Newborn Screening for Biliary Atresia: A Review of Current Methods

Abstract

Purpose of Review

Biliary atresia is a serious neonatal liver disease due to obstructed bile ducts that has better outcomes when detected and treated in the first 30–45 days of life. This review examines different methods to screen newborns for biliary atresia as well as discusses observations from ongoing screening programs implemented in parts of the United States.

Recent Findings

Screening strategies for biliary atresia include detecting persistent jaundice, examining stool color, testing fractionated bilirubin levels, or measuring bile acid levels from dried blood spot cards. The stool color card program is the most widely used screening strategy worldwide. An alternative approach under investigation in the United States measures fractionated bilirubin levels, which are abnormal in newborns with biliary atresia. Fractionated bilirubin screening programs require laboratories to derive reference ranges, nurseries to implement universal testing, and healthcare systems to develop infrastructure that identifies and acts upon abnormal results.

Summary

Biliary atresia meets the disease-specific criteria for newborn screening. Current studies focus on developing a strategy which also meets all test-specific criteria. Such a strategy, if implemented uniformly, has the potential to accelerate treatment and reduce biliary atresia’s large liver transplant burden.

Introduction

Biliary atresia (BA), a liver disease of infancy characterized by obstruction of the extrahepatic biliary tree, fulfills all the disease-specific criteria of a condition that would benefit from newborn screening (Table 1). First, BA is a significant health condition, affecting 1 in 8,000–18,000 live births and accounting for the most pediatric liver transplantations worldwide [1]. Second, the natural history of BA is well-described, starting with a latent preclinical phase in which infants have abnormalities but are often unrecognized as affected (Figure 1). Visible symptoms of cholestasis develop after the first weeks of life, followed by rapidly progressing liver injury and end-stage liver disease in the first year of life. Third, BA has a well-established treatment which can delay or even avoid the need for liver transplant. This treatment, the Kasai portoenterostomy (KP), directly connects the intestines to the liver to restore bile flow. A critical factor predicting KP outcomes is the time at which the operation is performed. KPs performed before 30–45 days of life (often requiring detection within the preclinical phase) have the greatest chances of delaying or avoiding liver transplant [2]. Unfortunately, in the United States (US), without screening the average age at time of KP is after 60 days of life and there have been no recent improvements [3].

Table 1.

Principles of Newborn Screening, Adapted by the Wilson and Jungner Criteria^*

Wilson and Jungner Criteria [14]		Interpretation in the context of BA
Disease-specific	The condition sought should be an important health problem.	BA is the leading indication for pediatric LT, accounting for 60% of LTs in infants <1 year old and 30% of all LTs in children [37]. Identifying and treating infants earlier, i.e., in their latent or “preclinical” phase, can improve outcomes.
	There should be an accepted treatment for patients with recognized disease.	The KP is the only treatment proven to delay or even prevent the need for LT [38]. The KP bypasses the bile duct obstruction by connecting the liver directly to the intestines. Performing the KP earlier, i.e., before 30–45 days of life, has been associated with better outcomes [2].
	Facilities for diagnosis and treatment should be available.	BA can be diagnosed and treated in centers with pediatric specialists, including surgeons and gastroenterologists/hepatologists. These centers are available in the US and most developed countries.
	There should be a recognizable latent or early symptomatic stage.	In BA, a latent or “preclinical” phase occurs in the first weeks of life. Biliary obstruction is present, but signs such as jaundice or pale stool may not be apparent. As a result, BA is often recognized later than the ideal time for treatment [3].
	The natural history of the condition, including development from latent to declared disease, should be adequately understood.	The natural history of BA has been well-described, and includes a likely in-utero start, symptoms appearing in the first weeks of life, and end-stage liver disease and need for liver transplant by age 2 years if not treated [39,40]. BA is a neonatal disease which never starts at later time points.
	There should be an agreed policy on whom to treat as patients.	The standard of care is to treat all patients with BA surgically. Infants recognized in a timely manner receive the KP. Infants recognized too late for the KP, as well as infants who do not benefit from the KP, receive LT.
Test-specific	There should be a suitable test or examination.	Screening for BA using the physical exam, stool color cards (SCCs), fractionated bilirubin levels, and bile acid measurements have been studied.
	The test should be acceptable to the population.	Regions in Japan, Taiwan, Canada, Switzerland, Germany, and Brazil have adopted the SCC and multiple healthcare systems in the US have begun screening with fractionated bilirubin (see text for full discussion).
	The cost of case-finding (including diagnosis and treatment of patients diagnosed) should be economically balanced in relation to possible expenditure on medical care as a whole.	More studies are needed to evaluate the costs of case-finding. The cost-effectiveness of the SCC has been shown to be acceptable, and the cost-effectiveness of fractionated bilirubin measurements is still under investigation [27].
	Case-finding should be a continuing process and not a “once and for all” project.	Screening for BA is a continuous process based on the data available. For example, the SCC began in Japan has since been modified in different countries with innovations such as a phone app [11,12]. In addition, after fractionated bilirubin levels were discovered to be elevated at birth in BA, newborn screening for BA was initiated [22,26]. Both the SCC and fractionated bilirubin screening continue to be evaluated and improved, and new modalities such as bile acid screening have been proposed.

