Diagnostic methods utilized to distinguish BA from other causes of neonatal cholestatic jaundice include sonography, liver biopsy and intraoperative cholangiography. In the last two decades, a few sonographic observations have been described to aid in the diagnosis of BA [5–12] including the triangular cord sign (TC>4 mm), decreased gallbladder contractility, a small gallbladder length less than 15 mm described by Park et al. [6] or less than 19 mm described by Tan Kendrick et al. [11]. The latter authors previously described the “gallbladder ghost triad” (atretic gallbladder length <19 mm, thin or absent mucosal lining or indistinct wall, and an irregular gallbladder contour) [11].

Inclusion criteria for the study were preoperative sonograms at our center and definitive diagnosis of BA proven by liver biopsy, intra-operative cholangiogram and surgical pathology. Exclusion criteria included Kasai portoenterostomy or liver transplant prior to initial sonogram and lack of preoperative sonographic imaging at UCSF. Of the 20 BA infants, five were excluded because they had a Kasai procedure prior to their first US exam. The remaining 15 BA cases were included in this study. Histopathological confirmation of BA from the surgical specimens was used as the diagnostic gold standard.

Although in the past, preoperative nuclear imaging has been performed at our institution to assess patency of the biliary tree, we have abandoned this approach as both biliary atresia and severe intrahepatic cholestasis can demonstrate lack of isotope excretion into the intestine. Therefore, a HIDA scan was not performed in any of the infants included in this study.

The 20 infants with neonatal cholestasis with causes other than BA presenting during the same time frame were used as controls, called the non-BA group. A retrospective review of case and imaging records of these infants identified the underlying etiology in these cases, which included Alagille syndrome, alpha-1 antitrypsin deficiency, neonatal sclerosing cholangitis, neonatal hepatitis, cryptogenic cirrhosis, giant cell hepatitis and cholestasis of unknown etiology. Although the control infants were not age-matched by exact day of life to those in the study group, the average age in these groups was 2.15 and 2.13 months, respectively. All infants were NPO at a minimum of 4 h at the time of the sonographic evaluation.

US studies were retrieved and viewed from PACS (picture archiving and communication systems). These sonograms were performed using Sequoia US machines (Siemens Medical Solutions, Erlangen, Germany) using multifocused vector and linear transducers ranging from 4 to 8 MHz. Sonographic images of both groups (BA and non-BA infants) were reviewed by two experienced sonologists (RBG and SA) who have 31 years of combined post-residency experience in pediatric US. Images were assessed concurrently by both reviewers with consensus reached at the time of the review. Images were evaluated for the presence or absence of a gallbladder or a gallbladder-like structure including its two-dimensional measurements, for the presence of a gallbladder wall and the triangular cord sign.

On subsequent histopathological examination of surgical liver tissue of the 11 PsGB cases, no gallbladder was identified in 5, 4 had hypoplastic or atretic gallbladders and 2 had tissue that was not consistent with a true gallbladder/no epithelium (Table 2). Sonographically, all the pseudo gallbladders were smaller than or equal to the discriminatory length of 15 mm except for two (17 mm and 21 mm length). However, the two longer PsGBs were not normal-appearing on US exam, had irregular wavy contours and did not have a normal wall (Fig. 3). The 17-mm structure correlated with an atretic gallbladder on pathological evaluation, while no gallbladder was seen on pathology for the 21-mm case. Hence, a normal gallbladder was not seen by US in any infant with proven BA.

Moreover, a normal gallbladder wall was not identified in 9/11 (82%) cases demonstrating the PsGB sign. Of the two PsGB patients (18%) who did have a normal-appearing wall, the sonographically measured lengths were very small at 8 mm and 7 mm, with the first shown to be an atretic gallbladder at pathology, while no gallbladder tissue was seen in the other.

In the control group (Table 1) a normal-size gallbladder was demonstrated in 15/20 (75%) non-BA cases. Of these, there was one case where a normal echogenic wall was not clearly observed but the gallbladder measured 19×10 mm, much greater than the 15-mm cutoff limit established by Park et al. [6]. This was therefore interpreted as a normal gallbladder. The remaining five (25%) non-BA patients had a gallbladder length less than 15 mm, but a normal echogenic gallbladder wall and shape was present in all five cases. Thus, a PsGB sign was not seen in any of the non-BA controls.

