A nomogram for predicting early- and mid-term native liver survival after surgery in patients with type III biliary atresia: a single-center retrospective study

Abstract

Background

Biliary atresia (BA) is a severe obstructive biliary disease. This study aimed to explore the independent risk factors influencing the prognosis of BA in children who underwent Kasai portoenterostomy (KPE) and to construct a nomogram for predicting the prognosis.

Methods

The clinical data of patients with type III BA who underwent KPE at Guangzhou Women and Children’s Medical Center from June 2016 to May 2024 were retrospectively analyzed. Univariate and multivariate Cox regression analyses were used to screen for independent risk factors. A nomogram was constructed using the screened variables to predict postoperative early- and mid-term native liver survival (NLS). The predictive efficacy of the model was evaluated using the concordance index (C-index), receiver operating characteristic (ROC) curve and area under the curve (AUC), and calibration curves.

Results

Two hundred patients with type III BA whose clinical data were available were randomly divided into training and validation cohorts at a ratio of 7:3. Age at surgery, recurrent cholangitis, early bile drainage (EBD), and aspartate aminotransferase (AST) level within 1 month after surgery were identified as independent risk factors (all P < 0.05), and a nomogram was constructed using these factors. The C-index was 0.76 (95% CI: 0.71, 0.80) for the training cohort and 0.69 (95% CI: 0.57, 0.75) for the validation cohort. Excellent discrimination and calibration were observed for both groups.

Conclusion

The nomogram constructed using variables such as age at surgery, recurrent cholangitis, EBD, and postoperative AST level can accurately predict the probability of NLS after surgery, providing a quantifiable assessment tool for clinical individualized clinical management.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12893-026-03985-y.

Introduction

Biliary atresia (BA) is a severe hepatobiliary disease that occurs in neonates and is characterized by progressive inflammation and fibrosis of intrahepatic and extrahepatic bile ducts [1, 2]. The morbidity of BA is 1/5 000 ~ 1/10 000, and most untreated patients die of liver failure before the age of 2 years [2, 3]. Kasai portoenterostomy (KPE) is currently the preferred surgical procedure for patients with BA; KPE aims to restore bile drainage and relieve obstruction by removing the extrahepatic biliary system and reconstructing it with an intestinal loop connected to the porta hepatis region (Roux-en-Y anastomosis), thereby improving the prognosis [2]. The success of KPE is determined by the restoration of bile drainage and the clearance of jaundice. However, owing to persistent inflammation and fibrosis in the liver after KPE, approximately 70% of patients with BA eventually progress to cirrhosis and liver failure, ultimately requiring liver transplantation to sustain life [2, 4].

Currently, many factors are related to the prognosis of BA patients after KPE, including age at surgery, liver stiffness measurement (LSM), serum fibroblast growth factor 19 (FGF19) level, cytomegalovirus (CMV) infection status, aspartate aminotransferase (AST) level, early jaundice clearance and postoperative cholangitis [5–10]. However, these influencing factors have relatively limited use in clinical practice and cannot be used to visually predict the prognosis of BA. A nomogram is a graphical tool that can combine multiple clinical factors. It is advantageous because it allows the results of regression equations to be visualized so that an individual’s disease risk or survival probability can be calculated, thereby providing a more accurate prediction of disease prognosis [11]. Some scholars have constructed relevant nomogram models for jaundice clearance and native liver survival (NLS) after KPE [12–15]. However, the models or indicators used in previous studies have certain limitations in terms of availability and stability and can predict the prognosis for only up to 1 ~ 2 years after surgery. Currently, effective clinical prediction models for predicting early- and mid-term NLS rates after KPE are lacking.

The aims of this study were to investigate the independent risk factors affecting early- and mid-term NLS rates after KPE and to construct a nomogram that can be widely used. This model can be applied to predict early- and mid-term NLS rates after KPE in patients with BA, thereby providing a theoretical basis for clinical practice.

