A Phase Ib Study of AMG 102 in Combination with Bevacizumab or Motesanib in Patients with Advanced Solid Tumors

Interactions between the c-Met receptor tyrosine kinase and its ligand hepatocyte growth factor/scatter factor (HGF/SF) result in signaling events that promote cellular proliferation, survival, migration, and morphogenesis (1–4). The expression of c-Met and HGF/SF has been shown to predict poor prognosis in several cancer types (5), and dysregulated c-Met activation has been implicated as a contributing factor in multiple human malignancies (5–8). For these reasons, several strategies targeting HGF/SF and c-Met are being explored in clinical trials (6). These include antagonists of HGF/SF and c-Met, monoclonal antibodies directed against HGF/SF and c-Met, and small-molecule tyrosine kinase inhibitors of c-Met (6, 9). In addition to these approaches, the HGF/SF antagonist NK4 and decoy Met have been shown to inhibit tumor growth and angiogenesis in human xenograft models (10, 11).

AMG 102 is a fully human immunoglobulin G2 monoclonal antibody that has been shown in preclinical studies to induce regression of established tumor xenografts, inhibit in vitro tumor cell proliferation, and promote tumor cell death (12). In a phase I, first-in-human study (13), AMG 102 had acceptable toxicity up to the maximum tested i.v. dose of 20 mg/kg every 2 weeks in patients with advanced solid tumors. In addition, prolonged stable disease was seen in several patients.

Preclinical studies have suggested that there are interactions between the HGF/SF and vascular endothelial growth factor (VEGF) signaling pathways that result in enhanced tumor growth (14, 15). Therefore, simultaneous targeting of the c-Met and VEGF signaling pathways may offer benefits that exceed the inhibition of either target alone. A number of studies have shown that inhibition of angiogenesis is an effective therapy for a variety of solid tumors, either alone or in combination with other anticancer treatments. Bevacizumab combined with chemotherapy has proven effective in the treatment of a variety of different tumor types (16–19). Motesanib is a highly selective oral inhibitor of VEGF receptors (VEGFR) 1, 2, and 3; platelet-derived growth factor receptor; and Kit (20). In phase I and II studies, motesanib has shown acceptable toxicity and displayed antitumor activity following once-daily oral administration (21–23).

We hypothesized that the combination of AMG 102 with antiangiogenic agents might have additive or synergistic antitumor effects. The primary objective of the current study was to determine the safety and tolerability of the combination of AMG 102 and either bevacizumab or motesanib in the treatment of refractory advanced solid tumors.

Patients age ≥18 years with pathologically documented advanced solid tumors refractory to standard treatment or for which no curative therapy was available, an Eastern Cooperative Oncology Group performance status of ≤2, and adequate hematologic, renal, and hepatic function were eligible for the study.

Exclusion criteria included history of lymphoma or leukemia; squamous cell tumors of the lung or head and neck; ovarian cancers; large central hilar or mediastinal lesions (≥3 cm); symptomatic or untreated central nervous system metastases requiring concurrent treatment; unresolved toxicities from prior anticancer therapy; myocardial infarction within the previous 12 mo; symptomatic congestive heart failure (class >II, New York Heart Association); uncontrolled hypertension; unstable angina; cardiac arrhythmia; history of thrombotic events (including arterial and venous thrombosis and pulmonary embolism) within 1 year of study; history of active upper gastrointestinal hemorrhage, peptic ulcer disease, bleeding diathesis, gastrointestinal disease or surgery of the gastrointestinal tract, or iron deficiency anemia; history of pulmonary hemorrhage or gross hemoptysis within 6 mo before enrollment; positive test for HIV, hepatitis C virus antibodies, or hepatitis B surface antigen; or a positive test for hepatitis B core antibody in the presence of a negative test for hepatitis B surface antibody. Patients were also excluded if they had prior exposure to motesanib or AMG 102, treatment with anticancer therapy or hormonal antitumor therapy within 30 days before study day 1, or radiation treatment within 14 days before enrollment. For cohorts with motesanib, patients were excluded if they had concurrent or prior treatment with rifampin, phenobarbital (within 14 days before enrollment), or strong cytochrome P450 3A inhibitors. Patients were not excluded for previous treatment with bevacizumab. Institutional review board approval was obtained for all study procedures, and all patients provided informed consent before enrollment in the study.

