Eligible subjects at a single academic institution had HCC by virtue of either: i) the European Association for the Study of the Liver (EASL) diagnostic imaging criteria [19
]; or ii) biopsy. All subjects were clinical candidates for TACE with at least one lesion ≥ 3 cm and no lesion ≥ 15 cm, and no more than three lesions total. Tumor location had to permit embolization of all tumor nodules on initial TACE. Subjects awaiting OLT were eligible if their model for end-stage liver disease (MELD) priority score was < 28 points at entry. The MELD score cutoff was based on a median MELD score of 30 at the time of transplant at our institution during the period that the study was conducted. This allowed subjects to complete the three scheduled angiograms before being withdrawn from study.
Subjects were excluded if they had Child's class C liver dysfunction, Eastern Cooperative Oncology Group (ECOG) Performance Status (PS) >2, bilirubin >2.5 mg/dL, INR > 1.5, extrahepatic disease, or thrombosis of the main portal vein. Subjects were excluded if they demonstrated contraindications to bevacizumab. Initially, subjects were excluded if they had platelets < 100,000/μL. After the first three subjects enrolled in the TACE-BEV arm completed week 10 without significant bleeding (defined as bleeding requiring transfusion), additional subjects were excluded if they had platelets < 75,000/μL. After safety was confirmed using the 75,000/μL platelet cut-off, the protocol was amended in consultation with the Food and Drug Administration (FDA) to exclude subjects with platelets < 60,000/μL, in an effort to improve accrual. The protocol was approved by the Medical Institutional Review Board at the University of California, Los Angeles. Written informed consent was obtained and the study was conducted according to federal and institutional guidelines, observing the standards set by the Helsinki Declaration.
Study design and treatment procedures
The study design is depicted in Figure . A one to one randomization was performed with a computer-generated allocation, distributing 30 subjects to either TACE-O or TACE-BEV.
Hepatic angiograms were scheduled at day 8, week 10, and week 14. The day 8 procedure included TACE of all lesions. The week 10 and 14 angiograms were performed to assess tumor vascularity. Two observation points were chosen because it was unknown how long it would take to develop a treatment effect, and 10 and 14 weeks were specifically selected to approximate the 2 to 3 months separating the first and second scheduled TACE procedures in previous clinical trials (8,9). TACE was not performed during the week 10 angiogram, justified by the opportunity for TACE at week 14. At week 14, a second TACE was performed if: i) a previously embolized feeding vessel had recanalized; or ii) there was residual tumor blush in a treated area, and a feeding vessel was visualized; or iii) new lesions had developed that were amenable to TACE.
TACE was performed using doxorubicin 25 mg/m2 (emulsified with lipiodol, 10-12 mL), cisplatin 50 mg/m2, and mitomycin-C 5 mg/m2, with Embosphere® microspheres (Biosphere Medical, Rockland, MA). Chemotherapy and the embolic agent were administered through the hepatic artery to segmental and/or subsegmental branches feeding the tumor. Stump occlusion of segmental or subsegmental feeding branches was performed with microfibrillar collagen (Avitene, Davol, Inc., Cranston RI) as needed to achieve stasis. Following TACE, subjects were admitted to the hospital for at least 24 h, where they received hydration, pain control, and antiemetics as needed. Subjects also received metronidazole and ciprofloxacin for 7 days after TACE.
Subjects randomized to TACE-BEV received bevacizumab 10 mg/kg intravenously (IV) on day 1, one week prior to the first TACE. Post-TACE, bevacizumab administration resumed at 10 mg/kg every 2 weeks, as long as serum transaminases had returned to pre-TACE levels, or within normal range. Subjects receiving bevacizumab with no signs of progressive disease were allowed to continue bevacizumab therapy until: i) they experienced unacceptable toxicity; or ii) they developed tumor progression (as defined below); or iii) their MELD score increased to > 28 points; or iv) they requested discontinuation of drug.
Subjects randomized to TACE-O were eligible to cross-over to bevacizumab at week 16 if they had evidence of progressive disease. Subjects who crossed over were required to meet the original eligibility criteria, and the criteria for bevacizumab administration status post TACE, as outlined above.
Pretreatment and follow-up studies
All subjects underwent baseline procedures including a history and physical, 12-lead electrocardiogram, and tumor assessment with triple phase computed tomography scan (CT) and/or contrast enhanced magnetic resonance imaging scan (MRI) of the liver within 4 weeks of starting protocol treatment. Additional staging investigations were performed at the discretion of the investigator. Subjects with documented grade III varices, or a history of upper gastrointestinal bleeding, were required to undergo endoscopic evaluation prior to study treatment. Within 2 weeks of starting protocol treatment, a physical exam, complete blood count (CBC), comprehensive metabolic panel (CMP), international normalized ratio (INR), urinalysis, and AFP were performed.