^*Does not require permission for use and reproduction.

Figure 1.

Natural History of BA

The natural history of BA is well-described. The disease starts with a “preclinical” phase in the first weeks of life. This follows the disease onset, which occurs at an unknown time but may be in utero, given gallbladder abnormalities early in the second trimester, low amniotic fluid gamma-glutamyl transferase levels in the mid-second trimester, and high fractionated bilirubin levels at birth [39]. After the preclinical phase, infants enter the clinical phase where they appear jaundiced and develop pale colored stools characteristic of biliary obstruction. Liver injury ensues, and, if untreated, BA will invariably lead to end-stage liver disease and death/need for liver transplant by 2 years of age [40].

Although the disease-specific criteria for neonatal screening are fulfilled, to date there is no accepted BA screening approach in the US which fulfills the test-specific criteria outlined initially by Wilson and Jungner (Table 1) [4]. Several recent studies have examined potential options that could identify infants earlier and help ensure a timely KP. This review discusses the different BA screening programs that have been proposed and investigated, as well as shares the experiences of implementing one of the methods in four areas across the US.

Different Methods of Screening

Physical Exam

One screening approach is to detect jaundice, which develops in BA when bile flow is blocked and conjugated bilirubin (Bc) retained. A screening program to detect neonatal liver diseases using jaundice was introduced by the Children’s Liver Foundation in the United Kingdom in 1993 [5]. The program identified infants older than 14 days who had jaundice and evaluated them with urinary bilirubin and serum conjugated bilirubin measurements. Similarly, American Academy of Pediatrics (AAP) guidelines use jaundice to identify liver disease, by recommending further testing if jaundice is present at birth, accompanied by “dark urine or light stool,” requires re-admission for phototherapy, or present after two weeks of life [6–8].

Screening programs based on detecting jaundice alone present important challenges. First, at the initial routine well visits, infants with BA may not appear jaundiced despite having elevated fractionated bilirubin levels. This is because in the first weeks the levels may still be below the visual threshold (5 mg/dL) [9]. Second, many more infants appear jaundiced from non-hepatic reasons, such as physiologic jaundice or breast-milk jaundice. For example, in the United Kingdom study, up to 15% of infants appeared jaundiced at 14 days in part because of breast-feeding [5]. Third, there is only moderate interobserver agreement (pairwise kappa 0.48) among clinicians detecting jaundice. The interobserver agreement improves when total bilirubin (TB) levels are above 12 mg/dL [10].

Stool Color Card

Another important screening approach uses the stool color card (SCC), to identify infants with pale stools that develop over time when bile flow to the intestine is obstructed in BA. SCC programs give caregivers a card with color photographs of normal and pale stools. Caregivers are asked to call a central phone number if they notice pale stools as well as to discuss observations of their infant’s stool color at the routine 1-month health visit. Newer innovations include an app, which evaluates stool color using a smartphone’s camera and internal color recognition software [11].

