Neurocrit Care (2008) 8:253–258
DOI 10.1007/s12028-007-9020-4
ORIGINAL RESEARCH
Perioperative Neurological Complications After Liver
Transplantation are Best Predicted by Pre-transplant Hepatic
Encephalopathy
Rajat Dhar Æ G. Bryan Young Æ Paul Marotta
Published online: 10 October 2007
Ó Humana Press Inc. 2007
Abstract
Introduction Liver transplant (LT) recipients are at
significant risk for the development of neurological
complications, such as altered mental status and seizures, in
the postoperative period. Identifying accurate predictors of
these events may allow optimal selection and preparation of
candidates, and minimize risk after transplantation.
Methods One hundred and one consecutive adult LT
recipients were evaluated retrospectively for neurological
morbidity occurring in the first 30 days postoperatively.
These events were analyzed in relation to specific predic-
tive variables including preoperative complications of liver
failure, such as hepatic encephalopathy (HE).
Results Median age was 50 years, 63% were male and
hepatitis C was the most common indication for LT
(n = 36). Median Child-Pugh score was 9 with 45% being
Class C. Over half (n = 52) had experienced clinical HE
prior to LT, while one quarter (n = 26) were encephalo-
pathic at the time of LT. Neurological complications
occurred in 31 patients in the postoperative period, with
encephalopathy occurring in 28 and seizures occurring in
4; drug toxicity was responsible for neurological morbidity
in 12 patients (39%). Length of hospital stay was
R. Dhar (&)
Department of Neurology, Washington University School of
Medicine, 660 S. Euclid Ave, Campus Box 8111, Saint Louis,
MO 63110, USA
e-mail: dharr@wustl.edu
G. B. Young
Department of Clinical Neurological Sciences,
University of Western Ontario, London, ON, Canada
P. Marotta
Multi-Organ Transplant Unit, London Health Sciences Centre,
London, ON, Canada
significantly prolonged (median 19 vs. 12 days, P = 0.005)
and all mortality (n = 3) occurred in those with neurolog-
ical complications. There was no association between
etiology of liver failure and complications; logistic
regression identified active preoperative HE as the stron-
gest predictor of postoperative morbidity (OR 10.7 95% CI
3.8–29.9).
Conclusion Neurological events, manifesting most often
as encephalopathy, occurred in almost one-third of patients
after LT. Those suffering from HE at the time of LT may
be more vulnerable to the metabolic stresses of surgery and
the neurotoxicity of the drugs used, and were at highest risk
for such complications.
Keywords Liver transplantation
Hepatic
encephalopathy
Complications
Postoperative

Seizures
Drug toxicity
Introduction
Although liver transplantation (LT) extends survival and
improves quality of life for patients suffering complica-
tions of severe hepatic dysfunction [1], postoperative
complications remain a significant source of morbidity and
mortality [2]. The brain, functionally and structurally
altered in patients with liver disease [3, 4], may be pref-
erentially vulnerable to perioperative insults and the
toxicity of the immunosuppressive regimens utilized.
Adverse neurological events are known to complicate 10%
or more of LT procedures [5–8], with the majority occur-
ring in the first month postoperatively [9, 10]. Common
complications include postoperative delirium, encepha-
lopathy, and seizures; drug-related neurotoxicity from
calcineurin inhibitors (cyclosporine or tacrolimus) plays a
254 Neurocrit Care (2008) 8:253–258

role in a significant proportion of these events, with a considered present if the measured value rose by 10 mmol/l
greater risk after LT than after other transplants [11]. or more in the perioperative period. The immunosuppres-
Few clinical studies have both accurately assessed the sive regimen used was recorded.
