Journal Logo

Original Clinical Science—Liver

Living Donor Liver Transplantation in Children: Perioperative Risk Factors and a Nomogram for Prediction of Survival

Lu, Yu-Gang MD1,2; Pan, Zhi-Ying MD2; Zhang, Song MD2; Lu, Ye-Feng MD3; Zhang, Wei PhD4,5; Wang, Long MD6; Meng, Xiao-Yan MD7; Yu, Wei-Feng MD, PhD2

Author Information
doi: 10.1097/TP.0000000000003056



Living donor liver transplantation (LDLT) for children was first introduced to China in June 1997 with the help of Professor Tanaka from Kyoto University. To date, >1500 pediatric LDLT operations have been performed in mainland China. Benefiting from the innovation and development of surgical techniques, immunosuppression, and perioperative management during the past few decades, pediatric LDLT has achieved promising short- and long-term outcomes. Nowadays, LDLT has become the mainstream method for pediatric LT in China.

Over the past 3 decades, there have been many studies reporting outcomes of LDLT.1-7 However, only a few large-scale studies analyzed the prognostic factors for outcomes after LT. Factors that have been repeatedly emphasized include Pediatric End-stage Liver Disease (PELD) score,5,8 graft-to-recipient weight ratio (GRWR),6 bodyweight,7 and other factors including renal function, warm ischemia time, graft type, and ABO incompatibility were also reported to be associated with mortality for pediatric LT. Notably, most of these studies were carried out over a decade ago and included various types of transplantation procedures. Thus, some of these potential perioperative prognostic factors need further validation, and a scoring system for accurate prediction of the possible mortality risk is needed.

The aims of this retrospective study were to go over our experience in pediatric LDLT in recent years, to analyze perioperative prognostic factors affecting morbidity and mortality following LDLT in children, and to build a novel scoring system to predict the risk of postoperative death. Furthermore, these results may provide clinicians with valuable references to lower postoperative mortality of pediatric LDLT recipients, aiming to prolong patient survival.


Study Design

We analyzed the medical data for all living donors and recipients who underwent primary LDLT surgery from January 1, 2014, to December 31, 2016, at Renji Hospital, Shanghai, China. Clinical follow-up was applied in each month in the first half-year after surgery, and every 3 months after that, the last follow-up record was up to June 2018. This study was approved by the Renji Hospital Ethics Committee (Approval number: [2018] 218).

Patients were enrolled in this cohort if they (1) were under 12 years old (pediatric patients); (2) underwent LDLT for end-stage liver disease (ESLD); or (3) had complete clinical records of the peritransplantation procedure. Patients were excluded if they (1) underwent LT for acute liver failure and cancer; (2) underwent combined organ transplantation, heterotopic transplantation, and retransplantation; or (3) had incomplete clinical and instrumental follow-up.

The collective data of all recipients were obtained and presented. For each patient, the baseline characters were screened: gender, age, height, weight, diagnosis for primary liver disease, neutrophil lymphocyte ratio (NLR), platelet lymphocyte ratio, PELD score, white blood cell, donor’s age, gender, donor-recipient blood group compatibility (ABO-identical/compatible/incompatible), body mass index, implementation of Kasai procedure. Intraoperative factors included operation time, anhepatic phase, blood loss, fluid infusion volume, blood transfusion, GRWR, and norepinephrine (NE) application. Postoperative outcomes were postoperative hemoglobin, blood glucose level, lactic acid level, central venous pressure, and hospital stay-time.

Immunosuppressive Protocol

Tacrolimus (TAC) and prednisone were used as the principal immunosuppressive regimen. TAC was administered orally twice a day with an initial dose of 0.15 mg/kg/d, and subsequently, the target blood concentration was adjusted to 10–15 ng/mL for the first month, 8–12 ng/mL for the following 5 months, and 5–8 ng/mL thereafter. Methylprednisolone was intravenously administrated during the anhepatic phase and on the first day after transplantation at a dose of 5–10 mg/kg, then the dose was gradually decreased to 0.5–1 mg/kg/d at a rate of 0.5–1 mg/kg/d. Finally, the dose of methylprednisolone was maintained at 2.5–5 mg/d for 3 months. Generally, repeat or multiple postoperative infections was considered as over-immunosuppressive, whereas an acute cellular rejection was considered as under-immunosuppressive, the dose of TAC would be adjusted accordingly.

