Full text of DOI:10.1007/BF03020372

GENERAL ANESTHESIA
1075
Intraoperative blood losses and transfusion
requirements during adult liver transplantation
remain difficult to predict
[Les pertes sanguines peropératoires et les besoins transfusionnels restent toujours
difficiles à prédire pendant la greffe du foie]
A. Steib MD,* G. Freys MD,* C. Lehmann MD,* C. Meyer MD,† G. Mahoudeau MD*
Purpose: To identify preoperative factors associated with high
blood losses during liver transplantation for chronic end- stage liver
disease.
Objectif : Découvrir les facteurs préopératoires associés à d’impor-
tantes pertes sanguines survenant pendant la greffe du foie pour ma-
ladie chronique du foie au stade ultime.
Methods: Four hundred and ten consecutive patients were includ-
ed in this retrospective study. Blood losses were calculated, based
on transfusion requirements. The population was divided into two
groups: the upper quartile was defined as the high blood loss (HBL)
group and the lower three quartiles as the low blood loss group.
Fourteen preoperative variables were collected. Qualitative vari-
ables consisted of the type of hepatopathy, Child-Pugh’s classifica-
tion, sex, the surgical team’s experience, previous abdominal
surgery and portal hypertension. Quantitative variables were age,
hemoglobin concentration Hb, platelet count, prothrombin time,
activated partial thromboplastin time, fibrinogen concentration, fib-
rin degradation products (FDP) and euglobulin lysis time. Univariate
analysis and stepwise multivariate analysis were conducted.
Méthode : Quatre cent dix patients successifs ont participé à une étude
rétrospective. Les pertes sanguines ont été calculées d’après les besoins
transfusionnels. Les participants ont été divisés en deux groupes : le quar-
tile supérieur étant le groupe de perte sanguine élevée (PSE) et les trois
quartiles inférieurs, le groupe de perte sanguine faible (PSF). Des 14 va-
riables préopératoires rassemblées, les variables qualitatives ont été : le
type d’hépatopathie, la classification de Child-Pugh, le sexe, l’expérience
de l’équipe chirurgicale, les interventions abdominales antérieures et l’hy-
pertension portale ; les variables quantitatives ont été : l’âge, le taux d’hé-
moglobine (Hb), la numération plaquettaire, le temps de prothrombine,
le temps de céphaline activé, la concentration en fibrinogène, les produits
de dégradation de la fibrine (PDF) et le temps de lyse des euglobulines.
L’analyse univariée et l’analyse multivariée séquentielle ont été réalisées.
Results: Patients in the HBL group required 12 units of red blood
cell or more to maintain a Hb $100g·L–1. HBL was associated with
severe liver disease, previous abdominal surgery, use of a venove-
nous bypass and little surgical experience in orthotopic liver trans-
plantation (OLT). In the HBL group several hemostatic parameters
were more disturbed before surgery. The multivariate analysis dis-
closed three independent variables associated with HBL: Hb and
FDP concentrations and previous upper abdominal surgery. When
combined, these resulted in a high specificity (98%) but low sensi-
tivity to predict blood loss.
Résultats : Les patients du groupe PSE ont eu besoin de 12 unités de
globules rouges ou plus pour maintenir une [Hb] $100g·L–1. La PSE a
été associée à : une maladie hépatique sévère, une intervention
abdominale antérieure, l’utilisation d’une dérivation veino-veineuse et
peu d’expérience en greffe du foie orthotopique (GFO). Dans le groupe
de PSE, certains paramètres hémostatiques étaient plus instables
avant l’opération. L’analyse multivariée a révélé trois variables
indépendantes associées à la PSE : les concentrations d’Hb et de PDF
et les interventions abdominales hautes antérieures. Les résultats
combinés présentent une forte spécificité (98 %) mais une faible sen-
sibilité à prédire les pertes sanguines.