frequency of neurological complications in this high-risk
population as well as tried to identify factors that may predict
Outcome Measures
these adverse events [12]. Knowledge of these factors may
help optimize selection and preparation of candidates for LT,
The primary outcome was the development of any neuro-
in order to minimize the risks for those undergoing this
logical dysfunction in the 30-days after LT. This duration
procedure. In this study, we assessed the incidence, severity,
of follow-up was selected because it encompassed the
and implications of neurological complications in the
majority of complications in a number of previous studies
immediate postoperative period after LT. We then analyzed
[9, 10], and comparable reliable data were available for all
the relationship of these events to important preoperative
transplant recipients over this well-defined period. The
variables, including history of hepatic encephalopathy (HE),
complications ascertained included: (1) Encephalopathy:
to identify factors that place recipients at higher risk for
including delirium, psychosis, or diminished level or con-
postoperative neurological dysfunction.
sciousness, not necessarily related to hepatic dysfunction;
(2) Seizures; (3) Drug neurotoxcity: specifically that
ascribed to use of calcineurin inhibitors or corticosteroids;
Methods
(4) Vascular insults including ischemic or hemorrhagic
stroke, and hypoxic brain injury; (5) CNS infections; (6)
Patients
Central pontine myelinolysis (CPM). Complications were
also qualitatively graded as being either mild (if isolated
One hundred and one consecutive adult recipients from our
and short-lasting, requiring minimal intervention) or severe
regional transplant center who underwent liver transplan-
(morbidity that was persistent or required active interven-
tation over an 18-month period between September 2000
tion, such as recurrent seizures or significantly reduced
and March 2002. Approval for this study was obtained
level of consciousness). Other outcomes evaluated inclu-
from our institution’s Research Ethics Board.
ded time to extubation, length of stay in the hospital after
LT, and mortality prior to hospital discharge.
Data Collection
Statistical Analysis
Pre-specified data were abstracted from hospital records and
a transplantation database. This included patient demo-
Descriptive statistics were used to present baseline and
graphics, etiology of liver failure, urgency status and time
operative variables, along with specific complications.
on the wait list, preoperative metabolic status, including
Subjects were divided into those experiencing postoperative
calculation of Child-Pugh score (based on assessment of
neurological complications and those not, and these groups
preoperative liver function, ascites, and HE) [13]. Com-
were compared using the v?-statistic (for categorical vari-
plications of liver failure, including ascites, variceal
ables) and the student’s two-tailed t-test (for continuous
bleeding, HE, and hepatorenal syndrome were categorized
variables). Significant pre- and postoperative variables are
as either absent, mild if well-controlled, or severe when
presented in tabular format showing univariate odds ratio for
persistent and/or prominent (for example requiring repeat
complications, with 95% confidence intervals. All preoper-
hospital admissions). The occurrence of active HE at the
ative variables associated at P < 0.010 with major outcomes
time of LT was specifically noted. Operative variables
were entered into a backward stepwise logistic regression
included length of procedure, blood transfusion require-
model with significance set at P < 0.05 (corrected for
ments, significant intraoperative hypotension (systolic
multiple analyses using the Bonferroni correction). Analyses
blood pressure <80 mm Hg for 10 min or more) or the
were performed using SPSS for Windows (version 11.5;
occurrence of the reperfusion syndrome [14]. Postoperative
SPSS, Chicago, IL, USA).