Statistical Analysis

Normally distributed data were represented as mean ± SD, non-normality data were represented as median and the interquartile range (IQR). Categorical data were presented as number and ratio. Cox regression analyses were used to explore the independent factors associated with the mortality of pediatric LDLT patients. Variables with P < 0.1 in the univariate analysis were then included in the multivariable model, using the iterative process of backward selection. Nonsignificant variables (P > 0.05 on likelihood ratio test) were removed in a stepwise procedure, and model fitting was then evaluated using the Akaike information criterion (AIC), as the model with the smallest AIC was selected as the final model. A coefficient-based nomogram was developed in predicting postoperative mortality; a calibration curve was performed and presented for internal verification. Time-dependent area under the curve (AUC)9 and decision curve analysis10 were used to determine the clinical net benefit associated with the use of the novel model in comparison to the PELD model. The mortality rates of indicators with different stratification were also presented by Kaplan–Meier curve; difference was tested by log-rank test. The significance level was set as P < 0.05 at 2 tails. All statistical analyses were performed using the SPSS version 25.0 software program (IBM SPSS, Chicago, IL) and R language version 3.5.2 (Foundation for Statistical Computing, China), the package of “readxl,” “party,” “stdca,” “survival,” “rms,” and “timeROC” were used.


Recipient Characteristics and Outcomes

Between the years 2014 and 2016, 435 pediatric LDLTs were performed in our hospital. We excluded 5 patients, 2 of whom underwent LDLT for cancer and 3 of whom had no complete follow-up data. The characteristics and part of the surgical outcomes of the resultant 430 recipients are summarized in Table 1. There were 189 female (43.9%) and 241 male (56.1%) recipients with a median (IQR) age of 7 (6, 10) months, height of 70.4 ± 13.5 cm (mean ± SD), and bodyweight of 8.5 ± 4.1 kg (mean ± SD). Biliary atresia was the most common reason for pediatric LDLT in our cohort (391 patients, 90.9%), and the inherited metabolic diseases ranked as the second cause (with 34 patients, 7.9%). The PELD score for them was 19 (12, 24) (median [IQR]). The preoperative NLR of recipients was 0.9 (0.6, 1.3) (median [IQR]). The majority of donors were male relatives (269 in 430, 62.6%), with a median (IQR) age of 29 (26, 34) and body mass index of 22.2 ± 3.1 (mean ± SD). Sixty-four (14.9%) recipients received ABO-incompatible grafts. The operating time of these LDLT was 6.8 hours (6.0, 7.7) (median [IQR]). GRWR for them was 3.27% (2.71%, 3.93%) (median [IQR]). NE was applied during surgery in 85 (19.8%) recipients.

Recipient characteristics and surgical outcomes

The median follow-up time was 887 days (658, 1108) (median [IQR]). The Kaplan–Meier curve of overall survival for the recipients is presented in Figure 1. Thirty-seven death events happened throughout the follow-up period, with 36 (97.3%) dying within a half-year after transplantation; 1 child survived 439 days before graft failure. Overall survival of our patients at 1 and 2 years was 91.6% and 91.4%, respectively. The causes of recipients’ deaths are summarized in Table 2. Infection and organ failure were the main causes of death among pediatric LDLT recipients after surgery.

Cause of recipient’s death
Kaplan–Meier survival curve of 430 pediatric LDLT recipients in this cohort. LDLT, living donor liver transplantation.

Prognostic Factors for 180-day Overall Survival

Because death events mainly occurred in the first half of the year after LDLT, we carried out the subsequent analysis based on the 180-day follow-up data. All the factors shown in Table 1 were analyzed for overall survival of recipients. The result of the univariate analysis is shown in Table 3. According to the univariate analysis, 7 possible risk factors were identified, including preoperative PELD score and NLR, operation time, blood loss, GRWR, intraoperative NE infusion, and the level of hemoglobin after the surgery. Kaplan–Meier curves of 180-day overall survival in pediatric LDLT recipients for all the 7 variables in the unadjusted model are shown in Figure 2. Significant differences were found in PELD score (P = 0.003), NE infusion (P < 0.001), GRWR (P = 0.033), NLR (P = 0.036), and blood loss (P = 0.030, log-rank test for trend) in log-rank test (continuous variables of PELD score and NLR were transferred as categorical variables based on clinical experiences; Table 3). For the other 2 variables, different trends across groups also existed as all P < 0.1.

