Alimentary Pharmacology & Therapeutics
Volume 37, Issue 12, pages 1132–1156, June 2013
J. H. Lewis*, J. G. Stine
Article first published online: 3 MAY 2013
© 2013 John Wiley & Sons Ltd
Most drugs have not been well studied in cirrhosis; recommendations on safe use are based largely on experience and/or expert opinion, with dosing recommendations often based on pharmacokinetic (PK) changes.
To provide a practical approach to prescribing medications for cirrhotic patients.
An indexed MEDLINE search was conducted using keywords cirrhosis, drug-induced liver injury, pharmacodynamics (PDs), PKs, drug disposition and adverse drug reactions. Unpublished information from the Food and Drug Administration and industry was also reviewed.
Most medications have not been adequately studied in cirrhosis, and specific prescribing information is often lacking. Lower doses are generally recommended based on PK changes, but data are limited in terms of correlating PD effects with the degree of liver impairment. Very few drugs have been documented to have their hepatotoxicity potential enhanced by cirrhosis; most of these involve antituberculosis or antiretroviral agents used for HIV or viral hepatitis. Paracetamol can be used safely when prescribed in relatively small doses (2–3 g or less/day) for short durations, and is recommended as first-line treatment of pain. In contrast, NSAIDs should be used cautiously (or not at all) in advanced cirrhosis. Proton pump inhibitors have been linked to an increased risk of spontaneous bacterial peritonitis (SBP) in cirrhosis and should be used with care.
Most drugs can be used safely in cirrhosis, including those that are potentially hepatotoxic, but lower doses or reduced dosing frequency is often recommended, due to altered PKs. Drugs that can precipitate renal failure, gastrointestinal bleeding, SBP and encephalopathy should be identified and avoided.
The use of medications in patients with cirrhosis frequently raises several concerns, especially for gastroenterologists and hepatologists who are often asked for their opinion regarding the safety of drugs in this setting.[1-3] The chance that acute drug-induced liver injury (DILI) might develop and worsen the underlying liver disease is one of the most common fears of prescribing prescription as well as over-the-counter (OTC) agents. However, the risks of precipitating or worsening renal failure, inducing gastrointestinal (GI) bleeding, and provoking hepatic encephalopathy (HE) are, in fact, more likely scenarios in the cirrhotic patient. The changes that may occur in drug disposition, metabolism, excretion and elimination as a result of cirrhosis, the presence of ascites or the influence of transjugular intrahepatic portosystemic shunts (TIPS) or other factors on hepatic blood flow, present unique challenges in determining how best to prescribe the multitude of agents that are used to treat both hepatic and nonhepatic-related disorders in this population, as will be discussed.
Frequently encountered situations where drugs are used to manage the various complications of cirrhosis include the use of diuretics, beta blockers, medications for the treatment of insomnia and sleep reversal, alcohol withdrawal, acute and chronic pain relief, and the use of sedatives and anaesthetics for endoscopic and surgical procedures, in addition to the routine use of medications to treat a variety of other comorbid medical conditions, including short- and long-term use of acid-suppressive medications, and various antimicrobial agents. Herbal and other complementary and alternative medications are commonly used by patients with chronic liver disease (CLD), including those with advanced cirrhosis, and may be associated with unanticipated adverse effects.
Among the many drugs that may be used in cirrhosis, there are a number of potentially hepatotoxic medications that are often of particular concern in patients with end-stage liver disease. Chief among them are paracetamol (acetaminophen) and other OTC pain relievers.[4-6] Years ago, Zimmerman, the father of modern day drug-induced hepatotoxicity, opined that most patients with pre-existing liver disease were not more likely than others to experience hepatic injury on exposure to drugs that can cause idiosyncratic liver damage. However, he reminded us that despite the lack of data suggesting that most hepatotoxic drugs were harmful in the setting of CLD, should a DILI reaction occur, the consequences could be more dire in patients with impaired hepatic function. A similar view was also voiced by Andreasen who noted that while nonpredictable hepatotoxic reactions did not appear to occur more frequently in patients with liver disease than in others, hepatotoxicity could still be masked by the underlying liver disease or the effects of alcohol.
As a result, many clinicians and patients avoid the use of drugs, such as paracetamol for headaches or other pain relief, as well as the use of statins for treating hypercholesterolaemia in patients with cirrhosis, and oftentimes in patients with lesser degrees of hepatic impairment. The reality, however, is that the number of agents actually reported to increase the risk of hepatotoxicity in CLD remains relatively small (Table 1).
|Antituberculosis drugs (e.g. isoniazid, pyrazinamide, rifampicin)|
|HAART (e.g. nevirapine)|
In particular, the use of antituberculosis drugs (ATDs) and highly active antiretroviral therapies (HAARTs) in patients with chronic viral hepatitis are among the most commonly cited as having an increased risk of acute DILI.[8, 9] There are rare instances and case reports of other agents causing injury in this setting, but in general, most drugs can be and are used safely. In contrast, several clinical examples exist where the treatment of an underlying liver disease is safely conducted with drugs that have the potential for hepatic injury. This includes methotrexate in patients with primary biliary cirrhosis, the use of the glitazones in the management of non-alcoholic steatohepatitis and statins, which are being increasingly used in a number of CLD settings with beneficial effects, among more recent examples. While increased vigilance on the part of prescribers should be conducted whenever drugs are used in CLD, the risk of decompensation or hospitalisation among cirrhotics taking over-the-counter analgesics (OTCAs) was not shown to be increased in two recent studies.[13, 14]
In this review, we summarise the potential risks associated with the use of various medications and drug classes used in cirrhosis based on potential changes in drug metabolism and disposition. In addition, the recommendations for the safe use of some of the most commonly prescribed agents, as made by several authors and research groups, will be offered. Although metabolic changes have been studied for many drugs in patients with cirrhosis going back several decades, the recommendations for safe use in this setting in early reviews mainly involved dose reduction along with monitoring of plasma concentrations  and liver function testing. Not until more recently, however, have formal guidelines for the evaluation of pharmacokinetics (PKs) and dosing adjustments in patients with hepatic impairment been proposed by both the Food and Drug Administration (FDA) as well as the European Medicines Agency (EMEA). Their recommendations are based largely on the Child-Pugh classification to categorise the degree of liver dysfunction in cirrhosis. In the case of the FDA guidance, hepatic impairment PK studies are recommended if hepatic metabolism and/or drug excretion accounts for 20% or more of the absorbed drug, or if the drug has a narrow therapeutic range. PD endpoint assessments should also be completed in such cases; the type and nature of which need to be selected and discussed with FDA staff. Dose reductions are usually required if there is a twofold or greater increase in area under the curve (AUC) in hepatic impairment studies. The EMEA makes similar recommendations in their guidance document, although relatively few agents have been studied adequately in patients with advanced decompensated cirrhosis (e.g. Child-Pugh class C). As a result, expert opinion based on anticipated pharmacological changes often remains the mainstay of current clinical recommendations.
Effects of cirrhosis on drug metabolism
PK and pharmacodynamic changes
As summarised by Delco et al., Verbeeck and others who have studied the disposition of drugs in CLD,[15, 22-26] a number of significant PK changes are known to occur in patients with cirrhosis that often require dose adjustments to use those medications safely (Table 2).
|Pathophysiological factor||Clinical consequence|
|Reduced hepatic blood flow/lower first-pass extraction and portosystemic shunting||Higher bioavailability/serum levels|
|Hypoalbuminaemia||Less protein binding (increased serum concentrations)|
|Ascites/oedema||Increased volume of distribution for hydrophilic drugs|
|Portal gastropathy||Altered (increased or decreased) drug absorption|
|Loss of CYP metabolic activity||Reduced first-pass metabolism/clearance|
|Reduced glutathione stores||Increased toxicity|
|Impaired biliary excretion||Increased serum concentrations|
|Impaired renal excretion||Increased serum concentrations|
As a result of modified PKs due to reduced drug clearance, upregulation of drug receptors and changes in receptor sensitivity, a number of agents may produce adverse clinical effects due to higher plasma drug concentrations in cirrhotic patients and should be used with caution (Table 3). These and other specific factors affecting drug disposition will be discussed below.
|Enhanced pharmacodynamic effects|
|Precipitate encephalopathy||Opioid analgesics, anxiolytics, sedatives|
|Precipitate renal failure||NSAIDs|
|Worsen or precipitate GI bleeding||NSAIDs|
|Decreased therapeutic response seen with|
|Beta adrenoreceptor antagonists|
|Diuretics (furosemide, triamterene, bumetamie)|
Hypoalbuminaemia must also be taken into consideration, in particular for drugs that have a high binding profile, defined as a drug that exists in circulation in bound form >90% of the time. Hypoalbuminaemia can result in important variations in the amount of drug circulating in an unbound form, which would be responsible for drug efficacy and potential toxicity.[20, 21, 24, 27] Reduced protein binding seen with hypoalbuminaemia leads to an increased volume of distribution (Vd). It follows that the larger the fraction of unbound drug, the larger the Vd, implying that the chronic changes seen with cirrhosis will lead to increased Vd for hydrophilic drugs. This is of particular concern when prescribing drugs that are only found in the extracellular compartment (e.g. antimicrobial agents, including Beta-lactams, certain aminoglycosides and both glycol- and lipopeptides). It is unclear if compensatory mechanisms can fully counteract the increased levels of the drug (i.e. theoretically the higher drug concentration could be rapidly distributed and eliminated to obtain a new equilibrium). Acidic drugs (Table 4) have an affinity for albumin and are more likely to be affected by hypoalbuminaemia.[27, 28] Temperature and pH are also known to affect protein binding in vitro, although the clinical effects of these variables remain undetermined for most drugs.