medical complications were recorded including renal fail-
ure (threefold rise in serum creatinine or requiring renal
replacement therapy), pneumonia, myocardial infarction Results
(MI), sepsis (systemic inflammatory response with hypo-
tension or positive cultures), and graft dysfunction Descriptive features of the 101 patients comprising the
(encompassing primary non-function, acute rejection, and study population as well as operative variables are outlined
portal vein thrombosis). A major rise in serum sodium was in Table 1. The majority of LT recipients had at least one
Neurocrit Care (2008) 8:253–258 255

Table 1 Patient characteristics Table 2 Complications pre- and post-transplantation

Variables Frequency (n = 101) Complication Frequency (mild/severe)

Age, years (median, range) 50 (17–72) Preoperative complications

Sex (male/female) 64/37 Hepatic encephalopathy 52 (28/24)
Status at time of transplant Preoperative encephalopathy 26 (11/15)a
Status 1–2 84 Ascites 69 (28/41)
Status 3–4 17 Variceal bleeding 36 (12/24)
Waiting time, d (median, range) 146 (1–1212) Hepatorenal syndrome 13 (5/8)
Etiology of liver failure Postoperative complications
Hepatitis Ca 36 Neurological complications
Primary biliary cirrhosis 13 Encephalopathy 28 (17/11)
Alcoholic cirrhosis 10 Seizures 4 (3/1)
Primary sclerosing cholangitis 9 Drug toxicity 12 (10/2)
Autoimmune hepatitis 7 Medical complications
Fulminant hepatic failure 6 Renal failure 23
NASH 6 Pneumonia 7
b
Others 14 Myocardial infarction 5
Child-Pugh score (median, range) 9 (5–14) Sepsis 4
Class A (score 5–6) 17 Graft dysfunction 6
Class B (score 7–9) 39 a
Eight patients had HE only in the immediate preoperative period
Class C (score 10–15) 45 without any prior history
Operative variables (median, range)
Duration of surgery (minutes) 395 (260–850)
infection, or CPM. There was significant overlap between
RBC transfusions intraoperative 4 (0–18)
confusional states and drug neurotoxicity, with 11 of the 12
FFP transfusions intraoperative 8 (0–24)
patients felt to have drug-related morbidity presenting with
Intraoperative hypotension 21
altered mental status. Conversely, seizures were not com-
Reperfusion syndrome 2
monly associated with drug toxicity (one of four patients)
a
Alcohol was implicated as an additional contributing factor in 8 of or even persistent mental status changes. Length of stay
these
b
was significantly longer in those with postoperative neu-
Other causes incl. cryptogenic, Budd-Chiari, alpha1-antitrypsin
rological events (median 19 vs. 12 days, P = 0.005). The
deficiency
three patients who died prior to discharge all experienced
NASH = non-alcoholic steatohepatitis
postoperative encephalopathy. Mortality was also strongly
associated with postoperative sepsis, renal failure, graft
preoperative complication of liver failure (Table 2). dysfunction, as well as intraoperative hypotension.
Notably, just over half had a history of HE and 26 were
encephalopathic at the time of the LT (including seven in
hepatic coma). Major shifts in perioperative sodium levels Variables associated with Neurological Complications
were noted in 26 patients (with a maximal rise of 0.5–
1.0 mmol/l per hour in most cases). Factors associated with neurological complications are
shown in Table 3. Although a history of HE was strongly
associated with morbidity, on further analysis, it was only
Outcomes the group with episodes of severe HE that had a higher risk
of complications (13 of 22 vs. 7 of 28 if only mild HE).
Time to extubation in the ICU was a median 22 h (range There was no association between etiology of liver failure
12–468). Length of hospital stay was a median 12 days or status on the transplant list and occurrence of neuro-
(range 6–101). The vast majority (87%) were discharged logical morbidity. Those with alcoholic cirrhosis did not
within a month of LT. Overall, 31 patients suffered at least have a higher rate of postoperative confusion (3/10) than
one neurological complication after LT. The most frequent those with other causes for cirrhosis (11/36 in those with
are shown in Table 2. One encephalopathic patient suffered hepatitis C). All four patients with sepsis experienced
a hypoxic–ischemic brain injury prior to death from sepsis. postoperative encephalopathy, associated with a seizure in
There were no confirmed cases of stroke, hemorrhage, CNS one.