Results of univariate Cox regression analysis for recipient’s survival
Kaplan–Meier survival curves of 180-d overall survival in pediatric LDLT recipients. Overall survival was significantly worse in recipients with preoperative PELD score > 15, P = 0.005 (A), in recipients with GRWR > 4%, P = 0.033 (B), in recipients with preoperative NLR ≥2, P = 0.033 (C), and in recipients with intraoperative application of NE, P < 0.001 (D), whereas operation time (E) and postoperative Hb levels (F) did not make any difference in 180-d overall survival of recipients. Overall survival was significantly worse in recipients with intraoperative blood loss >400 mL, P = 0.03 (G). GRWR, graft-to-recipient weight ratio; Hb, hemoglobin; LDLT, living donor liver transplantation; NE, norepinephrine infusion; NLR, neutrophil lymphocyte ratio; PELD, pediatric end-stage liver disease score.

Factors with P < 0.1 in the univariate analysis were then included in the multivariate analysis. The result of multivariate Cox regression analysis for recipients’ survival is shown in Table 4. By using the iterative process of backward selection, nonsignificant variables (P > 0.05 on likelihood ratio test) were removed in a stepwise procedure. Three models were obtained after that. The final model was selected as the one with the smallest AIC (AIC = 411.06; Table 4); the other 2 models with relatively large AIC and small C index are presented in Table S1 (SDC, The PELD score, GRWR, NLR, and intraoperative NE infusion were found to be significant predictors of the overall survival (Table 4).

Results of multivariate Cox regression analysis for recipients’ survival

Development of a Nomogram for 180-day Overall Survival

The PGNN (named according to the initials of PELD score, GRWR, NLR, and NE infusion) nomogram for estimating 90-, 180-day overall survival is shown in Figure 3, while 90- and 180-day calibration plots are shown in Figure 4A and B, respectively, as predicted survival in our model is generally coincident with actual survival. Decision curves for the overall survival model are also shown in Figure 4Cand D. The plots show that PGNN model–based decisions are more beneficial than PELD score system for patients in the range of threshold probabilities of about 5%–50% at both 90 and 180 days. The time-dependent AUC analysis was then used to determine if the novel PGNN nomogram was better at predicting survival than PELD score alone (Figure 4E). The median time-dependent AUC with a range for the PGNN nomogram and PELD score is 0.797 (0.763, 0.829) and 0.656 (0.649, 0.669), respectively. These results suggest that our final model is reliable, and the predictive capacity of the PGNN model is higher than PELD score.

Novel nomogram (PGNN nomogram, named according to the initials of PELD score, GRWR, NLR, and NE infusion) predicting the 90- and 180-d overall survival of pediatric LDLT recipients. GRWR, graft-to-recipient weight ratio; NE, norepinephrine infusion; NLR, neutrophil lymphocyte ratio; OS, overall survival; PELD, pediatric end-stage liver disease score.
Internal validation for the PGNN nomogram (named according to the initials of PELD score, GRWR, NLR, and NE infusion). A and B, Calibration plots for the validation sample of the PGNN nomogram, estimated at 90 and 180 d. The average predicted probability (nomogram-predicted overall survival; x-axis) was plotted against Kaplan–Meier estimate (observed overall survival; y-axis). 95% confidence intervals of the Kaplan–Meier estimates are indicated with vertical lines. C and D, Decision curves for overall survival at 90 and 180 d applied to the PGNN nomogram. E, Time-dependent AUC curve for overall survival applied to the PGNN nomogram. AUC, area under the curve; PELD, pediatric end-stage liver disease score; PGNN, named according to the initials of PELD score, GRWR (graft-to-recipient weight ratio), NLR (neutrophil lymphocyte ratio), and NE (norepinephrine) infusion; OS, overall survival.


