Review article: thrombocytopenia in chronic liver disease and pharmacologic treatment options


  • This review was prepared from an invited lecture held at the 12th United European Gastroenterology Week, Prague, Czech Republic, September 25–29, 2004.

Dr E. G. Giannini, Department of Internal Medicine, Gastroenterology Unit, University of Genoa, Viale Benedetto XV, No. 6, 16132 Genoa, Italy. E-mail:


In patients with liver disease, thrombocytopenia is a clinical feature that may represent an obstacle to invasive diagnostic or therapeutic procedures, chemotherapy, and anti-viral treatment. Stimulation of the bone marrow is the most promising therapeutic intervention for thrombocytopenia in patients with chronic liver disease.

The description of thrombopoietin and its (de)regulation in patients with chronic liver disease have disclosed new treatment opportunities. Indeed, pharmacologic treatment options for thrombocytopenia can be divided into treatments targeted at the thrombopoietin receptor (synthetic thrombopoietins and thrombopoietin-mimetic agents), and use of cytokines with general thrombopoietic potential. Unfortunately, use of synthetic thrombopoietin was hampered by the development of neutralizing antibodies, and thrombopoietin mimetic agents have not yet entered clinical studies. Interleukin-11 proved to be useful in increasing platelet count in patients with chronic liver disease, although its use is limited by side-effects.

Erythropoietin has shown promising results in improving thrombocytopenia in cirrhotic patients. In patients with chronic liver disease, safe and well-tolerated treatments aimed at improving thrombocytopenia are still lacking. Larger studies are needed to evaluate and better characterize the thrombopoietic potential of erythropoietin. Human studies with thrombopoietin-mimetic agents are eagerly awaited in order to assess both effectiveness and safety of these drugs.


Prevalence of thrombocytopenia in patients with chronic liver disease (CLD) may vary according to various factors, such as the criteria used to define this haematologic abnormality and the severity of the underlying liver disease. From a practical point of view, thrombocytopenia is defined as any decrease in platelet count below the lower normal limit (i.e. <150 × 109/L). However, it is often operatively considered as the level below which performing invasive manoeuvres (e.g. liver biopsy) or administering interferon therapy could be dangerous (i.e. <50–75 × 109/L), or, lastly, as a threshold below which platelet transfusion is indicated (i.e. <10 × 109/L).1–4 As far as severity of disease is concerned, the prevalence of thrombocytopenia in patients with acute hepatitis is higher in those with liver failure when compared with those without liver failure (52% vs. 16%),5 while among patients with CLD the prevalence of thrombocytopenia is higher in those with cirrhosis when compared with chronic hepatitis patients (64% vs. 6%).6 Lastly, in a large cohort of cirrhotic patients, we observed that ‘any thrombocytopenia’ has a prevalence of 76%, while 13% of the patients reach the ‘liver biopsy or interferon (IFN) therapy’ threshold, and 1% of the patients alone reach the ‘transfusion trigger’ threshold.7 Furthermore, we observed that the prevalence of the various degrees of thrombocytopenia in these patients was distributed differently depending on severity of cirrhosis, with an increasing trend in patients with more severe disease.7 These features likely explain why the reported prevalence of thrombocytopenia in the published series ranges from 15% to 70%, being lower in patients with compensated CLD and higher in patients with end-stage liver disease.

In CLD, severe thrombocytopenia is often associated with the life-threatening complications of end-stage liver disease, and it is likely that the latter factor rather than the thrombocytopenia itself eventually determines the patients’ prognosis. Despite the presence of co-existing coagulation defects,8 thrombocytopenia per se is rarely a critical clinical problem in these patients, unless particular situations occur. Nevertheless, some studies identified thrombocytopenia as a parameter that is independently associated with the occurrence of complications of cirrhosis and patients’ prognosis.9, 10 Therefore, in well-defined and specific clinical settings, the availability of a tool that is able to safely increase platelet count in CLD patients would probably help the clinicians to manage their patients more appropriately.

The target population

As mentioned above, thrombocytopenia can become a clinically relevant problem in particular situations. Well-defined clinical situations in which thrombocytopenia may become a significant clinical issue in CLD patients include: performing invasive diagnostic or therapeutic procedures, IFN treatment, bleeding oesophageal varices, management of patients on orthotopic liver transplantation (OLT) waiting lists, chemotherapy for solid tumours or haematological malignancies, and surgery.

