Phase 2 study of bevacizumab plus erlotinib in patients with advanced hepatocellular cancer




Epidermal growth factor receptor (EGFR) and vascular endothelial growth factor (VEGF) are rational targets for therapy in hepatocellular cancer (HCC).


Patients with histologically proven HCC and not amenable to curative or liver directed therapy were included in this 2-stage phase 2 trial. Eligibility included an Eastern Cooperative Oncology Group (ECOG) performance status (PS) of 0 or 1 and Child's Pugh score of A or B, and 1 prior systemic therapy. Patients received erlotinib 150 mg daily and bevacizumab 10 mg/kg on days 1 and 15 every 28 days. Objective tumor response was the primary end point.


Twenty-seven patients with advanced HCC (median age, 60 years) were enrolled in this multi-institutional study. The proportion of patients with Child's A classification was 74%. One patient had a confirmed partial response and 11 (48%) achieved stable disease. Median time to disease progression was 3.0 months (95% confidence interval [CI], 1.8-7.1). Median survival time was 9.5 months (95% CI, 7.1-17.1). Grade 3 toxicities included rash, hypertension, fatigue, and diarrhea.


In this trial, erlotinib combined with bevacizumab had minimal activity in patients with advanced HCC based on objective response and progression-free survival. The role of targeting EGFR and VEGF in HCC needs further evaluation in molecularly selected patients. Cancer 2012. © 2011 American Cancer Society.


Hepatocellular cancer (HCC) is the most common primary neoplasm of the liver, accounting for approximately half a million deaths annually worldwide.1 The rising incidence in the United States, as well as in other parts of the western world and Japan, is largely attributed to a significant increase in the prevalence of hepatitis C-related chronic liver disease.2 Choice of treatment for patients with HCC largely depends on the stage of the disease, presence or absence of chronic liver disease, and degree of hepatic dysfunction. Unfortunately, most patients with HCC are not candidates for curative treatments because of large and multicentric tumors, macrovascular invasion, or extrahepatic spread at the time of diagnosis. Conventional cytotoxic therapy has been disappointing and associated with significant toxicity.3, 4

Among the most prominent pathologic and radiologic features of HCC in addition to its multicentricity is its hypervascularity and early invasion of vascular structures. Vascular endothelial growth factor (VEGF) is the most potent and specific of angiogenic growth factors, and its expression is increased in surgical specimens of HCC compared with nontumoral liver tissue.5-8 The degree of VEGF expression has also been shown to increase according to disease stage and histologic tumor grade.9-11

Epidermal growth factor receptor (EGFR) and its ligands EGF and TGF-alpha are important in cell proliferation, as well as motility, adhesion, invasion, survival, and angiogenesis.12 EGFR is actively expressed in human hepatoma cell lines, and EGF is a potent mitogen in these cells.13-15 There is also an increased coexpression of TGF-alpha and EGFR that occurs frequently in human HCC. In addition, evidence supports a role for hepatitis C virus (HCV) core proteins to directly activate the MAP kinase cascade downstream of EGFR and to prolong its activity in response to mitogenic stimuli in HCV-infected liver cells.16

The multiplicity and complexity of molecular aberrations in HCC necessitates a rational multitargeted approach in therapy. In addition to the VEGF-VEGFR pathway, other signaling molecules may be considered as rational targets for new drug development in HCC. The role of EGFR in HCC's pathogenesis and the early clinical reports of the potential benefit of targeting EGFR-mediated signaling prompted further development of this class of agents in HCC. The studies by Philip et al17 and Thomas et al18 demonstrated a potential benefit for erlotinib in disease stabilization in patients with advanced HCC. The early reports of single-agent bevacizumab inducing regression of vascularity and tumor growth in advanced HCC19 were proof of principle in targeting VEGF in HCC. The current study was therefore built on experiences from these pilot trials in advanced HCC that tested single-agent activities of either erlotinib or bevacizumab. In addition to the multitargeted approach using bevacizumab with erlotinib, there was evidence from preclinical models to support an antiangiogenic effect by targeting the EGFR pathway that would potentiate the effect of bevacizumab.26

Taken together, preclinical evidence on the role of EGFR and VEGF/VEGFR in the pathogenesis of HCC and the early reports of activities of anti-EGFR17, 18 and anti-VEGF/VEGFR19, 20 drugs in patients with advanced HCC provided a rationale for investigating the dual inhibition of angiogenesis and EGFR as a multitargeted treatment in patients with advanced HCC. The primary objective of this study was to evaluate the response rate of the combination of erlotinib and bevacizumab in patients with advanced HCC in a multi-institutional setting.


