemia is not warranted [2, 7]. In comparison to other scoring systems, the merits of using the SBASS are that it consists of a few variables that are readily obtained in any initial workup of an infant with cholestatic jaundice [1, 2, 4, 5, 14, 15]. To our knowledge, the SBASS is the simplest scoring system available with high sensitivity and accuracy.
When an SBASS score of < 3 is obtained, other diagnostic tests can be used prior to deciding on surgery, such as PTC and liver biopsy. In our study, PTC was used to exclude biliary atresia in 6 (28%) of the cases, and this approach is similar to others such as Jensen et al who reported a sensitivity of 100% and specificity of 86% of PTC in diagnosing BA [16]. This workflow helps to relieve strain on operating theatre resources and also reduces the risks associated with surgical procedures under general anaesthesia in small infants [16, 17]. However, strict selection criteria must be used to avoid failure of this procedure, including ultrasound identification of gallbladder size.
Even in institutions where there is a low threshold for surgical exploration for cholestatic jaundice in infants, a low SBASS score (< 3) can still be helpful in guiding preoperative preparations. For example, the anaesthetic team may opt not to insert central venous and intra-arterial lines at the beginning of the procedure, anticipating a low likelihood of BA, thus avoiding the patient safety risks associated with these invasive cannulations. We have proposed an algorithm, as shown in Fig. 2, based on the findings from this study.
HIDA (hepatobiliary iminodiacetic acid), PTC (percutaneous cholangiogram), EHBT (extrahepatic biliary tree), GB (gall bladder), BA (biliary atresia), KPE (Kasai portoenterostomy).
Figure 2: Proposed algorithm for workup to differentiate between biliary atresia and other causes of neonatal cholestatic jaundice.
Several limitations of the present study should be acknowledged. Firstly, ultrasound is an operator-dependent procedure, and non-standardised ultrasound protocols across multiple institutions and referral sources can impact reports and interpretations. Hence, establishing a standardized ultrasound protocol for investigation of cholestatic jaundice in infants is required. Secondly, as BA is a progressive illness, biochemical and anatomical morphology can evolve along the course of the disease. It is possible that very early patient referrals at lower gestational age may result in falsely low scores even in the presence of BA and vice versa. Our study was not designed to account for this factor and longer-follow up for these cohorts might be required. We are also mindful of the presence of referral bias, particularly in patients referred by a paediatric gastroenterology team who are likely to have worked up and excluded a number of patients, eventually forwarding only those more likely to have BA. To mitigate this bias in the future, we can apply our scoring method in the primary health care setting to assess whether it yields comparable outcomes.