Materials and methods

Patient selection

The clinical case data of children with type Ⅲ BA who underwent KPE at Guangzhou Women and Children’s Medical Center, which is affiliated with Guangzhou Medical University, from June 2016 to May 2024 were retrospectively analyzed in this study. Laparoscopic intraoperative cholangiography was routinely performed in all patients. For those with severely fibrotic and obliterated gallbladders where contrast medium could not be administered, the final diagnosis was established based on laparoscopic macroscopic observations combined with postoperative pathological results. The inclusion criteria used in this study included the following: (1) all children who were diagnosed with type III BA by routine laparoscopic intraoperative cholangiography combined with laparoscopic biliary exploration and postoperative pathological examination; and (2) children who, with their families, cooperated with the doctors, took medicines regularly after the operation, had regular reexaminations as needed, and had complete follow-up date. The exclusion criteria were as follows: (1) children with type I or type II BA; and (2) patients with incomplete or missing follow-up data, or who refused to participate in the study. The Ethics Committee of Guangzhou Women and Children’s Medical Center, affiliated with Guangzhou Medical University, reviewed and approved this study (No. 309B01). All the patients’ families provided informed consent and signed informed consent forms in accordance with the principles of the Declaration of Helsinki.

Definitions of cholangitis and early bile drainage and the staging of liver fibrosis

In this study, cholangitis was defined as the presence of systemic inflammatory manifestations (fever ≥ 38.0 °C or elevated inflammatory markers levels such as white blood cell count and C-reactive protein levels), along with evidence of cholestasis (aggravated jaundice, lighter stool color or the reappearance of white stools, and elevated serum bilirubin levels). Cholangitis that occurred within 1 month after KPE was defined as early cholangitis, whereas cholangitis that occurred after 1 month was defined as late cholangitis [16]. Cholangitis occurring ≤ 3 times within 6 months after KPE was defined as occasional cholangitis, and cholangitis occurring > 3 times was defined as recurrent cholangitis. To better evaluate the regression of jaundice after KPE, the concept of early bile drainage (EBD) was introduced [17]. Early bile drainage-good (EBD-G) was defined as yellowing of the stool color, significant regression of the skin and scleral jaundice, and a one-third decrease in serum bilirubin levels compared with preoperative levels within 1 month after KPE. Conversely, early bile drainage-poor (EBD-P) was defined by the opposite characteristics. Liver fibrosis was assessed histologically by wedge biopsy during KPE and recorded in accordance with the METAVIR scoring system as follows: F0: no fibrosis; F1: portal fibrosis without septa; F2: portal fibrosis with few septa; F3: numerous septa or lobular distortion without cirrhosis; or F4: cirrhosis. When the liver was classified as F3 or F4, it was referred to as advanced hepatic [18].

Collection of clinical data and follow-up

The clinical data of the children, including sex, age at surgery, body weight, body mass index (BMI), aspartate aminotransferase-to-platelet ratio index (APRI), alanine aminotransferase (ALT) level, aspartate aminotransferase (AST) level, gamma-glutamyl transferase (GGT) level, total bile acid (TBA) concentration, and albumin (ALB)level, were collected before and within 1 month after KPE surgery. Follow-up was conducted through outpatient reexamination, telephone calls, and other methods. The follow-up endpoint was liver transplantation or death, and the native liver survival time after KPE surgery was calculated. Native liver survival time after KPE was defined as the time from the procedure to liver transplantation or death resulting from liver failure, hepatic decompensation or other liver-related adverse events. Early native liver survival was defined as ≤ 1 year postoperatively, and mid-term native liver survival as 1–5 years postoperatively. The follow-up deadline was December 2024.