In this study, we reported our experience with 430 pediatric LDLT recipients at a Chinese hospital and demonstrated that high preoperative PELD score (>15) and NLR (≥2), GRWR > 4%, and intraoperative NE infusion were perioperative-independent risk factors associated with postoperative mortality of pediatric LDLT recipients. We also established a novel PGNN nomogram for estimating the 6-month overall survival of these recipients. Postoperative infection and organ failure associated with infection and surgery were the main causes of death.

The PELD score was developed to prioritize pediatric LT candidates based on the risk of 90-day pretransplant death.11 Although a recent study has shown that PELD score significantly underestimated actual 90-day mortality,12 it still plays a crucial role in adjudicating allocation decisions. A retrospective analysis done by Bourdeaux et al13 suggested that high PELD scores were not associated with post-LT outcomes, while a recent study demonstrated that high PELD score was an independent risk factor for post-transplant morbidity but not for mortality.14 And another analysis done by Barshes et al8 concluded that pediatric patients with a PELD score of 17+ derived survival benefit early after LT, and increasing PELD scores were associated with increasing transplantation benefits after LT. In our study, we found that PELD score (>15) was an independent risk factor for mortality of pediatric LDLT recipients, with an adjusted hazard ratio of PELD 1.04 (1.01–1.07; Table 4). We deem that PELD score does correlate with post-transplant survival, but whether it is suitable as a predictive factor needs more convincing evidence.

Liver graft size matching is a major factor determining a successful outcome of LDLT. In this study, recipients with GRWR > 4% were at a significantly increased risk of poor outcome after LDLT. This finding is consistent with the results of previous studies,15-17 which reported that GRWR > 4% was an independent predictor of 30-day mortality, and large grafts in pediatric LDLT were associated with vascular complications and parenchymal necrosis due to insufficient blood supply to the grafts. And there are also studies showing that oversized grafts can increase intra-abdominal pressure after the LT surgery, resulting in decreased blood supply from the hepatic artery or portal vein and increased thrombosis events.18,19 On the contrary, LT using small-for-size grafts is implicated in an even more complicated scenario, where reperfusion injuries and immunological insults can also interfere with the fate of grafts. Therefore, it is important to evaluate the size of the graft (the part to be donated) by preoperative imaging data. And a reduction to the graft segment is highly recommended for pediatric LDLT with expected GRWR > 4%.

In the present study, we found that preoperative NLR ≥ 2 was an independent risk factor for the survival of pediatric LDLT recipients. The rationale for the NLR is to compare the inflammatory response (ie, neutrophils) to the host immunity (ie, lymphocytes). Excessively elevated NLR before transplantation represents systemic inflammatory immune disorders in these patients. Our findings suggested that preoperative inflammation-immune levels have an important impact on the outcome of pediatric LDLT. Recently, an elevated NLR has been reported as a prognostic indicator of poor overall survival in patients with a variety of solid cancers. Some of these patients received immune checkpoint inhibitors (targeting the cytotoxic T-lymphocyte–associated protein-4 and programmed death-1 receptors) treatment, which was similar to the post-transplant immunosuppressive strategy.20,21 Studies among cirrhotic patients have also shown that patients with NLR > 2.72 had significantly lower survival,22 NLR was an independent predictor of 3-month mortality in acute-on-chronic liver failure patients,23 and NLR was associated with the severity of cirrhosis.24 Our results were similar to these previous findings. Therefore, the pediatric LDLT recipients might benefit from early intervention of the systemic inflammatory response before transplantation.

In this study, we also found that intraoperative NE infusion was associated with postoperative mortality of pediatric LDLT recipients. NE was usually used to minimize transfusion volume and fluid administration during the preanhepatic and anhepatic phase and to reduce the occurrence of postreperfusion syndrome during the neohepatic phase of LT. However, intraoperative infusion of NE to improve hemodynamics might not be a wise choice for recipients with ESLD undergoing LDLT surgery. It has been reported that the features of the cirrhotic cardiomyopathy were observed in <70% of infants with ESLD listed for LT.25 Recent findings suggest that elevated bile acids, by acting on specific receptors, affect the metabolism and function of cardiomyocytes, vascular endothelial cells, and smooth muscle cells, and reduce the β-adrenergic receptor density and membrane fluidity in the myocardium.26,27 For these ESLD patients, NE might further increase cardiac load and negatively affect the prognosis.