|Assessment||Degree of abnormality||Score|
|Prothrombin time (s > control)||0–3.9||1|
|Total score group severity||5–6 A Mild = MELD 1–10|
|7–9 B Moderate = MELD 11–20|
|10–15 C Severe = MELD >20|
As most drugs that cause hepatotoxicity do so unpredictably via idiosyncratic mechanisms, higher blood levels are not expected to increase the risk of DILI for most agents. It is of interest, however, that even so-called ‘idiosyncratic’ agents may have a dose relationship. Indeed, Lammert et al. found that drugs given in a daily dose of 50 mg or greater were more likely to be hepatotoxic than those given in daily doses of 10 mg or less. Similarly, those agents undergoing extensive (>50%) hepatic metabolism were more likely to be hepatotoxic compared with agents undergoing lesser degrees of hepatic metabolism. Extrapolating these findings to patients with CLD, the changes in metabolism and drug clearance that occur in cirrhosis could theoretically affect a drug's risk of causing DILI; as a result, dose reductions in patients with cirrhosis (with or without renal impairment) have been the general rule dating back more than three decades.[15-17, 20-24]
Drug disposition in patients with ascites
Cirrhosis can have effects on drug absorption, distribution, bioavailability, cytochrome P450 (CYP) metabolism and hepatic and renal clearance mechanisms, resulting in pharmacodynamic (PD) consequences.[2-24, 26] Drug absorption of orally administered agents may be altered by impaired or increased permeability in the presence of mucosal inflammation and oedema seen in portal hypertensive gastropathy, or by delays in gastric emptying. The presence of ascites and oedema can affect the Vd and as a result, both bioavailability and elimination half-life may be affected.[21, 32] The use of serial large volume paracenteses should also be kept in mind when considering dose adjustment of drugs based on ascites and Vd.
There is a rather limited body of information on drug disposition in ascites. Antineoplastic agents that accumulate in ascites (e.g. doxorubicin) have been shown to be cleared much more slowly and have a prolonged half-life. On the other hand, gemcitabine is not distributed in ascitic fluid and its PKs appear unchanged in the presence of ascites.
Other drugs also remain unaffected by ascites and a larger Vd, despite being active predominantly in the extracellular compartment. The half-life of amikacin is similar in both cirrhotic and noncirrhotic patients with normal renal function, despite the presence of ascites. Importantly, multiple reports suggest that the effects of ascites on tacrolimus disposition is negligible, ; only 0.01–0.09% of the dose administered is found in ascites.
Diuretic therapy is commonly used to treat decompensated cirrhotics with ascites. Furosemide and torasemide are two such loop diuretics, both of which have altered PK and PD profiles in cirrhosis. When given intravenously, both drugs had a reduced elimination rate through both renal and nonrenal routes in addition to having a Vd and t1/2 twice normal at steady state  and when compared with healthy subjects. More importantly, the natriuretic potency of both furosemide and torasemide is markedly reduced in the cirrhotic. A poor response to natriuresis appears to be related to altered furosemide kinetics due to decreased delivery of drug to the renal site of action. The concurrent administration of midodrine to enhance the loss in natriuretic response had no measurable clinical effect in one study.
In contrast to loop diuretics, data from Dao and Villeneuve suggest that for triamterene, a potassium-sparing diuretic, despite having markedly elevated plasma levels in cirrhotics with ascites as compared with healthy volunteers, there was no difference in the magnitude of its diuretic response.
Effects of cirrhosis on P-glycoprotein and membrane transporters
P-glycoprotein (P-gp) is a transmembrane transporter encoded by the ABCB1 multidrug resistance gene (MDR1) found on the apical surface of epithelial cells of several tissues, including the small intestine, bile canaliculi and renal tubular cells as well as the luminal surface of endothelial cells of capillaries in the brain and testes. It has many functions, including the absorption and excretion of multiple drugs, and when its activity is altered due to induction or inhibition, there are wide-ranging PK and PD effects that have been seen, as well as drug–drug interactions (DDIs).[41-47] Statins and angiotensin-II receptor blockers (ARBs) have been among the most widely studied agents in this regard,[43, 48-51] although studies on P-gp specifically performed in cirrhotic patients are lacking to our knowledge. Nevertheless, it might be assumed that this transporter would be impaired in cirrhosis in concert with reduced activity of CYP3A4, or in patients with extrahepatic cholestasis, leading to changes in drug disposition. Strong inhibitors of CYP3A and P-gp have the potential to increase the concentration of certainly concomitantly administered drugs, such as statins (as seen with certain protease inhibitors used to treat chronic hepatitis C, and should be used cautiously, if at all, to avoid potentially serious DDIs, particularly in cirrhosis
|St John's Wort||Erythryomycin|
Dose adjustments in cirrhosis
Decreased renal elimination in cirrhosis has been reported for fluconazole, ofloxacin and lithium, among others, and dose adjustments are warranted.[20, 21, 24] Similarly, the anti-viral agents used to treat or as prophylaxis in chronic HBV (entecavir, tenofovir, etc.), are given in reduced dosage (as decreased frequency) for patients with impaired renal function.
The issue of whether CYP enzymes are altered in concentration or impaired in function in cirrhosis is important in terms of phase 1 metabolism of many drugs, especially acetaminophen and various opioid analgesics.[4, 25, 55] With respect to acetaminophen (paracetamol), which is metabolised to its reactive benzoquinone-imine metabolite (NAPQI) by a number of CYP P450 enzymes, including 3A4, 2E1 and 2D6, and which serves as the basis for recommendations to avoid its use in cirrhosis, Benson et al. point out that not only is CYP-450 activity not increased in severe liver disease patients, but several studies have demonstrated that CYP-450 content is either unchanged or is reduced, therefore, potentially leading to a lesser degree of NAPQI formation.
Similarly, the concentration of hepatic glutathione has important consequences for preventing hepatocellular injury to agents such as paracetamol, and concern has been raised that glutathione depletion in cirrhosis could place patients at an increased risk of injury, especially with concomitant chronic alcohol ingestion.[57, 58] However, not all studies have demonstrated decreased glutathione concentrations in this setting. Moreover, Benson et al. note that significant reductions on the order of 30% or greater would be needed before covalent binding of NAPQI or other toxins to hepatocytes would occur. As a result, they opine that concerns about having to avoid therapeutic doses of paracetamol in the setting of cirrhosis are often misplaced.
Drugs with a normally high rate of first-pass extraction by the liver will typically have lower bioavailability. However, as a result of decreased hepatic blood flow, and the presence of portosystemic shunts in cirrhosis, such drugs have increased bioavailability and serum concentrations, and often require dose reductions [e.g. beta adrenergic blockers, calcium channel antagonists, cisapride and other prokinetic agents, antipyschotics, antianxiety and sedative agents, antiparkinson drugs, antidepressants, sumatriptan, certain statins (fluvastatin, lovastatin), morphine].
For intravenously administered agents, it is suggested that a normal initial dose can be given, but subsequent doses should be reduced according to hepatic clearance. However, caution must be exercised with bolus parenteral administration as the initial unbound drug concentrations can be disproportionately high in the setting of hypoalbuminaemia for highly protein-bound drugs as it takes time for plasma binding proteins to exert their effect and re-establish a state of equilibrium. Delco et al. proposed using Doppler flow measurements of portal and hepatic vessels or measuring serum bile acid levels to help gauge the extent of portosystemic shunting and the degree of dose reduction – although this remains only a rough estimation.
For drugs that undergo low first-pass hepatic extraction (<30%), bioavailability is generally not affected in cirrhosis, but clearance may be reduced in the setting of hypoalbuminaemia for those agents undergoing high protein binding, and dose reductions may be needed. For drugs undergoing intermediate first-pass extraction [e.g. codeine, amiodarone, ciprofloxacin, erythromycin, itraconazole, atorvastatin, pravastatin, simvastatin, tricyclic antidepressants (TCAs), omeprazole], the influence of portosystemic shunting in cirrhosis is less than that seen with high extraction agents. As a result, treatment can be initiated with a low normal dose, but maintenance dosing should be adjusted downwards.[20, 21]
For antimicrobial agents affected by a larger Vd due to changes in protein binding and ascites, larger initial doses are suggested in the first 24–48 h, especially in the critically ill patient. Ertapenem and daptomycin are examples of antibiotics where higher initial dosing may be needed. After the initial phase, dosing should be guided by antibiotic clearance, with close attention paid to changes in the GFR of cirrhotic patients.