As far as chemotherapy is concerned, patients with liver disease may experience more severe degrees of thrombocytopenia because of both treatment-induced myelosuppression and to the possible decrease in thrombopoietin (TPO) liver production caused by the hepatic toxicity of some treatment regimens. Transcatheter arterial chemo-embolization is increasingly being used in the treatment of cirrhotic patients with hepatocellular carcinoma, and the presence of thrombocytopenia can be a criteria for exclusion from treatment because of both the invasive nature of the technique and the concomitant use of one or more chemotherapeutic drugs. Furthermore, thrombocytopenia can be an important obstacle to surgical procedures or even simply to dental extraction. Although platelet transfusion is considered the ‘golden standard’ therapeutic option in all of these settings, patients can become refractory to platelet transfusion if they receive multiple transfusions over time. They may also have reactions to the transfusions, and they can be exposed to the risks of infections, especially when pooled products are administered.

In cirrhotic patients, unequivocal features of advanced CLD are often present when the degree of thrombocytopenia is such that performing liver biopsy may be considered unsafe. Therefore, this should not be regarded as a limit, as histological diagnosis would often be redundant in this setting. However, liver biopsy may be required in cirrhotic patients to diagnose hepatocellular carcinoma, and although current guidelines limit the role of liver biopsy in this situation,11 recent studies have demonstrated that liver biopsy of focal liver lesions in these patients may reveal unexpected diagnoses.12 Furthermore, in the future, indications to liver biopsy in cirrhotic patients could paradoxically increase. In fact, liver biopsy can be used to evaluate particular histological features that may be useful for identifying patients at higher risk of developing hepatocellular carcinoma.13 In this scenario, patients may also require repeated biopsies over time, thus increasing the likelihood of procedure-related adverse events.

Thrombocytopenia can be an important limit to IFN therapy because of the fact that it is both an ineligibility criteria and a reason for treatment discontinuation.14 Therefore, it represents an important reason for denying or discontinuing treatment in patients who are most often in need of anti-viral therapy. IFN administration is known to decrease platelet count because of a direct, dose-dependent effect on bone marrow.15 The use of new pegylated-IFN in patients with chronic viral hepatitis has led to an increase in terms of therapeutic efficacy, although side-effects have increased as well. Indeed, in one of the largest pegylated-IFN therapeutic trials, 6% of the patients were excluded from treatment because of thrombocytopenia, and in the same study, thrombocytopenia was responsible for dose reduction and therapy discontinuation in approximately 20% and 3% of the patients, respectively.16 These features are likely higher in every-day clinical practice. Interestingly, it has been shown that during IFN treatment of chronic viral hepatitis there is a blunted TPO response to the decreasing platelet count, which is more evident in cirrhotic patients.17 Noteworthy, successful IFN treatment is associated with restoration of the correct TPO-platelet count feed-back mechanism, likely because of an improvement in liver function.18

Lastly, patients awaiting OLT consistently have severe degrees of thrombocytopenia. These patients are subject to a greater risk of bleeding because of the high incidence of large oesophageal varices and coagulation impairment. Bleeding episodes may deteriorate liver function to the point of rendering the patient an unsuitable candidate for OLT, and may lead to patient death. Furthermore, treatment of bleeding episodes may require platelet transfusions, which can induce allo-immunization and refractoriness to new transfusions, and this may turn out to be an important issue when platelet transfusion is later needed to reduce the bleeding during OLT. Finally, recent studies have shown that anti-viral therapy may play a role in cirrhotic patients with chronic hepatitis C awaiting OLT, although in this setting thrombocytopenia once again was confirmed as a critical issue, as it caused treatment discontinuation in more than half of the patients.19

