Phase II, Open-Label Study of Brivanib as First-Line Therapy in Patients with Advanced Hepatocellular Carcinoma

Hepatocellular carcinoma (HCC) is the fifth most common malignancy worldwide (1) and the third largest cause of cancer-related deaths (2). HCC is unique in that it represents 2 diseases, primary liver dysfunction/cirrhosis and malignancy. These 2 comorbid conditions create unique clinical challenges, particularly because agents with efficacy against malignant growth may worsen underlying cirrhosis, making it difficult to interpret composite end points (3). As HCC is a highly vascularized tumor, therapeutic concepts targeting key molecular pathways involved in tumor angiogenesis are of particular interest in targeting malignancy. Sorafenib, an inhibitor of VEGFR-2, c-Kit, and raf, has demonstrated activity and tolerability in advanced HCC (4, 5).

Fibroblast growth factor-2 (FGF-2) is a potent angiogenic factor in HCC (6–9). Upregulation of alternate proangiogenic signals, such as the FGF signaling pathway, may play a role in evasive resistance to VEGF-targeted antiangiogenic therapy (10). Moreover, preclinical data suggest that although VEGF inhibitors reduce primary tumor growth, they also promote tumor invasiveness and metastasis in 2 mouse models with melanoma or pancreatic cancer (11–13). Brivanib, the first oral selective dual inhibitor of FGF and VEGF signaling, is formulated as an orally administered l-alanine ester prodrug, brivanib alaninate (14). Brivanib has strong antiangiogenic effects, as well as potent direct effects, on tumor cells across a range of tumor types, including liver and colon(15–17). In addition, brivanib has demonstrated antitumor activity in xenograft HCC models expressing FGF receptors (17). Thus, targeting both the VEGF and FGF signaling pathways may provide clinical benefits to HCC patients. Brivanib has a half-life of 12 hours and is administered once daily at a dose of 800 mg, with no reported differences in pharmacokinetics between Asian versus nonAsian races and patients with mild hepatic impairment versus patients with no hepatic impairment (18, 19).

Here we present the results from a study of brivanib as first-line therapy in patients with unresectable, advanced HCC.

This was originally a randomized study of brivanib versus doxorubicin in newly diagnosed HCC patients. At the time the protocol was initiated, no agent was proven to extend survival in HCC and doxorubicin was used as a comparator arm based on historical data reporting response rates of 10% to 20%.(20) Based on positive sorafenib phase III results, the protocol was amended to discontinue the doxorubicin arm. Accordingly, this was a multicenter, open-label, phase II, single-agent study of once-daily brivanib alaninate 800 mg in patients with unresectable, locally advanced, or metastatic HCC who had received no prior systemic therapy (cohort A) or 1 prior regimen of angiogenesis inhibitor therapy (cohort B). This study was conducted in accordance with the Declaration of Helsinki and International Conference on Harmonization-Good Clinical Practice. The study protocol and any amendments were approved by the independent ethics committee of each center and by the authorities in each relevant country. Informed consent was obtained from each patient. Here we report results from cohort A.

Patients with measurable, unresectable, locally advanced, or metastatic HCC, those with biopsy-proven HCC, or those who had clinical evidence of HCC [α-fetoprotein (AFP) levels ≥ 400 mg/L and standard imaging criteria; ref.21] and were serology positive for hepatitis B or C, with a Cancer of the Liver Italian Program (CLIP) score 3 or less and Eastern Cooperative Oncology Group (ECOG) performance status of 0 to 2, and adequate hepatic function and renal function, were eligible for inclusion in the study (see online Supplementary section A for full criteria).

The primary objective was to estimate the 6-month progression-free survival rate. Six-month progression-free survival rate was defined as the proportion of treated patients who had not progressed or expired prior to 6 months from the first dose. Secondary objectives were median progression-free survival, tumor response rate, time to response, duration of response, disease control rate, time to progression, overall survival (see online Supplementary section B for full definition of efficacy endpoints), and safety and tolerability.