  

Statistical analysis

IBM SPSS Statistics 25.0 and R (4.4.1) statistical software packages were used for data processing and analysis. The count data are expressed as frequencies and composition ratios, and intergroup comparisons were performed using the χ² test or Fisher’s exact test. The measurement data conforming to a normal distribution in the two groups are expressed as the mean ± standard deviation (SD), and intergroup comparisons were performed using the t test. The measurement data with skewed distributions are expressed as M (Q1, Q3), and intergroup comparisons were performed using the Mann‒Whitney U test. The survival curves were plotted using the Kaplan‒Meier method, and the log-rank test was used. A Cox proportional hazards regression model was used for univariate and multivariate survival analyses, and the selected variables were used to create a nomogram for predicting 1-year, 2-year, 3-year, and 5-year probabilities of NLS rates after KPE. The calibration curves generated using the “rms” and “survival” R software packages were used to evaluate the discrimination and accuracy of the model. The optimal cutoff values of continuous risk factors were determined by the maximally selected log-rank statistic. A P value < 0.05 was considered to indicate statistical significance.

Results

Study population enrollment and baseline characteristics

Patients with BA were screened at the Women and Children’s Medical Center affiliated with Guangzhou Medical University from June 2016 to May 2024. Finally, according to the inclusion and exclusion criteria, 200 patients with type III BA who underwent KPE were included and randomly divided into a training cohort (n = 140) and a validation cohort (n = 60) at a ratio of 7:3 (Fig. 1). Until the end of follow-up, the median native liver survival time was 16.00 months (95% CI: 7.77, 24.23). The median follow-up time was 48.00 months (95% CI: 37.17, 58.83) for the training cohort and 48.00 months (95% CI: 23.27, 72.73) for the validation cohort. The training cohort comprised of 140 patients with BA, including 76 males and 64 females, with a median age at surgery of 67.00 days (56.00, 78.75). The validation cohort consisted of 60 patients with BA, including 26 males and 34 females, with a median age at surgery of 66.00 days (56.00, 75.25). There were no statistically significant differences in sex, age at surgery, weight, BMI, or APRI between the two groups (P > 0.05); moreover, there were no statistically significant differences in ALT, AST, GGT, TBA, or ALB levels before KPE or within 1 month after surgery (P > 0.05) (Supplementary Table 1). There were no statistically significant differences in baseline data between the two cohorts, which ensures the reliability of subsequent model construction and validation and avoids potential bias caused by uneven baseline characteristics.

Fig. 1.

Flow chart of patient selection for training cohort and validation cohort of the study

Construction of the prognostic prediction nomogram for the training cohort

Univariate Cox regression analysis of the training cohort revealed that age at surgery, recurrent cholangitis, EBD, preoperative ALT (Pre-ALT) level, postoperative AST (Post-AST) level, postoperative GGT (Post-GGT) level, and postoperative TBA (Post-TBA) level were significantly associated with native liver survival in patients with BA (all P < 0.05) (Fig. 2A). Multivariate Cox regression analysis revealed the following variables as independent prognostic factors: age at surgery (HR 1.013; 95% CI 1.001–1.025; P = 0.040), recurrent cholangitis (HR 2.093; 95% CI 1.252–3.497; P = 0.005), EBD-P (HR 2.597; 95% CI 1.610–4.190; P < 0.001), and post-AST concentration (per 10-unit increase; HR 1.025; 95% CI 1.006–1.044; P = 0.010) (Fig. 2B).

Fig. 2.

Forest plot of Cox regression analysis for predictors of the prognosis of BA children. EBD-G, early bile drainage-good; EBD-P, early bile drainage-poor; KPE, Kasai portoenterostomy; BMI, body mass index; APRI, aspartate aminotransferase-to-platelet ratio; Advanced hepatic fibrosis, F3 or F4; Pre, Pre-operative level; Post, 1month postoperative level; ALT, alanine transaminase; AST, aspartate transaminase; GGT, gamma-glutamyl transferase; TBA, total bile acid; Alb, albumin. ALT、AST、GGT、TBA、ALB were treated as a continuous variable with per 10-unit increase

