Orthotopic liver transplantation (OLT) is an established method of therapy for pediatric patients with end-stage liver disease [1,2]. Blood loss during OLT is highly variable, and massive hemorrhage is a well known complication of the procedure. Massive bleeding and transfusion of stored blood result in many complications, including episodes of acute hypovolemia, citrate loading with ionized hypocalcemia, hyperosmolality, and hypothermia. A significant relation between large blood transfusion and postoperative morbidity and mortality has been recognized in many transplant centers [3-5].
The goal of the present study is to identify pediatric OLT candidates at high-risk for intraoperative massive hemorrhage. Similar studies have been performed in adult OLT recipients, but the results may not be applicable to the pediatric population. Liver diseases leading to OLT are different, and other techniques such as reduced-size liver grafting are used. A similar study was done in pediatric OLT recipients by Lichtor et al.  in 1988, but the series was limited (21 OLTs in 17 recipients) and a multivariate analysis in a larger series was suggested by the authors. Lichtor et al. aimed at predicting the amount of blood transfusion and the risk of death, whereas we analyzed variables associated with a high blood loss.
Between February 1986 and January 1991, 106 OLTs were performed in 95 patients less than 15 yr of age. One patient received a combined liver and kidney transplantation. Retransplantations and patients with combined liver-heart-lung transplantation were excluded from the study.
Anesthesia was induced using intravenous thiopental or intramuscular injection of ketamine. Anesthesia was maintained with isoflurane and fentanyl. Pancuronium was used for muscle relaxation. Arterial and central venous catheters were inserted for hemodynamic monitoring and blood sampling. Two or three intravenous cannulae were secured for blood transfusion.
Blood volume replacement used a mixture of homologous citrated packed red blood cells (RBCs) with a hematocrit of 0.75, fresh frozen plasma (FFP), and crystalloids. An intraoperative autotransfusion device (Cell-Saver No. IV; Haemonetics Inc., Braintree, MA) was used in 16 patients . Central venous pressure was maintained at 8-10 mm Hg during the dissection period and prior to clamping of hepatic vessels. A lower central venous pressure was occasionally tolerated before abdominal closure. No venovenous bypass was used except in one 10-yr-old patient with pulmonary hypertension. The minimum acceptable hematocrit at the end of the procedure was designated to be 0.3. A simplified coagulation profile was used for guidance of component therapy: platelet count >80 times 109 /L, prothrombin time <1.5 times control value, fibrinogen >100 mg/dL. No antifibrinolytic drug was administered prophylactically to prevent hyperfibrinolysis. When oozing from the surgical field occurred, platelets and FFP were infused. epsilon-Amino caproic acid (20 mg/kg) was administered in seven children with oozing and an euglobulin lysis time less than 1 h.
Since size-matched pediatric donors are scarce, we often had to reduce the volume of young adult donor livers. We used two types of grafts: a left liver graft or a left lobe graft. The choice of the reduction technique depended upon two criteria: the donor-recipient body-weight ratio and the transverse internal basithoracic distance of the recipient . Parenchymatous transsection was performed along the main scissures using a forceps-crushing technique without any hilar dissection. Absorbable clips were used for hemostasis and biliostasis, and ligatures and running sutures were used for hemostasis of larger vessels. The graft was implanted orthotopically in the usual manner. The transsection area was sealed with a biologic glue. After reperfusion, sites of bleeding were controlled with stitches.
Intraoperative blood loss was evaluated by the quantity in milliliters per kilogram of body weight of RBCs infused to obtain a hematocrit ratio between 0.30 and 0.35 at the end of surgery. In the 16 patients with autotransfusion, the quantity of RBCs transfused has been evaluated according to the following calculation: Vt = Vh + Va times (0.50/0.75), where Vt = total RBCs volume transfused, Vh = homologous RBCs volume transfused, and Va = autologous RBCs volume transfused. This adjustment is based on an hematocrit of 0.50 in autologous RBCs. Other fluid input including crystalloids, FFP, and platelets were ignored because their use is affected by other fluid losses, such as transsudation from the abdominal cavity, and by the need to correct coagulation defects. The population was arbitrarily divided into two groups based on RBC requirements. The lower three quartiles were defined as mild or moderate blood loss (Group 1). The upper quartile was defined as the severe blood loss group (Group 2).