Causes of thrombocytopenia in CLD: the role of thrombopoietin

The occurrence of thrombocytopenia in patients with CLD can be considered an event with multiple aetiologies. Indeed, thrombocytopenia may be caused by various factors which are not mutually exclusive but, on the contrary, can act simultaneously.20 While the original theory by Aster suggested that thrombocytopenia is exclusively attributable to increased pooling of platelets in the enlarged spleen because of portal hypertension,21 many other factors are currently believed to be responsible for decreasing the platelet count in patients with CLD.20 As mentioned above, thrombocytopenia depends upon the stage of CLD, although aetiology of liver disease may also play a role. Two types of mechanisms may act alone or synergistically with splenic sequestration, thus determining thrombocytopenia. One is a central mechanism, which involves either myelosuppression because of hepatitis viruses or the toxic effects of alcohol abuse on the bone marrow, while the second mechanism is peripheral and involves the presence of antibodies against platelets.22–25 However, while both corticosteroids and immunosuppressants seemed to obtain successful results when thrombocytopenia was caused by immune-mediated phenomena,26 neither surgical nor non-surgical treatments aimed at relieving portal hypertension were effective in reducing thrombocytopenia.27, 28 In fact, well-conducted studies reportedly showed that even transjugular intrahepatic porto-systemic stent shunt was not able to reverse thrombocytopenia, although it did prove to be effective in reducing the porto-systemic pressure gradient.29 Consumption coagulopathy is likely to play a minor role, if any, in determining thrombocytopenia in patients with compensated liver disease, while it can be responsible for decrease in platelet count observed in cirrhotic patients with sepsis or shock. More recently, the characterization of TPO and the results of studies aimed at evaluating TPO pathophysiology in patients with liver disease have revealed a new scenario regarding the possible mechanisms of thrombocytopenia in the course of CLD.

Thrombopoietin has only been described recently,30–35 although its existence was postulated several years ago.36 TPO is a glycoprotein consisting of 353 aminoacids and has a molecular weight of 30 kDa. The TPO gene is located on chromosome 3q27. TPO structure can be divided into two domains, amino- and carboxy-terminal (Figure 1). The former domain binds to the c-Mpl receptor and shares remarkable homology with erythropoietin (EPO). On the contrary, deletion of the carboxy-terminal domain does not affect the activity of the protein in vitro, although it does decrease its bioavailability after parenteral administration. TPO acts at all levels of megakaryocytopoiesis together with other cytokines, while it is the sole regulator of platelet production and steady state.37–39 Following the description of this important thrombopoietic agent, researchers shifted their attention from initial studies aimed at evaluating the role of TPO in patients with haematologic disorders to studies aimed at evaluating the role of TPO in thrombocytopenia of liver disease.40–42 On the basis of studies carried out on TPO-knock out mice and based on evidence that TPO is almost exclusively produced by the liver in adults, the results of these studies convincingly demonstrated that the relatively low TPO serum levels observed in cirrhotic patients are mainly the result of the decreased hepatic production of this hormone.43–46 In fact, TPO is produced by the liver at a constant rate and is cleared from circulation upon binding to its receptor (c-Mpl) on both megakaryocytes and platelets.47 Thus, circulating TPO levels depend upon hepatic synthesis and peripheral uptake.48 It is evident that impairment of the liver functioning mass may therefore cause a decrease in TPO production. However, while haematological studies reached univocal conclusions regarding the relationship between TPO and platelet homeostasis, the results of TPO studies in hepatology may, at first glance, not appear as straightforward (Figure 2). In fact, various studies performed on different sub-sets of patients affected by CLD reportedly found slightly increased, normal, or decreased TPO blood levels.49–65 These apparent discrepancies are mainly the result of the heterogeneity of the populations that were evaluated, and of the assays (monoclonal vs. polyclonal antibodies) and medium (plasma vs. serum) that were used to evaluate TPO levels.47 Furthermore, one of the main pitfalls that led to considering the TPO levels of patients with CLD as ‘normal’ was that TPO levels were often interpreted separately rather than together with their respective platelet count. Indeed, what seems to emerge is that serum TPO levels in cirrhotic patients are ‘inappropriately low’ for the degree of thrombocytopenia that is observed (Figure 2). This has been elegantly demonstrated by studies that evaluated TPO kinetics and platelet count, as well as platelet function in cirrhotic patients before and after OLT.49–52 Further, it has been observed that TPO serum levels in patients with cirrhosis are decreased in parallel with a decrease in the liver functioning mass, and correlate inversely with the patients’ Child-Pugh score. Furthermore, interesting results were reported in a study which showed that even procedures that relieve portal hypertension, such as partial splenic embolization, produce an increase in platelet count that is mainly because of an improvement in liver function and consequently in TPO secretion.66 Lastly, it has also been shown that successful interferon treatment of patients with chronic hepatitis C and thrombocytopenia is accompanied by an improvement in platelet count, which is mediated by an increase in TPO serum levels.18 In conclusion, the discovery of the important role played by TPO in thrombocytopenia related to CLD has broadened the horizon of possible treatment options for this condition.