A nomogram was constructed based on the results of the multivariate Cox regression analyses to predict the probabilities of 1-, 2-, 3-, and 5-year native liver survival in patients with BA after KPE (Fig. 3A). According to the hazard ratios derived from the multivariate Cox model, each predictor was assigned a corresponding weight value. The native liver survival probability for patients with BA could be predicted by summing the scores of all the predictors. For example, consider a patient with BA who underwent KPE at 70 days of age (scoring 23 points), had an AST concentration of 130 U/L within 1 month after surgery (scoring 13 points), experienced good early bile drainage (scoring 0 points), and experienced fewer than 3 episodes of cholangitis within 6 months after surgery (scoring 0 points). This patient would have a total score of approximately 36 points, indicating that the 5-year native liver survival probability after surgery is approximately 60%.

Fig. 3.

Nomogram and the receiver operating characteristic (ROC) curve for predicting NLS after KPE in patients with biliary atresia. A Nomogram for predicting the probability of NLS in BA children within 5 years post-KPE; B The receiver operating characteristic (ROC) curve of the nomogram in the training cohort; C The receiver operating characteristic (ROC) curve of the nomogram in the validation cohort. Post-AST, Aspartate transaminase level at 1 month postoperatively, was treated as a continuous variable with per 10-unit increase

Evaluation and validation of the nomogram model

The nomogram model was assessed and validated using multiple methods. The C-index for the training cohort was 0.76 (95% CI: 0.71, 0.80), and the C-index for the validation cohort was 0.69 (95% CI: 0.57, 0.75), indicating good agreement with actual native liver survival. The AUC values for 1, 2, 3, and 5 years after KPE for the training cohort were 0.81 (95% CI: 0.74, 0.89), 0.83 (95% CI: 0.75, 0.90), 0.81 (95% CI: 0.73, 0.89), and 0.83 (95% CI: 0.73, 0.92), respectively, and the corresponding the AUC values for the validation cohort were 0.74 (95% CI: 0.61, 0.88), 0.77 (95% CI: 0.63, 0.91), 0.75 (95% CI: 0.60, 0.90), and 0.84 (95% CI: 0.71, 0.96), respectively (Fig. 3B and C). The calibration curves for 1-, 2-, 3-, and 5-year native liver survival probability after KPE in patients with BA were highly consistent between the actual survival status and the predicted probabilities for the training and validation cohorts (Fig. 4). The nomogram exhibited good predictive accuracy, high discrimination, and reliability for the training and validation cohorts.

Fig. 4.

Calibration curves of the training cohort and validation cohort of BA patients. A-D Calibration curves for 1-, 2-, 3-, and 5-year native liver survival in the training cohort of infants with BA who underwent KPE; E-H Calibration curves for 1-, 2-, 3-, and 5-year native liver survival in the validation cohort of infants with BA who underwent KPE

Construction of the risk score model

The 6-month, 1-year, 2-year, 3-year, and 5-year native liver survival rates of patients with BA after KPE were 69.5%, 53.0%, 44.2%, 42.0%, and 37.8%, respectively (Fig. 5A). Consistent with the results of the multivariate Cox regression analyses, the results of the Kaplan‒Meier survival analysis also revealed that the above variables (age at surgery, post-AST concentration, recurrent cholangitis, and EBD) were significantly associated with native liver survival in patients with BA (all P < 0.05) (Fig. 5B, C, D, and E). According to the Cox proportional hazards model, the survival risk score for each sample was calculated as follows:

The optimal cutoff value was determined by the survminer package based on the maximal selected rank statistics method; those with a risk score ≤ 0.165 were classified into the low-risk group, and those with a score > 0.165 were classified into the high-risk group. The results indicated that the native liver survival time of low-risk patients with BA was significantly longer than that of high-risk patients in both the training and the validation cohorts, and the difference was statistically significant (all P < 0.0001) (Fig. 5F, G).

Fig. 5.

