Pilot study of pentoxifylline in hepatopulmonary syndrome


  • Rajasekhar Tanikella,

    1. Liver Center, Division of Gastroenterology and Hepatology, University of Alabama at Birmingham, Birmingham, AL
    Search for more papers by this author
  • George M. Philips,

    1. Department of Medicine, University of Alabama at Birmingham, Birmingham, AL
    Search for more papers by this author
  • Dorothy K. Faulk,

    1. Liver Center, Division of Gastroenterology and Hepatology, University of Alabama at Birmingham, Birmingham, AL
    Search for more papers by this author
  • Steven M. Kawut,

    1. Department of Medicine, College of Physicians and Surgeons, Columbia University, New York, NY
    2. Department of Epidemiology, Joseph L. Mailman School of Public Health, Columbia University, New York, NY
    Search for more papers by this author
  • Michael B. Fallon

    Corresponding author
    1. Liver Center, Division of Gastroenterology and Hepatology, University of Alabama at Birmingham, Birmingham, AL
    • University of Alabama at Birmingham, 290 MCLM, 1918 University Boulevard, Birmingham, AL 35294-0005
    Search for more papers by this author
    • Telephone: 205-975-5676; FAX: 205-975-9777

  • The ClinicalTrials.gov registration number is NCT00593658.


Hepatopulmonary syndrome (HPS) results when chronic liver disease or portal hypertension causes intrapulmonary microvascular dilatation with hypoxemia. In experimental HPS, tumor necrosis factor alpha (TNF-α) overproduction contributes to vasodilatation, which is improved by pentoxifylline, a TNF-α inhibitor. The effectiveness of pentoxifylline in humans is unknown. The aim of this open-label, single-arm clinical trial was to assess the efficacy and tolerability of pentoxifylline in patients with cirrhosis and advanced HPS undergoing liver transplantation evaluation. Nine adults with cirrhosis and moderate to severe HPS were enrolled. All patients had an initial 2-week titration to a target dose of pentoxifylline of 400 mg by mouth every 8 hours, which was continued for 6 weeks. Baseline and follow-up arterial blood gases and TNF-α levels were evaluated. Adverse effects and tolerability were assessed. The 9 patients had a mean age of 55 ± 10 years, and 67% were female. The most common causes of cirrhosis were hepatitis C virus and alcohol (55%). The mean Model for End-Stage Liver Disease score was 11 (range, 6-19), and patients had advanced hypoxemia [mean partial pressure of arterial oxygen (PaO2) = 54 ± 12 mm Hg, mean alveolar-arterial oxygen gradient (A-a PaO2) = 57 ± 15 mm Hg]. Of the 9 patients enrolled, follow-up blood gases were done in 7. There was no significant change in PaO2 (P = 0.3) or A-a PaO2 (P = 0.3) with treatment. Pentoxifylline was poorly tolerated. Nausea (100%) and vomiting (56%) were the predominant side effects, and only a single patient was able to complete full-dose therapy. Treatment with pentoxifylline did not improve arterial oxygenation in advanced HPS, and tolerance was limited by gastrointestinal toxicity. Liver Transpl 14:1199–1203, 2008. © 2008 AASLD.

Hepatopulmonary syndrome (HPS) results from intrapulmonary microvascular dilatation that impairs arterial oxygenation in the setting of cirrhosis or portal hypertension.1 As many as 10%-20% of patients with cirrhosis being evaluated for orthotopic liver transplantation (OLT) have advanced HPS,2 and mortality is greater in those with HPS than in those without HPS.3 Currently, OLT is the only effective treatment, although postoperative mortality in HPS is increased with respect to patients with cirrhosis without HPS, with a 1-year survival of 68% to 80%.4 Therefore, effective medical therapy for advanced HPS could improve both preoperative and postoperative mortality.

Recent work in experimental models of HPS has revealed that both nitric oxide synthase–derived nitric oxide and heme oxygenase–derived carbon monoxide cause intrapulmonary vasodilatation. These alterations appear to be driven in part by tumor necrosis factor alpha (TNF-α) modulation of pulmonary blood flow and intravascular monocyte accumulation.5–7 Pentoxifylline is a nonspecific phosphodiesterase inhibitor with inhibitory effects on TNF-α and may be beneficial in a subset of patients with severe alcoholic hepatitis in which TNF-α overproduction contributes to liver injury.8, 9 In experimental HPS, pentoxifylline administration also decreases the severity of oxygenation abnormalities.10, 11 However, pentoxifylline therapy has been associated with dose-limiting side effects in patients with liver disease,8, 12, 13 and the tolerability of pentoxifylline in patients with cirrhosis and advanced HPS is unknown. Therefore, we conducted a clinical trial to evaluate the efficacy and tolerability of 8 weeks of pentoxifylline in patients with cirrhosis and advanced HPS who were being considered for OLT.