Fourteen preoperative items (listed in Table 1) were retrospectively collected from each patient's hospital chart. Patients were grouped according to their status for each discrete variable being evaluated. Three types of indication for OLT were considered: acute liver failure (fulminant or subfulminant), biliary atresia, and other types of chronic liver diseases. Previous abdominal surgery included major operations involving the liver, portal vein, or biliary tract. Two categories of preoperative medical support were considered: children in stable condition waiting at home for an organ to become available, and hospital-dependent children confined to the intensive care unit or hepatology unit (inpatient support). Ascites was considered present if clinically detectable. According to a rough clinical grading, encephalopathy was defined as drowsiness, stupor, or coma (Grades II, III, IV)  without any other cause than liver failure. Portal hypoplasia, suspected on preoperative duplex Doppler ultrasonography and confirmed by angiography, was defined as a portal vein caliber less than 3.5 mm . Intraabdominal malformations were polysplenia syndrome and mesenteric malrotation. Patients with decreased height for age, i.e., under 2 SD from the mean according to established growth charts, were considered as having poor nutritional status.
The distribution of the blood loss was highly skewed, thus this variable was expressed as median (range). Other results were expressed as mean +/- SD. Univariate analysis was performed through standard statistical tests. chi squared analysis was used for the comparison of the distribution of discrete variables in the two groups and unpaired Student's t-test for the comparison of two means. A logarithmic transformation was performed prior to comparison for assymetrically distributed variables. Since many variables were interrelated, we used a multivariate procedure as a final step. We chose a stepwise logistic regression with blood loss as the dependent variable (partitioned in two classes: severe blood loss (Group 2), and mild to moderate blood loss (Group 1). The model used all variables included in the univariate analysis as the independent variables. Calculations were made with the PLR procedure of the BMDP Statistical Software (BMDP Statistical Software Inc.; Los Angeles, CA). A P value <0.05 was considered as statistically significant.
There were 54 boys and 41 girls whose age ranged from 5 mo to 14 yr (median, 6 yr). Primary diagnoses leading to OLT were biliary atresia in 39, fulminant or subfulminant hepatic failure in 13, and other types of chronic liver diseases in 43 (familial cholestasis, 20; metabolic disorders, 13; miscellaneous diseases, 10). Twenty-three (24.2%) patients required inpatient support in the pediatric hepatology unit, and 11 (11.6%) were in the intensive care unit prior to OLT. Previous abdominal surgery included a portoenterostomy in 39 cases, a portosystemic shunt in 2, and a cholecystectomy in 1. Intraabdominal malformations were present in 9 patients (7 polysplenia syndromes and 2 mesenteric malrotations) and portal vein hypoplasia in 11 patients, all with biliary atresia.
Wide interindividual RBC requirements were observed, as shown in Figure 1. Median RBC requirement was 79 mL/kg (range, 4-586). The upper quartile of the population (Group 2, i.e., severe loss group) required 123 mL/kg or more (median, 161). Sixteen patients (6 in Group 2) received autologous blood (median, 45 mL/kg; range, 7-198 mL/kg).
The results of the univariate analysis are summarized in Table 2 and Table 3. Portal vein hypoplasia was dramatically different between groups: more than 12 times higher in Group 2. The distribution of pathologic diagnosis was significantly different between the two groups with a higher proportion of children with acute liver failure and with biliary atresia in Group 2. The hepatocellular functional reserve was significantly decreased in Group 2 as shown by a greater proportion of children requiring inpatient support, having ascites, encephalopathy, and a prolonged prothrombin time. Intraabdominal malformations were more frequent in Group 2. Reduced-size liver grafts and indirect signs of portal hypertension such as esogastric varices tended to be more frequent in Group 2, but were not statistically related to higher blood loss. Previous abdominal surgery, age, nutritional status, and platelet count had no prognostic value.
All the variables used in the univariate analysis were included. Results are shown in Table 4. Only the presence of portal vein hypoplasia, inpatient support, and use of a reduced-size liver graft were independently associated with a severe blood loss. By multiplying the variable (0 = no, 1 = yes) by their associated coefficient, the probability of a high blood loss may be calculated as: Equation 1 where Equation 2 With a cutoff point of 0.858 (P > 0.858, high risk; P < 0.858, low or moderate risk), the model correctly classified 79.2% of the patients with severe blood loss and 71.4% of the patients with mild to moderate blood loss.