Potential DDIs in cirrhosis
Franz et al.  recently examined the potential DDIs and adverse effects from polypharmacy in hospitalised patients with advanced liver disease in Switzerland. They retrospectively studied 400 patients with cirrhosis (more than two-thirds of whom were alcoholic), each with a mean of six comorbid conditions. They were taking a median number of five medications (range 0–18), of which a median of 3 (range 0–16) were hepatically eliminated. Among their results, 28% of these patients had one or more adverse drug reactions (ADRs); most of which involved diuretics or NSAIDs, and 21.5% had one or more DDIs, some of which caused severe reactions and/or the need for hospitalisation. The most common ADRs due to DDIs were hyperkalaemia (18.2%), hypoglycaemia (17.4%), an increased bleeding risk (12.9%), respiratory depression (7.6%), increased risk of nephrotoxicity (6.8%) and increased risk of cardiac toxicity (6.8%). Nearly 13% of all potential DDIs resulted in an ADR, and most ADRs occurred in patients with the most advanced cirrhosis (Child-Pugh C), often with renal impairment. Eight per cent of all ADRs directly resulted in hospitalisation. The most commonly responsible drugs were spironolactone, torasemide, furosemide and ibuprofen. Examples of the most common DDIs included potassium-sparing diuretics with ACE inhibitors; beta blockers with insulin; benzodiazepines (BZDs) with opiates and potassium-sparing diuretics with NSAIDs. Their study is a cautionary tale that emphasises the complexities that commonly affect pharmacotherapy and drug handling in the cirrhotic population as previously noted by others.[23, 24] For convenience, Table 10 is provided to show the conversion between Child-Pugh classification and MELD.
Effects of TIPS and other portosystemic shunts on drug metabolism and QTc prolongation
Transjugular intrahepatic portosystemic shunts and other surgical shunts (e.g. Denver shunt) are performed to manage complications from portal hypertension. Patients who have undergone TIPS appear to have changes in drug metabolism. As shown by Chalasani et al., there was a loss in first-pass metabolism of midazolam due to reduced intestinal CYP3A concentrations, which is typically responsible for up to 40% CYP3A activity, leading to increased bioavailability of this medication. Similar findings were found with nifedipine and isradipine, emphasising the need for possible dose adjustments in such patients.
Cirrhotic patients are frequently found to have baseline QTc interval prolongation, likely reflecting an altered ventricular repolarization due to the portosystemic shunting of splanchnic-derived cardioactive substances into the systemic circulation. In their cohort study, Trevisani et al. measured QTc intervals in 19 cirrhotic patients at baseline, 1–3 months and 6–9 months after TIPS and compared them to 10 noncirrhotic portal hypertensive patients. The authors found statistically similar baseline QTc intervals (453 ± 8 ms vs. 465 ± 6 ms); however, 1–3 months after TIPS, the QTc interval increased to 484 ± 7 ms (P = 0.042) and to 480 ± 6 ms (P = 0.03) at 6–9 months, despite relative preservation in liver synthetic function, plasma electrolytes and haemoglobin concentrations. Similarly, Vuppalanchi et al. described a cohort of eight cirrhotic patients with TIPS and six cirrhotics who had not undergone TIPS and observed that baseline QTc intervals were higher in both the TIPS (418 ms) and non-TIPS cirrhotics (431 ms) as compared with nine healthy controls. After a 7-day course of erythromycin (a known QTc interval prolonging medicine metabolised by CYP3A), cirrhotics with a TIPS developed a significantly greater prolongation in their QTc interval (180 ± 68 ms) compared with both the cirrhotics without TIPS (31 ± 10 ms) and with the healthy controls (38 ± 3 ms) (P = 0.03). These findings should prompt caution when prescribing any medications known to prolong QTc in cirrhotic patients who have undergone a TIPS procedure, especially those patients who are being prescribed a fluoroquinolone for SBP treatment or prophylaxis to best prevent potentially fatal ventricular arrhythmias.
Use of specific drug classes in cirrhosis
Many drugs are prescribed in cirrhosis for nonliver-related comorbidities, although how often they are used is less well studied. Lucena et al. from the Spanish Collaborative Study Group on Therapeutic Management in liver diseases have described the prescribing patterns along with the most commonly used medications. For those common conditions requiring treatment among 568 cirrhotic patients from 25 hospitals, medications were given most often for diabetes mellitus in 30%, various infections in 24%, cardiovascular disorders in 20% and for active alcoholism in 15%. Accordingly, the most commonly prescribed medications were sulfonylureas (such as gliblenclamide) for diabetes, fluoroquinolones and amoxicillin-clavulanate for infections, ACE and angiotensin-II receptor inhibitors and calcium channel blockers for hypertension, and lorazepam, chlormethiazole and tiapride (a dopamine receptor antagonist) for alcohol withdrawal symptoms. Most drugs (other than calcium channel blockers) were administered in a reduced daily dose (about two-thirds of normal) to reflect the potential changes in drug disposition in these patients, 82% of whom were Child-Pugh class B or C, with nearly half having ascites. In particular, NSAIDs, antidepressants and anxiolytics were all prescribed less at the time of discharge as compared with on admission, reflecting the treating physician's concern about these classes in cirrhosis over the long term.
These authors noted some interesting prescribing patterns in cirrhosis, namely that certain drugs, ACE inhibitors and ARBs in particular, were frequently used despite being not recommended for patients with cirrhosis and ascites. They noted that amoxicillin-clavulanate was commonly prescribed, despite being the leading cause of antibiotic DILI in general, and with few data to guide its use in cirrhosis. Paracetamol, which undergoes glucuronidation and is considered safe to use in cirrhosis, was prescribed to 30% of patients, including those with alcoholic liver disease, where the risk of DILI is potentially increased. However, it was prescribed in an average dose of not more than 3 g/day as needed.
The following sections describe the disposition of several classes of drugs in cirrhosis as found in the published literature, supplemented, in some cases, by the information describing PK changes seen in patients with hepatic impairment as described in product labelling. The list is not intended to be exhaustive and for specific drugs not addressed in this or other reviews, the interested reader is encouraged to consult the individual prescribing information contained in the Physicians' Desk Reference as well as the FDA website for possible labelling and other safety alerts.
Anaesthetics and sedation for surgery and endoscopy
The effects of anaesthetics and sedatives on the recovery period and mental status of patients with cirrhosis after endoscopy and surgery is a common clinical concern. The use of propofol for endoscopy has been relatively well studied and is not thought to precipitate HE.[67, 68] Given it has a short half-life and rapid action, it appears to be quite safe to use in cirrhosis. Indeed, it is preferable to the use of BZDs, which may lead to prolonged somnolence, encephalopathy and recovery, especially in decompensated cirrhotics.[67, 68] Indeed, the group from Indianapolis found that propofol sedation, even administered by registered nurses with adequate patient monitoring, was well tolerated in patients with cirrhosis undergoing upper endoscopy to screen for oesophageal varices. They noted that patients were more satisfied with the quality of the sedation, and had a quicker return to baseline function compared with patients receiving conscious sedation midazolam with meperidine. While propofol is therefore preferred, careful attention should be paid to potential adverse reactions, such as hypotension, tachycardia, hypoventilation and QTc prolongation that can be associated with its administration.[68-70]
Surgery performed in a cirrhotic patient, especially transplant procedures, has a unique set of challenges for the surgical team, as well as the anaesthesiologist. Among the choices for anaesthetics for liver transplant patients and those undergoing TIPS, factors such as the effects on hepatic blood flow figure prominently in the decision-making process. In a study of 10 patients who first received propofol then desflurane, and a group of another 10 patients receiving desflurane then propofol, Meierhenrich et al. found that propofol anaesthesia was associated with significantly greater preservation of hepatic blood flow compared with desflurane.
In a study of 21 patients with hepatitis C, free and hepatic venous wedge pressures measured using either propofol or desflurane were studied by Mandell et al. Desflurane was found to reduce blood pressure differences between the portal and systemic circulations, leading to errors in the measurements and assessment of the success of treatment for portal hypertension. Propofol had less of an effect on the difference and may be preferred in this setting. Their findings have important implications for patients undergoing TIPS procedure, especially where portal gradients need to be accurately assessed. Desflurane, however, is not metabolised by the liver and would still be preferred to other volatile anaesthetic agents or halothane.
Alonso Menarguez et al. found that there was no higher incidence of acute renal failure using sevoflurane among transplant recipients, although the drug is renally excreted. They suggest that in liver transplant anaesthesia, sevoflurane is as safe as propofol with respect to renal and hepatic function.
With regard to living donors undergoing partial hepatectomy, Ko et al. studied 64 adult donors and compared desflurane with isoflurane anaesthesia. They found better post-operative and renal function tests with isoflurane compared with desflurane at equivalent doses, suggesting that isoflurane may in fact be safer in the donor population, which has important implications for patient and graft survival in the recipients. At our own institution, the preferred agent for liver transplant recipients is isoflurane as well.
There continues to be a controversy over the use of HMG Co-A reductase inhibitors (statins) in patients with CLD and cirrhosis, mostly due to a concern of causing acute-on-chronic liver injury. Although the US FDA currently considers serious statin-related liver injury to be a rare and unpredictable event, it nevertheless remains an important consideration among physicians and patients alike. The FDA has reviewed postmarketing data, cases from US registries of DILI as well as ALF, and other sources of information on statin-related hepatotoxicity, and has concluded that the risk of serious liver injury from all currently marketed statins is ‘very low’ (including a risk of less than or equal to just 2 per 1 million patient-years in the Adverse Event Reporting System database). However, the risk of adverse hepatic events from statins receives renewed (but often undeserved) emphasis with each new case series that appears in the literature. Indeed, the proportion of patients, including those with cirrhosis, who are not being prescribed statins because of the potential for adverse hepatic events, but who might otherwise benefit from their cardioprotective effects, is substantial.