Patient Eligibility

Patients were to have histologically or cytologically confirmed HCC and not be candidates for resection, ablation, liver transplantation, or liver directed therapy. Other eligibility criteria included presence of measurable lesion(s), Eastern Cooperative Oncology Group (ECOG) performance status (PS) 0 or 1, Child's Pugh Classification A or B, and 1 prior regimen of either systemic or liver directed therapy. For patients with history of prior cryotherapy, radiofrequency ablation, or ethanol injection, there should have been >6 weeks since that therapy with indicator lesion(s) outside the area of treatment or, if the only indicator lesion was within the treatment area, there must have been definite evidence of disease progression. Patients must have completed any prior major surgery (eg, laparotomy) ≥4 weeks prior to registration and minor surgery (eg, insertion of indwelling vascular device) ≥2 weeks prior to registration and recovered from all surgery-related complications. Eligibility included the following laboratory parameters: absolute neutrophil count ≥1500/μL, platelet count ≥75,000/mm3, total serum bilirubin level ≤ upper limit of normal range (ULN), serum albumin ≥2.5 g/dL, serum AST/ALT ≤2.5 × ULN, alkaline phosphatase ≤5 ULN, INR ≤1.2, and serum creatinine <2 mg/dL. Excluded from the study were patients who received prior treatment with either an anti-VEGF/VEGFR or anti-EGFR drug or who had another malignancy with the exceptions of adequately treated basal cell or squamous cell skin cancer, in situ cervical cancer, or adequately treated stage 1 or 2 cancer from which the patient was in complete remission. Patients were also excluded if they received prior radioembolization when liver only target lesions were present or if they had a history of a cerebrovascular accident or transient ischemic attack <6 months prior to registration, uncontrolled hypertension, unstable angina pectoris within 6 months, symptomatic congestive heart failure, myocardial infarction ≤6 months prior to registration, serious uncontrolled cardiac arrhythmias, or gastrointestinal bleeding that required procedural intervention (eg, variceal banding) within 3 months prior to registration. No active bleeding or pathologic conditions that carried a high risk of bleeding (eg, tumor involving major vessels, gastroesophageal ulceration, or esophageal varices) were allowed. The institutional human investigation review boards of all participating institutions approved the study, and all patients provided a signed informed consent.


Bevacizumab 10 mg/Kg was administered intravenously on days 1 and 15. Erlotinib 150 mg orally once daily was administered on a continuous basis. Treatment cycles were every 28 ± 3 days. In the event of missed doses of any of the drugs for any reason, the cycle days continued to be counted as if the drug(s) were administered. Only the dose of erlotinib was modified for toxicity to 100 mg or 50 mg per day. Patients in whom one agent was held continued to receive the other agent. Treatment was continued until disease progression, undue toxicity, or consent withdrawal by the patient. Subjects were withdrawn from the study if they failed to recover to grade 0-1 or tolerable grade 2 (or within 1 grade of pretreatment toxicity) from a treatment-related toxicity within 14 days or they experienced agent-related adverse events requiring dose modification despite 2 previous dose reductions.

Disease Assessment

Objective response to therapy was assessed using the Response Evaluation Criteria in Solid Tumors (RECIST).21 Measurable disease was defined as one or more lesions whose longest diameter could be accurately measured as >2.0 cm. Computerized tomography (CT) or magnetic resonance imaging (MRI) was used to measure response to therapy and repeated every 8 weeks, unless clinically indicated. Total disappearance of target lesions constituted a complete response (CR), whereas a minimum of a 30% decrease in the sum of the longest diameter (LD) of the lesions was classified as a partial response (PR). New lesions or a 20% increase in the sum of the LDs of the target lesions was considered progressive disease (PD). Otherwise, patients were classified as having stable disease (SD). Patients were reevaluated for disease status after a minimum of 4 weeks of achieving a CR or PR to confirm the assessment. Similarly, an SD was reassessed at a minimum interval of 8 weeks. Patients with global deterioration of health status requiring discontinuation of treatment without objective evidence of disease progression at that time, and not related to study treatment or other medical conditions, were considered to have PD due to symptomatic deterioration.

Time to progression (TTP) was calculated from study entry to disease progression. Time to death (ie, survival) was calculated from the date of study entry to death or last contact. Time to treatment failure was defined to be the time from the date of registration to the date at which the patient was removed from treatment.