Figure 1.

Structure of thrombopoietin (TPO) and erythropoietin (EPO). The receptor-binding domain of TPO shares 20% identity and 45% similarity with EPO.

Figure 2.

Summary of TPO serum levels and relative platelet counts in normals and in both haematological and liver diseases. In particular, TPO serum levels are appropriately high in patients with immune thrombocytopenic purpura and inappropriately low in patients with chronic liver disease and thrombocytopenia. Adapted from Ref.43

The sites and modalities of intervention

Stimulating the principal organ of megakaryocytopoiesis, i.e. the bone marrow, would appear to be the most intriguing and promising possible therapeutic intervention for thrombocytopenia in cirrhotic patients. Thanks to the description of TPO and its modality of action, we now have new insight into the possible treatment options for this stimulation. Thus, putative therapeutic interventions can be divided into two groups: (i) the use of treatment targeted at the TPO-receptor (c-Mpl), and (ii) the use of cytokines with general thrombopoietic potential.67, 68 This sub-division implies different sites and mechanisms of action. Moreover, agents acting on the TPO-receptor (c-Mpl) can be further subdivided into synthetic TPOs and TPO-mimetic agents (Figure 3).

Figure 3.

Main sites of action of putative pharmacologic agents for thrombocytopenia.

Treatments targeted at the TPO receptor (c-Mpl)

Synthetic thrombopoietins

Recombinant human TPO (rhTPO) and pegylated recombinant human megakaryocyte growth and development factor (PEG-rhMGDF) are the synthetic peptides available as synthetic TPOs. MGDF is the truncated form of TPO. Available pharmacodynamic data for PEG-rhMGDF and rhTPO derive from clinical studies carried out on patients undergoing non-myeloablative chemotherapy.69–72 Following PEG-rhMGDF administration, a dose-dependent increase in platelet count may be observed, although with a considerable inter-individual variation (range: 51–584%). Platelet count begins to rise 6 days after PEG-rhMGDF administration, peaks at between 12 and 18 days, and returns to pre-treatment values between 22 and 30 days administration. RhTPO has a similar modality of action. The platelets produced by both PEG-rhMGDF and rhTPO are morphologically normal and do not show any significant changes in aggregating activity. Even though platelet counts usually reach very high values, the incidence of thrombotic events is quite low.73

Although PEG-rhMGDF showed good therapeutic efficacy in favouring platelet recovery in patients undergoing chemotherapy, concerns were raised regarding the safety profile of this peptide, thus development of the drug was discontinued in 1998. In particular, subcutaneous administration of PEG-rhMGDF caused the appearance of neutralizing antibodies that cross-reacted with both recombinant and endogenous TPO and that were associated with the occurrence of thrombocytopenia and pancytopenia in platelet donors and patients on chemotherapy.74 Although the exact reasons for this side-effect remain to be established, it is likely that this phenomenon is caused by the mode of administration, as TPO is a potent stimulator of dendritic cells and the subcutaneous route may enhance immunogenicity by stimulating antigen-presenting cells. To date, rhTPO, which is administered intravenously, has not been associated with the appearance of neutralizing antibodies.73, 75 Similarly, development of promegapoietin, a genetically engineered dual receptor agonist for human interleukin-3 (IL-3) and c-Mpl receptors, was also discontinued following phase I trials after some patients developed antibodies to the molecule associated with severe thrombocytopenia.67, 68, 73 The relevance of these safety concerns as well as financial considerations regarding the potential application of these expensive agents in a ‘small’ market has further discouraged studies regarding the efficacy of PEG-rhMGDF and rhTPO in patients with cirrhosis and thrombocytopenia.75

TPO-mimetic agents

Thrombopoietin is the natural c-Mpl receptor ligand. However, both TPO-mimetic peptides and small molecular weight, non-peptide substances have been designed to target c-Mpl. These substances share no homology with the primary sequence of TPO and compete with the native molecule for c-Mpl binding. TPO mimetics are capable of binding to c-Mpl and of activating the receptor. They have been shown to stimulate the in vitro proliferation and maturation of megakaryocytes from human bone marrow cells and to promote an increase in platelet count when administered to animals.73, 76, 77

