Perioperative Management of Patients With Liver Di
Perioperative Management of Patients With Liver Di
doi: 10.1016/j.bjae.2021.11.006
Advance Access Publication Date: 31 January 2022
111
Perioperative management of patients with liver disease
Definition and classification of liver disease distortion of the hepatic architecture and eventually cirrhosis
(which strictly speaking is a histological diagnosis). Although
The term ‘liver disease’ comprises a large spectrum of path-
liver biopsy is now less commonly performed in patients with
ological processes that lead to impairment of the synthetic
cirrhosis, the term ‘advanced CLD’ is used where advanced
and metabolic functions of the liver. There are numerous
fibrosis/cirrhosis is suspected from radiological or endoscopic
definitions of hepatic pathology, including how long liver
findings. Patients with CLD may be termed ‘compensated’
function has been impaired, whether the liver is failing, or in
(asymptomatic, often undiagnosed) or ‘decompensated’
the case of cirrhosis, if the patient’s condition is compensated
(defined by the development of either jaundice, ascites,
or decompensating.
hepatic encephalopathy, and/or variceal haemorrhage).7
Cardiovascular Hyperdynamic circulation with high cardiac output and low systemic vascular resistance
Relative hypovolaemia secondary to systemic and splanchnic vasodilatation
Cirrhotic cardiomyopathy (diastolic dysfunction, prolonged Q-T interval and a blunted contractile response to stress)8,9
Respiratory Mechanical compression of lungs from large volume ascites and hepatomegaly can lead to impaired lung
movement, alveolar hypoventilation, decreased FRC and atelectasis
Hepatic hydrothorax
Hepatopulmonary syndrome (dyspnoea and hypoxaemia in the setting of chronic liver disease caused by
intrapulmonary arteriovenous shunting)10
Portopulmonary hypertension (an increase in intrapulmonary vasoconstriction and marked vascular remodelling
in patients with portal hypertension in the absence of other causes of arterial or venous hypertension)11
Gastrointestinal Portal hypertension (secondary to an increased pressure gradient between the portal vein and hepatic veins),
manifesting as varices, splenomegaly and ascites12
Spontaneous bacterial peritonitis
Central nervous Hepatic encephalopathy (a potentially reversible neuropsychiatric abnormality thought to result from
system decreased metabolism of neurotoxins such as ammonia, short-chain fatty acids and mercaptans)14
Nutrition Malnutrition
Muscle wasting
Hypoglycaemia
Hypoalbuminaemia
Impaired wound healing
Haematological Coagulopathy
Anaemia
Thrombocytopenia
Hypofibrinogenaemia
Haemostatic Balance
Fig 1 The haemostatic balance in chronic liver disease. The simultaneous altered synthesis of both procoagulant and anticoagulant factors seen in chronic liver
disease results in a stable haemostatic balance. FRC, functional residual capacity; t-PA, tissue plasminogen activator.
consequence of reduced synthesis of clotting factors. More more recently, the Mayo Clinic postoperative mortality risk in
recent evidence suggests that such patients are actually in a patients with cirrhosis calculator, and the VOCAL-Penn
dynamic state of ‘rebalance’ whereby distorted synthesis of model.
anticoagulant factors is offset by coinciding alterations in
procoagulant factors (Fig. 1).15 The ChildeTurcotteePugh score
Thrombocytopenia secondary to splenic sequestration and The CTP score is calculated using the patient’s serum bilirubin
impaired hepatic thrombopoietin synthesis is also commonly and albumin concentrations, INR and severity of encepha-
seen in patients with cirrhosis; however, platelet function is lopathy and ascites. The patient is consequently assigned to
normally preserved because of an increase in endothelial- one of three CTP classes (A, B or C), which are used to predict
derived von Willebrand factor (vWF) concentrations. The in- postoperative morbidity and mortality in both hepatic and
ternational normalised ratio (INR) therefore provides a poor non-hepatic intra-abdominal surgery.18 Weaknesses of the
representation of the haemostatic balance and risk of scoring system include the subjective nature of ascites and
bleeding in patients with CLD, and patients who are stable encephalopathy grading, the equal weighting given to all the
should not be assumed from the outset to have a tendency to variables and the arbitrary selection of cut off values for its
bleed. If intraoperative bleeding does occur, it is more typi- parameters.