A-a PaO2, alveolar-arterial oxygen gradient; ALT, alanine aminotransferase; AST, aspartate aminotransferase; HCV, hepatitis C virus; HPS, hepatopulmonary syndrome; INR, international normalized ratio; IQR, interquartile range; MELD, Model for End-Stage Liver Disease; NA, not available; NASH, nonalcoholic steatohepatitis; OLT, orthotopic liver transplantation; PaO2, partial pressure of arterial oxygen; TNF-α, tumor necrosis factor alpha.


Study Design and Screening

This was an open-label, single-arm clinical trial of pentoxifylline in subjects with HPS. Patients undergoing evaluation for liver transplantation at the University of Alabama at Birmingham were screened for eligibility. The institutional review board at the University of Alabama at Birmingham approved the study.

Inclusion Criteria

We included subjects ≥ 18 years of age with cirrhosis defined by a combination of clinical, laboratory, radiologic, and pathologic findings. All enrollees had advanced HPS as defined by the European Respiratory Society Task Force.14 Each patient had a contrast echocardiogram consistent with intrapulmonary shunting and an arterial oxygen tension [partial pressure of arterial oxygen (PaO2)] of <70 mm Hg. Subjects had to be able to provide informed consent.

Exclusion Criteria

We excluded patients with active infections, known malignancy, a history of intolerance to pentoxifylline, and intrinsic cardiopulmonary disease, which was defined as any of the following: (1) an elevated estimated right ventricular systolic pressure or right or left ventricular systolic dysfunction by echocardiography, (2) a pulmonary infiltrate or pleural effusion seen on chest radiography, (3) an obstructive ventilatory defect (forced expiratory volume in 1 second/forced vital capacity < 0.70), or (4) a restrictive ventilatory defect (vital capacity or total lung capacity < 70% predicted).


The primary endpoints for this study were the changes in the PaO2 and alveolar-arterial oxygen gradient (A-a PaO2) from baseline to follow-up and the incidence of adverse events. Secondary endpoints were the changes in the Model for End-Stage Liver Disease score, international normalized ratio, serum creatinine, liver function tests, and TNF-α from baseline to follow-up.

Data and Specimen Collection

Patient data, including demographics, etiology of liver disease, routine laboratory results, contrast echocardiography, pulmonary function test results, and abdominal radiological studies, were collected. Sampling for arterial blood gases was performed with the subject seated while breathing room air at baseline and at the time of discontinuation of pentoxifylline. A-a PaO2 values were calculated with the alveolar gas equation. Liver function tests were repeated at the time pentoxifylline was discontinued. In a subset of patients, serum levels for TNF-α were obtained before and after treatment.

Dosing Methods and Adverse Event Reporting

Subjects were instructed to take 400-mg extended-release tablets of pentoxifylline (TEVA Pharmaceuticals) once daily by mouth for 7 days followed by 400 mg twice daily for 7 days and then 400 mg thrice daily for 42 days. Each subject was given a log to record the number of tablets taken and any adverse effects on a daily basis. This log was reviewed weekly by telephone conference by the research coordinator. Patients had clinic visits at week 4 and week 8 or at discontinuation of pentoxifylline. Adverse events were reported to the investigator, and if it was necessary, doses were adjusted. All patients were given proton pump inhibitors prior to initiation of pentoxifylline, and antiemetics and antidiarrheal agents were used as needed for side effects.

Statistical Analysis

Data were summarized as mean ± standard deviation, median and interquartile range (IQR), or n (%) as appropriate. Signed rank tests were used to compare the PaO2, A-a PaO2, and secondary endpoints at baseline and after therapy. A 2-tailed P value of less than 0.05 was considered significant. With α = 0.05 and a sample size of 7 patients, there was more than 80% power to detect an effect size of 1.3 standard deviations. SPSS version 15.0 (SPSS, Inc., Chicago, IL) was used for all analyses.