Conclusion: Despite our efforts we were unable to identify pre-
dictive risk factors of bleeding during OLT even in a homogeneous
population. Centres should evaluate their practice individually in an
attempt to identify patients at high risk of being transfused.
Conclusion : Malgré nos efforts, nous n’avons pu déterminer les fac-
teurs de risque prédictifs de saignements pendant la greffe de foie
orthotopique chez une population homogène. Les hôpitaux doivent
évaluer leur pratique individuelle afin de découvrir les patients à risque
d’être transfusés.
From the Departments of Anaesthesia,* and Transplantation,† Hôpitaux Universitaires, Strasbourg, France.
Address correspondence to: Dr. Annick Steib, Departement of Anaesthesia, 1 Place de l’Hôpital, BP N 426, 67091 Strasbourg Cedex,
France. Phone: 03 88 11 54 10; Fax: 03 88 11 67 60; E-mail: Annick.Steib@chru-strasbourg.fr
Accepted for publication July 5, 2001.
Revision accepted August 29, 2001.
CAN J ANESTH 2001 / 48: 11 / pp 1075–1079

1076
CANADIANJOURNAL OF ANESTHESIA
LOOD loss represents a serious problem
during orthotopic liver transplantation
(OLT). Improvements in operative man-
agement, surgical techniques and graft
administered at a temperature of 36°C. Initially, the
surgical procedure was performed with the systematic
use of a heated venovenous bypass. From 1994
onwards, the piggyback technique, not requiring cross
clamping or transection of the inferior vena cava, was
the preferred method. The senior anesthesiologists
and surgeons involved in the procedures remained the
same during the entire duration of data collection.
B
preservation have widely contributed to a significant
reduction in transfusion requirements during the last
decade.1 However, blood losses remain highly vari-
able. As the quantity of blood products required con-
tributes to higher mortality and morbidity2–6
preoperative identification of patients at high risk of
massive intraoperative hemorrhage is of great interest.
Previous attempts to recognize these patients have
produced contradictory results. Most of these studies
have failed to define, preoperatively, biological factors
(specially those related to hemostatic disorders) pre-
dictive of hemorrhage.1,7–10 Surgical conditions such
as experience of the team, use of specific coagulation
devices and previous abdominal surgery have, more
often, been blamed.1,2 Moreover, the majority of these
trials included various types of liver disease in the
patient population under study, mixing acute and
chronic cases. In addition, surgical teams and transfu-
sion protocols sometimes changed during the course
of the studies. Thus, in the present report we attempt-
ed to identify demographic and preoperative factors
associated with intraoperative hemorrhage and trans-
fusion in a homogeneous group of patients with end-
stage chronic liver disease scheduled for OLT and
managed by the same medical team.
Replacement therapy
Packed red blood cells (RBCs) were administered to
maintain hematocrit levels at 30%. Fresh frozen plas-
ma (FFP) was infused in a ratio (RBCs/FFP) of 1 to
1.5. Platelets were transfused to maintain the platelet
count at around 100 x 109·L– 1. Cryoprecipitate was
indicated when the fibrinogen level decreased below
1g·L– 1. Intraoperative salvage of blood was used in the
absence of infection or cancer (Cell Saver nIV,
Haemonetics Inc, Brain tree MA, USA).
A bolus of aprotinin (5000 KIU·kg– 1) was injected
systematically before revascularization. Additional bolus-
es (200 000 KIU) were given whenever unexplained
generalized oozing occurred in the surgical field.
Evaluation of intraoperative blood loss
Intraoperative blood loss was evaluated by the quanti-
ty of RBC units required to maintain a hematocrit
ratio around 30% at the end of surgery. In patients
with autotransfusion, intraoperative blood loss was
evaluated according to the following calculation: total
RBC units required = homologous units + autologous
units. Autologous units were determined by the fol-
lowing formula: autologous volume transfused x
(0.50/0.75)/250. This adjustment was based on a
hematocrit of 0.5 in autologous RBCs (compared to
0.75 in RBCs) and a mean volume of 250 mL in each
pack of RBCs. The number of FFP and platelet units
transfused during the procedure was recorded. For
further analysis the population was arbitrarily divided
into two groups based on total intraoperative RBC
requirements. The upper quartile was defined as the
HBL and the lower three quartiles constituted the low
blood loss group (LBL). As a consequence of this
arbitrary definition, the cutoff point separating the
two groups was 12 RBC units.