This phase Ib study used an open-label, nonrandomized, dose-escalation design and was conducted at three centers. The study was designed to investigate the safety, pharmacokinetic, and pharmacodynamic profile of the combination of AMG 102 with either bevacizumab or motesanib. End points included the incidence of adverse events, the pharmacokinetic profile of AMG 102 in the presence of motesanib or bevacizumab, tumor response, and incidence of anti-AMG 102 antibodies. An exploratory pharmacodynamic analysis was done, which included assessment of circulating total HGF/SF and soluble c-Met.

Six dose cohorts of three to six patients each were initially planned, including two cohorts for the evaluation of 3 mg/kg or 10 mg/kg AMG 102 in combination with bevacizumab and four cohorts for the evaluation of AMG 102 in combination with motesanib. Four cohorts were enrolled in this study. Patients received AMG 102 in 2-week cycles along with bevacizumab in 2-weeks cycles or motesanib daily. Three patients were enrolled in cohort 1 and received 3 mg/kg AMG 102 (0.5- to 1.0-h i.v. infusion) every 2 weeks followed 30 minutes later by 10 mg/kg bevacizumab (0.5- to 1.5-h i.v. infusion) every 2 weeks. After completion of enrollment in cohort 1, three patients were enrolled in cohort 2 and received 10 mg/kg AMG 102 plus 10 mg/kg bevacizumab every 2 weeks. Subsequently, three patients enrolled in cohort 3 and received 20 mg/kg AMG 102 plus 10 mg/kg bevacizumab every 2 weeks. Cohort 3 was not originally planned but was added to investigate the maximum AMG 102 dose tested (20 mg/kg) in the first-in-human study. Although no dose-limiting toxicities (DLT) were observed in the initial three patients enrolled into cohort 3, the cohort was expanded with three more patients (for a total of six) to collect additional safety and pharmacokinetic data at this highest-tested dose level of AMG 102. Only two of the three planned patients were enrolled into cohort 4 and received 3 mg/kg AMG 102 every 2 weeks followed by 75 mg motesanib orally once daily. Three additional planned cohorts of AMG 102 in combination with motesanib were not enrolled as a result of concerns of an increased risk of cholecystitis and gallbladder enlargement observed in other studies with motesanib.

A DLT was defined as any grade 3 or 4 treatment-related hematologic or nonhematologic toxicity according to Common Terminology Criteria for Adverse Events version 3.0 (excluding alopecia; ref. 24) during the first 21 days of treatment. If one of three patients in a cohort experienced a DLT, three additional patients were to be enrolled at the corresponding dose level. If two or more patients experienced a DLT, then no further patients were to be enrolled in the cohort. Patients were enrolled into the next-highest AMG 102/bevacizumab dose cohort if ≤33% of patients in the previous AMG 102/bevacizumab dose cohort experienced a DLT. All patients continued treatment until disease progression, unacceptable adverse event, or withdrawal of consent.

Blood chemistry, coagulation, hematology, and urinalysis were assessed at baseline, predose, and postdose on days 1, 2, and 8; predose on days 15 and 16; predose every 2 weeks beginning on day 29; and at the end-of-study visit. A 12-lead electrocardiogram (ECG) was done at baseline, predose, and postdose on day 1; postdose on day 15; predose on day 57; predose at every 8 weeks thereafter; and at the end-of-study visit.

Blood samples for the assessment of anti–AMG 102 antibodies were collected predose on days 1, 29, and 57; at scheduled assessments (every 8 weeks); and at the end-of-study visit. Anti–AMG 102–binding antibodies were assayed using an electrochemiluminescent immunoassay and a receptor-binding bioassay (13).

Serum samples for the measurement of AMG 102 concentrations were collected predose, at the end of infusion, and at 6, 24, 96, and 336 h postdose in cycle 1 and cycle 5 for estimation of pharmacokinetic parameters. Serum concentrations of AMG 102 were determined by an AMG 102–specific ELISA (Amgen Inc.) as described previously (13). Plasma samples for the measurement of motesanib concentrations were collected predose and at 0.5, 1, 2, 4, 6, 12, 24, and 336 h postdose in cycle 1 and cycle 5. Plasma concentrations of motesanib were determined by a validated liquid chromatography/tandem mass spectrometry method (CEDRA Corporation) as described previously (22). The lower limits of quantitation for AMG 102 and motesanib were 31.25 and 0.2 ng/mL, respectively. The pharmacokinetics of bevacizumab were not assessed.