Patient Recruitment and Characteristics

We screened 12 consecutive subjects between 40 and 69 years old found to have hypoxemia and intrapulmonary shunting on contrast echocardiography between January 2005 and October 2006. Of these, 10 met inclusion and exclusion criteria and were offered enrollment. One declined study participation. A total of 9 subjects were enrolled into the study after written consent was obtained. Baseline clinical characteristics of the 9 subjects are summarized in Table 1. The mean age was 55 ± 10 years, and 67% were female. Hepatitis C virus and alcohol, alone or in combination, were the most common causes of cirrhosis (55%). The mean Model for End-Stage Liver Disease score was 11 (range, 6-19). All patients had advanced HPS on the basis of arterial blood gas analysis (PaO2 = 54 ± 12 mm Hg, A-a PaO2 = 57 ± 15 mm Hg).

Table 1. Baseline Characteristics of the Study Patients
VariableHPS (n = 9)
  1. NOTE: Data are shown as mean ± standard deviation or n (%).

  2. Abbreviations: A-a PaO2, alveolar-arterial oxygen gradient; ALT, alanine aminotransferase; AST, aspartate aminotransferase; HCV, hepatitis C virus; HPS, hepatopulmonary syndrome; INR, international normalized ratio; MELD, Model for End-Stage Liver Disease; NASH, nonalcoholic steatohepatitis; PaO2, partial pressure of arterial oxygen.

Age, years55 ± 10
Female6 (67)
Cause of cirrhosis 
 HCV3 (33)
 Alcohol1 (11)
 Cryptogenic2 (22)
 HCV, alcohol1 (11)
 Hemochromatosis1 (11)
 NASH1 (11)
Total bilirubin (mg/dL)2.5 ± 1.7
ALT (U/L)35 ± 24
AST (U/L)56 ± 32
Alkaline phosphatase (U/L)152 ± 69
INR1.4 ± 0.2
Serum creatinine (mg/dL)0.8 ± 0.2
MELD score11 ± 4
PaO2 (mm Hg)54.4 ± 12.5
A-a PaO2 (mm Hg)57.3 ± 15.5

Arterial Blood Gas Analysis

Of the 9 patients enrolled, arterial blood gas sampling was performed before and after therapy in 7 (Fig. 1). In the remaining 2 patients, arterial blood gases were not repeated after therapy as one underwent OLT at week 3 of treatment (patient 4) and the other withdrew from the study at 7 weeks (patient 5). Patient 1 discontinued therapy after 6 weeks and patients 6 and 9 discontinued therapy after 3 weeks because of side effects. Each of these patients returned for arterial blood gas sampling at the time pentoxifylline was stopped. In the 7 treated patients, there was no improvement in arterial oxygenation after treatment. The median (IQR) before and after therapy for PaO2 was 63 (45, 66) versus 61 (41, 63) mm Hg (P = 0.3) and for A-aPaO2 was 48 (42, 68) versus 49 (43, 70) mm Hg (P = 0.3).

Figure 1.

Individual A-a PaO2 values before and after pentoxifylline therapy. Abbreviations: A-a PaO2, alveolar-arterial oxygen gradient; NA, not available; OLT, orthotopic liver transplantation.

TNF-α Measurements

In a subset of 6 patients, baseline TNF-α values were obtained, and the mean baseline TNF-α level was 10 pg/mL (range, 6-17.5 pg/mL). In 3 patients (patients 1, 2, and 3), TNF-α levels were repeated after therapy and were not significantly different from pretreatment values, with a median increase of 0.2 pg/mL (IQR, 0-6.2). TNF-α levels before and after therapy were as follows: patient 1, 17.5 and 23.7 pg/mL; patient 2, 12.3 and 12.5 pg/mL; and patient 3, 8 and 8 pg/mL.

Adverse Effects and Tolerability of Pentoxifylline

Adverse effects were common, particularly gastrointestinal effects, with nausea and dyspepsia occurring in 100%, vomiting in 56%, and diarrhea in 44% (Table 2). Pentoxifylline therapy was not associated with significant changes in renal or hepatic function. The median serum creatinine level remained normal before [0.8 mg/dL (IQR, 0.7-0.8)] and after treatment [0.9 mg/dL (IQR, 0.7-1)]. Likewise, the liver enzymes [median change in alanine aminotransferase, 1.0 U/L (−10 to 17 U/L), aspartate aminotransferase, −0.5 U/L (−9 to 11 U/L), and alkaline phosphatase, −19 U/L (−34 to 58 U/L)], international normalized ratio [−0.01 (−0.1 to 0.06)], and Model for End-Stage Liver Disease scores [2.5 (−1.3 to 4)] were not significantly different before and after therapy. No patients reported bleeding, change in ascites, portosystemic encephalopathy, or infections during therapy.