The SCC was first tested in 1994 in Tochigi Prefecture, Japan [12]. Of the 313,230 live births, 264,071 SCCs (84.3%) were collected from infants at the 1-month health visit. There were 34 infants diagnosed with BA during the study period (birth prevalence 1:9,213 infants). The SCC identified 26 of these cases, with the remaining eight infants having stools which were normal at the 1-month health visit (n=1), overlooked because of a complicated NICU stay (n=2), not acted upon because clinicians did not notice accompanying jaundice (n=3), parent failed to be given SCC (n=1), or not observed because of loss to follow-up (n=1). At the 1-month health visit, the SCC had a sensitivity, specificity, and positive predictive value of 76.5%, 99.9%, and 12.7%, respectively. The average age at KP with the SCC program was 60 ± 19 days (54 ± 16 days for the 26 identified by the SCC), which was earlier than before SCC implementation (70 days). Transplant-free survival at 5, 10, and 15 years was 87.6%, 76.9%, and 48.5%, respectively, compared to 68.0%, 60.0%, and 51.0%, respectively, in Sendai, Japan, without the SCC [13].

The SCC was later tested in 2002–2003 across 96 hospitals in the 4 regions of Taiwan [14]. Of the 119,973 live births during the study period, there were 78,184 SCCs (65.2%) brought to the 1-month visit or mailed/faxed to the registry center. Twenty-nine infants were diagnosed with BA during the study period by the SCC (birth prevalence 1:2,696 infants), including 26 infants before 60 days of life. The program in Taiwan had a sensitivity, specificity, and positive predictive value for detecting BA by 60 days of life of 89.7%, 99.9%, and 28.6%, respectively. The screening program led to more KPs before 60 days of life (65.7% with the SCC vs. 49.4% in historical controls) and was associated with an improvement in the average age at KP to 48 days from 60 days without screening [15]. Furthermore, screening was also associated with an improved 3-year jaundice-free survival rate with native liver (56.9% with the SCC vs. 31.5% in historical controls) [16]. Taiwan subsequently used the SCC to start the first national BA screening program. Because of successes in Japan and Taiwan, other countries including Germany, Canada, Switzerland, and Brazil have adopted variations of the SCC program and have shown that it is a cost-effective way to detect BA earlier [17–20].

In the US, there are structural reasons which make implementing the SCC challenging. First, the healthcare system is decentralized, with no central call center to address parental concerns about pale stools. As a result, individual pediatrician offices would be responsible for answering phone calls and making judgements about stool color. Second, the SCC program is rooted in a standard 1-month health visit, when parents and primary care providers (PCPs) can review stool color. However, in the US, a standard 1-month health visit does not exist. Instead, the AAP recommends a “by 1 month” health visit, which many PCPs elect to schedule at 2 weeks of life at the time of the second state newborn screen [21]. Pale stools may not be present in many patients with BA at this early time point, as the Japanese study reported pale stools developing in only 76.5% of infants by 30 days of life [12].

Newborn Fractionated Bilirubin Measurements

A third screening approach uses fractionated bilirubin levels, which are elevated starting at birth in BA [22]. An early study in the United Kingdom recorded Bc and Bc to TB ratios in all infants in the first weeks of life (median age of 7 days old, range 4–28 days) [23]. There were 23,214 infants screened, of whom 107 infants had levels exceeding the 97.5^th percentile value for the population (Bc >18 μmol/L and Bc to TB ratio >0.20). These infants were followed further by measuring fractionated bilirubin, aspartate aminotransferase, alanine aminotransferase, alkaline phosphatase, and gamma-glutamyl transferase levels. There were 11 patients identified with liver disease including 2 with BA [24].