cally related to portal hypertension (with little relevance to
haemostatic factors), hypervolaemia, or mucosal or traumatic
The MELD and UK model for ESLD score
bleeding caused predominantly by hyperfibrinolysis. In pa-
Three variables are used to calculate the MELD score: creati-
tients with ALF and ACLF whose clinical state may be
nine, bilirubin and INR (with the addition of sodium to
complicated by sepsis and multiorgan failure, this haemo-
formulate the UK model for end-stage liver disease [UKELD]).19
static balance is somewhat less predictable.16
MELD scores of <10, 10 to 15, and >15 approximate to CTP
classes A, B, and C, respectively (Fig. 2); however, as always,
Preoperative risk assessment these are approximations and the balance of risks and bene-
fits should be assessed in every case on an individual basis.17
As a consequence of the loss of functional hepatic reserves
and associated systemic impairments, surgery in patients
with CLD is associated with a high morbidity and mortality.
Mayo Clinic postoperative mortality risk in patients with
The individual’s response to surgical stress is impaired and
cirrhosis calculator
The patient’s age, ASA class and aetiology of cirrhosis are used
they are at increased risk of a myriad of complications
in the web-based Mayo Clinic postoperative mortality risk in
including bleeding, infection, impaired wound healing,
patients with cirrhosis calculator, and serve to increase the
hepatic decompensation and acute renal failure.17 The deci-
predictive value of the test.20 However, the model is still not
sion to perform surgery should therefore be weighed care-
perfect as data from both elective and emergency surgeries
fully, and preoperative assessment should involve risk
were combined together in the prediction modelling (which
stratification and an informed consent process.
may confound results), and factors such as the severity of
portal hypertension (a major contributing factor to post-
Patient-related risk factors operative morbidity and mortality) are not included.
There are no universally accepted scoring systems used to
assess patients with CLD before surgery, but those most VOCAL-Penn model
commonly used include the ChildeTurcotteePugh (CTP) Only recently published (and encompassing data relating to
score, the model for end-stage liver disease (MELD) score, and age, serum albumin and bilirubin, platelet number, surgery
Preoperative patient
with liver disease
Fig 2 Preoperative considerations as to whether to proceed with elective surgery in patients with different degrees of impaired liver function. ALF, acute liver
failure; ALI, Acute liver injury; CTP, ChildeTurcotteePugh score; MELD, model of end-stage liver disease score.
category, emergency indication, fatty liver disease, ASA class with or without urgent transfer of the patient to a tertiary liver
and obesity), the VOCAL-Penn Model is thought to substan- centre. In elective cases, patients with CLD have an increased
tially improve postoperative mortality prediction in patients (but varying) risk of postoperative morbidity and mortality
with cirrhosis.21 This, however, remains to be determined. with all surgical procedures, and this risk increases in parallel
Figure 2 details a flowchart to help determine whether to with their CTP or MELD score.2 Those surgeries with particular
proceed with elective surgery in patients with different de- increased risk include cardiothoracic surgery, hepatobiliary
grees of liver impairment. Importantly it should be considered surgery and intra-abdominal procedures. Several case series
as a ‘rule of thumb’, for surgical risk is continuous and there have thus suggested that laparoscopic surgery should be
are no absolute cut-off values for excluding patients with performed whenever possible, as it is associated with less
cirrhosis from surgical procedures. In addition, the CTP and tissue trauma, traction of the abdominal viscera, disruption of
MELD scoring systems do not consider many of the extrahe- the splanchnic vasculature and collateral vessels, risk of
patic manifestations of liver disease e some of which are bleeding and infection; all of these significantly affect the risk
associated with much higher rates of decompensation after of postoperative decompensation.
surgery (e.g. portal hypertension). High-risk patients should Asides from major surgery, patients with CLD often require
always be discussed with a specialist liver centre. invasive procedures that may be associated with blood loss.
