Should cirrhosis change our attitude towards treating non-hepatic cancer?


  • Giuseppe Cabibbo,

    Corresponding author
    1. Dipartimento di Biopatologia e Metodologie Biomediche, University of Palermo, Palermo, Italy
    • Sezione di Gastroenterologia, DIBIMIS, University of Palermo, Palermo, Italy
    Search for more papers by this author
    • Both authors contributed equally to this study.
  • Laura Palmeri,

    1. Cattedra di Oncologia Medica, Dipartimento di Discipline Chirurgiche ed Oncologiche, University of Palermo, Palermo, Italy
    Search for more papers by this author
    • Both authors contributed equally to this study.
  • Sergio Palmeri,

    1. Cattedra di Oncologia Medica, Dipartimento di Discipline Chirurgiche ed Oncologiche, University of Palermo, Palermo, Italy
    Search for more papers by this author
  • Antonio Craxì

    1. Sezione di Gastroenterologia, DIBIMIS, University of Palermo, Palermo, Italy
    Search for more papers by this author


Dr Giuseppe Cabibbo, Sezione di Gastroenterologia,

Dipartimento Biomedico di Medicina Interna e Specialistica,

University of Palermo,

Piazza delle Cliniche 2,

90127 Palermo, Italy

Tel: +39 091 655 2280

Fax: +39 091 655 2156



Cirrhosis is a major cause of morbidity and mortality and is the end stage of any chronic liver disease. Cancer, a leading cause of death worldwide, is a growing global health issue. There are limited data in the literature on the incidence, prevalence and management of non-hepatic cancers (NHC) in cirrhotic patients. The aim of this brief review was to underline the main concerns, pitfalls and warnings regarding practice for these patients.

Survival of patients with compensated cirrhosis is significantly longer than that of decompensated cirrhosis and patients with NHC and in Child-Pugh class C should not be candidates for cytotoxic chemotherapy. It is important before starting cytotoxic chemotherapy to assess the aetiology and stage of liver disease and to screen these patients for portal hypertension and fluid retention. During cytotoxic chemotherapy, the effectiveness of cancer treatment, as well the appearance of early signs of hepatic decompensation, must be thoroughly monitored. Future phase 3 trial designs in oncology should include a share of patients with compensated cirrhosis to obtain specific information in this setting. Identification of tests able to measure the global degree of hepatic impairment caused by cirrhosis could help in the management of this particular clinical situation.


hepatitis B virus


hepatocellular carcinoma


hepatitis C virus


Model for End-Stage Liver Disease


non-hepatic cancers

Non-hepatic cancers (NHC) and cirrhosis are common conditions, each of them representing a leading cause of death worldwide. It has been estimated that there were 7.6 million cancer-related deaths worldwide in 2007 [1], while deaths attributable to chronic liver disease were 1.28 million in 1990 (2.5% of global deaths) [2]. Hence, by the pure laws of mathematics, one could estimate that at least 190 000 patients with solid organ malignant tumours also have cirrhosis.

The issue of comorbidity may be more complicated, since in the senior age fascia, when cirrhosis is more prevalent, colon or gastric cancer is distinctly more frequent. Other factors, either constitutional (male gender) or acquired (obesity, alcohol abuse), may act as further confounders. Nonetheless, only hepatocellular carcinoma (HCC) is typically associated with cirrhosis as a problem in clinical practice. It is arduous to estimate the incidence of cancer other than HCC in patients with cirrhosis. Little data can be retrieved from the literature [3-7] and they show that incidence and management were, and probably still are, controversial. However, an interesting recently published article shows that the overall risk for non-HCC malignancies is more than two-fold greater for patients with cirrhosis (mostly biliary and gastrointestinal malignancies) than for the general population [8].

When facing patients with NHC who have cirrhosis, clinicians may encounter some difficulties both in terms of choosing the appropriate treatment for cancer and of managing treatment-related hepatotoxicity and adverse liver events. In this brief review, we aimed to underline the main concerns, pitfalls and warnings regarding practice for these patients.

Cirrhosis and survival of the patient with non-hepatic cancer

It is difficult to estimate the real prevalence of cirrhosis in the general population, as the disease is mostly asymptomatic until the patient develops liver failure. The prognosis of cirrhosis is highly variable, being influenced by a number of factors, including disease stage, aetiology and feasibility of eradicating the aetiological factors and presence of complications and comorbidities. More than 30 years ago, Child and Turcotte designed a scoring system that, with a few modifications by Pugh, remains the most widely used for prognostic classification of cirrhotic patients [9]. Survival of patients with compensated cirrhosis is significantly longer than that of decompensated patients, with median survival times of >12 years and 2 years respectively [10].