256
Table 3 Factors associated
Variable, n (%) Complications (n = 31)
with neurological complications
Preoperative factors
Age (mean ± SD) 52.4 ± 9.9
Child-Pugh Class C 21 (68)
Hx of HE 21 (68)
Preoperative HE 18 (58)
Hepatorenal syndrome 8 (26)
Postoperative factors
Renal failure 11 (36)
Sepsis 4 (13)
*Variables entered into the Pneumonia 5 (31)
multivariate analysis
In multivariate analysis of the four preoperative vari-
ables associated with postoperative neurological
complications, only HE in the immediate preoperative
period (adjusted OR 10.7, 95% CI 3.8–29.9) remained
significantly associated (P < 0.0125). Patients with FHF,
many of whom (four of six) were in hepatic coma prior to
surgery, did not suffer an excess rate of complications
including encephalopathy (2/6 vs. 29/94, P = 1.00) and
they all survived to discharge. Excluding this low-risk
group, the rate of postoperative neurological complications
was 16 out of 21 amongst those with active preoperative
HE. No specific variables were associated with postoper-
ative seizures, likely as too few events occurred to
discriminate. There was an equivalent rate of drug-induced
neurotoxicity distributed between Tacrolimus (11/86, 13%)
and Cyclosporine (1/10, 10%, P = NS); the only factor
associated with drug toxicity was postoperative pneumonia
(OR 6.1, 95% CI 1.1–33.1).
Discussion
In this study of 101 consecutive patients undergoing LT,
neurological complications occurred in almost one-third in
the immediate postoperative period. This rate is concordant
with a number of previous reports, highlighting the con-
siderable frequency at which neurological morbidity occurs
in this setting [7, 10]. The burden associated with these
events is also significant; we found an almost doubling in
hospital stay after transplant surgery in those suffering such
complications. All deaths occurred in those with nervous
system dysfunction, likely related to their strong associa-
tion with sepsis and graft failure, which both induce
metabolic encephalopathy.
In fact, encephalopathy, the most common neurological
complication in our and most other series [6, 15], is usually
multifactorial, related to metabolic derangements, organ
failure, and drug toxicity. The latter, largely caused by
Neurocrit Care (2008) 8:253–258
No complications P value Univariate
(n = 70) OR (95% CI)
48.8 ± 12.6 0.16
24 (34) 0.002* 4.0 (1.6–9.9)
*
31 (44) 0.03 2.6 (1.1–6.4)
8 (11) <0.001* 4.1 (1.7–9.8)
5 (7) 0.02* 4.5 (1.3–15.2)
12 (17) 0.04
0 0.002
2 (3) 0.015
calcineurin inhibitors, occurred in almost one in eight
patients after LT and accounted for almost 40% of com-
plications. Alterations in the blood–brain barrier due to
preexisting HE may make these patients more vulnerable to
neurotoxicity than in other transplant settings [11]. The
incidence of seizures appears to have fallen in this and
other contemporary series [6, 16, 17], perhaps with closer
monitoring of drug levels and cautious dose initiation and
adjustment. We found only a single case of a seizure
attributable to drug toxicity in our series, compared to a
25% rate of seizures after LT in an older series, where most
were related to cyclosporine neurotoxicity [5]. We also did
not find any difference in rates of toxicity between cyclo-
sporine and tacrolimus. Intrinsic brain processes, such as
vascular insults and CNS infections did not contribute to
postoperative neurological dysfunction in this series.
Despite sodium shifts being relatively frequent, CPM was
not diagnosed in any of our patients. With a documented
rate of 1% in a number of studies [7, 10, 18], this may not
be surprising for a study of this size.
There was no increased rate of confusional episodes in
patients transplanted for alcoholic cirrhosis compared to
those with hepatitis C. This contrasts with a previous study
that showed an eightfold higher rate of encephalopathy in
the group with alcoholic cirrhosis [19]. In that report a
much higher rate of preoperative HE was present in the
alcoholic cirrhosis group; furthermore, those with acute
confusional states had higher preoperative ammonia levels.
It is more likely that it is not the precise etiology of liver
failure that contributes to postoperative confusion, but how
vulnerable the brain is prior to LT. Alcoholics with prior
episodes of HE, high levels of ammonia, and additional
neurotoxicity from their alcohol exposure, may simply
have more preexisting brain injury placing them at higher
risk for complications.