Toxicity Assessment

All patients who underwent protocol-based therapy were considered eligible for toxicity assessment. Evaluations were based on at least once per cycle clinical assessments and graded according to the National Cancer Institute Common Toxicity Criteria (NCI CTC) version 3.0.

Statistical Methods

The primary end point of the study was objective response. The point estimate of the success rate was calculated as the number of successes (CR + PR) divided by the number of evaluable patients, with a confidence interval (CI) calculated by the method of Duffy and Santner.22 All patients meeting the eligibility criteria who signed a consent form, began treatment, and had at least 1 post-baseline disease assessment were evaluable for response. The largest success proportion considered ineffective was 10% (ie, Ha: P ≤ 10%), whereas the smallest success proportion considered effective was 25% (ie, Ha: P ≥ 25%). The decision criteria for the efficacy of this treatment strategy in this patient population were based on a Flemming22 phase 2 study design with a planned interim analysis. If 9 successes were observed in 50 evaluable patients, this dual-agent strategy warranted further studies in larger groups of patients. The planned interim analysis occurred at the time the 21st patient became evaluable. Here, at most 2 observed successes would close the trial due to insufficient evidence of activity. Six successes were considered sufficient evidence of activity to recommend further testing in larger groups of patients. This study design yielded 87% power at .06 level of significance to detect a success rate of at least 25% if the true success rate was at most 10%. Secondary end points included adverse events associated with the study therapy, time to disease progression, duration of response, and time to treatment failure. The distributions of survival time, time to progression, and time to treatment failure were estimated using the method of Kaplan-Meier. Two-sided P-values ≥.05 were considered statistically significant. All analyses were performed using SAS Version 9.0 (SAS Institute, Cary, NC).


Patient Characteristics

Between October 2006 and March 2008, 27 patients with advanced HCC were accrued onto the study by the 6 membership institutions. Baseline patient characteristics are presented in Table 1. A majority of patients were male (74%) and white (78%) with a median age of 60 years (range, 36-79 years). Twenty (74%) patients had Child's Pugh Classification A (vs B). Sixteen (59%) patients had ECOG PS of 0. Twelve (44%) patients received prior surgery related to their tumor, 7 (26%) received liver directed therapy (transarterial chemoembolization [TACE] in 6 patients, and TACE and cryoablation in 1 patient), and 1 (4%) received systemic therapy (CCI-779). Fifty-six percent of patients had a history of chronic liver disease, and the majority had HCV infection. Seventy percent of patients had distant metastases including liver (13 patients), lung (12 patients), lymph nodes (8 patients), subcutaneous tissue (1 patient), bone (1 patient), abdominal mass (1 patient), and other (3 patients).

Table 1. Characteristics of 27 Patients With Advanced Hepatocellular Cancer Treated With Erlotinib Plus Bevacizumab
  • ECOG, Eastern Cooperative Oncology Group; NOS, not otherwise specified; RUQ, right upper quadrant; TACE, transarterial chemoembolization; UWCCC, University of Wisconsin Carbone Cancer Center.

  • a

    Patients are reflected once in each classification.

Median age, y (range)60 (36-79)
 Male20 (74%)
 Female7 (26%)
 White21 (78%)
 Black or African American4 (15%)
 Asian1 (4%)
 Unknown1 (4%)
ECOG performance status 
 016 (59%)
 111 (41%)
Child's Pugh Class 
 A20 (74%)
 B7 (26%)
Previous surgery related to tumor12 (44%)
Previous liver directed therapy7 (26%)
 TACE and cryotherapy1
Previous systemic therapy1 (4%)
Presence of background liver disease15 (56%)
 Hepatitis B1 (7%)
 Hepatitis C8 (53%)
 Alcoholic cirrhosis2 (13%)
 Alcoholic cirrhosis + hepatitis C1 (7%)
 Alcoholic cirrhotic + hepatitis B and C1 (7%)
 Cirrhosis NOS2 (13%)
Registration location 
 UWCCC11 (41%)
 Mayo Rochester4 (15%)
 Mayo Arizona4 (15%)
 Wayne State4 (15%)
 Washington3 (11%)
 Abbott-North1 (4%)
Status of primary tumor 
 Resected with no residual6 (22%)
 Unresected17 (63%)
 Recurrent4 (15%)
Extrahepatic metastatic sites, median (range)2 (0-4)
Sites of extrahepatic metastasesa 
 Lung3 (12%)
 Peritoneal-RUQ1 (4%)
 Lung and liver6 (23%)
 Nodal and lung1 (4%)
 Liver and gastrohepatic and celiac  lymphadenopathy1 (4%)
 Nodal and abdominal1 (4%)
 Nodal and liver2 (8%)
 Nodal and lung and liver2 (8%)
 Nodal and liver and retroperitoneal adenopathy1 (4%)
 Nodal and subcutaneous and bone and liver1 (4%)