Brivanib was administered orally on a continuous daily schedule of 800 mg. Patients continued treatment until first evidence of disease progression on imaging by World Health Organization (mWHO) criteria or earlier if unacceptable toxicity was observed. Doses were reduced or delayed according to protocol-specified criteria or, for lower-grade toxicities, if the investigator deemed dose reduction or delay was required in the interest of the patients' safety (online Supplementary section C).

Tumor response was assessed every 6 weeks using the tumor response criteria via computed tomography/magnetic resonance imaging. Tumor assessment was performed by an Independent Radiological Review Committee (IRRC; primary analysis) and by study investigators. In addition, we also performed a retrospective exploratory analysis to reassess tumor response and progression using the new proposed modified Response Evaluation Criteria in Solid Tumors (mRECIST) for HCC criteria. Indeed, this study was designed in April 2006 before the publication of the American Association for the Study of Liver Disease-Journal of the National Cancer Institute (AASLD-JNCI) guidelines for HCC that recommend assessment of tumor responses be performed using the new proposed mRECIST for HCC (3, 22). The original RECIST and mWHO criteria did not account for new targeted therapies that are cytostatic and aim to achieve necrosis of the tumor, which may not be paralleled by tumor shrinkage. Thus, the new proposed modified guidelines for HCC were introduced to measure reductions in viable tumor volume recognized as enhancing areas using dynamic imaging techniques. All atypical intrahepatic and all extrahepatic lesions were assessed, using conventional RECIST criteria as outlined in the recent proposed mRECIST for HCC publication (22).

Safety was assessed via adverse events (AE), vital signs, electrocardiograms, echocardiograms, physical examinations, and clinical laboratory tests (see online Supplementary section D).

A total of 50 treated subjects were required in the first-line daily cohort. This was the number of subjects needed so that the lower bound of the 95% CI of the 6-month PFS rate would exclude 35% given the observed PFS rate at 6 month is 50% (see online Supplementary section E for full details).

All analyses described in the manuscript included all subjects with no prior systemic therapy who were treated with 800 mg brivanib alaninate every day (QD).

Statistical analysis (see online Supplementary section E for full details).

The 6-month PFS rate for the first-line QD cohort, as per IRRC using the mWHO criteria, was the primary endpoint of this study. The 6-month PFS rate was defined as the number of subjects who had not progressed or died prior to 6 months from the date of their first dose, divided by the number of subjects in the first-line daily cohort. An exact binomial 2-sided 95% CI for the PFS rate was computed. The PFS rate at 6 months was also estimated using the Kaplan–Meier method and 95% CI was calculated using the Greenwood formula. For details see online Supplementary section E.

A total of 14 patients were initially treated with doxorubicin; 6 of these patients at progression crossed over to receive brivanib treatment. These data are not included in this manuscript. Of 58 enrolled patients, 3 withdrew consent before the start of the study and 55 were enrolled from March 7, 2007, to May 4, 2009 (Table 1). As of Dec 2010, 2 survivors remain on treatment. The majority of patients were Asian (64%); 91% of patients were Child–Pugh grade A, and 89% were Barcelona Clinic Liver Cancer (BCLC) stage C. Seventy-three percent of patients were heavily pretreated with local or locoregional therapies. Median and mean duration of treatment (95% CI), as estimated using Kaplan–Meier analysis, were 1.74 months (1.41, 2.76) and 3.89 months (2.35, 5.44), respectively. One subject was still on treatment at time of analysis.

Using mWHO criteria, the 6-month progression-free survival rate calculated using the exact binomial method was 18.2% (95% CI, 9.1%-30.9%) as determined by IRRC assessment (primary end point; Table 2). The 6-month progression-free survival rate calculated using Kaplan–Meier estimates was 21.1% (95% CI, 9.6%–32.5%) as determined by IRRC assessment and 15.9% (95% CI, 5.8%-26.0%) as determined by investigator assessment.

Median progression-free survival was 2.7 (95% CI, 1.4–3.0) months by IRRC assessment (Fig. 3A).