This is supported by the major finding of our analysis,
that the strongest predictor of postoperative neurological
morbidity was the presence of active HE at the time of
Neurocrit Care (2008) 8:253–258
transplantation. This association has been suggested in a
prior report: a prospective study of 84 patients seen by a
neurologist before and serially after LT found that an
abnormal neurological examination prior to LT was the
strongest predictor of postoperative neurological compli-
cations [12].
There is a strong and coherent rationale for why those
with active HE would be at particularly high risk for
postoperative encephalopathy. It is now clear that liver
failure with even subtle or subclinical HE induces signifi-
cant changes in cerebral function and morphology,
primarily affecting astrocytes [20]. As the site of ammonia
detoxification, astrocytes accumulate the osmotically active
metabolite glutamine in the setting of liver failure. This
promotes intracellular swelling with depletion of the
compensatory organic osmolyte, myo-inositol serving as a
marker, as measured by magnetic resonance spectroscopy
(MRS) studies [21]. Increased astrocyte water content can
disrupt cellular function including maintenance of the
blood–brain barrier and neurotransmitter clearance. As
important constituents of the glio-neuronal network,
changes in astrocyte activity can induce widespread neu-
ronal dysfunction. Increased GABA-ergic tone is also
found in HE, likely mediated through increased production
of neurosteroids as a consequence of astrocyte swelling.
Elevations of extracellular glutamate, which accumulates
in this setting, can activate NMDA receptors and induce
nitric oxide synthesis, cerebral vasodilatation, as well as
seizures [22]. There is also a correlation between grade of
HE and MRI abnormalities before transplantation, with
evidence of subtle brain edema manifested in increased
ADC values even in those with subclinical HE [3]. Ele-
vation in ADC correlated with venous ammonia levels,
while more overt T2-abnormalities were only seen in those
with grade 2 or greater HE.
Even though liver transplantation ultimately improves
the selective cognitive deficits seen in mild HE [23], the
stress of the procedure itself can trigger worsening cerebral
dysfunction. Administration of benzodiazepines, shifts in
osmolality, and cytokine release known to occur after
major surgery, are all well known triggers of HE; their
contributions to postoperative encephalopathy may be
explained by their modulation of astrocyte swelling.
Patients with limited reserves of organic osmolytes,
depleted in attempting to compensate for active preopera-
tive HE, are unable to counteract the increased swelling
precipitated by these additional perioperative stressors and
so are at greater risk for cerebral dysfunction after LT.
Limitations of this study include its retrospective nature.
Extraction of relevant complications depended on adequate
recognition and documentation by the transplant and inten-
sive care services looking after the patients. However, these
patients are followed very closely in the post-transplant
257
period by a dedicated and trained team; most persistent and
significant changes in neurological status were likely noticed
while subtle symptoms may have been missed. We also
could only classify pre-transplant HE qualitatively into mild
or severe forms in this retrospective series, and did not utilize
the four-stage HE grading scale, which may have allowed a
more refined analysis of the association between degree of
HE and postoperative complications.
A prospective study examining the role of preoperative
HE in the pathogenesis of postoperative neurological
morbidity would allow a more accurate quantification of its
role and impact of its severity; examining interventions to
stabilize neurological status prior to LT and their effects on
the rate of these complications may also be warranted.
Reducing the frequency of these morbid events may reduce
the length of stay after LT and improve the neurological
status of these patients at discharge.
References
1. Moore KA, Jones R, Burrows GD. Quality of life and cognitive
function of liver transplant patients: a prospective study. Liver
Transpl. 2000;6:633–2.
2. Mazariegos GV, Molmenti EP, Kramer DJ. Early complications
after orthotopic liver transplantation. Surg Clin N Am
1999;79:109–29.
3. Lodi R, Tonon C, Stracciari A, et al. Diffusion MRI shows
increased water apparent diffusion coefficient in the brain of
cirrhotics. Neurology 2004;62:762–6.