Several recent retrospective and two prospective trials all suggest that patients with compensated liver disease are not at significantly increased risk of statin-induced hepatotoxicity.[12, 76-78] In a large randomised, prospective, placebo-controlled trial with high-dose pravastatin (which, in particular, does not undergo P450 metabolism) in hypercholesterolaemic patients with compensated liver disease due to mostly non-alcoholic steatohepatitis or HCV, the cumulative risk of doubling an elevated baseline ALT value over 36 weeks was in fact lower in the pravastatin-treated group compared with untreated controls (7.5% vs. 12.5%). In a post hoc analysis of the prospective, randomised GREACE trial, statins were found to be cardioprotective in 123 of those patients with moderately abnormal liver tests (mostly from non-alcoholic fatty liver disease). Cardiovascular events occurred in just 10% of the subjects with abnormal LFTs who received statins, compared to 30% of those with abnormal liver tests not receiving a statin (68% relative risk reduction). Moreover, de novo elevations in ALT values were seen infrequently in the study, and those receiving atorvastatin had ‘substantial’ improvement in abnormal baseline LFTs compared with those not treated with the statin.
Thus, the use of statins appears safe in CLD and may in fact have a beneficial effect on fatty liver and viral hepatitis.[12, 76, 78-84] Indeed, in both animal and human models, statins significantly lowered portal hypertension pressure and did not cause any untoward hemodynamic effects, thus improving hepatic perfusion. Moreover, when combined with nonselective beta blockers, the effects of statins were additive.[85, 86] The role of statins in preventing angiogenesis and inducing apoptosis has also been studied recently, perhaps presenting an opportunity for reducing the development of hepatocellular carcinoma.[87-91]
While patients with normal liver tests without underlying liver disease no longer require routine liver test monitoring with statins,[12, 75] it is still considered prudent to monitor ALT when statins are to be started in the setting of advanced fibrosis or cirrhosis. Whether one statin is safer than another in cirrhosis has not been formally studied – although pravastatin, in contrast to the others, does not undergo metabolism by the CYP-450 mixed function oxidase system, which may be impaired in severe hepatic disease.
Regardless of their apparent safety, statins remain largely underprescribed in cirrhotic patients. Franz et al. reported that fewer than 10% of their cohort of 400 patients were being treated with statin therapy, but of those cirrhotics taking a statin, none suffered an ADR.
Herbal products in cirrhosis
The safe use of herbal agents is an important area of concern among patients with cirrhosis and other CLDs, especially given the frequency of their use[93-95] and the potential risk of hepatotoxicity described for many of these products[96-98] Studies by Seeff et al. in cirrhotic patients who were enrolled into the HALT-C trial have provided updated information on the frequency of use of herbal agents in this population. A total of 60 herbal compounds were reported as being used by the 1145 study participants, of whom 23% were currently using them, 21% had taken herbals in the past and 56% reported that they had never used herbals. Silymarin (milk thistle) accounted for nearly three-quarters of all herbal therapy usage (17% of patients), despite no beneficial effects being found on ALT or HCV levels in the study. However, improvements in fatigue, anorexia, nausea and general health were found to be significant among users compared with non-users. No other herbal products accounted for more than 3% of usage, including green tea, garlic, ginseng, Echinacea, grape seed, melatonin, St John's wort, saw palmetto and kava kava. The study was not designed to identify whether further hepatotoxicity developed in any of the herbal therapy users, but relatively few individuals took agents that have been implicated in DILI (e.g. kava kava, valerian).[96, 97] And it should be noted that despite a case series suggesting that green tea may be hepatotoxic, others have questioned the analysis,[101, 102]. The adequacy of the causality assessment process has also been called into question for other herbals, including kava and black cohosh.[103, 104] Bunchorntavakul and Reddy remind us that herbal formulations are often poorly defined in terms of individual constituents, and may contain harmful contaminants, such as lead, mercury, arsenic or other heavy metals that can further confound the issue.
In a further analysis of the HALT-C dataset, Freedman et al. found that silymarin was associated with reduced progression from fibrosis and cirrhosis, although viral outcomes were unaffected. Nevertheless, it appears to be safe in cirrhosis. The herbal agent, glycyrrhizic acid, which is used in the treatment of chronic HCV, is best avoided in cirrhosis because of the risk of hypermineralocorticoid effects producing sodium retention, oedema, hypertension and potassium loss, which may worsen the clinical course of patients who may be receiving diuretics. The traditional Chinese herbal medicine Sho-saiko-to is a mixture of seven herbal preparations that has long been used in the treatment of CLD for its possible antifibrotic effects and appears to protect against the development of hepatocellular carcinoma in cirrhotic patients, perhaps through a reduction in hepatocyte necrosis via inhibiting the activation of stellate cells. Unfortunately, use may be limited by an increased risk of interstitial pneumonia and acute respiratory failure, which can be compounded when given concomitantly with pegylated interferon, and instances of hepatotoxicity have been reported.
Posadzki et al. reviewed the literature for potential interactions that can occur between various herbal agents and medications used for a multitude of disorders. While most herbals were not found to be associated with clinically serious consequences, they did note a number of circumstances where interactions with herbals, such as St Johns wort (Hypericum perforatum) and mistletoe (Viscum album), resulted in transplant rejection, delayed emergence from anaesthesia, renal and hepatic toxicity and a number of cardiovascular insults. St John's wort, in particular, is a strong inhibitor of CYP3A4, and as such is contraindicated in patients receiving a variety of agents, including the new protease inhibitors for the treatment of chronic hepatitis C.
Moderately severe interactions were found for several other herbal agents, including ginko biloba, Ginseng, kava (Piper methysticum), saw palmetto (Seronoa repens) and green tea (Camellia sinensis). They also mention that most of the interactions occurred when antiplatelet and anticoagulant drugs are given concomitantly. Echinacea has been described in multiple reports to have been consumed in large quantities in liver transplant recipients who subsequently presented with episodes of acute rejection, and in a patient with primary biliary cirrhosis who presented with worsening of her liver-associated enzymes and jaundice. The question of hepatotoxicity when Echinacea is combined with methotrexate was recently raised; however, no specific mention is made as to whether the cirrhotic patient is at similar or higher risk. The role of Echinacea as an individual hepatotoxin remains poorly established at this time. As most reports of herbal-related toxicity and drug interactions are of poor quality, it is difficult to draw firm conclusions about such potential interactions. As a result, the clinician (and patient) should always identify all medications, including herbals, that are being taken in the setting of cirrhosis.
Cardiovascular system drugs
Both reversible and fatal cases of DILI from labetalol have been reported.[113, 114] Labetalol is metabolised predominantly by glucuronosyl transferase, and PK data in rats suggest that in NASH, CYP P450 induction leads to increased hepatic removal of labetalol and other cationic beta blockers (e.g. propranolol, metoprolol and atenolol) leading to lower hepatic content of the drug. However, once the liver is cirrhotic, NASH-induced fibrosis and steatohepatitis lead to decreased hepatocyte permeability and increased hepatic sequestration of labetalol. On the other hand, oral administration of labetalol in CLD leads to higher plasma concentrations secondary to decreased first-pass metabolism. Bioavailability also correlates inversely with serum albumin levels leading to clinical implications, including greater falls in heart rate and supine blood pressure. As all beta blockers undergo first-pass metabolism, it can be inferred that CLD may lead to increased plasma concentrations of this class of medications and dose reduction should be considered. In general, labatolol should be used only when there are no other alternative therapies available due to the risk of particularly severe hepatotoxicity, and monitored frequently.
Not surprisingly, cirrhotic patients appear to be more widely prescribed beta blockers than most other classes of drugs with roughly 40% of patients on this type of medication. In their cohort, Franz et al. reported that of the 146 patients taking beta blockers, 7 patients suffered an ADR (with possible causality), one of which was due to a DDI. Only one patient was hospitalised as a result. However, in the series by Lucena et al.  only 7 of 568 patients were prescribed cardioselective beta blockers. The authors did not discuss non cardioselective beta blockers used for portal hypertension.
Eriksson et al. found small, but significant reductions in arterial blood pressure after the administration of captopril in patients with cirrhosis. Aldosterone concentrations decreased, while renin activity increased after captopril. They concluded that while captopril inhibits the renin–angiotensin system in patients with cirrhosis, it fails to significantly decrease portal venous pressure, and was therefore unlikely to protect against variceal bleeding in this population. Enalapril was found to have its bioactivation to enalaprilat significantly impaired in cirrhosis, but its PD effects were unaffected. Similarly, lisinopril had higher serum concentrations in cirrhosis, possibly due to increased drug absorption, and time to peak concentration was longer than for enalapril. The PD effects of these two agents, however, have not been well studied.
As a class, ACE inhibitors appear to be relatively well tolerated in cirrhosis.[59, 63] ADRs are rare. Five per cent of these are mostly due to hyperkalaemia (especially when combined with potassium-sparing diuretics such as spironolactone), urinary system disorders, including kidney injury and worsening liver function. ACE inhibitors accounted for nearly 1/3 of all antihypertensive medications prescribed to cirrhotics in the series by Lucena et al.