As the outcome in a patient with non-hepatic cancer (NHC) and decompensated (particularly Child-Pugh C) cirrhosis is related mostly to the hepatic functional impairment rather than to the neoplastic disease, it seems reasonable to say that these patients, in general, should not be candidates for cytotoxic treatments (Fig. 1). For Child-Pugh B patients, the effective degree of hepatic disfunction must be carefully evaluated, together with the tumour/patient characteristics (e.g. chemo-sensitivity, site of disease, kind and degree of symptoms), as suggested by the experience with hepatocellular carcinoma (HCC) patients treated with sorafenib [11]. It must be stressed that no formal experience or dosing recommendations are available in this setting, even for the most common cytotoxic agents, and that problems related to altered hepatic drug metabolism may be overwhelming.

Figure 1.

Proposed decisional algorithm for patients with cirrhosis and non-hepatic cancer.

Conversely, the oncological management of a patient with compensated (Child A) cirrhosis developing NHC is a matter of interest and poses various unresolved clinical questions. Few quality data are available, for a number of reasons:

  1. Most clinical trials in oncology exclude patients with any degree of impaired hepatic function or with cirrhosis. To date, neither the regulatory agencies nor the worldwide industry has ever prompted the inclusion of such patients in trials of oncological agents. This dramatically reduces the possibility of managing this comorbidity with an evidence-based approach.
  2. Many cytotoxic drugs are metabolized by the liver, where some drugs are inactivated to non-toxic, excretable metabolites or else activated if they are pro-drugs. Hence, there are potential hazards in the administration of anti-cancer therapy to patients with an abnormal liver.
  3. As no single test reliably measures the global degree of hepatic impairment, biochemical estimation of liver function in patients with cancer may under- or overestimate the risk of hepatic toxicity.
  4. Patients with cancer may carry a huge variety of metabolic paraneoplastic alterations (for instance, a condition of hypercoagulability) that could be of increased relevance in a cirrhotic patient and must be taken into account when approaching such patients. Moreover, many recent studies have indicated a prothrombotic state in patients with cirrhosis that can frequently induce portal vein thrombosis (PVT) during the course of the disease [12]. Furthermore, cytotoxic chemotherapy and malignancies per se can determine a thrombophylic diathesis, which may facilitate, in turn, the occurrence of PVT in patients with slow portal flow because of portal hypertension.

Reactivation of hepatitis viruses [especially hepatitis B virus (HBV)], the main cause of cirrhosis worldwide, can be induced by chemotherapy and cause severe liver dysfunction in patients with cirrhosis. So, the existence of markers of the HBV infection must be systematically verified before starting chemotherapy. This problem is carefully reviewed elsewhere [13-15].

Some major concerns arise for the clinician facing an oncological patient with compensated cirrhosis.

Should that patient be treated?

As said, this depends mainly on the prognosis of the liver disease. The answer to this fundamental question must be affirmative when the patient's life expectancy is not, or only marginally, influenced by the liver disease. In other words, it is our opinion that NHC should always be treated when it affects the ultimate prognosis of a patient with compensated cirrhosis and a life expectancy of more than 3 months. The choice of treatment must, however, take into account the effect that the oncological drugs by themselves may have on liver disease and mostly how hepatic decompensation ensuing from them will affect the possibility of the completion of the intended protocol. Clearly, other conditions potentially affecting pharmacokinetics and pharmacodynamics of the antineoplastic agents, increasing toxicties and/or reducing effectiveness, should be carefully taken into account. For example, elderly patients can have a decrease in volume of distribution, glomerular filtration rate, hepatic metabolism, intestinal absorption or levels of enzymes like dihydropyrimidin dehydrogenase, but also an increase in the expression of multidrug resistance gene, as well as decreased apoptosis, and decreased cell proliferation rate [16].

Which drugs are most appropriate?

After a thorough assessment of the liver status, otherwise healthy, compensated cirrhotic patients with NHC should be treated according to the standard of care for their specific neoplastic disease. The pharmacokinetic and pharmacodynamic characteristics of the drugs, together with the tolerability profile, must be kept in mind. Two interesting reviews on the topic of liver function in oncology have been recently published [17, 18].

