Eighty percent of the pediatric deaths caused by liver disease occur in children less than 2 years old,1 and the majority of liver transplants performed in this population involve partial liver grafts: reduced size liver transplantation, split liver transplantation, and living donor liver transplantation (LDLT). Living donation offers several advantages over deceased donation,2 such as a reduction in the time on the waiting list, procurement under optimal conditions from a healthy donor, a shorter cold ischemia time (CIT), elective scheduling of the operation,3 and better recipient outcomes after transplantation.4
The availability of living donors allows transplant teams to perform the procedure early in the course of liver disease5 before the occurrence of life-threatening complications. However, late referral to transplant centers is still a global problem that can compromise the success of the transplant procedure and increase the costs of patient care.6 It is important, therefore, to recognize the factors that negatively affect patient survival in order to improve results and minimize costs.
The largest series of pediatric LDLT procedures (more than 600 so far) has been reported by the Kyoto group,7 but publications with significant numbers of patients are limited unless they result from analyses of transplant registries (the United Network for Organ Sharing registry as well as European, Japanese, and Korean registries). The aim of this study was to examine the perioperative factors associated with patient and graft survival for 430 pediatric LDLT procedures (413 primary transplants). An analysis of the relationships of perioperative variables and the occurrence of vascular complications was also performed.
BA, biliary atresia; CI, confidence interval; CIT, cold ischemia time; GRWR, graft-to-recipient weight ratio; HAT, hepatic artery thrombosis; HR, hazard ratio; HV, hepatic vein; INR, international normalized ratio; LDLT, living donor liver transplantation; LL, left lobe; LLS, left lateral segment; MS, monosegment; PELD, Pediatric End-Stage Liver Disease; PV, portal vein; PVT, portal vein thrombosis; RL, right lobe; V5, segment V drainage; V8, segment VIII drainage.
PATIENTS AND METHODS
Between October 1995 and April 2011, 430 LDLT procedures (including 413 primary transplants) were performed for patients less than 18 years old at Sirio-Libanes Hospital/A. C. Camargo Hospital (São Paulo, Brazil). The data for primary LDLT recipients were analyzed through a retrospective examination of medical records from a prospectively collected database. Since the beginning of this program, a single physician was responsible for collecting data from charts and entering the data into the computer database. This study was approved by the hospital's ethics committee.
Recipient and Donor Selection
The evaluation of potential recipients of liver transplantation followed standard procedures, which included an investigation of the liver disease etiology and the degree of liver failure; upper gastrointestinal endoscopy, Doppler ultrasound, computerized tomography scanning, or magnetic resonance imaging; and dental, psychological, and nutritional evaluations. Recipient-donor selection was based on medical need and ABO compatibility. Pediatric patients with end-stage liver disease who were accepted by the transplant team for liver transplantation were initially placed on the waiting list for deceased donor liver transplantation.
The voluntary intent of the donor was first assessed, and informed consent was mandatory. All donors underwent a psychological evaluation to rule out any psychological disturbances, coercion, or commercial purposes. The preoperative evaluation of the candidates was previously described.5 Doppler ultrasound was performed to evaluate the vascular anatomy, liver echogenicity (the presence of steatosis and the detection of parenchymal lesions), and liver volumetry for left lateral segment (LLS) donation. For left lobe (LL) and right lobe (RL) transplantation, an anatomic evaluation was performed with magnetic resonance imaging and cholangio-magnetic resonance imaging. The liver weight/recipient body weight ratio (g/g) was used to estimate the preoperative graft-to-recipient weight ratio (GRWR) for LL and RL transplantation. Candidates with diabetes, hypertension, or any other comorbidity were excluded. Only donors less than 50 years of age with a body mass index less than 28 kg/m2 were accepted for surgery.
