Infants with BA invariably present with cholestatic jaundice [16]. However, the accurate diagnosis of BA can be challenging due to many overlapping features between BA and other causes of infantile cholestasis. In the last decade, many investigators have proposed novel diagnostic scoring systems [17–21] to risk-stratify infants with cholestasis. These scores used various combinations of clinical, laboratory, imaging and histology modalities. A few of them have demonstrated exceptional accuracy [17, 19]. However, their results are difficult to replicate as the scores required invasive or costly investigations such as liver biopsy and hepatobiliary scintigraphy. In addition, these modalities require the input of expert pathologists or radiologists.
To develop a practical diagnostic score that is highly sensitive in detecting BA, we conducted a review of infants with cholestasis, who were referred to the two exclusive pediatric surgical units in Singapore. We named this scoring system “Simple Biliary Atresia Score” (SBAS), as it is practical and easy to use.
The strength of our score lies primarily in its simplicity and reproducibility. This was the result of deliberate selection of relevant variables based on cumulative clinical experience. The five variables in our score rely only on objective serum liver function test and hepatobiliary ultrasonography. These two tests are non-invasive, readily available and they are often the first few investigations ordered by primary physicians for cholestasis.
In the same vein, our score does not require costly or advanced investigations such as hepatobiliary scintigraphy, magnetic resonance cholangiopancreatography (MRCP) or liver biopsy. These studies are resource intensive. In our institution, infants undergo MRCP under general anesthesia and the interpretation of MRCP images can be difficult, even for experienced radiologists. While liver biopsy has been shown to be highly accurate [7], the histopathologic features for many differential diagnoses can be nonspecific and unhelpful at the early stages of each disease [21]. We believe these challenges in using MRCP or liver biopsy are also encountered by many other centers in our region. Therefore, while the inclusion of these modalities into the scoring system may potentially enhance its accuracy, the score’s applicability will be restricted.
Our score also did not include subjective variables such as stool color assessment. While acholic stool has been shown to be associated with BA [22], this self-reporting variable is highly subjective and susceptible to observer bias [23].
Our study results are in concordance with previous studies that have confirmed the effectiveness of ultrasonography in distinguishing BA, namely using gallbladder length, pre-PV echogenicity and CBD diameter. Other specific sonographic findings, such as the triangular cord sign and subcapsular hepatic flow, have been proposed in previous scoring systems [10, 17]. However, these findings are highly operator-dependent and may not be reliable in centers where BA is not commonly encountered. Furthermore, as BA is a progressive obliterative cholangiopathy, sonography findings can change as the disease progresses [24]. For these reasons, we echoed previous investigators that a normal ultrasonogram alone cannot reliably rule out BA [12, 13] and weightage should also be assigned to biochemical markers.
With regards to laboratory markers, several investigators included elevated GGT in their diagnostic scores with various cut-off values [25–27]. Our results demonstrated that GGT is distinctly elevated in BA. Liao et al concluded that direct bilirubin level was better than D/T bilirubin ratio at detecting BA [28]. In contrast, our study has shown that there is no clear association between direct or total bilirubin levels and BA. In fact, average bilirubin levels were shown to be higher in our non-BA group. There was however, an association between higher D/T ratio and BA. Although the D/T ratio did not reach statistical significance, it appeared to have clinical significance and was therefore included in our score.
Notably, for patients with repeated ultrasonograms or liver function tests before surgery, we observed in some but not all, an increasing trend of their score closer to surgery. This is expected for a progressive disease. While the score may increase with time, it has been designed to prioritize sensitivity even in the initial investigative period with no postulated age limit. Bearing in mind that the score can be dynamic, the optimal timing to apply the score will be "as soon as all parameters are available" and we recommend using the most recent test results to calculate the score.
Informally, over the past five years, we have been applying this score to infants presenting with cholestasis and it has proven to be a valuable clinical aid. This study brings objectivity to our score and has tangibly evaluated its performance through external validation. The reproducibility and diagnostic efficiency of our score were further demonstrated by another independent study. Mahat et al modified this score for use in Malaysia, achieving a sensitivity of 96% (AUROC of 0.901)[1].
Nonetheless, there are several limitations of this study that need to be addressed. Firstly, this was a retrospective review of infants referred to a surgical unit. This inherently excluded a large number of cholestatic infants who had already been deemed by gastroenterologists as low-risk for BA. This was reflected in the high percentage of BA (67%) in our study, which may be prone to selection bias. Secondly, while our score was designed to prioritize high sensitivity and minimize delayed diagnosis of BA, its low specificity also implies that a considerable number of non-BA infants may undergo unnecessary procedures. Lastly, despite the score’s high sensitivity, one infant with BA was misdiagnosed in the validation cohort, as elaborated previously in the results section. To overcome the limitations, larger prospective studies are warranted to refine the scoring system and enhance its accuracy.