Here, we briefly examine the profiles of some of the most widely used anticancer drugs. The entire class of fluoropyrimidines can reasonably be taken into consideration as therapy because even if their primary metabolism is hepatic they are eliminated mainly through the kidney [19, 20]. Moreover, it has recently been shown that, in animal models, hepatic fibrosis does not affect the pharmacokinetics of 5-fluorouracil (5-FU) [21]. Antineoplastic antibiotics, such as doxorubicin, can be used as they rarely cause direct hepatic injury, which usually consists in transient aminotransferases and bilirubin increases on an idiosyncratic basis, while their cardiotoxic effects should be closely monitored [22]. Microtubule disrupting agents (e.g. taxanes, vinorelbine) should be used cautiously as they cause (generally reversible) direct hepatic injuries in about one-third of patients on high doses. However, two questions must be considered: (i) patients with elevated bilirubin, or abnormal aminotransferases and alkaline phosphatase levels are at increased risk for the development of grade 4 side effects; and (ii) severe fluid retention can occur in about 6% of patients despite use of an adequate dexamethasone premedication, which could be of major relevance in cirrhotic patients who avidly retain sodium and fluids [23].

With regard to combination regimens, very few specific data are available. Some information on their toxicity profile in cirrhotic patients can be obtained from published papers concerning chemotherapy combination regimens used in advanced HCC superimposed on cirrhosis. For example, gemcitabine-oxaliplatin combined with cetuximab, capecitabine-oxaliplatin with bevacizumab, 5-FU-oxaliplatin and leucovorin, 5-FU-cisplatin-mitoxantrone and capecitabine-oxaliplatin was recently evaluated for advanced HCC [24-28]. Major toxicities (grade 3 or 4) included myelosuppression (neutropenia and thrombocitopenia), skin toxicity, neurotoxicity, fatigue, hand-foot syndrome, diarrhoea, renal toxicity, bleeding and infection. Only with the combination of capecitabine and oxaliplatin was elevation of transaminases and/or bilirubin reported. Overall, toxicities observed with these regimens are manageable in cirrhotic patients and liver toxicities are unusual.

Not much is known about molecular-targeted therapies in cirrhotic patients. Most data come from studies on the treatment of HCC with the multitargeted tyrosine kinase (TK) inhibitors sorafenib and sunitinib [29, 30]. Though HCC is associated with cirrhosis, at least in the Western world, in more than 90% of patients [31, 32] in phase 3 studies [33, 34] used to register sorafenib for this indication, no specific attention was paid to the role of cirrhosis as a possible modifier of response and toxicity. An unspecified proportion of patients enrolled in these studies had cirrhosis, but most were reported as Child A. A post hoc analysis showed comparable toxicity and efficacy in the few Child B patients [35]. Moreover, in the phase 2 trial in which 99 Child-Pugh A and 38 Child-Pugh B patients were enrolled, the pharmacokinetics of sorafenib was not influenced by the Child-Pugh class [36].

Treatment with sunitinib was instead found to be associated with a high proportion of patient (about 10%) death from treatment-related causes. Gastrointestinal haemorrhage was also reported.

As concerns anti-epidermal growth factor receptor TK inhibitors, few data are available from a phase 1 study that evaluated the drug's pharmacokinetics in patients with liver dysfunction (defined as aspartate aminotransferase ≥3 times above normal, with or without albumin <25 g/L, or bilirubin 17–120 μmol/L, not necessarily related to cirrhosis)[37]. Longer half-life, reduced clearance and increased proportion of dose-limiting toxic effects were observed. The lack of data in this specific field makes it difficult to reach definitive conclusions and competitive research will likely continue over the coming years.

Finally, interesting data have been published concerning the use of bevacizumab in non-cirrhotic patients with liver metastases. In fact, decreased severity of the sinusoidal obstruction syndrome and no impact on hepatic steatosis and fibrosis have been reported, suggesting a positive effect of the drug on the liver tissue [38, 39].

An unanswered question concerning the use of antiangiogenics in cirrhotic patients is, in fact, their effect on portal hypertension, though recent experimental work with sorafenib seems to suggest a possible role of this drug in reducing portal hypertension, according to data from animal models [40].

Despite the absence of solid evidence in this particular clinical setting, we believe that antineoplastic agents can be used, though with caution, in patients with compensated cirrhosis and NHC. Today, little specific data are available on the use of molecular-targeted therapies.

Are normally expected adverse events of the chosen regimen worse in cirrhotic patients, and if so, how to manage and/or prevent them?