Klotz has reviewed the metabolism and PK changes of several antiarrythmics and recommendations for their use in cirrhosis are presented in Table 6. CYP-mediated phase 1 pathways are affected by several agents and require dose reductions as they often have narrow therapeutic indices. PK changes of other agents suggest that the half-life of flecanide is prolonged in cirrhosis, leading to possible accumulation and therefore its use requires dose reduction. In contrast, while the elimination of encainide was seen in cirrhosis, plasma levels of its active metabolites were largely unaffected.
|Drug||Metabolic changes||Recommendations for dosing in cirrhosis|
|Lidocaine||Prolonged elimination half-life|| |
Reduce IV dose by factor of 2 or 3
(Reversible changes after recovery from acute viral hepatitis)
|Quinidine||Half-life increased by 50%, oral clearance unchanged||Reduced dose may be necessary|
|Propafenone||Bioavailability triples||Reduce dose two to threefold|
|Metoprolol||Increased AUC and half-life||Reduce dose based on clinical response|
|Carvedilol||Clearance reduced, half-life prolonged by threefold; bioavailability by fourfold||Start at 20% of usual dose|
|Satolol||No first-pass effect; normal elimination||No dose adjustment needed|
|Amiodarone||Normal half-life is 25–53 days and is predicted to be prolonged in cirrhosis||Dose based on clinical response|
|Dronedarone||Limited clinical experience in cirrhosis||No dose adjustment needed in CP A or B|
|Diltiazem||Half-life prolonged by >50%||May need lower dose|
|Verapamil||Extensively metabolised, clearance reduced two to threefold, bioavailability doubled; half-life prolonged fourfold||Reduce dose by 50%|
|Flecanide||Fourfold increase in half-life||Reduce dose to avoid accumulation|
|Encainide||Fourfold decrease in oral clearance, increased bioavailability of parent, but no change in active metabolites||Dose based on clinical effect|
|Enalapril||Increased Cmax and AUC, but no appreciable difference in inhibition of ACE activity vs. controls||Dose based on clinical effect|
|Lisinopril||Less effect on serum concentration than reduced dose in renal failure with enalapril||Dose based on clinical effect|
|Propranolol||Steady state clearance decreased||Dose based on clinical effect|
|Nadolol||Not metabolised by the liver||Dose adjustment for renal impairment|
Disease modifying antirheumatic drugs
While there is a well-described risk of HBV reactivation with both biologic and nonbiologic DMARD use,[123, 124] much less is known about the safety of these medications in cirrhosis as these patients are generally excluded from clinical trials. Thus, we are left with expert opinion to best guide our therapeutic decisions.[125, 126] For nonbiologic disease modifying antirheumatic drugs (DMARDs), the 2008 American College of Rheumatology (ACR) guidelines advised that despite patients receiving treatment with anti-viral agents for HBV, leflunomide and methrotrexate were contraindicated for all Child-Pugh classifications, and minocycline and sulfasalazine were contraindicated for Child-Pugh Class C. In treatment-naïve hepatitis B patients, the ACR recommended against using hydroxychloroquine in advanced cirrhosis (Child-Pugh Class C), although no specifics were provided. While the ACR acknowledged that TNF-alpha blockade has been used when antihepatitis B therapy has been given prophylactically, they still recommended against the use of any biologics in treated or untreated chronic hepatitis B or hepatitis C patients or those with Child-Pugh class B or C cirrhosis of any aetiology. The updated 2012 ACR guidelines are silent on how to manage rheumatoid arthritis RA patients with hepatitis B receiving nonbiologic DMARDs, and still advise against the use of any biologic agent for RA patients with untreated chronic HBV (due to what they list as ‘contraindications to treatment or intolerable side effects’) and in RA patients with treated hepatitis B if there is evidence of advanced cirrhosis (Child-Pugh class B or C). These guidelines do not address the issue of whether or not RA patients who are receiving concomitant anti-viral therapy for hepatitis B could be treated with biologics, or whether patients with a history of hepatitis B and a positive hepatitis B core antibody would be candidates for biologics. In contrast, the revised guidelines do recommend the use of etanercept for patients with hepatitis C (although there is no mention of the degree of hepatic impairment that might be present). Patients with underlying hepatitis B undergoing cancer chemotherapy (especially those with lymphoma receiving rituximab) should receive prophylactic therapy using one of the nucleoside or nucleotide analogues during and after the course of chemotherapy to prevent reactivation and symptomatic flaring of the hepatitis.[129, 130] Clearly not all societal recommendations are consistent when it comes to these patients with chronic viral hepatitis or cirrhosis, and we await consensus guidelines to be agreed upon.
Historically, biguanides, such as metformin, and other oral hypoglycaemic agents, including the sulfonylureas were often avoided in CLD due to the risk of acute DILI,[1, 113, 131] presenting a conundrum to physicians treating diabetic patients with comorbid liver disease, including NAFLD, which is commonly found in this population. Contrary to what was previously thought, Brackett has opined that metformin is not associated with exacerbation of liver injury or was a significant cause of DILI, and in fact, may be beneficial in the treatment of steatohepatitis and may protect against the development of hepatocellular carcinoma, as reported by others.[133, 134]
Brackett notes that reports of lactic acidosis in cirrhotics are rare, with lactic acidosis being much more likely due to the effects of encephalopathy or hypoxaemia. Furthermore, most of the case reports occurred in cirrhotics using alcohol, there is no reason to withhold metformin from patients with liver disease, and that routine monitoring of liver-associated enzymes during its use is not supported by published evidence. As a case in point, a report of possible hepatotoxicity associated with metformin, involved a 61-year-old obese male who presented with jaundice, nausea, fatigue and weight loss 2 weeks after beginning metformin. Aminotransferase values were elevated in a range of 10–15 times the upper limit of normal. The patient was also receiving an extended release niacin formulation as well as simvastatin, and all medications were discontinued. His symptoms and signs resolved with liver tests returning to normal approximately 2 months later. No rechallenge was performed. While metformin was listed as the possible causative agent, extended release niacin and the statins, certainly could have contributed and remain confounders.
Antidepressants, anticonvulsants and other psychotropic and central nervous system agents
The PK and PD effects of agents in these classes have been studied to varying degrees in patients with cirrhosis.[136, 137] Tables 7-9 summarise the effects of hepatic impairment on the use of these drugs; much of the information on their metabolism and recommendations for dosing can be found in the individual drug labels and patient medication guides that detail prescribing information, either directly from the manufacturers or accessed via the FDA website. While valproic acid is contraindicated in patients with liver dysfunction, all of the other agents listed in Tables 7-9 can be used with caution in cirrhosis.
|Fluoxetine (SSRI)||Reduced clearance of parent and active norfluoxetine metabolite; prolonged half-life||Lower doses or reduced frequency in cirrhosis|
|Duloxetine (SNRI)||85% reduction in clearance in CP class B cirrhotics; threefold increase in half-life; fivefold increase in mean exposure||Current label advises it should not be prescribed to patients with substantial alcohol use or evidence of chronic liver disease|
|Escitalopram (SSRI)||Clearance reduced 37%, half-life doubled with increased plasma concentration||Do not exceed 10 mg/day|
|Venlafaxine (Effexor) (SNRI)||Half-life prolonged by 30%; clearance reduced 40–50% in CP A and B and up to 90% in CP class C||Reduce dose by 50%|
|Amitriptyline||Increased sedative effect in cirrhotics with portocaval shunts||Reduce dose accordingly|
|MAOIs Tranylcypromine (Parnate)||Half-life may be prolonged||Should not be used in patients with abnormal LFTs or history of liver disease|
|Bupropion (Wellbutrin)||Half-life 50% longer in alcoholics; PKs unchanged in CP class A and B, but altered threefold in class C cirrhotics||Use with extreme caution in severe cirrhosis; use lower doses (not to exceed 75 mg/day) and reduced frequency|
|Mirtazapine (Remeron)||Clearance reduced by 30%||Reduced dose may be necessary|
|Lamotrigine||Half-life increased 50% in CP class A and B and by 300% in CP class C||No dose adjustment in CP A, reduce by 25% for CP B and C without ascites, reduce by 50% in CP class C with ascites|
|Phenytoin (Dilantin)||Hypoalbuminaemia increases plasma concentrationChronic alcohol use causes reduced serum levels; acute alcohol causes increased serum levels|| |
Avoid with alcohol
Adjust dose based on serum albumin concentration and monitored blood levels
|Carbamazepine (Tegretol)||Not well studied, but primarily metabolised in the liver|| |
Dose based on monitored blood levels; periodic CBCs recommended;
HLA-B*1502 typing recommended in Asians to predict risk of serious skin reactions
|Valproic acid (Depakote)||Hyperammonemia may confound hepatic encephalopathy; may also cause thrombocytopenia||Contraindicated in patients with liver disease or significant hepatic dysfunction|
|Levetiracetam (Keppra)||PKs unaffected may cause somnolence, fatigue|| |
Use with caution in patients with prior history of liver disease; discontinue immediately for significant hepatic dysfunction
No dose adjustments for use in liver disease
|Haloperidol (butyrophenone)||PKs unaffected|| |
Jaundice reported, avoid alcohol due to additive effects
No specific dose adjustment recommended
|Lithium||Reduced clearance||Toxicity does not include hepatic dysfunction; no specific dosing recommendations for liver disease|
|Diazepam (Valium)||Half-life increased two to fivefold (up to 500 h); clearance decreased by 50%||Use with caution|
|Zolpidem (Ambien)||Half-life increased twofold||Use lowest dose possible|
Selective serotonin reuptake inhibitors (SSRIs), neuroleptics and antiepileptics are prescribed in up to 10% of cirrhotics, and ADRs are not uncommon and typically occur when these medications are prescribed in combination with other potentially harmful agents. For example, aspirin and venlafaxine use together have been reported to cause GI bleeding in cirrhotics; amitriptyline and bactrim prolong the QTc interval and are potentially pro-arrhythmic; imipramine and venlafaxine were reported to also cause QTc prolongation. The combination of SSRIs plus a neuroleptic or antiepileptic also predisposes to serotonin syndrome in the cirrhotic (4 out of 132 DDIs in Katz et al. cohort).