Although these may be grouped into those associated with
low and major risk of bleeding (Table 2), this risk dichotomy is
Procedure-related risk factors based purely on expert opinion: there are no specific cut-off
In tandem with the patient’s individual risk factors, the type values for INR or platelet concentration based on definitive
and anatomical site of the proposed surgical procedure, data that indicate reliably when the risk of bleeding is
including whether it is an elective or emergency procedure, increased.
has major implications for postoperative hepatic dysfunction
and overall outcome. Emergency surgery is associated with a
four- to fivefold increase in the relative risk of morbidity and
History, examination and investigations
mortality: this will not be directly represented by MELD-based A thorough preoperative assessment including a history, ex-
predictive models. Unless life- or limb-threatening (and amination and investigations is mandatory for all patients
should the time allow), surgery should therefore be deferred if with liver disease e the extent of which will be determined by
at all possible, to allow for optimising the patient’s condition, the urgency of the surgery required. Particular focus should
Low risk (major bleeding risk <1.5%) High risk (major bleeding risk >1.5%)
Peripheral and central venous catheter (CVC) insertion and removal Organ biopsies
Coronary angiography Tumour ablation
Diagnostic oesophagogastroduodenoscopy and variceal ligation Transjugular intrahepatic portosystemic shunt
Gastrointestinal endoscopies Endoscopic polypectomy
Transoesophageal echocardiography Therapeutic coronary angiograph
Diagnostic bronchoscopy Gastrostomy placement
Dental cleaning and non-extraction procedures Dental extraction
determine the following questions. (i) What is the severity of does not reduce the risk of bleeding, is not associated with
the liver disease? (ii) Does the patient have any extrahepatic increased thrombin generation, and may even tip the scale
manifestations of CLD? (iii) Is the patient decompensating, in towards hypercoagulability because of the procoagulant
ACLF, or is there any evidence of concurrent sepsis? Disease- changes in fibrin structure and platelet hyperreactivity.
specific considerations should also be taken into account, for Recent studies therefore suggest that although rotational
example what dose of steroids is a patient with autoimmune thromboelastometry (ROTEM) or rotational thromboelastog-
hepatitis taking (as stress-dosed steroids may be required raphy (ROTEG) only assess ex vitro aspects of clotting, they do
intraoperatively)? provide a better assessment of coagulation status than con-
Important signs and symptoms include those suggestive of ventional coagulation studies. The new guidelines represent a
advanced CLD in general (e.g. jaundice, caput medusae, spider significant change to standard practice whereby the routine
naevi, hepatosplenomegaly, gynaecomastia, asterixis and use of prophylactic FFP before surgery is no longer recom-
encephalopathy) and those that may compromise anaes- mended unless indicated by a viscoelastic haemostatic assay
thesia such as large volume ascites and pleural effusions. (VHA).16
Such findings may inform subsequent investigations (Table 3). A low fibrinogen concentration has been associated with
an increased risk of bleeding in patients with decompensated
disease, particularly ACLF; absolute concentrations >1.0 g L 1
Preoperative optimisation before invasive procedures are thus a reasonable target. If
All patients with liver disease should be physiologically opti- hypervolaemia is a concern, then cryoprecipitate should be
mised before surgery, and early involvement of an experi- used rather than FFP.
enced multidisciplinary team has been shown to improve A platelet count >50109 L 1 is adequate to allow clot for-
outcomes. Please refer to Table 1 for the relevant aspects to mation in most patients with cirrhosis. Prophylactic trans-
consider for optimisation, and to Supplementary Table S1 fusions to raise the values higher than this have not been
(online only) for more details on specific management. shown to be beneficial, and they expose the patient to po-
tential complications of transfusions, including portal vein
thrombosis.21
Haematological optimisation It should also be remembered that in the event that blood is
In light of recent evidence supporting the notion of ‘reba- required, patients with CLD who have had previous variceal
lanced haemostasis’ in the patient with CLD, the traditional bleeding have often received multiple previous transfusions,
approach of correcting of a coagulopathy (indicated purely by and may therefore have atypical antibodies. Blood for these
a deranged INR) before an invasive procedure by giving patients will need extended cross-matching and blood prod-
coagulation factors, in particular fresh frozen plasma (FFP) ucts from regional blood banks, all of which take time and
and cryoprecipitate, has been superseded.16 Blood product require advanced preparation.