Primary Efficacy End Point

The primary end point of this study was objective response using the RECIST (Table 2). Twenty-three (85%) patients were evaluable for objective response assessment. Four patients were inevaluable because they were off the study prior to first radiologic evaluation for objective response. One patient, a 68-year-old white female, demonstrated a PR with reduction of serum α-fetoprotein levels after 3 cycles, which was confirmed at the next disease assessment; thereafter, treatment was concluded during the 10th cycle due to grade 3 diarrhea that was poorly controlled. The estimated success rate based on this number of confirmed responses was 5% (95% CI, 0%-23%). Eleven (48%; 95% CI, 27%-69%) and 11 (48%; 95% CI, 27%-69%) patients had a best response of either stable disease or disease progression, respectively.

Table 2. Objective Tumor Response by RECIST Criteria in 27 Patients With Advanced Hepatocellular Cancer Treated With Erlotinib Plus Bevacizumab
  • CI, confidence interval; CR, complete response; PR, partial response; RECIST, Response Evaluation Criteria in Solid Tumors.

  • a

    Patients having completed 8-week restaging or died due to progression within 8 weeks of going on study.

  • b

    Kaplan-Meier method.

Best objective response 
 Evaluablea23 (85%)
 Not evaluable4
 PR1 (4%)
 Stable disease11 (48%)
 Progressive disease11 (48%)
 Confirmed CR/PR1
Time to progression 
 Medianb (95% CI)3.0 mo (1.8-7.1)
 % progression-free 
  4 mo45% (29%-71%)
  6 mo35% (20%-62%)
  9 mo15% (5%-42%)
 Medianb (95% CI)9.5 mo (7.1-17.1)
 % alive 
  4 mo85% (72%-100%)
  6 mo73% (58%-92%)
  9 mo58% (42%-80%)
  12 mo42% (27%-66%)
  15 mo36% (24%-63%)
  18 mo31% (17%-55%)
Time to treatment failure 
 Medianb (95% CI)2.0 mo (1.8-4.6)
 % on study 
  4 mo33% (20%-57%)
  6 mo22% (11%-45%)
  9 mo11% (4%-32%)

Secondary Efficacy End Points

Twenty-one (78%) patients experienced disease progression. Sites of progression included liver (9 patients), lung (7 patients), lymph nodes (4 patients), abdominal mass (1 patient), bone (1 patient), brain (1 patient), and other (2 patients). Median time to disease progression was 3.0 months (95% CI, 1.8-7.1 months). Forty-five percent, 35%, and 15% of patients were progression free at 4, 6, and 9 months, respectively. Eighty-five percent, 73%, and 58% of patients were alive after 4, 6, and 9 months, respectively. Median survival was 9.5 months (95%CI, 7.1-17.1) with a median follow-up of 23.4 months (range, 1.8-29.1 months). The distributions of time to progression and overall survival are shown in Figure 1. Five of the patients died without documentation of progressive disease, and the other patient remained alive without progression.

Figure 1.

Progression-free and overall survival in patients with advanced hepatocellular cancer treated with erlotinib and bevacizumab are shown.

Treatment Administration

Twenty-seven patients completed a total of 108 cycles of treatment (median, 2 cycles; range, 1-12 cycles). Patients received a median of 72% (range, 21%-108%) of the planned dose of erlotinib, with 11 patients having 14 cycles of dose reductions. Bevacizumab was held in 4 patients for 6 cycles and 10 patients for 13 cycles on days 1 and 15, respectively. Erlotinib was held during 20 cycles in 16 patients. The estimated time to treatment failure was 2.0 months (95% CI, 1.8-4.6 months). Events precluding further study treatment included disease progression (59%), adverse reactions (15%), refusal (11%), and miscellaneous reasons (15%, eg, death during treatment [1 patient], alternative treatment [1 patient], or physician discretion [2 patients]).


All patients were evaluable for adverse event analysis (Table 3). The maximum grade of treatment-related adverse events was 3 and experienced by 15 (58%) of patients. The most common toxicities included (grade 1/2/3): rash (6/8/6 patients), hypertension (3/4/1 patients), fatigue (6/4/2 patients), and diarrhea (13/2/4 patients). A 79-year-old white female died of acute respiratory failure due to pneumonia that was considered unrelated to the study drugs, occurring day 81 during the treatment cycle, after having received full bevacizumab and 75% of planned erlotinib.