Factors influencing blood loss in OLT have been examined in many studies, most of them performed in adult OLT recipients and focusing on preoperative coagulation abnormalities. Intraoperative hemorrhage in OLT is bound to be multifactorial . In the present study, using a stepwise logistic regression model, the relative contribution of several factors was determined in a consecutive series of 95 first pediatric OLT. The results highlight the importance of portal vein hypoplasia, inpatient support, and use of a reduced-size liver graft as independent risk factors for a high blood loss.
Blood loss has been estimated indirectly through blood transfusion requirements, an approach which has certain limitations. Blood and fluid therapy is aimed primarily at providing satisfactory circulatory response, not at strict replacement of true blood loss. Nevertheless, this approach is commonly used and has direct practical implications on bank blood use. The alternative method, direct estimation of blood loss based on surgical suction and weighing surgical sponges, may be even less accurate in liver transplantation as an abundant transsudation of the peritoneal cavity may lead to overestimation of erythrocyte loss. In the present study, patients were arbitrarily divided in two groups according to blood requirements. Whether the 123 mL RBC/kg cutoff point (implying loss of approximately four blood volumes) is significant in terms of mortality and morbidity has not been specifically studied herein. However, similar amounts of blood transfusion have been suggested to be associated with a worse prognosis in adult liver transplant recipients .
The choice of factors which might predict increased blood loss in pediatric OLT recipients was based on previous studies in adults and children [6,12] and factors suggested to be of interest by clinical experience, such as portal vein hypoplasia.
Many studies have focused on the influence of preoperative coagulopathy. In the present study, the prothrombin time and platelet count were the only coagulation tests taken into account and were found to be poorly related to blood loss in the multivariate analysis. A relationship between pretransplantation recipient coagulation abnormalities and intraoperative blood loss has been found in adults [5,13]. Although a comparable correlation has been found in a limited number of patients by Lichtor et al. , others have questioned the prognostic value of preoperatively obtained coagulation data in children [11,12]. Reviewing the records of 238 OLTs in 200 children, Carlier et al.  could not find any correlation between a coagulation abnormality score and the mean blood requirements in different etiology groups. This is in accordance with the results of the present multivariate analysis showing that anatomic and technical factors are major determinants of a high blood loss in pediatric OLT.
The etiology of the underlying liver disease is the most important bleeding risk factor in a previous study in children . In the present study, retransplantations were excluded and three etiology categories were considered. Children with biliary atresia are usually considered at high risk of bleeding due to adhesions, as most of them have undergone previous abdominal surgery for a portoenterostomy . In the present study, the proportion of children with biliary atresia was only slightly larger in the high blood loss group. This etiology category actually was heterogeneous with regard to bleeding. Interestingly, previous abdominal surgery (mainly portoenterostomy), in itself, was not associated with a high blood loss. This apparent contradiction with previously reported studies [12,14] may be due to the design of the present study. In fact, the median blood loss was higher in children with previous abdominal surgery, but the proportion of previously operated children was not significantly higher in the high blood loss group. The higher proportion of children with acute hepatic failure in the high blood loss group is in keeping with findings by Lichtor et al.  in children and by Mor et al.  in adults. The average blood loss in this category is close to that reported by Carlier et al. . This life-threatening condition is associated with severe coagulation abnormalities due to hepatic and renal failures and the frequent use of a reduced-size graft.
Volume reduction of the liver is expected to result in bleeding from the raw surface of the graft after reperfusion. The reduced-size liver transplantation technique has been correlated with blood loss in the univariate analysis by Lichtor et al. , but has not been found to be an independent predictor in the multivariate analysis. As, in earlier times, the technique has been used predominantly, if not exclusively, in emergent situations, an overlapping of this variable with acute liver failure may have occurred.
The level of medical support is a logical and simple marker of the degree of liver failure. Hospital dependency overlaps with other conditions reflecting an advanced degree of illness (such as ascites, encephalopathy, prolonged prothrombin time). The relationship between intraoperative blood loss and advanced liver dysfunction confirms previous findings in pediatric OLT .