The operative technique for the donor hepatectomy followed principles described elsewhere.8, 9 Parenchymal transection was accomplished with a Cavitron ultrasound surgical aspirator (Cavitron, Stanford, CA) after dissection of the vascular inflow to the liver. The transection line on the liver varied with the graft type [RL, LL, LLS, or monosegments (MSs)]. Intraoperative cholecystectomy and cholangiography were mandatory for determining the position at which to cut the biliary tree. MSs were further reduced during the donor operation before graft removal [Fig. 1 (dotted line)]. Conventional clamping, cutting, and suturing of the major vessels were performed during graft removal. After the hepatectomy, the graft was flushed with Euro-Collins solution or histidine tryptophan ketoglutarate solution at 4°C, and it was prepared for implantation.
Each graft was implanted with the piggyback technique. The donor's hepatic veins (HVs) were anastomosed to the confluence of the recipient's 3 HVs in LL/LLS transplantation. When the graft's HV was larger than the recipient's HV in LL/LLS and RL transplants, a triangulation technique was used for HV implantation.10, 11 The portal vein (PV) was anastomosed in an end-to-end fashion with 7-0 Prolene sutures and with interposition vein grafts when they were necessary [sclerotic PVs in cases of biliary atresia (BA)]. The hepatic artery was always reconstructed with microvascular techniques, and 9-0 or 10-0 nylon sutures were used (Ethicon, Edinburgh, United Kingdom). When feasible, a duct-to-duct anastomosis was preferred. A parachute technique with a continuous 7-0 Prolene suture was used for the posterior wall of the biliary-enteric anastomosis, and 7-0 interrupted polydioxanone sutures without stenting were used for the anterior wall.
After the recipient operation, a Doppler ultrasound scan was routinely performed on postoperative day 1 to evaluate the vascular patency. Immunosuppression was based on tacrolimus (Prograf) and steroids. The targeted tacrolimus trough level for the first month was 8 to 10 ng/mL. Patients were weaned from steroids (when this was possible) after the third postoperative month.
The preoperative variables associated with the recipient's clinical status included the following: recipient age and body weight, Pediatric End-Stage Liver Disease (PELD) score, z score for height/age, bilirubin, albumin, international normalized ratio (INR), hemoglobin, sodium, presence of ascites, and previous surgery. The analyzed technical aspects included the GRWR and the use of vascular grafts. The measurement of the laboratory parameters was assessed the week before transplantation in all cases. Except for GRWR (4%), recipient body weight (10 kg), and recipient age (1 year), the cutoff used for each variable was the median value in our series. The posttransplant variables included the incidence of hepatic artery thrombosis (HAT), portal vein thrombosis (PVT), biliary leaks, and biliary stenosis. Further studies were conducted in order to determine the association of pretransplant parameters (body weight ≤ 10 kg, presence of ascites, PELD score, total bilirubin level, GRWR, z score for height/age, and use of vascular grafts) and the incidence of vascular complications (HAT and PVT).
MedCalc 220.127.116.11 (MedCalc Software, Mariakerke, Belgium) was used for statistical analysis. The Kaplan-Meier method was used to estimate overall survival. For the assessment of prognostic factors, log-rank tests and Cox proportional hazards models in univariate and multivariate analyses were used. All factors that attained a P value < 0.1 in the univariate analysis were included in the multivariate model. All models were built with a forward stepwise method. The data were unadjusted and were based on the univariate analysis results. Only factors that attained a P value < 0.05 were kept in the final model. The association between categorical variables was verified with chi-square tests. The Mann-Whitney test was used to analyze the numerical variables without normal distributions.