Subsequent studies examined screening with fractionated serum bilirubin levels in the newborn period, i.e., in the first 24–48 hours of life (ratios were not used, because most infants with BA still have ratios ≤0.20 at this early time point) [22]. In a pilot study of 11,636 newborns, 1.1% (n=121) of infants had fractionated bilirubin levels exceeding their laboratory’s 97.5^th percentile value [25]. Of those, 9.1% (n=11) had elevated repeat levels by two weeks of life including the two infants in the study with BA. This two-step newborn fractionated bilirubin screening approach had a sensitivity, specificity, and positive predictive value of detecting BA in the first days of life of 100.0% (95% CI, 19.8–100.0), 99.9% (95% CI, 99.8–99.9), and 18.2% (95% CI, 3.2–55.2), respectively. In a larger study of 123,279 infants, 1.1% (n=1,354) of newborns had an elevated initial fractionated bilirubin level [26]. Of those, 8.8% (n=119) had elevated repeat levels by two weeks of life including the seven infants in the study with BA. Comparable to the pilot study, the sensitivity, specificity, and positive predictive value of detecting BA in the first days of life were 100.0% (95% CI, 56–100), 99.9% (95% CI, 99.9–99.9), and 5.9% (95% CI, 2.6–12.2), respectively. Screening was associated with significantly younger age at KP (36 days with screening vs. 56 days before screening implementation), and screening also identified other liver diseases such as Alagille Syndrome and alpha-1-antitrypsin deficiency.

Limitations to newborn fractionated bilirubin screening include: (1) direct bilirubin (DB) assays varying across laboratories and thus requiring each hospital laboratory to derive their own reference interval; (2) a higher positive rate with the initial test at 24–48 hours of life in Black infants versus infants from other races; and (3) a high level of coordination needed between nurseries and PCPs to ensure infants with elevated levels initially are re-tested by 2 weeks of life (see below). In addition, the cost-effectiveness of newborn fractionated bilirubin screening remains unknown. A recent study suggested that newborn bilirubin screening would be significantly more expensive than the SCC per life-year gained [27]. However, this study was performed before data for fractionated bilirubin screening was available, and assumed screening had a specificity of only 98% and required complete evaluations for screen positive patients (many patients only needed a third repeat fractionated bilirubin measurement [26]). The study also did not assess the cost savings the screening may accrue with early identification of other, non-BA neonatal liver diseases.

Bile Acids

A screening method currently under investigation is measurement of bile acids s using dried blood spots. If successful, this would allow BA screening to be conducted alongside each state’s standard newborn screen for inborn metabolic conditions. In the initial study, bile acid levels measured by tandem mass spectrometry failed to distinguish between healthy newborns and those with liver disease [28]. A more recent study found that taurocholate levels in dried blood spots collected 3–4 days after birth were significantly elevated in patients with BA (n=8) compared to that of jaundiced (non-cholestatic) infants (n=17) and normal healthy infants (n=292) [29]. Using a taurocholate cutoff of 0.63 μmol/L, this test had a sensitivity of 79.1% and specificity of 62.5% for BA.

Implementing Screening in the United States

The authors have initiated a two-stage screening approach using newborn fractionated bilirubin measurements in different areas of the US (Houston and the Rio Grande Valley, Salt Lake City and areas of Utah, New Orleans, and San Antonio; see Table 2 for overview of these programs). Fractionated bilirubin measurements include either Bc or DB testing (see below for a more complete discussion). The first stage involves testing all infants before discharge from the newborn nursery. Infants with initial levels exceeding their laboratory’s 97.5^th percentile value are considered positive. In the second stage, infants who were positive in stage 1 are re-tested by two weeks of life. Infants with repeat fractionated bilirubin levels >1 mg/dL are urgently referred to a specialist for further evaluation. Infants with a normal level, or a level ≤1 mg/dL and less than or equal to the initial level, in stage 2 are not evaluated further for BA. Finally, infants with a level greater than the initial level but ≤1 mg/dL in stage 2 undergo repeat testing in one week and are referred if the levels do not decrease.

Table 2.