Benzodizepines and alcohol withdrawal syndrome management
The management of alcohol withdrawal syndrome (AWS) is commonly required in the in-patient setting. Benzodiazepines (BZDs) are the cornerstone of therapy; however, the impaired metabolism of BZDs in cirrhotics is problematic as their oxidation is reduced. MacGilchrist et al. demonstrated significant impairment of midazolam metabolism with just a single dose with resultant greater sedation observed 6 h after intravenous administration. The impairment was independent of albumin or serum bilirubin and the authors did not distinguish between compensated and decompensated cirrhotics. Flumazenil may be administered for BZD-induced encephalopathy, however, its use carries an independent risk of inducing seizure activity.
In their cohort study, Franz et al. found that BZDs and related drugs were prescribed in more than 25% of cirrhotics (clorazepam 7%, oxazepam 3.5%, diazepam 3.3%). Similar use was reported by Lucena et al. for a number of these psychotropic drugs. In their multicentre observational series of 568 hospitalised cirrhotics, 44% (n = 248) were patients with alcoholic cirrhosis; however, only 18 patients were hospitalised for alcohol withdrawal. Thirteen of these patients were discharged home on medication for alcohol deprivation syndrome, the majority of patients (81%) received chlormethiazole, an alternative to neuroleptic BZDs. BZDs were however being taken for other indications, including insomnia amongst others. Lorzepam accounted for one-third of the 26 different BZD prescriptions; clorazepate (10%), diazepam (7%) and oxazepam (7%). At discharge, prescribing practices were a little different and reflected a switch to BZDs with short or intermediate half-lives; lorazepam accounted for half of the discharge prescriptions. Other BZDs prescribed included oxacepam, bromazepam, chlorazepam and lormetazepam (8%) each. The authors attribute this change in prescribing to a growing awareness of anxiolytics (such as BZDs) having a risk of precipitating encephalopathy.
Cirrhotics who were prescribed BZDs at discharge were given low doses in accordance with recommendations that if BZDs are to be used, only those agents not metabolised by the liver (lorazepam and oxazepam) with short to intermediate half-lives should be used, in the lowest dose possible. However, regardless of how they are metabolised, BZDs may be associated with an increased risk of excessive sedation, memory deficits, respiratory depression in patients with liver impairment, and abuse and dependence liability. Not surprisingly, the majority of DDIs and ADRs reported by Franz et al. were related to respiratory depression when used in combination with opiates, and to CNS depression/encephalopathy when used as monotherapy.
Recent data in decompensated cirrhotic patients suggest a potential role for gamma hydroxyl butyrate (GHB), a short-chain fatty acid that exerts an ethanol-mimicking effect on the central nervous system by acting on its own receptors and on the GABAB receptors which moderate AWS. However, further study is necessary before recommendations can be made regarding this particular treatment. Of note, GHB is metabolised by the liver, although the relatively short half-life of 4–6 h suggests that it may be well tolerated, even in decompensated cirrhosis.
Maintaining alcohol abstinence is particularly challenging, especially in the cirrhotic. Disulfiram has long been the mainstay of therapy; however, multiple case reports of fulminant hepatitis, some of which described the need for liver transplantation in patients with alcoholic cirrhosis, have been published.[140-143] Such reports suggest that disulfiram should be used very cautiously if at all in cirrhosis. Newer data from Italy point to a potential role for baclofen in the maintenance of alcohol abstinence. A muscle relaxant and GABA inhibitor, baclofen is metabolized to only a limited extent (15%) by the liver. In a randomised double-blind placebo-controlled study of 84 alcohol-dependent cirrhotics, Addolorato et al. found that of the 42 patients allocated to receive baclofen, 71% achieved and maintained abstinence at 12 weeks compared to 29% in the placebo arm. Moreover, the duration of cumulative abstinence was twofold greater for the baclofen group. No adverse hepatic events were reported, and the authors suggest a role for baclofen in this setting.
Antineoplastic and immunomodulating agents
The effects of hepatic impairment on the disposition of antineoplastic agents have been studied to a variable extent.[145, 146] While the metabolism of many classes of anticancer drugs is altered in cirrhosis, it has been noted that for many compounds, dose reductions are not needed when only moderate impairment is present (i.e. Child-Pugh class A and B cirrhosis). For certain agents, such as vincristine, vinblastine, paclitaxel and docetaxel, lower doses are recommended to prevent excessive neutropenia and neurotoxicity. Antineoplastic and immunomodulating agents were prescribed to 15% of cirrhotic patients in one cohort.
Antimicrobials and antiretrovirals
The effects of cirrhosis on the metabolism of various antibiotics, antiretrovirals and ATDs vary according to drug class, and even within medicinal classes. For drugs such as ampicillin, PK parameters are unchanged in cirrhosis, but dose reductions are recommended for patients with renal impairment. Fluoroquinolones are among the most commonly used antibiotics in cirrhosis, especially to treat and prevent spontaneous bacterial peritonitis (SBP). No significant changes in plasma levels or half-life were seen with ciprofloxacin, and no dosing adjustments are necessary in cirrhosis. Ofloxacin metabolism is altered by renal dysfunction in patients with ascites, although the penetration of ofloxacin and pefloxacin into ascitic fluid is excellent, achieving therapeutic levels.[149, 150] However, fluroquinolones may prolong QTc intervals in patients who have undergone TIPS and appropriate care should be taken.
Macrolide antibiotics, including erythromycin, azithromycin, clindamycin and chloramphenicol should be avoided in cirrhotic patients (Table 10). Tetracycline has a prolonged half-life, which corresponds to dose-related hepatotoxicity and should also be avoided. Beta-lactam antibiotics should be used with caution, given their propensity for leucopenia as cirrhotics are predisposed to infection from impaired reticular endothelial cell function and phagocytosis leading to ineffective hepatic destruction of bacteria. Aminoglycosides and vancomycin are generally contraindicated given their relatively high risk of inducing renal failure.
|Nitrofuantoin (chronic use)|
The recommendations for use of anti-TB drugs in cirrhosis are based largely on the known risk of hepatotoxicity in patients with CLD. Rifampicin is eliminated in the bile and can cause elevations in bilirubin due to competitive inhibition of excretory pathways. As a result, it can potentially worsen jaundice in a cirrhotic, and its risk of hepatotoxicity is increased when used concomitantly with isoniazid (INH). Therefore, caution is advised in the patient with hepatic impairment. In patients with CLD, ofloxacin has been used safely as a substitute for rifampicin and may be less hepatotoxic when combined with pyrazinmide compared with rifampin and pyrazinamide. INH has an increased risk of DILI, especially in slow acetylators (slow inactivators) among other genetic polymorphisms, attributed to increased blood levels in cirrhosis.[153, 154] An excess incidence of elevated LFTs has been recorded in patients with hepatitis B and hepatitis C,[155, 156] and while it is generally contraindicated in severe liver disease (PDR), it has been used safely with frequent (every 2–4 weeks) liver enzyme monitoring in cirrhotic patients awaiting liver transplant. Concomitant chronic alcohol use appears to increase the risk of hepatotoxicity. Importantly, INH has been successfully used to treat posttransplant tuberculosis.[159, 160] Pyrazinamide is not recommended for treatment of latent TB in nonhepatically impaired patients, and given the increased half-life seen in hepatic impairment, and the known increase in hepatotoxicity in CLD, it should be monitored very closely in cirrhosis. As noted, when combined with oflaxacin, its risk of liver injury is less compared with rifampin and pyrazinamide. Current recommendations suggest that the Child-Pugh Class A cirrhotics be treated in the same manner as noncirrhotic patients, however, pyrazinamide should be avoided in Class B disease. Ethambutol, a flurouquinolone and a second-line agent from the RIPE regimen, may be used in Child Class C disease.
Antifungal agents, including ketoconazole, miconazole, fulocnazole and itraconazole should be used with caution in cirrhosis, largely due to variable effects on CYP enzyme activity. Voriconazole must be dose-reduced for Child-Pugh Class A and B disease; however, it has not been studied in Class C patients.