The median age of the living donors was 30.1 years (range = 16-51 years). The median body mass index was 24.0 kg/m2 (range = 16.8-27.7 kg/m2). The majority of the donors were parents (85%). There were no ABO-incompatible transplants. The focus of this study was recipient outcomes, and donor-related complications are partially described in another publication.9
There were 294 girls and 119 boys with ages ranging from 3.6 to 217 months (median = 14 months). The median patient weight at the time of transplantation was 8.4 kg (range = 3.9-75.4 kg). The indications for liver replacement were BA [273 (66.1%)], tyrosinemia [15 (3.6%)], Alagille syndrome [8 (1.9%)], primary liver neoplasms [14 (3.4%)], fulminant liver failure [10 (2.4%)], alpha-1-antitrypsin deficiency [11 (2.7%)], primary sclerosing cholangitis [10 (2.4%)], Budd-Chiari syndrome [11 (2.7%)], cryptogenic cirrhosis [22 (5.3%)], autoimmune hepatitis [8 (1.9%)], progressive familial intrahepatic cholestasis [6 (1.5%)], chronic cholestasis [6 (1.5%)], other metabolic diseases [12 (2.9%)], and other diseases [7 (1.7%)].
The pretransplant variables included in this study are listed in Table 1. The median z score for height/age was −1.5 (range = −7.39 to 2.18); 40% of the patients presented with a z score for height/age < −2, and this indicated severe malnutrition at the time of transplantation. The PELD score was ≥20 (median = 15, range = −8 to 58) for 29.7% of the patients; 258 patients (62.5%) had a body weight ≤ 10 kg. One hundred seventy-four patients (42.1%) had ascites before transplantation, and 169 patients (41%) had undergone a previous operation (mostly Kasai portoenterostomy for the treatment of BA).
Table 1. Demographics of 413 Recipients of Primary LDLT at Sirio-Libanes Hospital/A. C. Camargo Hospital (São Paulo, Brazil)
Body weight (kg)
z score for weight/age
z score for height/age
The liver grafts included 348 LLSs, 5 MSs, 51 LLs, and 9 RLs (Fig. 1). The median GRWR was 3.15% (range = 0.6%-7.32%). The median intraoperative packed red blood cell transfusion was 200 mL (range = 0-1100 mL), and the median CIT was 55 minutes (range = 5 minutes to 7 hours). The median intensive care unit stay of recipients was 2 days (range = 1-77 days), and the median hospital stay was 13 days (range = 3-174 days).
Eighteen patients (4.3%) developed HAT, 26 (6.2%) had biliary leaks, 37 (8.9%) had biliary stenosis, 11 (2.6%) had early PVT (<30 days after transplantation), 24 (5.8%) had late PVT (>30 days after transplantation), and 8 (1.9%) had an HV outflow obstruction. Seventy-two patients (17.4%) underwent reoperation in this series because of HAT (n = 3), unconfirmed HAT (n = 7), intestinal perforations (n = 11), bleeding (n = 13), biliary leaks (n = 18), intestinal obstructions (n = 3), PVT (n = 12), or other causes (n = 5). 109 patients 26.4% had at least 1 postoperative complication. 16 patients (3.9%) underwent retransplantation (50% because of HAT). Sixty-two patients (15%) died because of sepsis (n = 12), respiratory failure (n = 14), liver necrosis (n = 6), tumor recurrence (n = 3), cerebral edema (n = 2), lymphoma (n = 9), chronic rejection (n = 3), Kaposi's sarcoma (n = 1), intraoperative hemorrhaging (n = 1), primary nonfunction (n = 1), acute pancreatitis (n = 1), or other causes (n = 9). Twenty-three patients died more than 1 year after transplantation (10 because of lymphoproliferative disorders and sarcoma).
The actuarial patient and graft survival rates at 12, 60, and 120 months were 90.6%, 83.6%, and 82.5% and 88.7%, 80.8%, and 80.0%, respectively (Fig. 2A). Patient survival was studied in 2 eras: 1995-2002 and 2003-2011 (Fig. 2B). There was no difference in patient survival between the study periods [P = 0.15, hazard ratio (HR) = 0.66, 95% confidence interval (CI) = 0.34-1.27].
