The anti-viral effect of sorafenib in hepatitis C-related hepatocellular carcinoma


Correspondence to:

Dr R. Cabrera, 1600 SW Archer Road, Room M440, PO Box 100214, JHMHSC, Gainesville, FL 32610, USA.




Sorafenib is currently the only approved systemic therapy shown to have efficacy in the treatment of advanced hepatocellular carcinoma (HCC). Recent studies suggest that hepatitis C (HCV)-related HCC patients derive more clinical benefit from sorafenib than other subgroups, but the mechanism for this effect is unknown. In vitro data suggest that sorafenib may exert anti-viral properties, and thus our aim in this study was to evaluate potential anti-viral activity of sorafenib in patients with HCV-related HCC.


To evaluate potential anti-viral activity of sorafenib in patients with HCV-related HCC.


We prospectively enrolled patients with HCV-related HCC treated with sorafenib for up to 6 months. Baseline clinical, viral and oncologic data were collected. Patients' HCV viral loads were obtained at various time points, and compared with their baseline viral levels. No patients received any known anti-viral therapy during this time.


Thirty-three patients were identified with baseline and subsequent HCV levels available for analysis. Six patients completed 6 months of full dose sorafenib, and comparisons of their HCV viral loads showed no significant change at week 24 (difference of means = 0.3500, CI: −0.1799–0.8799, P = 0.150), or the interim time points. Similarly, the HCV viral loads of all patients who received sorafenib and the viral loads of those patients who had tumour response to sorafenib showed no significant changes at any time point.


Despite preclinical data and previous subgroup analyses suggesting that sorafenib has an anti-viral effect against HCV, this study suggests that sorafenib lacks significant anti-viral activity in HCV patients with HCC.


Hepatocellular carcinoma (HCC) represents a major global health challenge as the fifth most common cancer in the world and the fastest rising cause of cancer-related death.[1] In most cases, HCC arises in the setting of cirrhosis. In the United States, the most common cause of cirrhosis is infection with chronic hepatitis C virus (HCV).[1, 2] In patients who are not candidates for or do not respond to HCV treatment, the sole approach to reducing cancer-related death in these patients is early detection.[3] Unfortunately, the majority of patients with HCC present with advanced cancer when they are not candidates for curative treatments.

For patients with advanced HCC, sorafenib has become the standard treatment recommended by practice guidelines after a survival benefit was shown with two pivotal trials.[4, 5] The clinical success of sorafenib led to a number of sub-analyses to clarify the benefit by subgroups. A post hoc retrospective subgroup analysis of HCV-positive patients in the Sorafenib Hepatocellular Carcinoma Assessment Randomized Protocol (SHARP) study showed an improvement in median overall survival and time to progression over non-HCV-infected patients.[6] In addition, preclinical studies indicate that sorafenib may inhibit HCV infectivity and replication.[7-9] These data spurred efforts to determine the explanations for this finding.

If sorafenib has anti-viral activity against HCV, these patients who develop HCC in the setting of HCV could have an added benefit in treatment with sorafenib, by reducing the HCV viral load. In this study, we determine if sorafenib has anti-viral activity in HCV-infected patients by prospectively following a cohort of patients with HCV-related HCC receiving treatment with sorafenib.


Study design

This study prospectively followed patients treated in the University of Florida Liver Cancer Clinic for intermediate or advanced stage hepatocellular carcinoma (HCC) with an underlying diagnosis of chronic hepatitis C infection (HCV). The study was approved by the Institutional Review Board at the University of Florida.

Study population

Eligible patients were 18 years of age or older, with chronic HCV infection, cirrhosis and stage A, B orC HCC by the Barcelona Clinic Liver Cancer (BCLC) staging system. Cirrhosis was diagnosed by clinical, laboratory, radiographic and/or endoscopic criteria. Patients were excluded if they had severely decompensated cirrhosis (Child's-Pugh score C), an Eastern Cooperative Oncology Group (ECOG) performance status of greater than 2, or were unable to give informed consent. Each patient's clinical history and imaging were discussed in a weekly Multidisciplinary Hepatobiliary Conference consisting of participants from hepatology, diagnostic and interventional radiology, pathology, medical oncology and transplant surgery. The diagnosis of HCC was reached according to American Association for the Study of Liver Diseases (AASLD) practice guidelines[3] and a consensus treatment plan was formed for each patient.