Methods
Four hundred and ten consecutive patients suffering
from chronic liver disease who received transplants in
our center between January 1988 and December
1998 were included in this retrospective study. Any
acute situations involving retransplantation for prima-
ry graft failure and transplantation for fulminant
hepatic failure were excluded.
Anesthetic procedure and monitoring
The anesthetic procedure was standardized using a
combination of thiopentone, fentanyl or sufentanil
and pancuronium. Patients were ventilated with an
air/oxygen/isoflurane mixture. Radial and pulmonary
arterial catheters were inserted for hemodynamic mea-
surements and blood sampling. Two iv cannulae were
secured for rapid blood transfusion. Calcium chloride
was administered to maintain ionized calcium concen-
trations between 0.9–1.2 mmol·L–1, and sodium
bicarbonate (1 mEq·kg– 1) was injected when base
excess was below -10. Isotonic saline solution was
infused at a rate of 10 mL·kg–1. Patients were kept
warm using a heated blanket, and iv fluids were
Variables
Fourteen preoperative variables, quantitative and quali-
tative, were collected retrospectively. Chronic
hepatopathy was divided into seven categories: alco-
holic cirrhosis, chronic hepatitis, biliary cirrhosis, meta-
bolic disease, cancer, cholangitis and cryptogenic
cirrhosis. Three surgical periods were identified as relat-

Steib et al.: PREDICTING HEMORRHAGE DURING OLT
1077
TABLE I Demographic data and blood product requirements
(mean SD)
liver disease. Preoperative portal hypertension, present
in cases of ascites and/or esophageal varices, was con-
sidered as another possible risk factor. Other preopera-
tive variables were: sex, age, hemoglobin concentration
Hb, platelet count, prothrombin time (PT), activated
partial thromboplastin time (aPTT), fibrinogen con-
centration, fibrin degradation products (FDP) and
euglobulin lysis time (ELT).
HBL
LBL
(n=110)
(n=300)
Age (yr)
48 11
50 11
Weight (kg)
70 14
71 14
Sex M/F (%)
Duration of surgery (min)
RBCs (units)
68/32
65/35
537 138
18.2 6.9
23 9.5
17.8 11.7
373 102***
5.9 2.9***
10.7 4.8***
9.1 9.4***
FFP (units)
Statistical analysis
Platelets (units)
Aprotonin (KIU)
Univariate analysis was performed using a chi-square
test for the comparison of discrete variables and an
unpaired Student’s t test for the comparison of two
means. Results were expressed as mean SD. A P value
<0.05 was considered statistically significant. A stepwise
logistic regression analysis, with blood loss representing
the dependent variable, used all the variables included
in the univariate analysis as independent variables.
Calculations were performed by a Statview 5 program
(Abacus Concepts, Inc). Sensitivity (Se), Specificity (Sp)
and positive (PPV) and negative (NPV) predictive val-
ues were calculated for each isolated predictive factor
and then in combination.
507 407 251 984 338 333 153 707***
HBL=high blood loss; LBL=low blood loss; RBC=packed red
blood cells; FFP=fresh frozen plasma. ***P <0.001.
TABLE II Preoperative biological variables
HBL
LBL
(n=110)
(n=300)
– 1
Hb beginning of operation(g·L )
93 17
94 16
116 73
51 14
1.4 0.4
105 18***
105 15***
132 83
60 17***
1.3 0.2
2.3 1***
25 59***
96 62
– 1
Hb end of operation(g·L )
– 1
Platelets (x 109· L )
Prothrombin time (%)
aPTT (ratio)
– 1
Fibrinogen (g·L )
2
0.8
– 1
FDP (mmoL·L )
56 118
106 59
Results
ELT (min)
Demographic data and blood product requirements
are summarized in Table I. Figure 1 shows the distri-
bution of RBC transfusions in the population.