This is conceivably the principal concern when treating a cirrhotic patient with NHC. The main expected toxic effects of the most widely used anti-neoplastic drugs should be considered when choosing treatment and promptly managed when they manifest. Many drugs particularly 5-FU, oxaliplatin and irinotecan have an intrinsic liver toxicity (mainly increased aminotransferases and bilirubin) that may exacerbate the underlying liver disease and, in principle, should be avoided or administered cautiously. Clearly this is only feasible when there are possible alternative regimens.

Many usually expected, easily manageable adverse events may have increased relevance in cirrhotic patients. For instance, leucopenia and thrombocytopenia, which are resulting from splenic sequestration secondary to portal hypertension, can magnify the effects of bone marrow suppression caused by antineoplastic agents. Similarly, it must be kept in mind that the equilibrium of the clot cascade is always at risk of alteration in these patients, as a consequence of paraneoplastic syndromes, chemotherapeutic agents and cirrhosis itself, among other factors. All of these can induce PVT, which in turn may worsen the prognosis of patients with compensated cirrhosis. So, because early diagnosis of acute PVT and anticoagulation are probably the main determinants of improved survival [41], in our opinion, such patients should have frequent (every 2–3 months) ultrasound evaluation and pulsed-Doppler assessment of the portal flow to start an appropriate anticoagulation treatment as early as possible, though this is not current practice.

How, and how often, should the patient's liver status be assessed?

As no specific guidelines exist, close co-operation between the oncologist and the hepatologist is recommended, together with a cautious approach during the decision-making process, to give the best care to cancer patients with cirrhosis.

Liver function itself (an imperfect concept, encompassing hundreds of different activities performed by hepatic parenchymal and non-parenchymal cells) may not be stable over time. In clinical practice, it would be preferable to evaluate liver status as a whole through a complex evaluation that encompasses clinical evaluation and biochemical and instrumental parameters. The Child-Pugh score and the Model for End-Stage Liver Disease (MELD) are useful tools and adopted daily by hepatologists [42]. These should be added to the common pretreatment laboratory examinations and recalculated before each chemotherapy cycle. However, these traditional scoring systems have several shortcomings. The variables in the Child-Pugh score are limited by a lack of consistency and reproducibility; for example, ascites and hepatic encephalopathy are graded subjectively and may be altered substantially by medical interventions (e.g. the use of diuretics for ascites or lactulose and rifaximin for encephalopathy). Bilirubin levels may depend on the aetiology of liver disease (e.g. primary biliary cirrhosis or primary sclerosing cholangitis), while low albumin plasma levels could also be related to the nutritional status and to a catabolic state secondary to malignancy.

On the other hand, the MELD score has been validated almost exclusively in advanced liver disease and for short-term prognosis. It must be remembered that aminotransferases, though a sensitive index of hepatocellular necrosis, can be misleading as an indicator of hepatotoxicity, as they are bound to fluctuate, sometimes widely, also in relation to the aetiology of the underlying liver disease, especially in viral (HBV and HCV) cirrhosis. With specific reference to the liver function and/or dysfunction evaluation in the field of oncology, the National Cancer Institute Organ Dysfunction Working Group (NCI-ODWG) suggested different criteria that utilize two objective, readily measurable, laboratory parameters, specifically total bilirubin (TB) and aspartate aminotransferase (AST).

National Cancer Institute Organ Dysfunction Working Group criteria classify liver dysfunction into four classes: normal [TB and AST < upper limit of normal (ULN)] and mild (TB > ULN to 1.5 × ULN or AST > ULN) function, in which dose modification of chemotherapeutic agents is usually not necessary, or moderate (TB > 1.5–3 × ULN, any AST) and severe [TB > 3–10 × ULN, any AST(43)] in which dose modification may be necessary (Table 1). Interestingly, a prospective comparison between the NCI and the Child-Pugh score has been carried out in a phase 1 study [44]. Unfortunately, no definitive results emerged and urgent calls for a more comprehensive, as well as specific, evaluation tool are expected in the literature. Moreover, the use of the NCI-ODWG criteria has never been validated for the assessment of liver function in cancer patients with cirrhosis.

Table 1. The NCI-ODWG (43) criteria for stratifying patients according to liver dysfunction into five groups, from normal liver function (A) to need for liver transplant (E)
GroupGroup AGroup BGroup CGroup DGroup E
Liver functionNormalMildModerateSevereLiver transplant
  1. A, normal; B, mild dysfunction; C, moderate dysfunction; D, severe dysfunction; E, liver transplant.