Informed consent was obtained, and patients were prospectively followed once they started treatment with sorafenib. Most patients were started at full dose sorafenib (400 mg twice daily), but some patients were started at a reduced dose (200 mg twice daily) at the discretion of the treating provider. Patients were evaluated in the Liver Cancer Clinic regularly to monitor for adverse effects of sorafenib or concurrent liver-directed therapy (LDT) for HCC. Dose reductions of sorafenib due to adverse effects were conducted, in accordance with our institutional protocol,[10] upon the development or worsening of hand-foot skin reaction, diarrhoea, hypertension, nausea, vomiting and/or fatigue. All adverse effects and dose reductions were documented in the patient database. Patients were evaluated for concurrent LDT by the Multidisciplinary group, and underwent these treatments as indicated. No patients received anti-viral therapy during the study period.

Outcome measures

The log10 HCV ribonucleic acid (RNA) levels were determined using polymerase chain reaction (PCR) amplification, followed by quantitative transcription-mediated amplification (TMA) if needed (Heptimax; Roche Diagnostics, Pleasanton, CA, USA) (lowest level of detection = 5 IU/mL). Baseline HCV RNA levels were determined at the start of sorafenib therapy, and at time points corresponding to clinically relevant time points in HCV treatment (week 4, corresponding to rapid virological response and week 12, corresponding to early virological response). In addition, interval time points (week 2 and week 8) were analysed to delineate viral kinetics.

The primary outcome measure was a change in HCV RNA levels after 12 weeks of full dose sorafenib therapy (‘full dose group’). Secondary outcome measures were a change in HCV RNA levels after 12 weeks of any dose sorafenib therapy (‘complete treatment group’), and a change in HCV RNA levels in patient who had tumour response to sorafenib (‘tumor response group’). Tumour response was defined as either complete or partial response by the modified Response Evaluation Criteria in Solid Tumors (RECIST) criteria.[11]

Statistical analysis

Demographic and clinical characteristics were characterised with descriptive statistics. For each time point, mean log10 HCV RNA levels, with standard deviations, were determined using descriptive statistics. Any missing HCV RNA values were determined missing at random, and therefore list-wise deletion was employed to calculate results. The difference between the baseline mean HCV RNA level and the mean HCV RNA level at each subsequent time point was determined with paired-sample t-tests, with 95% confidence intervals. A P-value of 0.05 or less was considered statistically significant.


Baseline clinical factors

The study cohort comprised 33 patients with HCV-related HCC (Table 1). Most were male (85%) and the average age of the patients was 60 years (50–80 years) at the start of treatment. Overall, the majority of the patients were Caucasian (90.9%). One patient had cirrhosis secondary to both HCV and chronic alcohol use, and the remainder (97%) had cirrhosis due to HCV alone.

Table 1. Baseline demographics of all patients treated with sorafenib (= 33)
Subject demographicsAgeMean (range)59.81 (50–80)
  1. AFP, alpha-fetal protein; BCLC, Barcelona Clinic Liver Cancer staging system; ECOG, Eastern Cooperative Oncology Group; EBRT, external beam radiation therapy; ETOH, alcohol; HCV, hepatitis C virus; IRE, irreversible electroporation; MELD, Model of End Stage Liver Disease; RFA, radiofrequency ablation; TACE, trans-arterial chemoembolisation; Y-90, Ytrium-90.

 GenderMale28 (84.8%)
  Female5 (15.2%)
 RaceCaucasian30 (90.9%)
  African-American2 (6.1%)
  Hispanic1 (3.0%)
 HCV genotype114 (42.4%)
  22 (6.1%)
  35 (15.2%)
  Unknown12 (36.3%)
 HCV treatment historyRelapser5 (15.2%)
  Partial Responder2 (6.1%)
  Nonresponder7 (21.2%)
  Naïve12 (36.3%)
  Unknown7 (21.2%)
 Baseline HCV viral load≥800 000 IU/mL22 (66.7%)
  <800 000 IU/mL11 (33.3%)
Liver functionChild-Pugh classA20 (60.6%)
  B13 (39.4%)
 MELDMean (range)9 (6–16)
 ECOG019 (57.6%)
  111 (33.3%)
  23 (9.1%)
Tumour characteristics and historyBCLC stageA1 (3.0%)
  B16 (48.5%)
  C16 (48.5%)
 AFPMean (range)1685 (3.7–10562.2)
 Liver-directed therapy prior to start of studyRFA2 (6.1%)
  TACE23 (69.7%)
  Resection1 (3.0%)
  IRE1 (3.0%)
  None8 (24.2%)
 Liver-directed therapy during studyRFA0
  TACE13 (39.4%)
  Y-906 (18.2%)
  EBRT2 (6.1%)
  None10 (30.3%)