It should be noted that although we had eliminated as many confounding factors as possible, some surgical factors such as major blood loss, prolonged surgery, reperfusion syndrome, etc., should not be ignored. So, regarding this point, intraoperative NE infusion should be considered as a surrogate composite end-point for multiple intraoperative risk factors in the final mortality equation. Our study was the first to question the safety of the NE infusion in pediatric LT patients; however, there is no doubt that follow-up studies on this subject are urgently needed.

Based on the analysis of risk factors, we also established a novel PGNN nomogram for estimating the 6-month overall survival of the LDLT recipients. The C index of this model was 0.764. Compared with the PELD system, this model indicated a higher net benefit. The most remarkable feature of this model is that it not only considers preoperative factors but also adds some intraoperative variables that may potentially have an impact on prognosis. Our findings support the use of a more comprehensive scoring system to assess the prognosis of pediatric LDLT recipients. Undoubtedly, the practicability and clinical value of this model needs to be further demonstrated by more comprehensive and systematic clinical trials.

There are several limitations in our study. First, this is a single-center retrospective study, with potential selection biases including race, indications for surgery, and surgical technique. Second, the relatively low mortality is favorable, but it may also limit the universality of our conclusions. Third, as we only applied internal validation, formal external validation is necessary before implementation of our nomogram into clinical practice. In addition, some other risk factors for poor outcomes might have been overlooked due to unavailable data, including preoperative echocardiography, plasma B-type natriuretic peptide levels, and pre-existing renal dysfunction, etc. Our study did not consider the analysis of strict adherence to immunosuppression medication and medical follow-up, but it is important to note that these factors have a great impact on long-term survival.


Our study found that preoperative high PELD score, high NLR, and high GRWR were independent perioperative risk factors for mortality of pediatric LDLT recipients, and intraoperative NE infusion was associated with poor overall survival in children after LDLT. In addition, we developed a novel PGNN nomogram for estimating the 6-month overall survival of these patients. Internal validation of the present model exhibited satisfactory performance characteristics. Further research is needed to validate these factors and to verify the practicality of the PGNN model.


The authors thank Dr Ryan Dowsell from the Imperial College London for editing.