According to IRRC assessment with mWHO criteria, 1 patient had a complete response, 3 patients achieved a partial response, and 24 had stable disease. Two of these patients with stable disease had unconfirmed partial response (no repeat scans showing partial response). The disease control rate (complete response + partial response + stable disease ≥42 days) was 51% [(1 + 3 + 24/55, respectively)]. Twenty patients had progressive disease and tumor responses were not assessable in 7.patients (Fig. 1; Table 2). As assessed by the investigator, 4 patients achieved confirmed partial response and 24 had stable disease (of these 1 had an unconfirmed PR). The disease control rate (partial response + stable disease) was 51% [(4 + 24)/55 treated, respectively] by investigator. Fifteen patients had progressive disease. Tumor response was not assessable in 12 patients by investigator. Five patients had a baseline and at least 1 follow-up scan but no scan while on treatment. As a result, IRRC gave a best response for those 5 patients but were not assessable by the investigator.

Some patients achieved rapid objective responses whereas others showed slow improvement over time; the time to PR/unconfirmed PR ranged from 1.14 to 12.45 months (5 to 54.1 weeks) and the time to the CR was 20.67 months (89.9 weeks) per IRRC according to mWHO criteria.

IRRC time to progression was 2.8 months (95% CI, 1.4–3.5) and investigator time to progression was 2.7 months (95% CI, 1.5–2.8).

Using the proposed mRECIST criteria for HCC, the 6-month progression-free survival rate was 35.6% (95% CI: 21.0–49.4), median progression-free survival was 4.7 (95%CI: 2.8–5.7) months, and median time to progression was 5.4 months (95% CI, 2.8- –; Table 2).

Upon mRECIST criteria for HCC, 26 of 55 patients had improved responses compared with mWHO criteria per IRRC. Among the 3 patients classified as partial response by mWHO, 1 was reclassified with complete response. Among the 24 patients classified as stable disease by mWHO, 3 patients were reclassified with complete response, 6 were reclassified with partial response (Table 2). In the original study protocol, treatment was terminated in patients who had an assessment of progressive disease. However,16 of 20 patients who were assessed as having a progressive disease, using mWHO criteria actually had some evidence of tumor response by mRECIST and may have benefited from continued treatment. Of these, 2 were considered partial response and 14 were considered stable disease (including unconfirmed partial response and unconfirmed complete response). Using mRECIST, 2 of 55 patients had downgraded responses compared with mWHO (1 stable disease was reclassified as progressive disease; 1 partial response became stable disease). The disease control rate (complete response, partial response, and stable disease) using mRECIST was 78.2% (95% CI, 65.0%-88.2%; 43/55 treated). Patient examples are shown in Figure 2.

Median overall survival was 10 (95% CI, 6.8–15.2) months (Fig. 3B; Table 2). Overall survival was longer in patients with disease control versus those without disease control—15.2 months (95% CI, 10.0–19.7) versus 5.1 months [95% CI, 3.6–8.9; hazard ratio = 0.56 (95% CI, 0.29–1.08)].

In a post hoc analysis of overall survival, 45 patients had BCLC stage C and Child–Pugh grade A cirrhosis. Median overall survival in this patient subgroup was 8.9 months (95% CI, 5.6–15.2).

Seventy-six percent of patients had elevated AFP levels [greater than the upper limit of normal (ULN)] at baseline, 42% had an AFP more than 400 ng/mL. Among patients with an elevated (>ULN) AFP level at baseline (n = 41), 56% had a more than 50% reduction in AFP levels from baseline (Fig. 3C; online Supplementary section Table A1).

Brivanib was generally well tolerated, with a manageable safety profile (Table 3). Eleven patients (21%) withdrew due to treatment-emergent AEs. Twenty-two patients (41%) had at least 1 dose reduction. Reasons for dose reductions were nonhematologic toxicities related to the study drug in 21 patients (95.5%).

