Introduction

Therapeutic strategies for biliary atresia (BA) have advanced, and the number of long-term native liver survivors has been increasing [1]. However, some of these survivors experience liver dysfunction and long-term complications [2], and they might require liver transplantation (LTx) if the hepatic function deteriorates drastically [3].

The determination of the optimal timing of LTx is one of the most important issues in BA patients, and in stable patients with mid-term jaundice-free native liver survival (NLS), this issue has not been resolved. The identification of the prognostic factors for long-term NLS would aid in the determination of the optimal timing of LTx in these patients.

Therefore, in this study we evaluated the prognostic factors for long-term NLS in BA patients after the Kasai procedure.

Materials and methods

A total of 200 patients with BA underwent surgery at Tohoku University Hospital between 1972 and 1995. Of these patients, 140 (70 %) showed jaundice disappearance after the surgery and 98 (49 %) had NLS for over 10 years. Among these patients, 31 were excluded (11 patients were born before 1983 and required early LTx according to indications, such as secondary pulmonary perfusion abnormalities, and 20 patients, who were mainly followed up by other medical providers, had insufficient medical information). Finally, 67 patients whose detailed medical records were available until the age of 20 years were included in this retrospective study. This study was approved by the ethics committee of Tohoku University (No. 2015-1-793).

We retrospectively evaluated the clinical parameters, including the type of BA, age at the Kasai procedure, medical conditions, and treatments. The medical conditions were cholangitis, gastro-esophageal varices (GEV) requiring endoscopic injection sclerotherapy (EIS), and hypersplenism requiring intervention until the age of 10 years. The type of BA was divided into the following four categories: type 1, atresia of the common bile duct; type 1-cyst, type 1 BA with a cyst of the extrahepatic bile duct; type 2, atresia of the hepatic duct; and type 3, atresia of the porta hepatis. Cholangitis was diagnosed when the patient had fever with elevated direct bilirubin and C-reactive protein levels.

The interventions for hypersplenism included splenectomy and partial splenic artery embolization, and these were performed when the patient had thrombocytopenia and clinical bleeding symptoms. EIS was performed when bleeding varices were confirmed or when the varices had a high risk for bleeding.

Statistical analysis

A descriptive statistical analysis was performed for each parameter. For each categorical variable, the Fisher exact test was used, and the crude odds ratios were obtained for 20-year NLS. When two or more factors in a regression model are highly correlated, multicollinearity exists, and this can cause issues in the regression model. Therefore, multicollinearity among the factors was examined using the kappa statistic, and we evaluated whether each factor was suitable for multiple logistic regression analysis. The adjusted odds ratios (aOR) were then obtained for 20-year NLS using multiple logistic regression analysis. We finally performed receiver operating characteristic curve analysis for an adopted model, and the area under the curve was obtained. All statistical analyses were performed using JMP Pro 12.2 (SAS Institute Inc., Cary, NC).

Results

Of the 67 study patients, 62 patients (92.5 %) had jaundice-free NLS at the age of 20 years, 4 patients died before the age of 20 years from liver failure, and 1 patient underwent living-related LTx because of progression of decompensated cirrhosis.

The patient characteristics and results of univariate analysis are presented in Table 1. The median age of the patients at the Kasai procedure was 63 days (range 27–148 days).

Table 1 Patient characteristics and univariate analysis
Full size table

In univariate analyses of clinical statuses, GEV requiring EIS and hypersplenism until 10 years of age were significantly different between patients with 20-year NLS and those without 20-year NLS. However, the type of BA, sex, age at the Kasai procedure, and incidence of cholangitis were not significantly different between the patient groups.

Multicollinearity among the factors was examined before multivariate logistic regression analysis, and we found a significant consistency between GEV requiring EIS and hypersplenism (p < 0.0001; Table 2). Therefore, we analyzed GEV requiring EIS and hypersplenism separately in multivariate analysis.

Table 2 Consistency between GEV requiring EIS and hypersplenism
Full size table

With regard to the type of BA, all five patients without 20-year NLS had type 3 BA, and there was no significant difference in univariate analysis. Therefore, the type of BA was excluded as an independent factor in multivariate analysis.

The results of the multivariate logistic regression analysis are presented in Table 3. In model 1, which included GEV requiring EIS as a factor, GEV was found to be a significant prognostic factor (aOR 33.8; p = 0.0033). Moreover, there was a significant difference between model 1 and a reduced model without any effects, except the intercepts (p = 0.0404).

Table 3 Multivariate analysis
Full size table

In model 2, which included hypersplenism as a factor instead of GEV, hypersplenism was found to not be a significant prognostic factor (p = 0.0575). Moreover, there was no significant difference between model 2 and a reduced model without any effects, except the intercepts (p = 0.2888).