The most common HCV genotype was 1 (42.4%), with two patients being genotype 2, and five patients being genotype 3. The genotype was unknown in 12 patients, as this was not available in their clinical records. There was a fairly even distribution of prior HCV treatment responses with most being prior nonresponders (21%) or naïve (36%) as would be expected in a majority genotype 1 US population. The majority of patients (67%) had a high viral load at baseline (greater than or equal to 800 000 IU/mL).

The majority of patients were Child-Pugh A (51.5%) or Child-Pugh B (39.47%). There were no Child-Pugh C patients. The average laboratory Model of End Stage Liver Disease (MELD) score was 9 at the start of treatment with sorafenib. Mean baseline asparate aminotransferase, alanine aminostransferase, alkaline phosphatase and total bilirubin were 88 U/L, 64 U/L, 166 U/L and 1.5 mg/dL respectively. Just over half the patients (58%) had a performance status of 0, while the remainder of the patients had a performance status of 1 (33%) or 2 (9%).

Most patients had unresectable HCC compatible with BCLC stage B (48.5%) and C (48.5%). The average serum alpha-fetoprotein level was 1685 (3.7–10562.2). A large number of patients received LDT for HCC before (76%) and/or during (61%) the 24 weeks of treatment with sorafenib. Trans-arterial chemoembolisation (TACE) comprised the majority of pre-sorafenib (70%) and during-sorafenib (39%) LDT.


Of the 33 patients in the study cohort, 23 (70%) were started on full dose sorafenib. Of the entire treatment group, only 12 patients (36%) were able to tolerate full dose sorafenib for 12 weeks or greater. Eight patients (24%) were started on full dose sorafenib and tolerated this without dose reduction for 24 weeks or longer. Of 10 patients who were initially started on low-dose sorafenib, only four patients (40%) were subsequently able to tolerate full-dose sorafenib for at least 12 weeks.

Viral kinetics

The average change in HCV viral load in log10 was assessed in the full-dose group, the complete treatment group and the tumour response group. In the full-dose group (n = 11), there was a slight increase in the mean HCV viral load from baseline to week 12 of 0.38 log (n = 8), which was not statistically significant (P = 0.07) (Figure 1). There was an increase in the viral load at week 2 (n = 3, +0.23 log, P = 0.25), at week 4 (n = 6, +0.20 log, P = 0.01) and week 8 (n = 6, +0.20 log, P = 0.10).

Figure 1.

Mean log10 HCV RNA of all patient treated with sorafenib, with vertical bars indicating standard deviation at each time point.

In the complete treatment group (n = 33), the mean HCV viral load log was slightly higher than baseline at 12 weeks (n = 17, 0.25 log), but this difference was not statistically significant (P = 0.16) (Figure 2). There was a slight decrease in mean viral load at week 2 (n = 15, −0.09 log, P = 0.72), a slight increase in mean viral load at week 4 (n = 18, +0.01 log, P = 0.91) and a slight decrease in mean viral load at week 8 (n = 20, −0.16 log, P = 0.45).

Figure 2.

Mean log10 HCV RNA of patients who received at least 12 weeks of full dose sorafenib, with vertical bars indicating standard deviation at each time point.

Viral kinetics and tumour response

In the tumour response group (Figure 3), the mean viral load steadily decreased from baseline to week 2 (n = 10, −0.26 log, P = 0.48), week 4 (n = 14, −0.03 log, P = 0.82) and week 8 (n = 16, −0.23 log, P = 0.38). None of these changes were statistically significant. Ultimately, at week 12, the mean viral load increased slightly at week 12 (n = 15, 0.24 log, P = 0.20).

Figure 3.

Mean log10 HCV RNA of patients who had a tumor response to sorafenib, with vertical bars indicating standard deviation at each time point.