BA is characterized by the obliteration of all or part of the extrahepatic bile ducts and/or the most central part of the intrahepatic bile ducts. The precise etiology of BA has not been established, although several pathogenetic mechanisms have been proposed including a developmental disruption of the biliary tree occurring antenatally (fetal form in approximately 20% of cases of BA) and an immunological response to a viral pathogen (perinatal form 70–80% of cases) [15, 16]. Current thinking suggests an association of an autoimmune response to a viral infection such as cytomegalovirus, group C rotavirus or reovirus type 3 [16–20]. Others have described the additional possible role of human papilloma virus [21] and retroviruses [22] as well as several other risk factors including advanced maternal age, increased parity, small for gestational age, low birth weight and prematurity [18].

Sonography is the initial imaging modality of choice when evaluating the neonate with cholestasis and helps to exclude overt anatomical causes of obstruction such as a choledochal cyst. Initial US research in BA focused on the presence or absence of the gallbladder and whether the gallbladder contracted following feeding [23–25]. The presence of a normal gallbladder (without other abnormal findings in the biliary tree) is usually thought to essentially exclude the diagnosis of BA. In this study we describe a potential important imaging pitfall of a pseudo gallbladder sign, which if misinterpreted as a normal gallbladder could delay important diagnostic testing and the correct diagnosis.

Since BA can occur at any level of the biliary tree, the presence of a gallbladder, cystic duct and a distal bile duct does not completely exclude a more proximal ductal atresia [9]. Clinical and histopathological suspicion of BA can be confirmed by an intraoperative cholangiogram [1], which shows failure of contrast agent reflux into the proximal/intrahepatic ducts.

Attention has also been directed to the identification and dilatation of bile ducts, echotexture of the liver and the TC sign. Choi and co-workers [5] first described the TC sign in 1996 in BA patients as a tubular or triangular echogenic tissue (fibrous ductal remnant) adjacent to the portal vein and correlated this sonographic finding with a biopsy-proved diagnosis of BA (Fig. 4). However, they recommended close follow-up of all patients given the possibility of false-positive and false-negative findings. Lee et al. [7] found that the TC sign has a 94% PPV and NPV, 98% specificity and 80% sensitivity. Compared to 73% sensitivity and 100% specificity for the TC sign, demonstrated by Humphrey and Stringer [26], our study shows 60% sensitivity and 95% specificity (Table 4). Li et al. [27] suggest that the TC sign should not be viewed as the only imaging criterion for diagnosing BA. Tan Kendrick et al. [11] have reported high sensitivity and specificity for the diagnosis of BA using the gallbladder ghost triad.

In our study, the reviewers were not blinded to the diagnosis at the time of the review and although there is an inherent bias in this method, the case and control groups were so clearly segregated based on the findings that we do not believe this would be significant to the general results. Of the 15 infants with pathologically proven BA, we found that no gallbladder was seen in 27% of our BA cases (n=4), but that 73% (n=11) of our BA patients had a gallbladder-like structure—the pseudo gallbladder sign. This structure was ≤15 mm in length in the majority (9/11) of cases and is indicated by the presence of a small ovoid to tubular fluid-filled structure in the expected region of the gallbladder (Fig. 2). It is smaller than a normal gallbladder and can have irregular margins but without a well-defined or normal-appearing wall. The term “pseudo gallbladder” emphasizes the potential to misconstrue this finding as a normal gallbladder, which can distract from the diagnosis of biliary atresia.

The presence or absence of a gallbladder wall helps distinguish a normal gallbladder from a PsGB[11]. A normal wall is measurable with a central echogenic mucosal interface and a thinner peripheral fibromuscular layer (Fig. 1), whereas the PsGB does not display a normal wall but rather has only a thin nonmeasurable echogenic rim around the fluid collection (Fig. 2). A normal gallbladder wall was absent in 82% of our BA cases with PsGB but was clearly identified in 95% of the non-BA control group. Absence of the gallbladder wall had an 87% sensitivity and 95% specificity with a 93% PPV for BA (Table 4).

Among the 11/15 BA cases demonstrating the PsGB sign, surgical pathology showed no gallbladder in five cases and a hypoplastic gallbladder in three, and the remaining three were labeled as “not a true gallbladder, atretic gallbladder, and no epithelium seen,” respectively. Thus, there was no single pathological correlate to the PsGB in our study. Histologically, hyaline cartilage at the portal plate [28, 29], in the bile ducts [30] and in an atretic gallbladder [31], and squamous epithelium within biliary ductules [29] has been reported. However, the wide spectrum of previously described histopathological findings in BA is mostly without sonographic correlation. The complete absence of smooth muscle bundles in the wall of an atretic gallbladder, noted by Altamirano [31], might correspond to our observation of the absence of a sonographically normal gallbladder wall in most of our PsGB cases. The varied surgical pathology of the PsGB in our study and in other reported BA cases suggests the need for a larger clinical-radiological-pathological study in the future to provide a more definitive pathological correlate to the PsGB.