Thirteen patients experienced drug-related serious AEs, which were grade 3/4 in 10 patients (Table 4). Twenty-six percent of patients had a diastolic blood pressure greater than 100-mm Hg and only 1.8% of patients had a diastolic blood pressure greater than 110-mm Hg during the treatment course. In addition, 25 patients (46%) had any-grade hypertension as a reported AE; grade 3/4 in 6 patients (11%). Grade 2 hypertension led to treatment discontinuation in 1 patient. Six patients (11%) experienced minor (grade 1/2, except for one grade 3 esophageal varices hemorrhage; none requiring blood transfusion) gastrointestinal bleeding events, including 2 minor anal bleedings (online Supplementary section Table A2). Of note, only 2 patients (4%) experienced hand-foot syndrome, which was grade 1/2 in 1 patient and grade 3 in 1 patient, which were considered drug related. Three patients (5.5%) had grade 3 encephalopathy, which was considered drug-related in 1 patient. Of the 35 deaths that occurred (1 within 30 days of starting the trial and 6 within 30 days of the last dose), none were considered treatment related (30 patients died due to disease progression, 5 died due to other reasons). Eight patients (14.5%) had grade 3/4 hyponatremia; however, some patients may have entered the study with up to grade 3 hyponatremia; see online supplementary section D).

No-one continued brivanib beyond progression; 25 of 55 (45.5%) subjects received subsequent therapy and 30 of 55 (54.5%) did not receive any subsequent therapy. No patients received maintenance therapy with brivanib. Eighteen of 55 (32.7%) subjects received at least 1 targeted therapy (antiangiogenic agents, monoclonal antibodies, tyrosine kinase inhibitors, or other targeted therapies) as subsequent therapy, including 16 patients who received sorafenib.

Results from this phase II study are preliminary evidence that brivanib as first-line therapy has antitumor activity and a manageable safety profile in advanced HCC in a predominantly Asian population with a high incidence of hepatitis B. This study did not reach the planned primary end point with a 6-month PFS of 18.2%, and median progression-free survival was 2.7 months. However, median overall survival was 10 months, which is an encouraging finding for this advanced-stage population that had a high incidence of extrahepatic disease and prior local treatment, together with brivanib's manageable safety profile.

The observed PFS and 6-month PFS was lower than expected. However, these results must be interpreted in the context of recent study results and changes to guidelines on end points for phase II studies in HCC. First, several different drugs are currently being developed in HCC and, as stated recently by Llovet and Bruix (23), the best way to assess their efficacy is to use randomized phase II trials. Of note, the current study was originally a randomized study of brivanib versus doxorubicin in patients with newly diagnosed HCC but based on the positive sorafenib first-line phase III results (5), the study protocol was amended to discontinue the doxorubicin arm. Second, although the original statistical plan for this study stated that a positive outcome for this trial would be if the 6-month progression-free survival exceeded 35%, this plan was amended, based on the phase II study of sorafenib in the United States (4). However, as the Asian sorafenib patients had a lower overall survival of 6.5 months and time to progression of 2.8 months (24), it may be appropriate to use a lower threshold for a positive phase II trial when patients are from Asia. Third, this study was designed before the publication of the recent AASLD-JNCI guidelines, which recommend that phase II trials of targeted agents should be powered to test a time-to-event surrogate end point such as time to progression with composite endpoints such as disease and progression-free survival tested only as secondary or tertiary endpoints (3); as a result, our study objectives were different than those now recommended. This study was designed for progression-free survival, but in review of patient scans it became clear that mWHO assessments may have overcalled progression (e.g., several patients developed ascites or small new ≤ 1cm lesions, which do not reliably represent progression) and did not reflect the antitumor effect. Notably, 16 patients who were originally assessed as PD via mWHO on their first scan were taken off study and then stopped brivanib when in reality they may have been having some treatment benefit, as shown by the proposed mRECIST criteria for HCC assessments. Using the proposed mRECIST for HCC, more responses were observed—including 5 CRs—substantially increasing the disease control rate. In addition, median time to progression almost doubled with mRECIST. The proposed mRECIST for HCC criteria therefore merit further assessment in randomized Phase 3 studies and validation as a surrogate for long-term outcomes.