Pharmacokinetic estimates for AMG 102 were generated using noncompartmental methods and WinNonlin software (version 5.1.1, Pharsight Corp.). Estimated parameters included observed maximum concentration (C_max), time to attain C_max (t_max), area under the concentration versus time curve (AUC_τ) for a dosing interval (τ), systemic clearance (CL), and drug accumulation ratio (AR) estimated by AUC_τ in cycle 5 versus AUC_τ in cycle 1. Due to the sampling scheme, terminal elimination half-life was not assessed. Summary statistics were computed for each parameter and grouped by study cohort and treatment cycle. Dose proportionality of AMG 102 was tested using a one-way ANOVA model of the dose-normalized (C_max or AUC_{τ norm} = C_max or AUC_τ/dose) and subsequently log-transformed C_max and AUC_τ with dose as the factor of interest. A statistically significant dose effect at α = 0.05 indicates that the pharmacokinetics of AMG 102 is not dose proportional. Analysis was done using SAS/STAT software (version 9.1, SAS Institute Inc.).

For the assessment of HGF/SF and soluble c-Met as potential biomarkers of response, blood was collected predose at days 1, 2, 8, and 22; at the time of tumor assessment (every 8 weeks); and at the end-of-study visit. As described previously (13), total plasma HGF/SF levels were measured using a quantitative sandwich ELISA kit that detects free and antibody-bound HGF/SF (including both pro-HGF/SF and mature HGF/SF), and plasma soluble c-Met was measured by a Meso Scale Discovery electrochemiluminescence assay.

Tumor imaging was done by computed tomography and was evaluated by investigators at each site per modified Response Evaluation Criteria in Solid Tumors (RECIST) 1.0 guidelines (25) within 2 weeks before day 1, before day 57 dosing, every 8 weeks thereafter, and at the end-of-study visit. Independent assessments of tumor response were not done. To qualify as stable disease, tumor measurements must have met the criteria for stable disease at least once after enrollment in the study. Patients without progressive disease at the last disease assessment date were censored.

Descriptive statistics were used to summarize demographic, safety, pharmacokinetic, biomarker, tumor measurement, and response data. Descriptive statistics on continuous data include means, medians, SDs, and ranges. Categorical data are summarized using frequency count and percentages. To determine whether levels of HGF/SF were dependent on treatment received or time on treatment (visits from day 1 to day 57) and to determine whether there is an interaction effect between treatment and time, levels of HGF/SF concentration (on a natural log scale) were analyzed using a linear mixed model (adjusted for baseline levels), with treatment, time, and the interaction between treatment and time as fixed effects and patient as a random effect.

Fourteen patients with refractory advanced solid tumors were enrolled in the study from May 15, 2006 to April 2, 2007. Patient demographics and disease characteristics are summarized in Table 1. Most patients (93%) had stage IV disease, and 71% had received at least three lines of prior chemotherapy. Five patients had received prior therapy with bevacizumab, one of whom received 3 mg/kg AMG 102 plus 10 mg/kg bevacizumab (cohort 1) and four of whom received 20 mg/kg AMG 102 plus 10 mg/kg bevacizumab (cohort 3). Before enrollment, one patient had a partial response and four had stable disease as a best result with their prior bevacizumab regimens before eventually progressing.

All 14 patients (100%) received at least one dose of AMG 102, 8 (57%) of whom received at least four cycles of combination treatment. One patient (cohort 2) with endometrial cancer, who started treatment in August 2006, continued to receive AMG 102 until December 2008 and had progression-free survival (PFS) of 121.9 weeks. There were no dose reductions during the study, but dose interruptions occurred in two patients. The first patient (cohort 2; endometrial cancer) did not receive treatment at week 59 as a result of ongoing treatment-emergent leg pain (grade 1/2). The second patient (cohort 4; carcinoid in lung) did not receive treatment at week 69 as a result of being away on vacation. The reasons for patients discontinuing the study included disease progression (n = 8), adverse events (n = 4), alternative therapy (n = 1), and withdrawn consent (n = 1).

No DLTs occurred in any cohort during the 21-day DLT assessment period. Patient enrollment in cohort 3 was expanded from three to six to gather additional safety and pharmacokinetic data at the highest dose of AMG 102 tested, per protocol specifications. Patient enrollment in cohort 4 was discontinued after two patients were enrolled and three planned motesanib cohorts were removed due to concerns related to the incidence of cholecystitis in other motesanib studies (21, 22). There was no occurrence of cholecystitis in this study; however, one patient in cohort 4 had gallbladder wall edema considered related to motesanib treatment.