There were no significant differences in age, weight
and sex distribution between groups of patients. Chronic
liver diseases leading to transplantation were comparable
(Figure 2). Previous upper abdominal surgery was more
HBL=high blood loss; LBL=low blood loss; Hb=hemoglobin con-
centration; aPTT=activated partial thromboplastin time; FDP=fib-
rin degradation products; ELT=eugobulin lysis time. Prothombin
time is expressed as the % of control value (normal range:
70–100%) and aPTT is expressed as the ratio aPTTpatient/aPTT
control (normal value below 1.25)
***P <0.001.
frequent in the HBL group (P <0.001). HBL was also
TABLE III Sensitivity (Se), Specificity (Sp), positive predictive
value (PPV) and negative predictive value (NPV) of each putative
predictor of high blood loss (HBL) and their combination.
significantly associated with surgery using a venovenous
bypass (31% vs 13% in the piggyback technique), severe
liver disease (50% in class C) and the period of initial sur-
gical experience (50% during the period 1988–1990 vs
17% in the last period) The incidence of preoperative
portal hypertension was similar in both groups. Finally,
Table II shows that patients with HBL were more ane-
mic at the beginning of the procedure, had significantly
increased FDP values and decreased PT (%) and fibrino-
gen values before incision.
The stepwise multivariate analysis disclosed only three
variables which were independently associated with intra-
operative HBL: Hb, FDP values assessed before surgery
and previous upper abdominal surgery. When Se, Sp,
NPV and PPV were considered separately for each of
these three variables, they were not of relevant clinical
interest. However, when analyzed in combination, speci-
ficity increased to 98% when Hb <100 g·L– 1, FDP >24
mmol·L– 1 and previous upper abdominal surgery had
been performed.
Se (%) Sp (%) PPV (%) NPV (%)
(A) Hb <100 g·L–1
(B) FDP >24 mmoL·L– 1 57
71
60
71
73
40
43
29
85
81
85
(C) Previous abdominal
surgery
45
(A) + (B)
(A) + (C)
(B) + (C)
(A) + (B) + (C)
42
34
27
18
89
92
94
98
59
33
54
69
81
84
84
82
ing to the progressive experience of the surgical team:
1988–1990 corresponded to the beginnings of hepatic
transplantation, 1991–1993 was the period of maturity
and 1994–1998 was a period of innovation with the
introduction of the piggyback technique. Previous
upper abdominal operations were recorded. Child-
Pugh’s classification was used to assess the severity of

1078
CANADIANJOURNAL OF ANESTHESIA
detect the risk of perioperative hemorrhage would allow
anesthesiologists to choose which patients are more like-
ly to benefit from antifibrinolytic agents such as apro-
tinin, tranexamic acid or epsilon-aminocaproic acid.
Although the blood sparing effect of these agents is rec-
ognized,12,13 their systematic use may be hazardous due
to adverse effects such as severe allergic reactions and fatal
intraoperative thromboembolism.14,15 On the other
hand, the influence of transfusion on outcome and graft
survival has been known for several years.2,4–6 Patients
who are not transfused or receive less than five units4 have
a lower morbidity and mortality. Therefore, limiting the
use of blood products should improve patients’ outcome.
For all these reasons, several attempts have been made to
identify patients with a high risk of bleeding and to define
preoperative factors which could influence intraoperative
bleeding.5,7–10 Results have been contradictory, mainly
because of non- homogenous patient selection, surgical
conditions and intraoperative management.
FIGURE 1 Percentile plot for red blood cell (RBC) require-
ments (units) per patient. The upper quartile represents the high
blood loss group (HBL). These patients required 12 units of RBC
or more to maintain hemoglobin $ 100 g·L– 1.