Current Regions in the United States Screening for BA Utilizing Fractionated Bilirubin

	Houston and the Rio Grande Valley	Salt Lake City and areas of Utah	New Orleans	San Antonio
Hospital Type(s)	General, Public (n=1) General, Private (n=12) Pediatric, Private (n=1)	General, Private (n=4)	General, Private (n=2)	General, Public (n=1)
Number of Hospitals	14	4	2	1
Estimated Annual Number of Live Births	30,000	16,000	4,000	3,200
Test(s) Used	Bc, DB	DB	DB	DB
Reference Range(s) mg/dL	Bc: 0.0–0.2 (n=1) DB: 0.0–0.3 (n=1) 0.0–0.4 (n=2) 0.0–0.5 (n=7) 0.0–0.6 (n=1) 0.0–0.7 (n=2)	DB: 0.0–0.5 (n=4)	DB: 0.1–0.6 (n=2)	DB: 0.0–0.4 (n=1)
Date Started	July 2013	October 2020	March 2021	February 2021

While fractionated bilirubin screening programs currently occur in different healthcare systems serving diverse populations in the US, there are common experiences shared across all programs. This includes core milestones which had to be met as fractionated bilirubin screening was being implemented (Figure 2):

1. Establishing fractionated bilirubin reference intervals in birth hospitals. There are two assays used to measure fractionated bilirubin levels, the Bc assay and the DB assay [30]. The Bc assay measures only conjugated bilirubin using direct spectrophotometry, whereas the DB assays measures conjugated bilirubin, delta bilirubin, and some unconjugated bilirubin using the diazo chemical reaction. As a result, Bc levels will usually be lower than the DB levels from the sample. As shown in Table 2, most screening hospitals use the DB assay.

   An important reason for identifying the type of assay is that DB assays require reference intervals to be derived at each site. This is because DB levels vary across analyzers, depending on slight variations in reaction conditions such as temperature, pH, and time [31]. The standard way of deriving reference intervals is by measuring 120 presumed normal newborn values and identifying the middle 95% of values (between the 2.5^th and 97.5^th percentile values) [32]. Values above the 97.5^th percentile values are considered elevated. In contrast to DB levels, Bc levels will not vary across sites and a standard reference interval of 0.0–0.2 mg/dL can be used. Importantly, the ratios of Bc or DB to total bilirubin ratios are not used as they may miss many infants with BA [22].

2. Creating a workflow for automated testing in the newborn nursery. Nurseries currently employ various practices to assess need for phototherapy and prevent bilirubin encephalopathy, including testing (i) all infants with total and fractionated bilirubin measurements, (ii) all infants with total bilirubin measurements, (iii) selected infants with total and fractionated bilirubin measurements, (iv) selected infants with total bilirubin measurements, or (v) infants with transcutaneous bilirubin measurements. For screening, at least one fractionated bilirubin measurement is required for all infants prior to discharge regardless of whether jaundice or other clinical signs are present. This may require adding fractionated bilirubin onto total measurements (the two can be measured simultaneously from the same sample). In addition, although transcutaneous measurements are one method recommended by the AAP to assess jaundice and need for phototherapy, they are not able to determine Bc or DB levels [6]. As a result, nurseries with a policy of only using transcutaneous measurements will have to introduce an additional test into their workflow.

   To ensure that infants are tested universally, and no infants are missed, newborn order sets can be changed to include routine fractionated measurements. In the nurseries currently using the screening approach, phlebotomists or bedside nurses collect the blood samples in the first 24–48 hours of life. They can use a venipuncture or heel stick approach, and often collect the blood in coordination with the first state newborn screen. It is possible that phlebotomists or nurses may require additional training to prevent hemolysis, because excess free hemoglobin has been shown to decrease fractionated bilirubin measurements [33]. To assist this, national laboratory organizations have created freely-accessible online video tutorials demonstrating the effective heel stick technique [34].

3. Designating newborn nursery personnel to review results. Screening requires newborn nursery personnel to fulfill two responsibilities. First, the result for each infant needs to be reviewed and those higher than the laboratory’s 97.5^th percentile identified. In all current screening sites, a pediatric gastroenterologist or a trainee associated with a pediatric gastroenterologist is reviewing the results as part of a research/quality improvement initiative. In most sites, the pediatric gastroenterologist receives a daily or weekly spreadsheet with the initial fractionated bilirubin value for all newborns. The spreadsheet is automatically generated by programs which query the patient chart, and the program’s coding can be shared among hospitals using the same electronic medical record system. In addition, in two areas, pediatric gastroenterologists receive direct communication from research or nursery staff when an infant has an elevated level.