The close relationship between portal vein hypoplasia (PVH) and high intraoperative blood loss is the most original finding of the present study. This rare situation (11.6% in our series) could not have been taken into account in a study including a small number of patients. Despite the lack of a precise anatomic and pathogenic definition, PVH is undoubtedly a reality for surgeons involved in pediatric liver transplantation. Lerut et al.  mention it as one of the technical difficulties of venous reconstruction in OLT. PVH is more frequent in patients with biliary atresia (in 14 of 20 patients in the study by Lerut et al. , and in all cases in the present series). Whether PVH is malformative in nature is unknown, but 4 of the 11 children with PVH had coexistent anomalies recognized as "polysplenia syndrome" (situs inversus, polysplenia, intestinal malrotation, azygos continuation of an interrupted suprarenal inferior vena cava). Lerut et al.  have hypothetized that the development of collateral vessels in the adhesions resulting from a previous portoenterostomy increased the "steal" of venous blood from the splanchnic bed thereby promoting the involution of the portal vein. A higher blood loss in case of PVH may then result from more extensive dissection in order to anastomose the donor portal vein to the confluence of the splenic and mesenteric veins, to collateral vessels developed in the adhesions, or to both. Similar difficulties may be encountered in patients with portal vein thrombosis, a rare situation where excessive blood loss has been reported . Documentation of portal venous anatomy is performed during the evaluation process. Thrombosis or PVH may be detected preoperatively by several reliable imaging techniques, the easiest and least invasive being Doppler ultrasonography.
Prediction of intraoperative hemorrhage during OLT will undoubtedly remain difficult, as many intraoperative uncontrollable variables can influence blood loss . A model based on preoperative characteristics is bound to misclassify some cases. This analysis emphasizes that the preoperative condition of the patient plays a crucial role in the intraoperative course of OLT. Such findings do not help avoiding situations of intraoperative massive hemorrhage. However, prediction studies are useful to draw attention to the potential significance of preoperative information, such as the presence of portal vein hypoplasia, and to alert the blood bank to save more blood, to prepare adequate venous access, and monitoring, as suggested by Lichtor et al. . The use of intraoperative blood salvage, which is expensive and has little application to the pediatric patient, may be cost-effective and indicated. The clinical value of the risk factors described in the present study needs to be compared to those of other populations of pediatric OLT recipients. Comparison and pooling of studies from different centers would probably allow the derivation and validation of a prediction instrument.
We thank Dr. Andre De Wolf and Dr. John R. Klinck for helpful comments and suggestions.
1. Starzl TE, Esquivel C, Gordon R, Todo S. Pediatric liver transplantation. Transplant Proc 1987;19:3230-5.
2. Whitington PF, Alonso EM, Piper JB. Pediatric liver transplantation. Semin Liver Dis 1994;14:303-17.
3. Shaw BW, Wood RP, Gordon RD, et al. Influence of selected variables and operative blood loss on six-month survival following liver transplantation. Semin Liver Dis 1985;5:385-93.
4. Brems JJ, Hiatt JR, Colonna JO, et al. Variables influencing the outcome following orthotopic liver transplantation. Arch Surg 1987;122:1109-11.
5. Mor E, Jennings L, Gonwa TA, et al. The impact of operative bleeding on outcome in transplantation of the liver. Surg Gynecol Obstet 1993;176:219-27.
6. Lichtor JL, Emond J, Chung MR, et al. Pediatric orthotopic liver transplantation: multifactorial predictions of blood loss. Anesthesiology 1988;68:607-11.
7. Kang YG, Aggarwal S, Virji M, et al. Clinical evaluation of autotransfusion during liver transplantation. Anesth Analg 1991;72:94-100.
8. Houssin D, Soubrane O, Boillot O, et al. Orthotopic liver transplantation with a reduced-size graft: an ideal compromise in pediatrics? Surgery 1992;111:532-42.
9. Sherlock S. Diseases of the liver and biliary system. 7th ed. London: Blackwell, 1985.
10. Lerut J, Tzakis AG, Bron K, et al. Complications of venous reconstruction in human orthotopic liver transplantation. Ann Surg 1987;205:404-14.
11. Borland LM, Roule M. The relation of preoperative coagulation function and diagnosis to blood usage in pediatric liver transplantation. Transplant Proc 1988;20:533-5.
12. Carlier M, Van Obbergh L, Veyckemans F, et al. Hemostasis in children undergoing liver transplantation. Semin Thromb Hemost 1993;19:218-22.
13. Bontempo FA, Lewis JH, VanThiel D, et al. The relation of preoperative coagulation finding to diagnosis, blood usage, and survival in adult liver transplantation. Transplantation 1985;39:532-6.
14. Cuervas-Mons V, Rimola A, Van Thiel DH, et al. Does previous abdominal surgery alter the outcome of pediatric patients subjected to orthotopic liver transplantation. Gastroenterology 1986;90:853-7.
© 1995 International Anesthesia Research Society
15. Shaked A, Busuttil R. Liver transplantation in patients with portal vein thrombosis and central portocaval shunts. Ann Surg 1991;214:696-702.