1. Goss JA, Shackleton CR, Swenson K, et al. Orthotopic liver transplantation for congenital biliary atresia. An 11-year, single-center experience. Ann Surg. 1996; 224:276–284
2. Chen CL, Concejero A, Wang CC, et al. Living donor liver transplantation for biliary atresia: a single-center experience with first 100 cases. Am J Transplant. 2006; 6:2672–2679
3. Neto JS, Feier FH, Bierrenbach AL, et al. Impact of Kasai portoenterostomy on liver transplantation outcomes: a retrospective cohort study of 347 children with biliary atresia. Liver Transpl. 2015; 21:922–927
4. Fouquet V, Alves A, Branchereau S, et al. Long-term outcome of pediatric liver transplantation for biliary atresia: a 10-year follow-up in a single center. Liver Transpl. 2005; 11:152–160
5. Rhu J, Jung SM, Choe YH, et al. PELD score and age as a prognostic index of biliary atresia patients undergoing Kasai portoenterostomy. Pediatr Surg Int. 2012; 28:385–391
6. Li JJ, Zu CH, Li SP, et al. Effect of graft size matching on pediatric living-donor liver transplantation at a single center. Clin Transplant. 2018; 32
7. Kasahara M, Fukuda A, Yokoyama S, et al. Living donor liver transplantation with hyperreduced left lateral segments. J Pediatr Surg. 2008; 43:1575–1578
8. Barshes NR, Lee TC, Udell IW, et al. The pediatric end-stage liver disease (PELD) model as a predictor of survival benefit and posttransplant survival in pediatric liver transplant recipients. Liver Transpl. 2006; 12:475–480
9. Kamarudin AN, Cox T, Kolamunnage-Dona R. Time-dependent ROC curve analysis in medical research: current methods and applications. BMC Med Res Methodol. 2017; 17:53
10. Van Calster B, Wynants L, Verbeek JFM, et al. Reporting and interpreting decision curve analysis: a guide for investigators. Eur Urol. 2018; 74:796–804
11. McDiarmid SV, Anand R, Lindblad AS; Principal Investigators and Institutions of the Studies of Pediatric Liver Transplantation (SPLIT) Research GroupDevelopment of a pediatric end-stage liver disease score to predict poor outcome in children awaiting liver transplantation. Transplantation. 2002; 74:173–181
12. Chang CH, Bryce CL, Shneider BL, et al. Accuracy of the pediatric end-stage liver disease score in estimating pretransplant mortality among pediatric liver transplant candidates. JAMA Pediatr. 2018; 172:1070–1077
13. Bourdeaux C, Tri TT, Gras J, et al. PELD score and posttransplant outcome in pediatric liver transplantation: a retrospective study of 100 recipients. Transplantation. 2005; 79:1273–1276
14. Kitajima T, Sakamoto S, Sasaki K, et al. Living donor liver transplantation for post-Kasai biliary atresia: analysis of pretransplant predictors of outcomes in infants. Liver Transpl. 2017; 23:1199–1209
15. Cho CW, Lee S, Kim JM, et al. Independent factors predicting postoperative 30-day mortality in 101 infants following liver transplantation. Ann Transplant. 2017; 22:631–637
16. Akdur A, Kirnap M, Ozcay F, et al. Large-for-size liver transplant: a single-center experience. Exp Clin Transplant. 2015; 13Suppl 1)108–110
17. Shirouzu Y, Kasahara M, Morioka D, et al. Vascular reconstruction and complications in living donor liver transplantation in infants weighing less than 6 kilograms: the Kyoto experience. Liver Transpl. 2006; 12:1224–1232
18. Payen DM, Fratacci MD, Dupuy P, et al. Portal and hepatic arterial blood flow measurements of human transplanted liver by implanted doppler probes: interest for early complications and nutrition. Surgery. 1990; 107:417–427
19. Feng AC, Fan HL, Chen TW, et al. Hepatic hemodynamic changes during liver transplantation: a review. World J Gastroenterol. 2014; 20:11131–11141
20. Lalani AA, Xie W, Martini DJ, et al. Change in neutrophil-to-lymphocyte ratio (NLR) in response to immune checkpoint blockade for metastatic renal cell carcinoma. J Immunother Cancer. 2018; 6:5
21. Lorente D, Mateo J, Templeton AJ, et al. Baseline neutrophil-lymphocyte ratio (NLR) is associated with survival and response to treatment with second-line chemotherapy for advanced prostate cancer independent of baseline steroid use. Ann Oncol. 2015; 26:750–755
22. Biyik M, Ucar R, Solak Y, et al. Blood neutrophil-to-lymphocyte ratio independently predicts survival in patients with liver cirrhosis. Eur J Gastroenterol Hepatol. 2013; 25:435–441
23. Moreau N, Wittebole X, Fleury Y, et al. Neutrophil-to-lymphocyte ratio predicts death in acute-on-chronic liver failure patients admitted to the intensive care unit: a retrospective cohort study. Shock. 2018; 49:385–392
24. Zhang H, Sun Q, Mao W, et al. Neutrophil-to-lymphocyte ratio predicts early mortality in patients with HBV-related decompensated cirrhosis. Gastroenterol Res Pract. 2016; 2016:4394650
25. Desai MS, Zainuer S, Kennedy C, et al. Cardiac structural and functional alterations in infants and children with biliary atresia, listed for liver transplantation. Gastroenterology. 2011; 141:1264–1272, 1272.e1
26. Desai MS, Mathur B, Eblimit Z, et al. Bile acid excess induces cardiomyopathy and metabolic dysfunctions in the heart. Hepatology. 2017; 65:189–201
27. Voiosu A, Wiese S, Voiosu T, et al. Bile acids and cardiovascular function in cirrhosis. Liver Int. 2017; 37:1420–1430

Supplemental Digital Content

Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved.