Transient elastography is useful in diagnosing biliary atresia and predicting prognosis after hepatoportoenterostomy

SUBJECTS AND CLINICAL DATA

We recruited 48 cholestatic infants (31 males and 17 females) from the Department of Pediatrics of National Taiwan University Hospital (NTUH) from May 2015 to December 2017 to this study prospectively. All patients presented with cholestasis (serum direct bilirubin level >1 mg/dL and direct to total bilirubin ratio >20%). Subjects with ascites, septic shock, and previous abdominal surgery were excluded. All subjects underwent workup for cholestasis, including blood tests, urine test, metabolic workup, and abdominal sonogram. Liver biopsy was performed in 36 subjects (75%) for diagnostic purposes, and IOC was performed in 22 subjects (45.83%). The diagnosis of BA was confirmed by IOC in 15 subjects (31.25%), all of whom underwent HPE. All patients underwent regular follow-up after the workup and management in our institution. Tests for blood biochemical parameters (total and direct bilirubin levels; alanine aminotransferase [ALT], alkaline phosphatase [ALP], and gamma-glutamyl transferase [GGT] levels) and abdominal sonograms were performed regularly during the follow-up period. The study protocol was approved by the institutional review board of NTUH.

MEASUREMENT OF SPLEEN SIZE AND DIAGNOSIS OF ESOPHAGEAL VARICES IN BA PATIENTS

In subjects with BA post-HPE, we performed an abdominal sonogram every 6 months since 6 months of age or at the presence of palpated splenomegaly at physical examination. Longitudinal length of spleen was measured by the abdominal sonogram. Esophagogastroduodenoscopy was performed for the diagnosis of esophageal varices (EV) at the presence of positive stool occult blood with/without splenomegaly or hematemesis in these BA patients.

LF MEASUREMENT BY TE

Liver fibrosis measurement (liver stiffness measurement; LSM) by TE (Fibroscan 502 Touch; Echosens, Paris, France) was performed during the cholestatic workup before liver biopsy and IOC. We used the S1 probe (5 MHz) for LSM. Ten shots within 3-5 minutes were performed in each subject. Given the accuracy of TE, an interquartile range (IQR)/median LSM value <0.3 was defined as valid. In subjects with BA, LSM was performed weekly for 1 month post-HPE and then monthly until 6 months after the operation. Six months after the operation, LSM was performed in subjects with BA at 3-month intervals. A bedside abdominal sonogram was performed to define the ascites-free status before each LSM. All TEs were performed by W.J.F. using the same instrument.

LF ASSESSMENT

LF status was assessed histologically in specimens from 36 subjects obtained by needle biopsy during the cholestatic workup or wedge biopsy during the operation. The METAVIR score was evaluated by a pathologist (J.Y.M.) who was blind to the LSM data.20

STATISTICAL ANALYSIS

STATA (version 14; StataCorp LP, College Station, TX) and MedCalc (version 18; MedCalc Software, Ostend, Belgium) software packages were used for statistical analyses. The primary outcome was a diagnosis of BA, and the secondary outcome was the cholestatic complications (thrombocytopenia, splenomegaly, and EV 6 months post-HPE and the need for LT younger than 1.5 years of age in BA subjects). For continuous variables, the nonparametric Mann-Whitney U test was used to assess differences in IQRs and medians between the two groups. Fisher's exact test or the chi-square test was performed to assess differences in incidence between the groups. Receiver operating characteristic (ROC) curve analyses were performed to determine cutoff values. Logistic regression was used to assess the odds ratio (OR) and 95% confidence interval (CI) for predicting BA. Cox proportional analysis, Kaplan-Meier plot, and log-rank test were applied for survival analysis of the need of LT. Regression analysis was also used for data analysis. A P value <0.05 was regarded as statistically significant.

GENERAL CHARACTERISTICS OF THE SUBJECTS

BA was diagnosed in 15 subjects, neonatal hepatitis in 18 subjects, total parenteral nutrition (TPN)-related cholestasis in 3 subjects, neonatal intrahepatic cholestasis caused by citrin deficiency (NICCD) in 1 subject, urinary tract infection (UTI) in 2 subjects, choledochal cyst in 2 subjects, liver-congestion–related cholestasis in 2 subjects, cystic fibrosis in 1 subject, glycogen storage disease in 1 subject, type I progressive familial intrahepatic cholestasis in 1 subject, Alagille syndrome in 1 subject, and inborn error of bile acid synthesis in 1 subject in these cholestatic infants (n = 48).