GW395058 is a pegylated peptide TPO receptor agonist currently being evaluated for the treatment of chemotherapy-induced thrombocytopenia. Preliminary studies have shown that the potential for inducing neutralizing antibodies is low.78, 79 In a large-animal model of chemotherapy-induced thrombocytopenia, GW395058 administration reduced the thrombocytopenic effects of carboplatin and was not associated with side-effects.80

JTZ-132 is a non-peptidic, small molecular TPO mimetic that was identified by screening chemical library with TPO-dependent and EPO-independent cell lines.81 JTZ-132 binds to the c-Mpl and activates the conventional TPO-induced activation pathway. In a recent study, this molecule was found to reduce platelet nadir and to increase platelet recovery in mice models of both chemotherapy- and irradiation-induced thrombocytopenia.81

Although these substances have shown promising results and an apparent lack of immunogenicity in the experimental setting, appropriate studies are required before suggesting a possible role for treatment of thrombocytopenia in humans.

cytokines with thrombopoietic activity

Initial thrombopoietic cytokines

Interleukin-1 (IL-1), IL-3, and IL-6 are pleiotropic cytokines that act early in the phases of megakaryocytopoiesis, and they all have in vitro thrombopoietic activity. However, clinical application of these cytokines for the treatment of thrombocytopenia was hampered by side-effects and/or marginal efficacy, and therefore these agents did not undergo any further clinical development.8, 67, 73


Interleukin-11 acts synergistically with IL-3, TPO and stem cell factors to increase the number and maturation of megakaryocytic progenitors.82 In particular, recombinant human IL-11 (rhIL-11, Oprelvekin, Neumega, Wyeth/Genetic Institute, Cambridge, MA, USA) acts by increasing the number and maturation of megakaryocytic progenitors. Plasma IL-11 levels are normal in cirrhotic patients with thrombocytopenia and remain unchanged after OLT.51 The mechanisms by which rhIL-11 increases platelet count in patients with liver cirrhosis are not completely clear. Oprelvekin has been approved in the US to prevent severe thrombocytopenia and to reduce the need for platelet transfusion following myelosuppressive chemotherapy for non-myeloid malignancies.83–85 In these patients rh-IL-11 has proven to be fairly effective, but with significant side-effects.

Recently, the efficacy and safety profile of rhIL-11 administration in patients with CLD was assessed in a group of Child-Pugh class A and B cirrhotic patients and in a group of chronic hepatitis C patients undergoing IFN therapy.86 In the first study, subcutaneous administration of rhIL-11 for 10 days (50 μg/kg/day) to 10 cirrhotic patients with platelet counts ≤80 × 109/L led to an increase in their peripheral platelet counts after 4 days of treatment, with a peak at day 13 (median, 93 × 109/L; range: 60–206 × 109/L), and a return to pre-treatment values after a median of 19 days. A slight, transient increase in white blood cell count was also observed, while the haemoglobin concentration decreased starting from day 4 of rhIL-11 administration and returned to pre-treatment levels after the 19th day. The patients’ aminotransferase levels decreased during treatment and rose back to pre-treatment levels during the follow-up period. Interestingly, during rhIL-11 administration, serum bilirubin levels gradually decreased over the first 20 days after the beginning of treatment. The main side-effect reported in this series was fluid retention (60% of patients), which was managed by salt restriction and use of diuretics. Fluid retention was responsible for weight gain and dilutional anaemia, while no effects were reported on arterial pressure or renal function. Importantly, the authors also reported that rhIL-11 allowed safe, intra-arterial chemotherapy in a young patient with hepatocellular carcinoma, a procedure that otherwise could not have been carried out. In the second study, an abstract by other authors reported that rhIL-11 was effective at preventing dose reduction or therapy discontinuation in 25% of the patients undergoing IFN or pegylated-IFN therapy for chronic hepatitis C. It was also useful in order to safely allow anti-viral therapy to be started in 5% of these patients. The only side-effect reported in this study was fluid retention that occurred in 5% of the patients.87

Two other case-reports sustained the use of rhIL-11. One involved a thrombocytopenic patient with chronic hepatitis C who required IFN treatment, while the other involved a cirrhotic patient who needed to undergo surgery.88, 89 Although no side-effects were reported in the first case, the patient in the second report experienced transient weight gain and dilutional anaemia because of water and sodium retention.