Kaplan–Meier survival curves of NLS. A Overall native liver survival rate curve; B Age; C Post-AST; D Cholangitis; E EBD; F Based on risk scores of the nomogram in the training cohort; G Based on risk scores of the nomogram in the validation cohort

Discussion

Kasai portoenterostomy is the preferred surgical approach for treating BA and can effectively improve the quality of life and prolong NLS in patients. Nevertheless, approximately 70% of children with BA eventually require liver transplantation to sustain life after KPE because of the continuous progression of the disease or other factors [2, 4]. Therefore, the risk of death or liver transplantation after KPE changes significantly over time. Early and accurate prediction of the NLS rate after KPE, timely assessment of liver transplantation indicators, and personalized clinical decision-making are important for patients, families, and the society. Our study revealed that age at surgery, recurrent cholangitis, early bile drainage, and aspartate aminotransferase level within 1 month after surgery were independent risk factors for the prognosis of patients with BA. Using the above factors, we constructed a nomogram to predict early- and mid-term NLS probabilities for patients with BA. This nomogram can be used to successfully classify patients into high-risk and low-risk NLS groups. All the predictive indicators included in the model can be easily obtained through routine clinical evaluation. Verified by various statistical methods, this model shows good predictive efficacy.

Postoperative recovery of liver function and jaundice clearance are important indicators for evaluating the success of KPE [19]. The success of KPE can be determined by the complete clearance of jaundice, which is conducive to obtaining a good NLS outcome for children with BA [10, 19]. In our study, the Cox regression analysis revealed that early bile drainage carried the greatest weight, suggesting the importance of early bile drainage in the nomogram model. Early bile drainage-poor after KPE is likely due to the small diameter of the hilar bile duct and the advanced-stage histological features of severe hepatocellular injury [20]. Shortening the time to clear jaundice after KPE is highly important for reducing the risk of liver injury and slowing the development of hepatic fibrosis, thereby leading to a better NLS outcome [19, 21, 22]. AST is an enzyme present in hepatocytes. When hepatocytes are severely damaged, AST is released from the mitochondria, and its increased concentration reflects the degree of hepatocellular injury. Lu et al. [23] reported that early postoperative AST levels were an influential factor in the prognosis of children with BA. The results of our study indicated that the higher the AST level was 1 month after KPE, the lower the probability of NLS. A possible reason is that the total bilirubin remains at a high level within 1 month, causing continuous cholestatic liver damage, which leads to a decrease in the NLS rate for children with BA [19].

Age at surgery is considered an important factor influencing the prognosis of KPE [24–26]. KPE within 60 days after birth is considered beneficial for achieving an optimal prognosis, whereas patients who undergo KPE beyond 60 days after birth have a significantly reduced likelihood of achieving long-term NLS [27]. Moreover, a 20-year long-term follow-up revealed that compared with other ages (31–60 days, 61–90 days, and ≥ 91 days), KPE within 30 days after birth was associated with higher NLS [28]. In our study, we found that the optimal cutoff value for age at surgery was 62 days; KPE performed at or before this age was beneficial for improving NLS in infants with BA, whereas KPE after this age may increase the risk of death or the need for a liver transplant. However, Apfeld et al. [29] controlled for patients’ clinical, demographic, and socioeconomic characteristics, as well as institutional factors, and reported significant differences in the age at which KPE was performed across different hospitals, with more than 45% of children undergoing KPE after 60 days of age. Therefore, improvements in screening and diagnostic methods can improve the prognosis of BA.

The occurrence of cholangitis significantly affects the prognosis of BA patients [30]. Cholangitis is the most common complication after KPE and occurs in approximately 60% of patients with BA within 6 months after surgery [31]. Recurrent episodes of postoperative cholangitis are associated with a poor prognosis of BA [32]. Moreover, Zhang et al. [12] reported that an increase in postoperative cholangitis episodes is a predictor of a poor prognosis after KPE. Liu et al. [16] reported that if early cholangitis is not treated in time, it can develop into recurrent cholangitis, thereby reducing the probability of NLS. The reason may be that cholangitis can damage the bile ducts, affect bile drainage, and lead to cholestasis, thereby accelerating liver injury and fibrosis. Higher the number of times a patient experiences cholangitis, the more severe is the liver injury, and the lower is their NLS probability. The results of our study are consistent with the above findings, and the NLS rate for children with recurrent cholangitis was significantly lower than that for children with occasional cholangitis.