Subjects with BA (n = 15) had higher GGT, direct bilirubin levels, and LSM values than those subjects without BA of similar age (n = 33; P = 0.001, 0.004, and 0.0004, respectively; Table 1). There was no relationship between LSM value and total serum level of bilirubin (P = 0.76), LSM value and serum direct bilirubin level (P = 0.11), or LSM value and serum bile acid level (P = 0.68) during cholestasis workup in these 48 subjects. In the 15 BA patients, there was also no significant relationship between LSM value and serum total and direct bilirubin levels (P = 0.62 and 0.95, respectively) or serum bile acid level (P = 0.94) pre-HPE.

In total, 36 subjects received liver biopsy in this cohort, and the METAVIR score was graded as F0 in 8 subjects, F1 in 5 subjects, F2 in 12 subjects, F3 in 6 subjects, and F4 in the remaining 5 subjects. Among these 36 cholestatic infants with liver specimens obtained within 90 days of age, the median METAVIR score was F3 (range = F2-F4) in BA (n = 15), F1 (range = F0-F2) in neonatal hepatitis (n = 14), F1 (range = F0-F2) in UTI (n = 2), F0 in one choledochal cyst, F0 in one cystic fibrosis, F2 in one glycogen storage disease, F0 in one type I progressive familial intrahepatic cholestasis, and F2 in one TPN-related cholestasis. LSM value was significantly higher in METAVIR F3-F4 subjects (median = 10.5 kPa; IQR = 8.9-17.0 kPa; n = 11) than METAVIR F0-F2 subjects (median = 4.9 kPa; IQR = 4.0-6.4 kPa; n = 25; P = 0.002). LSM value was also correlated with the METAVIR score in these 36 subjects (P = 0.023; Fig. 1A).

There are 12 cholestatic subjects without liver biopsy received regular follow-up in our institution. Four of them were diagnosed as cytomegalovirus-related neonatal hepatitis and had a total serum level of bilirubin <2 mg/dL before 1 year of age. Two of the 12 cholestatic subjects were diagnosed as TPN-related cholestasis, 1 with NICCD according to genetic analysis, 1 with Alagille syndrome according to genetic analysis, 1 with inborn error of bile acid synthesis according to urine bile acid analysis, 1 with choledochal cyst, and 2 with liver congestion attributed to right heart failure.

DIAGNOSTIC UTILITY OF LSM IN BA

ROC curve analysis showed that an LSM cutoff >7.7 kPa was optimal for predicting BA measured by TE (sensitivity = 80%; specificity = 97%; area under the curve [AUC] = 85.30%; P = 0.0001; Fig. 1B). A GGT level >200 IU/L was used as a diagnostic predictor of BA in a recent study.8 A GGT level >200 IU/L and LSM >7.7 kPa were predictive of BA in univariate logistic regression analyses (OR = 10.21 and 128; P = 0.001 and <0.001, respectively; Table 2). In the multivariate logistic regression model that included both factors, only the LSM value >7.7 kPa was predictive of BA (OR = 83.62; P < 0.001; Table 2). The positive and negative predictive values (PPV and NPV) of an LSM value cutoff >7.7 kPa for BA were 92.31% and 91.43%, respectively.

RELATIONSHIP BETWEEN AGE AND LF

Pre-HPE, LSM value of BA subjects increased significantly with age (P < 0.001; Fig. 1C). Three subjects with BA (20%) had an LSM value <7.7 kPa, and all of them underwent TE before 21 days of age. Another 12 subjects with BA (80%) who underwent TE after 21 days of age had LSM values >7.7 kPa during the workup for cholestatic liver disease. In the non-BA cholestatic subjects (n = 33), the correlation between LSM value and age was not significant (correlation coefficient = 0.48; P = 0.80).

In all cholestatic infants more than 21 days of age (n = 43), the PPV and NPV for BA of an LSM value cutoff >7.7 kPa were 92.31% and 100%, respectively.

FOLLOW-UP PATTERN OF LSM PATTERN IN BA PATIENTS POST-HPE

The clinical data of the 15 subjects with BA are summarized in Table 3. Jaundice-free status (serum level of total bilirubin <2 mg/dL) ever achieved in 14 subjects at a median time of 35 days (IQR = 17.0-67.5 days) post-HPE. The follow-up period post-HPE in case 15 is 2 months to the end of this study, and no cholestatic complications at the final visit (4.5 months of age).