Table 3. Adverse Event Profile in 27 Patients With Advanced Hepatocellular Cancer Treated With Erlotinib Plus Bevacizumab
  1. ALT, alanine transaminase; AST, aspartate aminotransferase.

  2. Adverse event grade according to National Cancer Institute Common Terminology Criteria for Adverse Events (NCI CTCAE) version 3. Events considered at least possibly related to study treatment.

Skin rash686


The majority of patients with HCC present with incurable disease with a median survival of <1 year because of lack of effective systemic therapies. Recognition of the hypervascular nature of HCC stimulated the search for antiangiogenesis strategies against this disease. The only drug that has proven to prolong survival in a prospective randomized phase 3 setting (SHARP and Asia-Pacific trials) is sorafenib, which targets VEGFR and Raf kinase.24, 25 The efficacy of sorafenib underscored the potential benefits of targeting the VEGF/VEGFR pathway in HCC as a platform for further development of targeted therapies. Nevertheless, the modest benefit that was demonstrated indicated the need to develop further therapies targeting other signaling pathways known to be activated in this disease.

Results of this study demonstrated lack of an apparent clinically useful antitumor activity for the combination of erlotinib and bevacizumab as measured by RECIST and time to disease progression. The response rate of 4% and the time to progression of 3.0 months were within the range of what was seen with erlotinib alone17 and inferior to that of bevacizumab.19 One challenge to the interpretation of the results of this study is the choice of the primary end point that was used in the design of this trial. It is well recognized that objective response is difficult to determine with certainty in HCC and hence may be an inadequate surrogate for survival. The lack of a relation between improvement in tumor shrinkage and prolongation of survival was confirmed by the randomized phase 3 trails of sorafenib. However, the progression-free survival was significantly improved with sorafenib when compared with placebo, suggesting that it may be a better surrogate for overall survival. In the current study, progression-free survival of 3.0 months was similar to the placebo arm of the SHARP trial and was significantly lower than that reported for bevacizumab single-agent experience.19

The results of this study, when put in the context of reported outcomes of other studies, highlight the issues of patient selection in a disease with marked heterogeneity in etiologic factors. The marked heterogeneity in eligibility criteria, especially with respect to tumor burden and background liver disease, coupled by the lack of standard clinical practices with respect to the use of systemic agents in relation to liver directed therapies challenges the ability to extrapolate results from one study to another. In this respect, results of this study must be put in the context of a study recently reported by Thomas et al27 from MD Anderson Cancer Center using a similar dose and schedule of the combination of erlotinib and bevacizumab and also with the outcome of the SHARP and Asia-Pacific trials discussed previously. The reported objective response rate of 25%, median progression-free survival of 9 months, and median survival time of 15.7 months in the MD Anderson Cancer Center study were superior to what was seen in the current study. A 110-patient phase 2 randomized trial (NCT00881751) is now ongoing, testing sorafenib versus bevacizumab and erlotinib. Unlike the MD Anderson Cancer Center single institution experience, the current study has the advantage of being a multicenter study with a population of patients more similar to what is seen at cancer centers throughout the US. The differences in outcome may also be explained by higher proportion of patients with Child's A disease (87.5% vs 74%) and extrahepatic disease (69% vs 78%) in the MD Anderson Cancer Center study. Another major difference between the 2 studies is the higher percentage of patients with HCV infection compared with the MD Anderson Cancer Center population (37% vs 25%). There is some evidence to indicate activation of the Raf/MEK/Erk pathway downstream of EGFR by HCV16 that may reduce the benefit of erlotinib in HCV-infected patients. It is noteworthy that an unplanned analysis of the SHARP data suggested an improvement in survival in the HCV subgroup when compared with the overall study population by using sorafenib that inhibits the Raf kinase.

In conclusion, this multicenter phase 2 study did not demonstrate a signal of activity for the combination of bevacizumab and erlotinib that may be considered clinically worthwhile in patients with advanced HCC. Further work must be undertaken to determine the role of combined anti-EGFR and anti-VEGF/VEGFR treatment strategies in this disease in patients better selected by molecular markers.


Supported by the Phase 2 Consortium (P2C) through its contract with the National Cancer Institute (N01 CM62205). Erlotinib was provided to the NCI by OSI Pharmaceuticals, Inc. (Farmingdale, NY).


The authors made no disclosures.