HAART agents have historically been associated with a higher risk of DILI in patients with chronic hepatitis B or C,[9, 162-164] and a correlation between the degree of hepatic fibrosis and hepatotoxicity has been found. While not all investigators have found chronic viral hepatitis to be associated with a higher risk of DILI with certain ARTs (e.g. tipranavir plus ritonavir; or from atazanavir/ritonavir), it has been shown that the disposition of several ARTs is significantly altered in CLD or cirrhosis.[168-170] Given that higher serum concentrations of NNRTIs (nevirapine, efavirenz, etc.) seen in cirrhosis are associated with adverse effects, including hepatotoxicity, drug level monitoring has been recommended as a means of preventing injury.[171-174]
Among anti-virals used to treat chronic hepatitis B, the nucleotide and nucleoside analogs are safe for use, including in patients with decompensated cirrhosis, in whom chronic therapy is advised to prevent disease progression. A reduced frequency of dosing is recommended for all of these agents in renally impaired patients. However, hepatic impairment does not alter the PKs of lamivudine or any of the newer nucleoside and nucleotide analogs for chronic hepatitis B as discussed in the drug labels of adefovir, entecavir, telbivudine or tenofovir. In contrast, it is generally advised not to treat patients with advanced cirrhosis from hepatitis C infection with interferon therapy due to the risk of hepatic decompensation (although clinical trials are currently underway in patients who are awaiting liver transplant).
Acid-suppressant medications are prescribed quite frequently in patients with cirrhosis. Indeed, PPIs were the most frequently prescribed class of medication prescribed to cirrhotics (given to approximately 40% of 400 patients) as reported by Franz et al. Lucena et al. reported that nearly one-fourth of the 568 patients in their series were prescribed these medications on admission and 35% on discharge. Ulcer-healing drugs accounted for 12% (n = 2377) of all medications prescribed, with both proton pump inhibitors (PPIs) and H2 blockers being utilised chronically.
Acid-suppressive therapy, including H2-blockers and PPIs, appears to be a potential risk factor for both spontaneous bacterial peritonitis (SBP),[180-183] infection with Clostridium difficile and other serious infections in patients with cirrhosis. The odds ratios for the various infectious associations range from about 1.5 to 3 for PPIs and somewhat less for H2 blockers.[181, 182, 185] While the precise mechanism by which acid suppression increases the risk of infection in this setting is uncertain, these agents are known to facilitate enteric bacterial colonization, overgrowth and gut translocation[183, 184], possibly through changes in intestinal permeability. The studies on which these associations of an increased risk of SBP, etc are based, have been retrospective and observational in nature,[180-185] and also have been criticized for other methodological reasons, including the large number of SBP patients excluded or invalidated, and the fact that many SBP patients were already receiving antibiotics when admitted, making the analysis even more challenging. For example, in the study by Goel et al. from the Cleveland Clinic, Terg notes that only about 10% of more than 1309 patients were eligible for analysis, most of whom had inadequate records and medications to lists to review. Moreover, the risk was apparent only for those patients taking PPIs in the previous 7 days , and why the investigators included patients already on antibiotics was not clearly stated. Terg also mentions that the data on the association of PPIs and bacterial overgrowth have been conflicting.
Although daily fluoroquinolone use has been shown to substantially reduce the risk of SBP when given as primary prophylaxis to high-risk patients with cirrhosis as reported in various meta-analyses,[188, 189] it is unknown whether or not the use of a fluoroquinolone (or other antibiotic) in this setting would mitigate the SBP risk associated with acid suppression in the same population. Since about two-thirds of cirrhotics were found to have no documented indication for the use of a PPI, we join the recommendation by Terg that acid-suppressive medications be used only when clinically indicated in the cirrhotic patient at high risk for SBP and other infections (especially those with ascites). While the risk of SPB was significantly lower in those taking PPIs for more than 90 days, the risk may not be completely eliminated in such patients, and the safety of chronic use of PPIs and H2 blockers in cirrhosis deserves additional study, both in terms of design and duration.
Miscellaneous drugs in cirrhosis
Elthrombopag is an oral thrombopoeitin-receptor agonist that has been in use for management of idiopathic thrombocytopenic purpura (and which was recently approved to allow patients with chronic hepatitis C to receive pegylated interferon-containing anti-viral treatment regimens. It has also demonstrated efficacy in reducing the need for platelet transfusions in patients with cirrhosis and thrombocytopenia who are undergoing invasive surgical and endoscopic procedures. Preapproval PK studies in patients with cirrhosis receiving a single 50 mg dose demonstrated an increased plasma concentration (AUC) and prolongation of half-life with increasing degrees of hepatic impairment. Compared with healthy controls, patients with moderate-to-severe cirrhosis (Child class B or C) had a 93% increase in AUC and more than a doubling (to 114%) of half-life measurements. In contrast, Cmax values decreased with worsening hepatic function, but there was high PK variability between subjects. It is recommended that the initial dose be reduced to 25 mg/day for those with Child class B or C cirrhosis, but that no dose adjustment is needed for renally impaired patients.
Pain management in the cirrhotic patient
Use of OTCAs in cirrhosis
Paracetamol (acetaminophen) is probably the single most feared drug by patients and nonhepatologists alike with respect to its use in patients with liver disease. A study of physician recommendations on the use of OTCAs by Rossi et al. suggested that when used in low doses (2 g or less), it is safe in cirrhosis. While it has been demonstrated that doses of 4 g daily over the course of 2 weeks in healthy volunteers can lead to marked, but clinically silent, elevations in ALT and AST, other groups have not shown significant ALT elevations with doses up to 4 g daily, including in patients who have used alcohol  when used for shorter periods. The issue of ‘therapeutic misadventure’, a termed coined by Maddrey and Zimmerman in the mid-1990s, is still a cautionary tale, in that unsuspecting chronic alcohol users (which can include cirrhotics) may experience an inadvertent overdose from paracetamol (in doses far lower than the traditional 10 g implicated in most intentional overdoses) leading to acute on chronic liver failure. However, no instances of acute liver failure were seen with daily doses less than 2.5 g. As a result, most clinicians suggest that their patients not use more than 2 g of paracetamol in a 24-h period and to limit the overall duration of treatment.[3, 4, 192] In the survey conducted by Lucena et al. in Spain, paracetamol was safely used when doses in general were kept to 3 g/day, even in those with alcoholic cirrhosis.
The work of a number of investigators has helped clarify the usage patterns of a number of OTCAs, including paracetamol and various NSAIDs, with respect to their safety in cirrhotics. Khalid et al. found that among noncirrhotic control patients in their liver clinic, 70% used paracetamol or NSAIDs. In their patients with compensated cirrhosis, over half mentioned the use of OTCAs with 25% using paracetamol and 31% using NSAIDs. Among those with decompensated cirrhosis who ended up hospitalised for a number of common causes of hepatic decompensation, only 35% mentioned the use of OTCAs with 19% saying they used acetaminophen and 16% using NSAIDs. Of interest was the fact that hospitalisation rates were not increased among cirrhotics using OTCAs as reported by Fenkel et al.
Physicians' attitudes about prescribing OTCAs in cirrhosis
Data suggest that paracetamol is prescribed rather sparingly in cirrhosis. Franz et al. reported its use in only 6% of cirrhotic patients, and its use was even less in another series. A study of physicians' attitudes regarding the use of OTCAs in CLD was undertaken by Rossi et al. The results of their web-based questionnaire survey found that internists and family physicians were significantly more likely not to recommend the use of acetaminophen in patients with compensated cirrhosis compared with gastroenterologists who felt that paracetamol (acetaminophen) would be safe. In patients with decompensated cirrhosis, 95% of family physicians and 70% of internists would not recommend the use of paracetamol compared to just 22% of gastroenterologists. Even among patients with mild chronic hepatitis without cirrhosis, 15–20% of general practitioners would not recommend paracetamol. In contrast, none of the gastroenterologists questioned in this survey would avoid paracetamol in that setting. Overall, the recommendation against the use of NSAIDs was significantly less common than recommendations against the use of paracetamol in both compensated and decompensated cirrhosis. Nongastroenterologists were more likely to recommend NSAIDs compared with gastroenterologists, who in turn, were more likely to recommend the use of paracetamol.
Given the very real possibility that ibuprofen and other NSAIDs might lead to or worsen GI haemorrhage in patients with underlying gastropathy and coagulopathy, the occasional use of paracetamol is in fact preferred over NSAIDs. In addition, ibuprofen has been associated with hepatotoxicity among patients with chronic hepatitis C, suggesting that it might not be safe in this setting, although others have challenged this notion. In a recent survey exploring the attitudes and preferences of medical students, residents and gastroenterology fellows regarding the use of paracetamol (acetaminophen) and NSAIDs in CLD patients with various degrees of hepatic impairment, paracetamol doses less than 2 g/day were preferred, although only the GI fellows thought it was safe to use paracetamol in patients with decompensated cirrhosis.[199, 200] These findings among medical students and residents were very similar to those of Rossi et al. that compared primary care physicians to practising gastroenterologists. Reasons for their preferences given by the trainees, who were worried about using paracetamol, reflected their concerns about worsening underlying liver disease, producing an inadvertent overdose, and the overall lack of evidence-based information on which to justify clinical decisions in this setting. Both surveys illustrate that a significant variability exists among health care providers regarding their recommendations for the use of OTCAs and affords another opportunity for physician as well as patient education on medication use in CLD.