In this study, we demonstrate that sorafenib has no significant impact on HCV viral loads, regardless of the dose of sorafenib used or the presence of a tumour response to sorafenib. The only statistically significant change in HCV viral load seen was actually a small increase very early in the course of treatment in the patients who received full-dose sorafenib for at least 12 weeks, but this change was not maintained after the 4-week time point.

Cirrhosis related to chronic HCV infection accounts for at least half of all patients with HCC in the US. Clearance of HCV in patients with cirrhosis reduces the risk of developing HCC.[12] However, many patients with cirrhosis are not candidates for HCV treatment or do not respond to HCV treatment.[13] Therefore, the majority of HCV patients who develop HCC are actively infected. Much interest is focused on the effect that HCC treatment might have on the underlying HCV infection. In patients with advanced HCC, the only FDA-approved therapy is sorafenib.[10, 14-16] To our knowledge, this is the first published study of the clinical impact of sorafenib on HCV viral loads.

To avoid skewing the findings, list-wise deletion was used for missing HCV RNA values. This technique resulted in smaller sample sizes at each time point, which could underestimate the effect of sorafenib on HCV viral load. However, the vast majority of early phase studies of novel agents for HCV treatment consist of 10 or fewer patients, which is comparable with our patient groups. As our study did not employ a placebo control group, we are unable to conclude if the small changes seen in HCV viral loads could have been related to a placebo effect or natural variations in HCV viral loads. An analysis of the SHARP trial data, which were placebo-controlled, could answer this question. Polymorphisms of interleukin 28B, which could play a role in any effects of sorafenib on HCV viremia, were not available in the majority of the patients. Finally, it is possible that any effects of sorafenib on HCV viral loads are seen only at doses greater than 400 mg twice daily, or that the observed decrease in sorafenib area under the curve concentrations with time[17] would require more frequent dosing to reveal the anti-viral effects. Both of these dosing strategies, however, would be clinically limited by patient tolerability.

Preclinical studies have demonstrated several methods by which sorafenib might inhibit HCV replication. In cell cultures models, sorafenib reduces the susceptibility of hepatocytes to HCV infection via anti-VEGF activity.[7, 8] Sorafenib can also directly inhibit HCV replication via nonstructural HCV replicon protein NS5A interaction with C-Raf.[9] Erlotinib, another multikinase inhibitor being studied for the treatment of HCC, has been reported to be associated with a decrease in HCV viral load.[18] Finally, post hoc analyses of the data from the SHARP trial indicated that patients with HCV-related HCC had improved survival and time to progression when treated with sorafenib than patients with non-HCV-related HCC.[6] This study, however, demonstrates that sorafenib does not have a significant impact on HCV viral loads, even when patients take the medication at full dose.

It appears that the preclinical evidence of sorafenib's inhibition of HCV replication is not clinically relevant. In HCV-infected patients, benefits in morbidity or mortality have not been associated with transient viral suppression. Thus, it would be hard to argue a survival benefit for sorafenib related to a minimal effect on viral replication. The improved survival noted in patients with HCV-related HCC on sorafenib is likely secondary to another factor. This improved survival could be related to an antiportal hypertensive effect of sorafenib, which has been demonstrated in preclinical models [19, 20] and in prospective clinical studies.[21-23] Other possibilities include an effect on the patient's antitumour and/or anti-viral immunity via the multiple pathways affected by sorafenib. At this time, there is no sufficient data to determine why patients with HCV-related HCC appear to have improved survival on sorafenib.

In conclusion, this study demonstrates that sorafenib does not have a significant impact on HCV viral loads in patients with HCV-related cirrhosis and HCC. These findings warrant confirmatory placebo-controlled studies, and further investigation such as a larger, multicentre trial is needed to determine which other factors are responsible for the apparent increased survival of HCV-related HCC patients on sorafenib.


Declaration of personal interests: Dr Cabrera has served as a speaker and consultant for Bayer AG and has received research funding from Bayer AG. Patrick Horne has received research funding from Bayer AG. Dr Nelson has served as a consultant for Bayer AG and has received research funding from Bayer AG and Human Genome Science.

Declaration of funding interests: This study was funded in part by National Institutes of Health/National Center for Research Resources Award UL1 RR029890 and National Institutes of Health/National Cancer Institute award K24CA139570.