The pretransplant variables associated with statistically significant lower patient survival are shown in Fig. 2. These patients had a z score for height/age < −2 (P = 0.04; Fig. 3A), ascites (P = 0.04; Fig. 3B), a body weight < 10 kg (P = 0.006; Fig. 3C), and an age < 1 year [P = 0.01; Fig. 3D (Kaplan-Meier curves, log-rank test); see also Table 2]. The survival rates at 12, 60, and 120 months are plotted on each curve. The univariate analysis correlating the pretransplant and posttransplant variables and posttransplant patient survival is shown in Table 2. Variables with a P value < 0.1 and the PELD score were used in the multivariate analysis (Cox proportional hazards regression). A body weight ≤ 10 kg (P = 0.003, HR = 3.33, 95% CI = 1.49-7.43) and the presence of HAT (P < 0.001, HR = 5.04, 95% CI = 2.26-11.26) were independently associated with worse patient survival in the final model.
Table 2. Univariate Analysis Correlating Pretransplant Variables and Patient Survival After LDLT
z score for height/age
The association of pretransplant and posttransplant parameters and graft survival was also examined (Table 3). The multivariate analysis included variables with a P value < 0.1 in addition to the PELD score. A body weight ≤ 10 kg (P = 0.002, HR = 2.8, 95% CI = 1.44-5.65) and HAT (P ≤ 0.001, HR = 10.72, 95% CI = 5.47-20.99) held statistical significance in this analysis.
Table 3. Univariate Analysis Correlating Pretransplant Variables and Graft Survival After LDLT
z score for height/age
The investigation of pretransplant variables and technical aspects and their association with the occurrence of HAT and PVT are shown in Tables 4 and 5, respectively. A body weight ≤ 10 kg was not associated with an increased incidence of HAT or PVT. Although there was a trend toward an association of larger grafts (≥4%) and PVT (P = 0.07), only the use of vascular vein grafts for PV reconstruction was statistically related to an increased risk of PVT (P = 0.001).
Table 4. Association of Perioperative Parameters and the Incidence of HAT
No (n = 395)
Yes (n = 18)
The data are presented as averages and standard deviations.
Multiple factors are related to patient survival after transplantation. Pretransplant scores have been developed to assess the waiting-list mortality of patients (eg, the PELD score)12 and objectively establish the severity of disease. The graft quality and type, the ischemia time, and the spectrum of complications after transplantation can also define the results. In this series, all patients received living donor liver grafts, which were uniformly healthy and were recovered with the same technique from stable donors by the same surgical team. Implantation was possible with a short CIT in the majority of cases. For these reasons, there was only 1 episode of primary nonfunction, and posttransplant morbidities could be attributed to the clinical status of the patients and the occurrence of technical complications after transplantation.
In the pretransplant scenario, malnutrition and hyperbilirubinemia are associated with an increased need for transplantation or death in patients with BA, even after Kasai portoenterostomy is performed.13 After transplantation, studies have shown that a pretransplant weight deficit is associated with late patient mortality (>1 year),14 and the baseline z score is related to late growth retardation15 as well as increased hospital stays and greater hospital costs.6 In this series, 40% of the patients were severely malnourished, and the univariate analysis demonstrated inferior patient and graft survival. This was probably due to the large number of patients with cholestatic liver diseases (73%) and especially BA. In fact, it is not possible to separate the relationship between hyperbilirubinemia and malnutrition in these cases, and another recent study has demonstrated the negative impact on outcomes for recipients with a body weight ≤ 10 kg and increasing bilirubin levels.5 The high prevalence of malnourishment in transplant candidates may indicate late referrals. However, even though the z score for height may be a reflection of the nutritional status, it may also be a manifestation of a systemic disease that influences growth independently of liver disease, such as Alagille syndrome.