Table 2. Adverse Effects Due to Pentoxifylline Among the Study Subjects
Adverse EffectNumber of Patients (%)
Nausea/dyspepsia9 (100%)
Vomiting5 (56%)
Diarrhea4 (44%)
Blurred vision2 (22%)
Dizziness1 (11%)
Headache1 (11%)
Fatigue1 (11%)

Adverse events significantly limited the tolerability of pentoxifylline treatment. Patient 4 was not included in tolerability assessments as this patient underwent OLT during treatment. As a group, patients took only 60% (range, 11%-100%) of the total intended study drug dose, and pentoxifylline was taken as directed in the protocol on only 44% (range, 11%-100%) of the study days. Only 1 patient (12%) could tolerate the prescribed full dose of pentoxifylline for 8 weeks. Table 3 presents the dose and duration of pentoxifylline treatment in each patient and whether arterial blood gases improved in the 7 patients for whom measurements were repeated after therapy. Of note, the single patient (patient 7) who completed full-dose therapy had no improvement in arterial blood gases at the end of treatment.

Table 3. Duration of Pentoxifylline Treatment and Dose Administered
Patient NumberDuration of Treatment (Days)Total Dose (Tablets)Improvement in Oxygenation
  1. Abbreviations: NA, not available (patient withdrew from the study); OLT, orthotopic liver transplantation.



This clinical trial was designed to obtain preliminary evidence on the efficacy and tolerability of pentoxifylline in patients with cirrhosis and advanced HPS. We found that pentoxifylline therapy did not improve arterial oxygenation in our patient group. However, tolerability of pentoxifylline was poor, and the majority of the patients were unable to take the prescribed dose. These findings support that pentoxifylline is not an effective therapy for advanced HPS in patients with cirrhosis.

Our finding that pentoxifylline did not improve oxygenation in human HPS differs from findings in an experimental model in which marked improvement in gas exchange was observed after treatment in association with a reduction in circulating TNF-α levels.10, 11 One potential reason for this result is that the frequent dose-limiting toxicity observed here impaired effects on TNF-α production. Baseline TNF-α levels, measured in 6 of our patients, were above established normal values but were lower than those reported in experimental HPS and in studies of alcohol-related liver disease.8, 13 In 3 of our patients, TNF-α measurements were repeated after pentoxifylline therapy, and levels did not uniformly decline. This finding is consistent with the concept that poor tolerance limited effectiveness. However, in the 1 patient in our study (patient 1) in whom oxygenation did improve on therapy, circulating TNF-α levels did not decline, and only 57% of the intended dose of pentoxifylline was delivered. In addition, in a large study in which pentoxifylline improved outcome in alcoholic hepatitis, no reduction in circulating TNF-α levels was seen.8 Finally, we have not excluded that the oxygenation response to pentoxifylline treatment may vary according to the severity of HPS (experimental HPS has mild oxygenation abnormalities) or that the underlying pathophysiologic mechanisms in human disease are distinct from those in animal models.

Adverse gastrointestinal effects due to pentoxifylline were common in our study and resulted in a 44% attrition rate. In other studies of patients with liver disease, the patient withdrawal rate due to adverse effects has ranged from 0% to 40%.8, 12, 13 We employed a dose escalation protocol in anticipation of adverse effects but were unable to avert gastrointestinal toxicity or improve tolerance. One potential explanation for the high incidence of adverse effects in our patients is the observation that pentoxifylline normally undergoes significant first pass metabolism in the liver to an active metabolite and that serum levels are significantly higher in cirrhosis.15 The observation that our patient withdrawal rate was similar to that in a prior study of patients with cirrhosis supports the idea that severity of liver disease may influence tolerance.13 We did not observe changes in serum creatinine or liver function before or after therapy with pentoxifylline that might have contributed to changes in pentoxifylline metabolism.

In summary, in this pilot trial of pentoxifylline in patients with cirrhosis and advanced HPS, we did not find improvement in arterial oxygenation with therapy. We did observe significant gastrointestinal toxicity with therapy, which limited the amount of drug administered, thereby diminishing effectiveness. However, we cannot exclude that pentoxifylline, if tolerated or if targeted to subgroups with less severe HPS or liver dysfunction, may be efficacious and have beneficial effects on oxygenation. However, our findings support the idea that pentoxifylline is unlikely to be a widely effective medical therapy for advanced HPS in patients with cirrhosis.