   Second, PCPs need to be told about newborns who had elevated fractionated bilirubin levels at birth, so they know to repeat levels by 2 weeks of life. This has been the most challenging step in implementation, with methods still evolving to improve efficiency. The most common method is the pediatric gastroenterologist directly calling the outpatient PCP. The advantages of this method are that PCPs can ask questions and learn about the screening protocol, which ultimately builds rapport and trust between specialists and outpatient providers. The disadvantages of this method are that it is time consuming and may not be feasible on larger scales unless clinical time is specifically allotted for making phone calls. Other options used to communicate with PCPs are messaging through electronic medical record systems, adding information in bold lettering to the discharge summaries, and/or sending information about the need for a repeat check directly with parents (with the risk of increasing parental anxiety; see Appendix materials for examples of materials that are distributed).

4. Designating personnel to follow-up outpatient results. Currently, a pediatric gastroenterologist, or a trainee associated with a pediatric gastroenterologist in the particular screening area, interpret the outpatient results. They receive these results through multiple mechanisms, including receiving return calls from the PCPs’ offices, reviewing the electronic medical record (when the PCP is in the same health care system), or calling the PCP office. In one area, the results are automatically returned because each positive stage 1 case is enrolled in a research study. Calling PCP offices requires the most time. Future initiatives could potentially merge fractionated bilirubin screening with the state newborn screening programs, which already have dedicated staff to follow-up other repeat laboratory test results efficiently.

   For various reasons, a small minority of infants who have elevated levels initially in the first 24–48 hours of life do not undergo repeat testing. In these cases, the pediatric gastroenterologist tries to follow these patients by phone call, to make sure no signs such as persistent jaundice or pale stools develop. The most common reasons why infants do not have repeat testing is because of lack of transportation for well visits, parents refusing further blood testing, PCPs refusing to test, and/or patients moving out of the US shortly after birth. Solutions to these problems include activating social work services to help parents with limited resources and providing more information to parents and PCPs about the benefits of repeat testing.

5. Developing a referral pipeline to pediatric gastroenterology/hepatology. Ideally, a pediatric gastroenterologist evaluates infants identified by screening within 7 days, to help ensure the KP is performed in the first 30–45 days of life. One challenge is that some patients live far away from pediatric specialty centers. To address this, initial telehealth consultations are used in some areas. A second challenge is that specialists will have to distinguish BA from a variety of other causes of neonatal cholestasis. The proposed algorithm has yet to be adopted [35,36].

In the future, medico-legal issues for fractionated bilirubin screening will also need to be carefully considered. Currently, each participating area conducts screening on a research and/or quality improvement basis. However, if fractionated bilirubin screening is adopted into routine clinical practice, the responsibilities of different individuals will need to be assigned. These include thoughtfully determining which healthcare providers are accountable for identifying abnormal results initially, conveying results from nursery to outpatient provider, retesting infants who tested high initially, and interpreting data from the repeat test. In addition, as with all newborn testing programs, the degree of autonomy parents have in refusing screening must be specified.

Figure 2.

Implementation of Newborn Screening for BA with Fractionated Bilirubin

Conclusion

BA is the leading indication for pediatric liver transplant and meets all disease-specific criteria warranting newborn screening. However, to date, there has not been a newborn screening approach widely accepted in the US for BA. Many studies have evaluated potential approaches which could detect affected infants earlier and potentially help accelerate treatments, including screening for jaundice, pale stools, elevated serum fractionated bilirubin levels, or elevated serum bile acid levels. Current and future studies will address efficacy, costs, and optimal implementation strategies, with the goal of intervening earlier and ultimately delaying or reducing the need for liver transplantation.

Appendix*

* Does not require permission for use and reproduction.