Our study focussed on chronic liver diseases and pre-
operative clotting data. Its first objective was to estimate
the actual quantity of blood loss. Blood losses are very
difficult to measure during liver transplantation. Most
research has quantified blood losses indirectly, using
blood transfusion requirements necessary to maintain
or reach a trigger value. Hb represents the most fre-
quent criterion for transfusion.1 1 In our study, the goal
was to maintain a Hb at around 100g·L–1 (Ht ~ 30%).
The second step consisted of determining which
patients could be considered to have sustained high
bleeding. As in another study,11 we arbitrarily divided
the population into two groups according to their intra-
operative RBC requirements. The cutoff point thus
obtained was approximately 12 units. Recently, a simi-
lar cutoff point was suggested by a study of 1,443 con-
secutive liver transplant recipients in which patients
requiring more than 12 units of RBCs intraoperatively
had a significantly decreased survival rate.5
Our results revealed that previous abdominal
surgery, preoperative low Hb and elevated FDP were
specific to predict HBL. This latter factor was also
identified as a predictive variable in a previous small
study of cirrhotic patients who experienced severe
intraoperative hyperfibrinolysis.7 Its preoperative
increase may reflect an imbalance between factors pro-
moting fibrinolysis or a low grade of disseminated
intravascular coagulation. The predictive role of Hb
found in our study was not surprising. Patients with an
initial low Hb, below 100 g·L–1, would require trans-
fusion in order to reach the selected trigger point.
However the difference in Hb between groups (93 vs
105 g·L– 1) was insufficient to explain the large differ-
ence in RBC requirements (5.9 vs 18.2 units).
FIGURE 2 Indications for liver transplantation in both groups
of patients. HBL=high blood loss; LBL=low blood loss; AC=alco-
holic cirrhosis; CH=chronic hepatitis; BC=biliary cirrhosis;
MD=metabolic disease; CC=cryptogenic cirrhosis. No significant
differences between groups.
Discussion
Despite recent progress in anesthetic and surgical care,
severe hemorrhaging remains a problem during liver
transplantation. The ability to predict intraoperative
bleeding would be of great use in the global management
and organization of the surgical procedure. Were this
information available, the blood bank could be alerted,
the most appropriate monitoring and venous access cho-
sen, and the surgeon better prepared.1 1 Being able to

Steib et al.: PREDICTING HEMORRHAGE DURING OLT
1079
2 Palomo Sanchez JC, Jimenez C, Moreno Gonzalez E, et
Surgical conditions seem to play an important role.
In a study of 164 recipients,8 experience of the surgi-
cal team was identified as an independent predictor of
transfusion. This was not the case in our study, despite
a significant decrease in the number of patients in the
HBL group between the first and the third observa-
tion periods. Deakin et al., before the piggyback tech-
nique era, advocated the role of technical factors such
as the use of a venovenous bypass and an argon beam
coagulator, both devices contributing to the decrease
of blood product requirements.1 Previous abdominal
surgery was independently associated with massive
transfusion in a recent report of 100 patients.2 Our
study corroborates this finding. All these results sug-
gest that surgical conditions are related, at least to
some extent, to the importance of blood loss. In par-
ticular, a history of previous abdominal surgery should
make every surgeon cautious.
Another variable commonly correlated with blood
loss is the severity of liver disease. Severity is assessed
either globally by Child-Pugh’s classification8 or, more
specifically, by hemostatic data (PT, aPTT, platelets,
ELT, FDP) and/or biological1,9 variables (urea, crea-
tinin, bilirubin). However, correlation is poor and most
authors have not been able to recommend specific pre-
operative screening tests. Surprisingly, clotting data
were not independent predictors of HBL in previously
published studies.5,8,10 The largest of these comprised
1,143 patients.5 As has been pointed out,1 3 the
observed discrepancies may be due to the heterogenity
of the recipient population. Patients with acute and
chronic liver disease are often analyzed together, blur-
ring the results. In addition, the choice of different trig-
gers for blood transfusions, associated to different
management protocols guided or not by coagulation
monitoring, with or without the use of antifibrinolytic
agents, may explain these conflicting results.
al. Effects of intraoperative blood transfusion on post-
operative complications and survival after orthotopic
liver transplantation. Hepatogastroenterology 1998;
45: 1026–33.