  2. AST, aspartate aminotransferase; NCI-ODWG, National Cancer Institute Organ Dysfunction Working Group; ULN, upper limit of normal range.

Total bilirubin≤ULNB1: ≤ULN
B2: >1.0–1.5 × ULN
>1.5–3 × ULN>3 × ULNAny value
B2: Any value
Any valueAny valueAny value

In our opinion, because there is still no specific indication for the management of cirrhotic patients treated with chemotherapy for NHC, a routine multidisciplinary approach, involving hepatologists and oncologists, is required to provide optimal care to these patients.

In other words, while waiting for a formal and validated scale to be used routinely for this setting of patients, oncologists are urged to add a consultation by a hepatologist to the usual prechemotherapy patient assessment (performance status, clinical examinations, biochemistry). Conversely, hepatologists are urged to consult an oncologist before deciding that a Child-Pugh A/B patient carrying an NHC is not a good candidate for an antitumoural treatment.

Conclusions and recommendations

To date, the appropriate strategy for effective and safe chemotherapy treatment of cancer patients with compensated cirrhosis has not been defined. It is well known that there is heterogeneity among cancer patients in terms of pharmacokinetics and pharmacodynamics. Generally, the overall effect of liver disease on drug pharmacokinetics is determined by the alterations produced in the processes of absorption, distribution and elimination. For these reasons, the oncologist prescribing anti-neoplastic therapy must carefully consider characteristics, route of administration, dose and dosing interval.

Since, as already mentioned, most clinical trials exclude patients with impaired hepatic function, available knowledge about individual chemotherapeutic agents in the setting of cirrhosis is based on small, retrospective studies or on trials investigating drugs for hepatocellular carcinoma (HCC) superimposed on compensated cirrhosis. Very few agents have, in fact, undergone formal phase 1 studies to test their pharmacokinetics and tolerability in patients with liver dysfunction, while empirical guidelines are frequently used in clinical practice because the phase 1 findings have not been further tested in phase 2 and 3 settings [45-50]. As previously noted, the lack of compensated cirrhotic patients in large phase 3 studies may lead to two different conditions: firstly, physicians might deny antitumour treatment to cirrhotic patients because of lack of evidence-based data; secondly, physicians empirically consider treating these patients without evidence and/or the possibility of comparing efficacy and safety of therapy. Consequently, the physician who plans treatment of a non-hepatic cancer in a patient with compensated cirrhosis is burdened, in the absence of any guideline for this specific type of patient, by the narrow therapeutic index, if any, of cytotoxic drugs, on one hand, and the complicated safety issues typical of such patients, on the other. The way to solve this clinical dilemma is to overcome, in principle, the prejudice that a cirrhotic patient cannot be treated if he/she develops a cancer other than HCC. Clinical researchers, health authorities and regulatory agencies must start including, or require inclusion of, a share of patients with compensated cirrhosis in large randomized controlled trials when investigating medical treatment of any cancer. Moreover, future clinical trial designs should include stratification of patients according to their basal liver function. An open question remains whether patients with tumour-related organ dysfunction should be approached differently than those with baseline (non-tumour-related) organ dysfunction.

A few final recommendations can be made (Fig. 1):

  1. Before starting treatment with cytotoxic chemotherapy, physicians should:
    1. carefully evaluate aetiology (virus, alcohol, others) and stage of liver disease (liver status with Child-Pugh and Model for End-Stage Liver Disease scores), screen for portal hypertension, evaluate fluid retention;
    2. prevent possible viral reactivation with lamivudine, when necessary (hepatitis B virus) and
    3. start correction of modifiable causal factors (alcohol, other hepato-toxins, diabetes).
  2. During cytotoxic chemotherapy, physicians should:
    1. evaluate the effectiveness of treatment of cancer and reassess cost-benefit frequently and at first signs of hepatic decompensation;
    2. recognize and treat the side effects associated with cytotoxic therapy by adjusting the dosage rather than suspending if possible and
    3. closely monitor, using clinical, laboratory and instrumental parameters, the underlying liver disease to recognize and treat early complications.
  3. As concerns liver disease, treatment and follow-up parameters should not change in terms of overall strategy except to accommodate the more frequent controls needed for cancer therapy. Treatment that may cause cytopenia, such as IFN-based antiviral therapies, must clearly be avoided or postponed in this setting.


The authors would like to thank Warren Blumberg for his help in editing the manuscript.

Conflicts of interest: No conflicts of interest exist.