transfusion is not a benign intervention and exposes the pa-
tient to numerous risks including transfusion reactions and
Intraoperative management
hypervolaemia with a consequential increase in portal hy-
pertension. Furthermore, it is now widely accepted that in At a minimum, full monitoring as per Association of Anaes-
patients with cirrhosis, FFP cannot correct an INR below 1.7, it thetists guidelines is required for all patients; however, in
light of the altered haemodynamics associated with CLD,
invasive blood pressure monitoring is advised. In major sur-
Table 3 Investigations in patients with liver disease. ROTEG, gery, in patients with a high CTP or MELD score, or both, a
rotational thromboelastography; ROTEM, rotational central venous catheter is recommended (although not
thromboelastometry without risk) to provide the ability to monitor and rapidly
correct electrolytes, measure central venous pressure and to
Investigations Includes infuse vasopressor drugs if required. Wide-bore cannulae
should also be placed if there are concerns about potential
Haematological Full blood count
Liver function tests
blood loss, and a rapid infuser device should be available if
Renal function and electrolytes necessary.
Coagulation profile (including fibrinogen) Fluid shifts secondary to drug-induced vasodilation or
Dynamic viscoelastic haemostatic assays drainage of large volume ascites should be anticipated, and
(e.g. ROTEG or ROTEM)
volume status needs to be constantly evaluated as patients
Cardiac ECG are at high risk of postoperative acute kidney injury (AKI) and
Transthoracic echocardiography
the development of hepatorenal syndrome. Observation of
(to quantify bi-ventricular function,
estimate pulmonary artery pressures and
traditional variables such as cardiovascular variables, urine
assess for cardiomyopathy) output and blood loss may be augmented by the use of cardiac
Cardiopulmonary exercise testing output monitoring. However, the presence of oesophageal
(particularly useful as a means to uncover varices is a relative contraindication to the placement of
systolic dysfunction which can be masked
oesophageal Doppler and transoesophageal echocardiogra-
by hyperdynamic circulation and
decreased afterload) phy probes. Alternative methods should be considered
including PICCO (pulse index continuous cardiac output),
Respiratory Arterial blood gas
Chest X-ray
FloTrac or pulmonary artery catheter monitoring. Isotonic
crystalloids should be used to replace fluids in the first
Hepatic Hepatic venous pressure gradient
instance (given via a fluid warming device), with FFP, cry-
(to assess degree of portal hypertension)
oprecipitate or fibrinogen as guided by VHA. Human albumin
solution may be used for plasma expansion if more than 5 L
ascites is drained from the abdomen, and packed red blood is also the preferred opioid for postoperative analgesia
cells if the haemoglobin concentration decreases. (including for use in patient-controlled analgesia), as it does
The volume of blood transfused correlates directly with not have an active metabolite and is excreted renally, although
postoperative complications and mortality; therefore, both it does accumulate when given in repeated or large doses.
meticulous surgical technique alongside optimal anaesthetic Other choices for postoperative analgesia include oxycodone
management should aim to minimise blood loss, and this in- and tramadol (in carefully titrated doses whilst monitoring for
cludes the avoidance of hypothermia, acidaemia and hypo- signs of sedation and encephalopathy). Non-steroidal anti-in-
calcaemia. Vasopressor drugs may be used once the patient’s flammatory drugs should be avoided because of the risk of
fluid status is optimised, and this may include stepwise treat- nephrotoxicity, platelet dysfunction and gastrointestinal hae-
ment with metaraminol, phenylephrine and noradrenaline morrhage. Paracetamol may, and furthermore should, be used
(norepinephrine). Terlipressin may be added if the patient’s in patients with CLD and ALF (unless it is the aetiology of the
systemic vascular resistance remains low. Prophylactic anti- latter) in an attempt to reduce opioid use; the dose should be
biotics should be given because susceptibility to infection is decreased (e.g. 1 g t.d.s.). Regional analgesia in the form of local
increased as a result of altered functions of the hepatic retic- infiltration or transverse abdominis plane blocks may be used
uloendothelial cells. Close monitoring of the patient’s glycae- to aid postoperative analgesia, and epidural placement can be
mic control is needed (hypoglycaemia should be avoided). considered assuming the INR is <1.5, the platelet count
If thought appropriate for the nature of surgery, and >100,000 mm 3 and/or the VHA is normal.