Brivanib as a first-line systemic treatment for advanced HCC demonstrated a median overall survival of 10 months, which is similar to the overall survival rate in studies of sorafenib as first-line therapy (4, 5) despite the fact that more than two-thirds of patients were enrolled in Asia in the brivanib study. In contrast, the phase III study of sorafenib in Asian patients reported an overall survival of 6.5 months (24). Experts believe that the lower overall survival in the Asian versus nonAsian sorafenib study population was due to the degree of pretreatment, the severity of the underlying disease, and the tumor burden (24). Additional differences in study design between the current study and the sorafenib studies, notably the SHARP study, make comparisons of overall survival difficult. For example, brivanib treatment was stopped in the current study in the case of progression when mWHO criteria were applied, whereas in the SHARP study mWHO criteria were used and sorafenib treatment was stopped in the case of both radiologic progression and symptomatic progression. The activity of brivanib in HCC is further supported by recently reported results demonstrating efficacy as second-line therapy in the post-sorafenib setting (25).

Importantly, the recent AASLD-JNCI guidelines recommend the use of time to progression as an end point in phase II clinical trials of targeted therapies and the use of proposed mRECIST for HCC criteria for assessing tumor responses. In the current study, 1 complete response and 3 partial responses were observed by mWHO criteria; however, with the proposed mRECIST for HCC criteria, 5 complete responses and 9 partial responses were observed. Notably, the time to progression of 2.8 months in the current study is comparable with the time to progression of 2.8 months reported in the Asian sorafenib study (24).

Responses and major AFP reductions in the current study also demonstrate encouraging signs of biologic activity in HCC patients. Until recently, the role of monitoring AFP response in HCC had not been rigorously evaluated. However, 2 recent retrospective analyses of AFP response in HCC patients undergoing systemic chemotherapy (26) or locoregional therapy (27) reported that achievement of an AFP response is an independent prognostic factor in determining survival (26, 27). In our study, more than half of patients with an elevated AFP (>ULN) level at baseline had a greater than 50% reduction in AFP levels from baseline following brivanib treatment.

Grade 3/4 AEs included fatigue in 16.4% of patients and hypertension in 10.9% of patients. The incidence of all-grade hand-foot syndrome was 3.6% in the current study compared with 9% to 82% of HCC patients treated with sorafenib (4, 5, 10, 24, 28–32), which led to dose reductions in 5% of patients in the SHARP study and 11.4% of patients in the sorafenib Asian Study (5, 24). Despite brivanib's antiangiogenic properties, there were no serious AEs usually associated with antiangiogenic agents, and its use was not associated with an unusual incidence of gastrointestinal hemorrhage, particularly of variceal origin as seen with bevacizumab (33).

In conclusion, brivanib, a dual inhibitor of VEGF and FGF signaling, has demonstrated preliminary evidence of antitumor activity and a manageable tolerability profile in patients with unresectable and/or metastatic HCC. These findings warrant development of brivanib in patients with unresectable and/or metastatic HCC and suggest that brivanib may be an effective treatment option as first-line therapy for these patients. To fully investigate the clinical activity of brivanib in HCC, these findings are being tested further in several randomized phase III studies in advanced HCC patients in the BRISK program, including studies of brivanib versus sorafenib as first-line treatment, brivanib versus placebo after sorafenib therapy, and transarterial chemoembolization with or without brivanib as adjuvant therapy.

I. Walters, C. Baudelet, and R. Harris are employees of Bristol-Myers Squibb. R. Harris and I. Walters have ownership of Bristol-Myers Squibb stock. F. Ismail has received commercial research grant support from Merck, Serono, Boehringer Ingelheim, and Novartis, and honoraria from a speakers bureau from Novartis, Roche, and Orient Europharma, and has served as a consultant for AstraZeneca, Novartis, and Merck. M. Thomas has received other commercial research support from Genentech. M. Karwal has received other commercial research support from Bristol-Myers Squibb. J-L. Raoul has received honoraria from a speakers bureau from Bayer Schering Pharma and Bristol-Myers Squibb and served as a consultant for Bayer Schering Pharma, Bristol-Myers Squibb, Taiho, and Novartis. R.S. Finn has received honoraria from a speakers bureau from Genentech and served as a consultant for Bayer. J-W. Park has served as a consultant for Bayer Healthcare, BMS, Pfizer, and Bukwang.

Editorial support was provided by Mark English, of PAREXEL, and was funded by Bristol-Myers Squibb.

Supported by research funding from Bristol-Myers Squibb (Study No. CA182006). This trial is registered at www.clinicaltrials.gov as NCT00355238.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.