Some scholars have recently constructed nomograms using CMV IgM-positive status, liver stiffness values, and liver gene expression status for the early prediction of prognosis of BA [12–15]. However, these models have certain limitations: on the one hand, some predictors are difficult to obtain in clinical practice, which limits their practical application; on the other hand, some studies use the occurrence of cholangitis or jaundice clearance as outcome indicators and cannot directly reflect NLS in children with BA after KPE. In addition, there is currently a lack of nomograms for predicting 5-year NLS after KPE. Therefore, we constructed a nomogram based on indicators before KPE and within 6 months after KPE to predict the 1-, 2-, 3-, and 5-year NLS probabilities after KPE in children with BA. The nomogram constructed in this study has significant advantages: all the predictors included can be conveniently obtained through routine investigation, which is conducive to the widespread roll-out and application of this nomogram in clinical practice. Moreover, our constructed nomogram demonstrated strong discrimination and calibration, providing theoretical advantages for early intervention after KPE in children with BA.

However, this study has several limitations: it is retrospective and single-center. In addition, the developed nomogram requires clinical data on cholangitis at 6 months postoperatively, restricting ultra-early prediction of early- and mid-term NLS. Moreover, the risk score cutoff of 0.165 is data-driven and population-specific, which may not be fully reproducible in external cohorts. For this reason, this threshold should be interpreted with caution in different patient populations. Notably, the model can be applied as a continuous risk prediction tool to reduce reliance on the fixed cutoff. Future studies should adopt prospective, multicenter designs to address the above limitations.

Conclusions

In conclusion, the multivariate Cox regression analysis revealed that age at surgery, recurrent cholangitis, early bile drainage, and aspartate aminotransferase levels within 1 month postoperatively were independent risk factors affecting the probability of NLS in children with BA after KPE. A nomogram constructed based on these factors can effectively predict early- and mid-term NLS probabilities in children with BA.

Abbreviations

ALT

Alanine aminotransferase

AST

Aspartate aminotransferase

ALB

Albumin

APRI

Aspartate aminotransferase-to-platelet ratio index

AUC

Area under the curve

BA

Biliary atresia

BMI

Body mass index

C-index

Concordance index

CMV

Cytomegalovirus

EBD

Early bile drainage

EBD-G

Early bile drainage-good

EBD-P

Early bile drainage-poor

FGF19

Serum fibroblast growth factor 19

GGT

Gamma-glutamyl transferase

KPE

Kasai portoenterostomy

LSM

Liver stiffness measurement

NLS

Native liver survival

ROC

Receiver operating characteristic

TBA

Total bile acid

Authors' contributions

S L: Study design. H Z, Y L, Z L, Z Z, L W and S G: Data collection and analysis. S L, B T, Z Y and M L: Statistics. S L, F L, J Z and X X: Writing of the manuscript. All authors read and approved the final manuscript.

Funding

This work was supported by the grant of National Natural Science Foundation of China [grant number 82170528], Natural Science Foundation of Guangdong Province [grant number 2022A1515012254], Municipality and University (Institute) Joint Funding Project of Guangzhou Science and Technology Plan [grant number 2024A03J1236], Guangzhou Clinical Characteristic Technology Project (No. 2023 C-TS48) and Guangzhou Municipal Health Science and Technology Project [grant number 20231A011040].

Data availability

The datasets generated and/or analysed during the current study are not publicly available due to patient privacy protection but are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The Ethics Committee of the Guangzhou Women and Children’s Medical Center, affiliated with Guangzhou Medical University, reviewed and approved this study(No. 309B01). The participant’s legal guardian or close relatives provided written informed consent to participate in this study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.