providing there is no coagulopathy, liver disease in itself is not
a contraindication for neuraxial anaesthesia; however, there is Postoperative considerations
also no strong outcome evidence to support its use over general
Patients with liver disease are at high risk of postoperative
anaesthesia. Both reduce mean arterial pressure and thus have
morbidity including wound infections and dehiscence, spon-
the potential to decrease hepatic blood flow. Intraoperative
taneous bacterial peritonitis, pneumonia, AKI, and hepatic
hypoxaemia is not uncommon in patients with severe liver
dysfunction and decompensation. Unless the patient has well
disease because of the presence of pleural effusions (which
compensated liver disease (i.e. CTP class A) and only a minor
restrict alveolar ventilation and reduce functional residual ca-
procedure has been undertaken, patients should therefore be
pacity), ventilationeperfusion mismatch associated with hep-
admitted to either the high-dependency unit (HDU) or ICU after
atopulmonary syndrome, or both. General anaesthesia with
surgery. If not extubated in the operating theatre, patients
the ability to optimise ventilation may be preferred, but care
should be warmed, woken and weaned from artificial ventila-
should be taken on induction of anaesthesia as patients are
tion as soon as possible to allow for a thorough neurological
susceptible to aspiration of gastric contents.
assessment; close observation for signs of encephalopathy,
Altered drug metabolism and decreased synthesis of
jaundice, ascites and coagulopathy is mandatory. Should any
plasma binding proteins dictates that perioperative medica-
of these signs of decompensation present after surgery, the
tions should be selected carefully, although the choice of
patient should be referred urgently to a tertiary liver centre.
which drug to use or avoid is much less important than the care
Prothrombin time is perhaps the single best indicator of syn-
with which they are given. In general, premedication with
thetic liver function (as increased serum bilirubin could also be
benzodiazepines is avoided as they may result in prolonged
the result of blood transfusions, blood extravasation or infec-
depression of consciousness and have the ability to precipitate
tion). However, an increased INR in a patient with CLD does not
encephalopathy. Propofol may be used for i.v. induction;
necessarily preclude them from being in a hypercoagulable
however, it should be given at a reduced dose as the cardio-
state, and nor does it appear to protect patients from hospital-
respiratory depressant effects are increased. Thiopental and
acquired deep venous thromboses or pulmonary emboli.
etomidate have also been used, although they seem to offer no
Venous thromboembolism prophylaxis must therefore be
selective advantage over propofol when titrated carefully.
considered on an individual basis after discussion with the
Owing to a lack of hepatic excretion, atracurium and cis-
surgical team. Sedatives and analgesia should be carefully
atracurium are commonly used neuromuscular blocking
titrated, with laxatives prescribed to diminish the chance of
agents; however, increased doses may be required because
hepatic encephalopathy. Renal function must be closely
their volume of distribution is increased and protein binding
monitored and careful attention paid to the i.v. infusions and
altered. Vecuronium and rocuronium have a prolonged elimi-
serum electrolytes. If AKI occurs, early consultation with renal
nation phase in severe liver disease, and the metabolism of
physicians and critical care is advised. Sepsis, often caused by
suxamethonium may take longer because of reduced pseu-
Gram-negative organisms, should be treated aggressively with
docholinesterase concentrations. Maintenance of anaesthesia
broad-spectrum antibiotics (such as piperacillin/tazobactam or
may be achieved using volatile agents or propofol via a target-
a third-generation cephalosporin). Hypo- and hyperglycaemia
controlled infusion (TCI). Although all volatiles can decrease
are both common and associated with higher morbidity and
cardiac output (and in turn hepatic blood flow), they undergo
mortality, and thus blood glucose concentrations should be
minimal hepatic metabolism and are regarded as safe (with the
monitored and managed appropriately.
exception of halothane, which is seldom still used in the UK).
Propofol-based TCI may be supplemented with remifentanil to
provide intraoperative analgesia, and this is perhaps regarded Declaration of interests
as the safest opioid to use as it is metabolised by tissue and red The authors declare that they have no conflicts of interest.
cell esterases, which e unlike plasma esterases e are pre-
served in patients with hepatic disease. Morphine should be
used with caution as its elimination is delayed in patients with
MCQs
liver disease as a consequence of both reduced hepatic blood The associated MCQs (to support CME/CPD activity) are
flow and extraction ratio, and carefully titrated short-acting accessible at www.bjaed.org/cme/home for subscribers to BJA
opioids such as alfentanil or fentanyl are preferred. Fentanyl Education.