Another finding in this study indicates a lower posttransplant survival rate for pediatric patients with pretransplant ascites (42.1%). The concern with ascites has always been more pronounced for adult transplant patients, for whom the matter is always taken under consideration when allocation policies are being discussed. A recent publication has shown that despite the subjective nature of ascites, moderate ascites informs risk models beyond the Model for End-Stage Liver Disease and the Model for End-Stage Liver Disease–sodium in the adult scenario.16 However, this has not been a common concern in the pediatric setting, even though the initial scores included the presence of ascites as part of the prognostic score,17 and previous publications showed lower patient survival for Child-Turcotte-Pugh class C pediatric patients.18 The developers of the PELD score did not consider the presence of ascites as a prognostic factor,12 probably because of the difficulty in establishing objective parameters for measuring it. On the other hand, 1 study showed a fair correlation between the Child-Turcotte-Pugh scoring system and the PELD score.19
Female sex and a high GRWR were implicated as independent risk factors for the development of HAT in a series with similar numbers of LDLT procedures20; the incidence of HAT was 6.7%. The authors debated whether recipients receiving livers with high GRWRs are at risk of developing large-for-size syndrome, with increased resistance to blood flow due to graft compression. This situation can jeopardize both arterial and portal inflows to the graft. Interestingly, because of aggressive surveillance and treatment, HAT did not significantly decrease patient survival in the aforementioned experience. In this series, a GRWR > 4% did not increase the incidence of HAT and PVT, nor did it affect patient survival. One can only speculate whether differences in the patient demographics or the high prevalence of pretransplant ascites (an enlarged abdominal cavity) in this study can explain these findings. Notably, a substantial number of children in this series with BA had not previously undergone a Kasai operation (169 patients with previous surgery/273 patients with BA), and they underwent transplantation with the association of ascites, hyperbilirubinemia, and malnutrition, which were markers of worse patient and graft outcomes in this cohort. However, the differences in age and body weight at the time of transplantation in the 2 series did not seem to be that discrepant (median values of 11 months and 7.3 kg for patients with HAT in the Japanese experience versus median values of 14 months and 8.4 kg for all the patients in this cohort).
In another publication from the same group,21 only a recipient body weight less than 6 kg was associated with an increased incidence of portal complications (8.5%). In this series with an 8.4% incidence of PVT, only the use of vascular grafts in PV reconstruction (reconstruction with a conduit) was associated with PV complications. It is probably unfair to blame the vascular conduits for the thrombosis because they were used in cases with very low portal calibers unsuitable for a primary anastomosis. Moreover, portal flow steal through collateral veins is difficult to quantify because it is supposed to be higher in sicker patients. This could be an additional cause of PV complications during follow-up (a 5.8% incidence of late thrombosis).
Patients with a body weight ≤ 10 kg (62.5%) had worse long-term patient and graft survival after transplantation. Even though many reports have correlated a low body weight with an increased incidence of vascular complications,22, 23 these were not the findings of this study within the analyzed range of body weights. The early posttransplant mortality rate (<12 months) in this series may be a reflection of the difficulties in managing these small infants postoperatively, even though there was not a significant difference in patient survival between the 2 eras. However, when we limited the analysis to patients with a body weight ≤ 10 kg, there was a definitive improvement in patient care over time, as demonstrated in a previous publication.5 Part of the long-term mortality observed in this cohort (47.8%) can be attributed to lymphoproliferative disorders and tumors; in contrast, a late mortality rate of 20% was attributed to malignancies in a study from the Studies of Pediatric Liver Transplantation database.14 Only recently (2008) was the monitoring of Epstein-Barr virus loads introduced to our clinical practice, and a reduction in the occurrence of Epstein-Barr virus–driven malignancies is expected.
The high incidence of malnutrition and ascites in this series may indicate late patient referral to transplant centers, particularly when LDLT is available for timely transplantation. The operative technique used in this large pediatric LDLT cohort showed that patients with a body weight ≤ 10 kg did not have an increased incidence of HAT or PVT. Only the use of vascular grafts for PV reconstruction increased the risk of PVT. A low body weight (≤10 kg) and the occurrence of HAT independently determined worse patient and graft survival.