Finally, a recent study evaluating the possible beneficial effects on liver biochemistry and histology in 20 chronic hepatitis C patients who were non-responsive to a previous course of anti-viral therapy indirectly confirmed the usefulness of rhIL-11 (5 μg/kg/day s.c. for 12 weeks) in improving platelet counts (143 × 109/L to 198 × 109/L, P < 0.001), and suggested that rhIL-11 may improve liver histology as well.90 The side-effects proved to be milder in this study, which used lower doses of rhIL-11 when compared with the previous studies.

The relatively smaller increase in platelet count and the hypothetically shorter duration of treatment needed by patients with CLD when compared with patients on myelosuppressive chemotherapy makes rhIL-11 an interesting peptide that deserves further evaluation on larger cohorts of patients. The main obstacle to its use is the fluid retention side-effect, whose management may become a significant problem especially in patients with decompensated cirrhosis.

Other agents

The amino-terminal domain of TPO shares a remarkable homology with EPO. Indeed, this domain has 20% sequence identity and 45% sequence similarity with human EPO.39, 91 Noteworthy, this domain is the one that binds to the c-Mpl receptor and is adequate for signalling and supporting cellular proliferation (Figure 2). Although in vitro EPO does not seem to have a direct effect on megakaryocyte production or maturation without TPO-co-stimulation, in vivo EPO treatment caused an increase in circulating megakaryocytic progenitor cells,92 and the EPO receptor is also expressed on megakaryocytes.93

Prior to the description of TPO, a study showed that in normal men undergoing repeated phlebotomies, platelet count increased during EPO treatment when compared with placebo.94 In another study specifically aimed at evaluating the thrombopoietic effect of EPO, rhEPO (4000 IU/day s.c.) was administered to 12 CLD patients with thrombocytopenia in a placebo-controlled trial.95 Criteria for inclusion were the presence of a platelet count <85 × 109/L, and after a mean treatment duration of 13 days (range: 7–20) eight patients reached a platelet count ≤100 × 109/L and were therefore considered responders to treatment (66%). Furthermore, a significant increase in platelet count when compared with pre-treatment values was observed in patients treated with rhEPO but not in controls. Noteworthy, no side-effects were recorded during rhEPO administration. Recently, a randomized, double-blind, placebo-controlled study showed that rhEPO significantly increased platelet counts in patients with alcoholic cirrhosis.96 In this trial, rhEPO (100 IU/kg s.c.) was administered every other day over a period of 5 days to 15 patients with alcohol-related cirrhosis and thrombocytopenia (platelet count <120 × 109/L), and the results were compared with those obtained in seven patients treated with placebo. Treatment with rhEPO determined an increase in platelet count by 20% on day 5 of the study, while on day 9 (i.e. 4 days after the last rhEPO administration), platelet count was 25% greater when compared with baseline. At the end of treatment, platelet counts of rhEPO-treated patients were significantly higher when compared with both pre-treatment values and end of treatment values of controls. Overall, although the exact mechanism(s) by which rhEPO increases platelet count in humans has not been elucidated, the results of these studies seem to suggest that rhEPO is able to increase platelet count in patients with CLD. This is especially important as the use of rhEPO has been suggested for the treatment of ribavirin-induced anaemia in patients with hepatitis C virus-related CLD undergoing IFN and ribavirin association anti-viral treatment. In fact, the use of rh-EPO in these patients is supported by trials showing that rhEPO treatment was able to decrease the need for dose reduction or discontinuation of ribavirin, and improved quality of life when compared with controls,97–99 and that, at least theoretically, its use should increase sustained viral response and be cost-effective as well.100 In this setting, the suggestion of using a single drug for the treatment of both anaemia and thrombocytopenia is intriguing although, unfortunately, studies on the use of rhEPO in hepatitis C patients undergoing IFN and ribavirin combination anti-viral treatment were not specifically aimed at evaluating platelet count behaviour during treatment, and therefore no sound data can be gleaned from these studies.