Receiver operating characteristic curve analysis was performed for model 1, and the area under the curve was 0.848 (Fig. 1).

Fig. 1
figure 1

Receiver operating characteristic curve for 20-year native liver survival in model 1. The area under the curve (AUC) is 0.848

Full size image

In Kaplan–Meier analysis, a significant difference in the NLS rate was noted between patients with GEV requiring EIS and those without GEV requiring EIS (p = 0.0017; Fig. 2).

Fig. 2
figure 2

Kaplan-Meier native liver survival curve. A significant difference in the native liver survival rate is noted between patients with gastro-esophageal varices (GEV) requiring endoscopic injection sclerotherapy (EIS) and those without GEV requiring EIS (p = 0.0017)

Full size image

Discussion

NLS with good quality of life is ideal for BA patients. However, there is a risk of deterioration in the hepatic condition, and LTx is required in cases of decompensated cirrhosis [3].

In Japan, living donors are widely available, while deceased donors are rare [4]. Generally, more time is needed to prepare for LTx in older BA patients than in infants with BA. Additionally, it is sometimes more difficult to prepare a living donor for older patients than for younger patients. The identification of prognostic factor in patients with long-term NLS can aid in the preparation of LTx in patients with worsening hepatic function. Therefore, the identification of prognostic factor in BA patients, especially long-term survivors, is important.

Previous studies have reported on the prognostic factors in BA patients [5–7]. The age at the time of Kasai procedure, type of BA, and incidence of cholangitis were generally recognized as significant short-term prognostic factors. The long-term factors include late-onset cholangitis and portal hypertension; however, these have not been confirmed as significant long-term prognostic factors.

In a previous study, all patients were included for analysis [6]. However, jaundice disappearance after the Kasai procedure is necessary for long-term NLS. If all patients who underwent the Kasai procedure are analyzed to assess the long-term prognostic factors, the factors that influence jaundice disappearance or the short-term outcomes can also influence the long-term outcomes. Therefore, it is necessary to use appropriate inclusion criteria for assessing the long-term prognostic factors. In this study, patients with jaundice disappearance after the Kasai procedure and NLS for over 10 years were included.

We found that the existence of portal hypertension (GEV requiring EIS) was a significant long-term prognostic factor. EIS and interventions for hypersplenism do not deteriorate hepatic function, although these therapies have some risk of complications [2, 8]. Generally, most variceal bleeding episodes can be controlled with adequate endoscopic therapies in the short term. In this study, variceal bleeding was not responsible for any deaths, and only hepatic failure was responsible for the deaths among our patients.

On the other hand, intervention for hypersplenism was found not to be a significant prognostic factor in multivariate logistic analysis. Our previous report found that partial splenic arterial embolization for hypersplenism resulted in a good prognosis in the mid-term, and hypersplenism was believed not to influence the prognosis at 20 years, if managed adequately [8].

There might be different effects with regard to portal hypertension and hepatic function between EIS and interventions for hypersplenism. However, limited information is available, and further studies are needed to clarify the effects.

Our study findings indicated that patients with deterioration of hepatic function might be present among BA patients with long-term NLS who require EIS or interventions for hypersplenism. Therefore, the hepatic condition should be examined more carefully in patients with portal hypertensive symptoms than in those without portal hypertensive symptoms, and if required, LTx should be prepared at the most optimal time.

In our study, cholangitis was not a significant prognostic factor. Cholangitis was believed to result in the deterioration of hepatic function in BA patients. However, our data did not indicate that cholangitis had a significant impact on 20-year NLS. Patients with 20-year NLS might need LTx later, and cholangitis might be responsible for the hepatic condition at that time.

This study had some limitations. First, approximately 20 % of the patients who were candidates for this study could not be analyzed because of insufficient information. Second, our findings cannot be applied to patients with deterioration of hepatic function 20 years after surgery. We found that some patients could not live with their native liver for more than 20 years. A meticulous follow-up program is necessary for long-term native liver survivors of BA, and further studies with accumulation of long-term patients will help establish the long-term prognostic factors in these patients. Third, other factors, such as laboratory data and imaging findings were not evaluated in this study. These factors might be important as long-term prognostic factors in BA patients, and, therefore, they should be evaluated in future studies.

In conclusion, the existence of symptomatic portal hypertension would influence long-term NLS in BA patients after the Kasai procedure. In such patients, accurate evaluation of hepatic function and adequate treatment for sequelae are needed, and if deterioration of hepatic function is noted, LTx should be performed at the optimal time.