4. Weissenborn K, Bokemeyer M, Ahl B, et al. Functional imaging
of the brain in patients with liver cirrhosis. Metab Brain Dis
2004;19:269–80.
5. Adams DH, Ponsford S, Gunson B, et al. Neurological complica-
tions following liver transplantation. Lancet 1987;1(8539):949–51.
6. Guarino M, Stracciari A, Pazzaglia P, et al. Neurological com-
plications of liver transplantation. J Neurol 1996;243:137–42.
7. Lewis MB, Howdle PD. Neurological complications of liver
transplantation in adults. Neurology 2003;61:1174–8.
8. Menegaux F, Keefe EB, Andrews BT, et al. Neurological com-
plications of liver transplantation in adult versus pediatric
patients. Transplantation 1994;58:447–50.
9. Borhani Haghighi A, Malekhosheini SA, Bahramali E, et al.
Neurological complications of first 100 orthotopic liver trans-
plantation patients in southern Iran. Transplant Proc
2005;37:3197–9.
10. Bronster DJ, Emre S, Boccagni P, Sheiner PA, Schwartz ME,
Miller CM. Central nervous system complications in liver
transplant recipients—incidence, timing, and long-term follow-
up. Clin Transplant 2000;14:1–7.
11. Bechstein WO. Neurotoxicity of calcineurin inhibitors: impact
and clinical management. Transpl Int 2000;13:313–26.
12. Pujol A, Graus F, Rimola A, et al. Predictive factors of in-hos-
pital CNS complications following liver transplantation.
Neurology 1994;44:1226–30.
13. Pugh RN, Murray-Lyon IM, Dawson JL, Pietroni MC, Williams
R. Transection of the oesophagus in bleeding oesophageal vari-
ces. Br J Surg 1973;60:646–9.
14. Aggarwal S, Kang Y, Freeman JA, Fortunato Jr FL, Pinksy MR.
Postreperfusion syndrome: hypotension after reperfusion of the
transplanted liver. J Crit Care 1993;8:154–60.
258
15. Vecino MC, Cantisani G, Zanotelli ML, et al. Neurological
complications in liver transplantation. Transplant Proc
1999;31:3048–9.
16. Glass GA, Stankiewicz J, Mithoefer A, Freeman R, Bergethon
PR. Levetiracetam for seizures after liver transplantation. Neu-
rology 2005;64:1084–5.
17. Wijdicks EF, Plevak DJ, Wiesner RH, Steers JL. Causes and
outcome of seizures in liver transplant recipients. Neurology
1996;47:1523–5.
18. Abbasoglu O, Goldstein RM, Vodapally MS, et al. Liver trans-
plantation in hyponatremic patients with emphasis on central
pontine myelinolysis. Clin Transplant 1998;12:263–9.
19. Buis IC, Wiesner RH, Krom RA, Kremers WK, Wijdicks EF.
Acute confusional state following liver transplantation for alco-
holic liver disease. Neurology 2002;59:601–5.
Neurocrit Care (2008) 8:253–258
20. Haussinger D, Kircheis G, Fischer R, Schliess F, vom Dahl S.
Hepatic encephalopathy in chronic liver disease: a clinical man-
ifestation of astrocyte swelling and low-grade cerebral edema? J
Hepatol 2000;32:1035–8.
21. Laubenberger J, Haussinger D, Bayer S, Guffler H, Hennig J,
Langer M. Proton magnetic resonance spectroscopy of the brain
in symptomatic and asymptomatic patients with liver cirrhosis.
Gastroenterology 1997;112:1610–6.
22. Haussinger D, Schliess F. Astrocyte swelling and protein tyrosine
nitration in hepatic encephalopathy. Neurochem Int 2005;47:
64–70.
23. Mattarozzi K, Stracciari A, Vignatelli L, D’Allessandro R,
Morelli MC, Guarino M. Minimal hepatic encephalopathy: lon-
gitudinal effects of liver transplantation. Arch Neurol 2004;61:
242–7.