3 Spahn DR, Casutt M. Eliminating blood transfusions.
Anesthesiology 2000; 93: 242–55.
4 Cacciarelli TV, Keeffe EB, Moore DH, et al. Effect of
intraoperative blood transfusion on patient outcome in
hepatic transplantation. Arch Surg 1999; 134: 25–9.
5 Cancemi E, Ramsay KJ, Ramsay MAE, et al. Liver
transplantation and blood use: predictors and impact
on outcome. Anesthesiology 1999; 89; A419
(abstract).
6 Schroeder RA, Johnson LB, Plotkin JS, et al. Total blood
transfusion and mortality after orthotopic liver trans-
plantation (Letter). Anesthesiology 1999; 91: 329–30.
7 Steib A, Gengenwin N, Freys G, Boudjema K, Levy S,
Otteni JC. Predictive factors of hyperfibrinolytic activi-
ty during liver transplantation in cirrhotic patients. Br J
Anaesth 1994; 73: 645–8.
8 Hendriks HG, van der Meer J, Klompmaker IJ, et al.
Blood loss in orthotopic liver transplantation: a retro-
spective analysis of transfusion requirements and the
effects of autotransfusion of cell saver blood in 164
consecutive patients. Blood Coagul Fibrinolysis 2000;
11(Suppl. 1): 87–93.
9 Findlay JY, Rettke SR. Poor prediction of blood trans-
fusion requirements in adult liver transplantations from
preoperative variables. J Clin Anesth 2000; 12:
319–23.
10 Gerlach H, Gosse F, Rossaint R, Bechstein WO, Neuhaus
P, Falke KJ. Die relevanz perioperativer gerinnungspa-
rameter für die indikation von bluttransfusionen
(German). Anaesthesist 1994; 43: 168–77.
11 Ozier YM, Le Cam B, Chatellier GI, et al.
Intraoperative blood loss in pediatric liver transplanta-
tion: analysis of preoperative risk factors. Anesth Analg
1995; 81: 1142–7.
Blood losses during OLT seem to vary from centre
to centre. Therefore, each centre should evaluate its
own practice in an attempt to identify patients requir-
ing specific attention in the area of transfusion and the
use of antifibrinolytics. In our centre, the combination
of a Hb below 100 g·L–1, FDP 24 mmol·L– 1and pre-
vious abdominal surgery is highly specific to predict
HBL during OLT for chronic liver disease.
Unfortunately, specificity is low, hampering our
efforts to correctly identify patients at risk of intraop-
erative hemorrhage.
12 Dalmau A, Sabaté A, Acosta F, et al. Tranexamic acid
reduces red cell transfusion better than - aminocaproic
acid or placebo in liver transplantation. Anesth Analg
2000; 91: 29–34.
13 Porte RJ, Molenaar IQ, Begliomini B, et al. Aprotinin
and transfusion requirements in orthotopic liver trans-
plantation: a multicentre randomised double-blind study.
EMSALT study group. Lancet 2000; 355: 1303–9.
14 O’Connor CJ, Roozeboom D, Brown R, Tuman KJ.
Pulmonary thromboembolism during liver transplanta-
tion: possible association with antifibrinolytic drugs and
novel treatment options. Anesth Analg 2000; 91: 296–9.
15 Laxenaire MC, Dewachter P, Pecquet C. Allergic risks of
aprotinin. Ann Fr Anesth Reanim 2000; 19: 96–104.
References
1 Deakin M, Gunson BK, Dunn JA, et al. Factors influ-
encing blood transfusion during adult liver transplanta-
tion. Ann R Coll Surg Engl 1993; 75: 339–44.