The first 16 weeks constituted the core treatment period. CBCs and CMPs were obtained daily for 3 days post-TACE, 1 week post-TACE, 2 weeks post-TACE, and every 4 weeks thereafter. This CBC and CMP schedule was repeated after a second TACE. An INR was obtained on day 3 post-TACE, then 3 weeks post-TACE, and every 4 weeks thereafter. This INR schedule was repeated after a second TACE. AFP was obtained every 2 weeks for the first 16 weeks, then every 4 weeks thereafter. Urinalysis or urine dipstick to screen for proteinuria was performed every 2 weeks. Physical exams were performed within 2 weeks prior to all angiograms and within 2 weeks following all angiograms. Toxicities were graded according to the National Cancer Institute common toxicity criteria (CTC), version 3.0. Follow-up CT or MRI was performed at weeks 8 and 16.
Subjects who received bevacizumab after week 16 were evaluated at least monthly with a physical exam, CBC, CMP, INR, and AFP. Urinalysis or urine dipstick was performed every 2 weeks. These safety evaluations were also performed at 30 and 60 days after the last dose of bevacizumab. CT or MRI for disease assessment was performed every 8 weeks while on bevacizumab.
Subjects who completed the 16 week core phase plus or minus the bevacizumab continuation phase were followed per institutional practice. All subjects were assessed for survival every 3 months.
Angiograms were assessed for changes in vascularity using the following parameters: 1) neovessel formation; 2) recanalization; 3) development of collaterals; and 4) number of vessels. All changes in vascularity were assessed with the catheter in the common hepatic artery, to allow standardized visualization. Neovessels, defined as fine vessels with disordered arborization patterns, were scored at weeks 10 and week 14 as follows: 0, no evidence of neovessel formation; 1+, some evidence of neovessel formation; 2+, obvious vessel formation (small vessels); and 3+, obvious vessel formation (small and large vessels). Recanalization was defined as the restoration of flow in a previously occluded vessel. Collateral vessels were characterized as small well formed vessels communicating between non-embolized and embolized vessels. The number of vessels was determined retrospectively in an unblinded fashion at the end of the study for all subjects who completed 2 or more angiograms by the trial interventional radiologist. This angiographic vessel count was performed using a single field measuring 2.43 cm in diameter (the size of an American quarter) at the site with the highest tumor vessel density. Although the arteriograms were obtained in several projections, all images were graded in anteroposterior projection. Accurate comparisons between studies and reproduction of magnification factors were achieved by performing the series of arteriograms in each individual subject with the same field size, object film distance, external fiducial marker, and contrast injection parameters as on their first TACE arteriogram. Vessel counts at 10 and 14 weeks were expressed as percent change from baseline. Vessel counts were compared between groups using the Wilcoxon rank sum test.
Cross-sectional imaging was considered along with angiography when assessing disease progression. Progressive disease was defined as any one of the following: an increase in the sum of the bidimensional products of all known disease by at least 25% by cross-sectional imaging, an increase in enhancement of a previously treated lesion by cross-sectional imaging, the appearance of a new lesion, or evidence of neovascularization by angiography at week 14. The angiographic demonstration of recanalization and/or the development of collaterals without neovascularization were not considered progressive disease. An objective response rate by cross-sectional imaging was not determined because most subjects on this investigator-initiated study were evaluated by CT, and lipiodol obscured tumor measurements by CT.
Progression free survival (PFS) at 16 weeks (end of the core phase) and overall survival (OS) were estimated by the Kaplan-Meier method, using JMP 6 statistical software (SAS institute, Cary, NC). The PFS times were censored based on the minimum time of four possible events: 1) last disease assessment; 2) orthotopic liver transplant; 3) second TACE if performed in the absence of progressive disease (e.g. embolization of a recanalized vessel without neovessel formation); 4) administration of a test dose of chemoembolization material, per institutional practice. OS was calculated based on intention to treat, and cross-over subjects were kept in the observation arm for the purpose of calculating OS. PFS and OS were compared between groups using log-rank tests.
Trough and peak blood samples were drawn from subjects randomized to the TACE-BEV arm for infusions one through five, and seven. Additional samples were drawn immediately prior to chemoembolization, and 72 h after chemoembolization. Bevacizumab serum concentrations were determined by an enzyme-linked immunosorbent assay (ELISA) with a minimum quantifiable concentration (MQC) in neat serum of 0.078 μg/mL [20
]. Bevacizumab concentration data was characterized by descriptive statistics at each time-point, with data described as means and 95% confidence intervals. To assess the difference of bevacizumab PK between the HCC subjects on this study and other oncology subjects, predicted concentrations for 500 patients were simulated using a population PK model of bevacizumab established on data from eight Phase I/II/III clinical trials (on file at Genentech). In this population PK model, significant PK covariates include weight, gender, and serum albumin on clearance (CL) and volume of distribution in the central compartment (V1); total protein on clearance; and baseline tumor burden on V1. Mean concentrations and associated 95% confidence intervals were calculated from these simulations and compared to the study subjects. These simulations were performed in NONMEM version VI beta, level 1.0 (LLC, Globomax, Ellicott City, MD).
Serum VEGF levels
Blood samples were obtained at the following time points: immediately prior to TACE; 1, 24, 48, and 72 hours after TACE; and 7, 15, and 21 days after TACE. VEGF-A was measured using an ELISA kit specific for human VEGF (R&D Systems, Minneapolis, MN, USA). Data was analyzed as the average for each timepoint and presented as fold-change from baseline. This was done for all subjects in the observation and bevacizumab arms that had at least one sample available for analysis. A one-sided t-test was performed for each timepoint.