Three patterns of the changes in LSM values post-HPE were evident (Supporting Table S1). LSM pattern 1 (n = 4, 26.67%; LSM resolving or stably low) involved a significant decline in LSM value and serum bilirubin level post-HPE that remained stable. The LSM value of case 9 pre-HPE was 10.1 kPa, and this patient underwent surgery at 50 days of age (METAVIR F3). His serum bilirubin levels and LSM declined and remained stationary after the 9 days after operation (Fig. 2A).

LSM pattern 2 (n = 10, 66.67%; LSM slowly elevated) involved elevated LSM values, even the jaundice-free status ever achieved post-HPE. The LSM value of case 4 was 9.1 kPa pre-HPE, and this patient underwent HPE at 48 days of age (METAVIR F3). Her LSM value remained elevated post-HPE, and her serum level of total bilirubin decreased to <2 mg/dL 37 days after the operation (Fig. 2B). She received LT at 11 months of age.

The interval between HPE and a jaundice-free status (total serum bilirubin level <2 mg/dL) in pattern 1 subjects (n = 4) was significantly shorter than that in pattern 2 subjects (n = 10; median = 16.5 vs. 36 days; IQR = 11-20 vs. 19.5-95.0 days; P = 0.03).

LSM pattern 3 (n = 1; 6.67%; LSM rapidly elevated) subjects showed persistent jaundice post-HPE and a rapid increase in the LSM values. The LSM value of case 10 was 10.5 kPa pre-HPE, and this patient underwent surgery at 60 days of age (METAVIR F3). Her jaundice persisted post-HPE, and her LSM value increased rapidly from 10.5 to 67.0 kPa (Fig. 2C). Now, she is waiting for LT.

The serial pattern of serum total bilirubin levels and LSM assessed before and after HPE in all BA infants are summarized in Supporting Table S1. Kaplan-Meier analysis demonstrated that need of LT is significantly higher in LSM patterns 2 and 3 BA subjects (n = 11) than pattern 1 BA subjects (n = 4; log-rank test, P = 0.04; Fig. 2D).

PREDICTIVE ROLES OF VERY EARLY SERIAL LSM MEASUREMENT POST-HPE ON CLINICAL OUTCOMES

In subjects with BA followed up for more than 6 months post-HPE (n = 14), LSM values declined or remained <7.7 kPa in 3 subjects with BA post-HPE and were elevated in another 11 subjects with BA compared to LSM values obtained during the cholestatic workup.

The LSM value measured 1 week post-HPE was positively correlated with longitudinal spleen length 6 months post-HPE (correlation coefficient = 0.74; P = 0.003; Supporting Fig. S1A), presence of EV 6 months post-HPE (correlation coefficient = 0.74; P = 0.003), and need for LT before 1.5 years of age (correlation coefficient = 0.74; P = 0.003; Table 4). Subjects with EV 6 months post-HPE (n = 5) had higher LSM values measured 1 week post-HPE than subjects without EV (n = 9; median = 26.0 vs. 10.8 kPa; IQR = 19.5-29.0 vs. 6.4-14.8; P = 0.006).

Five subjects with BA who required or had undergone LT before 1.5 years of age had significantly higher LSM values measured 1 week post-HPE than that in the other subjects with BA (median = 26.0 vs. 10.8 kPa; IQR = 19.5-29.0 vs. 6.4-14.8; P = 0.006). The LSM value of BA subjects measured 3 months post-HPE was also negatively correlated with platelet count 6 months post-HPE (correlation coefficient = –0.54; P = 0.04; Supporting Fig. S1B; Table 4).

ROC analysis yielded the cutoff of LSM >16 kPa measured 1 week post-HPE for best prediction of EV 6 months post-HPE and need of LT before 1.5 years of age (P < 0.001 and P < 0.001, respectively; Supporting Fig. S2A,B). BA infants with LSM >16 kPa measured 1 week post-HPE were noted to have a higher risk of developing EV 6 months post-HPE (risk ratio [RR] = 10; P = 0.005) and need of LT before 1.5 years of age (RR = 10; P = 0.005). A Cox proportional analysis demonstrated that need for LT was significantly higher in BA subjects with LSM >16 kPa than other BA subjects with LSM ≤16 kPa 1 week post-HPE (hazard ratio [HR] = 10.16; 95% CI = 1.13-91.72; P = 0.04).