Narcotic analgesic and other pain relievers in cirrhosis
The management of acute and chronic pain in cirrhosis often raises the fear of precipitating hepatic decompensation, worsening encephalopathy or creating addiction, especially in patients with a history of alcoholism or other addictive behaviours. In addition, the PKs of analgesics, in general, have been poorly studied in patients with the most severe degrees of hepatic impairment.[25, 202] Indeed, opioids were rarely prescribed among cirrhotics in Spain. The need for significant reductions in the dose or dosing frequency of a number of narcotic agents has been recommended since initial studies on morphine were conducted more than 50 years ago. For example, dose reductions have been recommended for many agents, including, hydromorphone, methadone, morphine, oxycodone, tramadol amitriptyline, bupivacaine, clonidine and gabapentin, especially in the presence of renal failure, and it is recommended that aspirin, dextropropoxyphene, NSAIDs and demerol be avoided in the presence of concomitant chronic renal failure. Murphy also suggests avoiding amitriptyline, methadone, carbamazepine and valproic acid, due to a potentially higher risk of acute liver failure occurring in cirrhosis.
Several comprehensive reviews of the PK and PD effects of opioids, antidepressants and other classes used to treat pain in the setting of cirrhosis are available[6, 25, 55, 202] and recommendations for their safe use have been recently summarised. Table 11 provides the important changes in metabolism that have been observed with the more commonly prescribed opioid and other classes of analgesics, which underlie the recommendations for their use in cirrhosis.[6, 25, 55, 202, 204]
|Drug||Potencyb||Active metabolite formed||Bioavailability in cirrhosis||Dose adjustment|
|Morphine||1×||Yes||Increased 100%||Reduce dose and frequency by half|
|Fentanyl||75–125×||No||Unchanged (CP A or B)||Usually none for single dosec|
|Meperidine (pethidine)||0.1×||Yes||Increased up to 80%||Generally avoid using, or reduced dose and avoid chronic used|
|Codeine (methylmorphine)||0.1 ×||Yes||Reduced||Poor analgesic effect and should be avoided|
|Methadone||1×||No||Largely unaffectede||None needed in compensated cirrhosis|
|Oxymorphone||7×||No||Increased 1.6–12.2 fold||Reduce dose to prevent accumulation|
|Oxycodone||1×||Yes||Increased 50–95%||Reduce dose to prevent accumulation|
|Hydromorphone||6–10×||Yes||Limited data||Limited data|
|Dextropropoxyphene||<1 ×||?||Increased||Avoid in cirrhosisf|
|Tramadol||0.1 ×||Yes||Increased two to threefold||Consider alternative agentsg|
As reviewed by Tegeder et al. and Smith, a number of opioids require hepatic biotransformation to active metabolites, which in the setting of impaired hepatic function can lead to disparate clinical effects. For example, a reduced analgesic effect has been seen with codeine, rendering it less useful as an analgesic in this setting. In contrast, due to a reduced first-pass effect, the bioavailability of meperidine is increased by as much as 80%, which can result in the accumulation of its neurotoxic norpethidine metabolite. Tramadol also requires metabolism to an active metabolite and its analgesic properties may not be fully effective in cirrhosis. Moreover, tramadol is known to lower seizure thresholds in epilepsy and is associated with serotonin syndrome when combined with SSRIs, TCAs, anticonvulsants or morphine.[204, 205] All opioids can aggravate or precipitate HE and should be used with caution.[6, 204] Fentanyl, methadone and hydromorphone often require reduced dosing, but do not have toxic metabolites and may be better tolerated.[25, 55, 204] While methadone appears to be safe in CLD patients with narcotic addiction, it is best avoided in alcoholics as ethanol interferes with its metabolism (and raises plasma concentrations).
Chandok and Watt have offered useful recommendations to treat acute and chronic visceral and neuropathic pain in patients with hepatic impairment, based in part on expert opinion as well as the known or expected changes in the metabolism and clearance of paracetamol, NSAIDs, opioids analgesics and TCAs. Table 12 provides the algorithmic approach they suggest to treat pain in cirrhotics who do not have renal impairment or active alcohol or substance abuse. Their recommendations include the use of acetaminophen (paracetamol) in a maximum dose of 2–3 g/day in cirrhotics as the first-line agent for the treatment of visceral or musculoskeletal pain, although they acknowledge that the safe maximal duration (>14 days) has not been determined. Paracetamol has a half-life in cirrhosis that is double that in healthy subjects, but CYP enzyme activity and glutathione levels do not appear to be significantly depleted with these sub-therapeutic doses in the non-alcoholic as previously discussed.[4, 56] Even among alcoholics, they note that small doses of paracetamol (<2 g) appear to be well tolerated when used for a short duration. In addition, the absence of GI or renal toxicity and its lack of an effect on platelets makes paracetamol preferable to NSAIDs as an analgesic/antipyretic.[6, 204] Tramadol is suggested as a second-line agent, with the caveats listed above, including not combining it with fentanyl or hydromorphone, which are listed as third-line agents for intractable pain. These authors recommend the use of TCAs in small doses or certain anticonvulsants (e.g. gabapentin, pregabalin) along with acetaminophen as first-line treatment for neuropathic pain, mindful of the sedation and anticholinergic side effects that they can cause.[204, 206]
|Type of pain||Sequence of medications recommended|
|Visceral/musculoskeletal||Acetaminophen <2–3 g/day or Tramadol 25 mg orally q8h|
|For intractable paina||Hydromorphone 1 mg orally q4h or Fentanyl 12.5 mg topically q72h|
|Neuropathic||Nortriptyline 10 mg orally at night or Desipramine 10 mg orally at night or Gabapentin 300 mg orally/day or Pregabalin 150 mg orally twice daily and Acetaminophen <2–3 g/day|
NSAIDs such as ibuprofen, naproxen and sulindac are excreted predominantly via the kidneys, are heavily protein-bound and metabolised by CYP-450 enzymes. As mentioned previously, they are not recommended in cirrhosis due to the risk of renal failure (from inhibition of prostaglandins leading to a decrease in renal blood flow, and sodium retention),[6, 207, 208] and GI mucosal bleeding (from underlying portal gastropathy or colopathy in the setting of coagulopathy).[6, 196] In addition, the hypoalbuminaemia of cirrhosis can alter their PKs with reduced protein binding.[20, 21] Moreover, some NSAIDs (e.g. diclofenac) have a significant hepatotoxic potential. The selective COX-2 inhibitors have not been well studied in cirrhosis and agents such as celecoxib, while potentially safer (as they do not affect platelet or renal function), should nevertheless be used with caution.
Ultimately the choice of an analgesic regimen should be based on an individualised approach taking into account all aspects of cirrhosis, including nutritional status, renal function, the potential for DDIs and aetiology of cirrhosis.
The safe use of medications in patients with CLD is an ongoing challenge for prescribers and patients alike. This becomes especially true in patients with cirrhosis, in whom significant changes can occur in the metabolism and handling of various agents, specifically those medications which undergo first-pass metabolism or are metabolised by the CYP3A enzymatic pathway. The presence of portosystemic shunts, including TIPS, may lead to increased bioavailability of drugs leading to QTc prolongation in cirrhotics, which may precipitate potentially fatal ventricular arrhythmias.
Very few drugs have been shown to have their hepatotoxicity potential enhanced by CLD with or without cirrhosis; nearly all of these involve antituberculosis agents given in the setting of chronic hepatitis B or C and among HIV/AIDS patients. Statins, in particular, appear to be beneficial in chronic hepatitis C and other causes of cirrhosis and may prevent progression to HCC.
Lower doses are recommended for the use of several drugs in cirrhosis, although in many cases, the rationale for this is based mostly on reduced clearance mechanisms rather than any known excess risk of hepatotoxicity or other adverse PD effects. NSAIDs, in general, should be used cautiously (or not at all) in cirrhotics due to their risk of precipitating renal failure and inducing or worsening GI bleeding. Nevertheless, OTCAs are widely prescribed, although they do not appear to increase the risk of hospitalisation. Paracetamol (acetaminophen) is safe in patients with CLD, including cirrhosis when used in small (2–3 g or less/day) doses for short durations, and is recommended as first-line treatment of visceral or musculoskeletal pain. Most narcotics and BZDs are best avoided as their half-lives can be prolonged in cirrhosis and may precipitate or worsen encephalopathy. Propofol (without BZDs or narcotics) is well tolerated among cirrhotic patients undergoing endoscopic procedures and isoflurane is a preferred inhalational anaesthetic for cirrhotics undergoing liver transplant. Many patients with CLD use herbal remedies, and clinicians need to be mindful that some can be hepatotoxic, making it essential to obtain a complete drug use history, including all prescription and OTC medicines and supplements.
Proton pump inhibitors and other acid-suppressive agents have been linked to a greater risk of SBP in cirrhotics and should be used with caution or avoided entirely. When required, shorter duration therapy should be considered, along with the use of prophylactic antibiotics to prevent bacteremia in patients with variceal or other GI bleeding undergoing endoscopy or other invasive procedures. Antibiotics are also recommended to prevent recurrent SBP in patients who have been successfully treated for a prior episode; whether or not this will reduce the risk of SBP from acid-suppressive agents has not been formally studied, but is likely a prudent recommendation.
Guarantor of the article: James H. Lewis.
Author contributions: Dr Lewis was responsible for the concept and design of the article and both Dr. Lewis and Stine performed the research, collected and analysed the data and wrote the paper. Both authors approved the final version of the manuscript.
Declaration of personal and funding interests: None.