Fourteen (100%) patients experienced treatment-emergent adverse events (Table 2). The most common (occurring in ≥5 patients) adverse events in patients who received AMG 102 plus bevacizumab (n = 12) included fatigue (n = 9; 75%), nausea (n = 7; 58%), constipation (n = 5; 42%), and peripheral edema (n = 5; 42%). Among patients who received AMG 102 plus bevacizumab, 5 (42%) experienced grade 3 adverse events, and 1 (8%) experienced a grade 4 adverse event (gastrointestinal hemorrhage and pulmonary embolism; see below). Both patients who received AMG 102 plus motesanib experienced grade 3 adverse events, and 1 (50%) experienced a grade 4 adverse event (pulmonary embolism; see below).

Four patients experienced serious adverse events (Table 2): one patient (cohort 1, day 87) with rectal cancer had grade 3 abdominal pain not considered related to study medication (patient was hospitalized); one patient (cohort 2, day 43) with non–small cell lung cancer had grade 4 pulmonary embolism and grade 2 superficial venous thrombosis (day 44) attributed to both AMG 102 and bevacizumab (patient discontinued treatment, was removed from the study, and subsequently had grade 4 gastrointestinal bleeding on day 52 considered related to bevacizumab); one patient (cohort 4, day 21) with bladder cancer had grade 4 pulmonary embolism attributed to motesanib and grade 3 deep vein thrombosis (day 27; patient discontinued study medication and was removed from the study); and one patient (cohort 3, day 1) with colon cancer who did not receive bevacizumab had grade 2 hypotension during infusion attributed to AMG 102 (patient discontinued treatment and was removed from the study). One other patient (cohort 3) with breast cancer who received prior bevacizumab before entering the study had grade 3 deep vein thrombosis on day 197 attributed to bevacizumab. None of the patients who received prior bevacizumab therapy before entering the study had serious adverse events attributed to bevacizumab.

In addition, a patient in cohort 3 discontinued treatment and was removed from the study following a treatment-related grade 2 ECG QTcF interval prolongation noted during the bevacizumab infusion on day 15. All prior ECGs were normal, and a repeat ECG on day 22 showed that the QTcF interval had returned to normal. The ECG abnormality in this patient was associated with low serum potassium (3.5 mmol/L).

There were no deaths or apparent dose-related trends in the incidence or severity of adverse events during treatment. Antibodies to AMG 102 were not detected in any patient.

AMG 102 serum concentration data were collected from 14 patients. However, data from one patient in cohort 3 (patient discontinued because of an adverse event) were insufficient to determine pharmacokinetic parameters. Pharmacokinetic parameter estimates of AMG 102 are provided in Table 3, and the pharmacokinetic profile of AMG 102 is depicted in Fig. 1A. To evaluate the effect of bevacizumab and motesanib on the pharmacokinetics of AMG 102, individual AMG 102 CL values in this study (at 3, 10, and 20 mg/kg) were compared with CL values estimated at the same doses in the first-in-human monotherapy study (13). The estimated median CL of AMG 102 in combination with bevacizumab or motesanib (0.112 mL/hour/kg; range, 0.086-0.205; n = 8) and AMG 102 monotherapy (0.128 mL/hour/kg; range, 0.061-0.265; n = 11) were similar. Because the CL of AMG 102 in combination with motesanib (0.140 mL/hour/kg) was available for only one patient, the effect of motesanib on AMG 102 pharmacokinetic estimates could not be evaluated. The accumulation ratio of AMG 102 exposure between cycle 1 and cycle 5 ranged from 2.1- to 2.7-fold in the presence of bevacizumab, which is consistent with the accumulation ratio (1.3-3.86) of AMG 102 monotherapy (13). The accumulation ratio in the presence of motesanib could not be calculated because of insufficient data.

Motesanib pharmacokinetic data were available for two patients who received 75 mg motesanib orally once daily in combination with 3 mg/kg AMG 102 every 2 weeks. Peak motesanib concentrations were achieved 1.5 hours postdose; the respective C_max values for the two patients were 0.361 and 0.415 μg/mL (day 1, cycle 1), whereas AUC_τ (τ = 24 hours) values were 2.17 and 2.03 μg × hour/mL (day 1, cycle 5), respectively. As a result of the small sample size, the effect of AMG 102 on motesanib pharmacokinetics could not be evaluated. Statistical analysis on the log-transformed C_max and AUC_τ of AMG 102 in combination with bevacizumab suggests that there was no evidence against AMG 102 dose proportionality in cycles 1 and 5.

Total plasma HGF/SF increased from baseline levels (mean ± SD, 1,358 ± 850 pg/mL; range, 200-3,280 pg/mL) to day 57 (mean ± SD, 9,739 ± 3,850 pg/mL; range, 5,350-17,700 pg/mL) in all cohorts (Fig. 1B). HGF/SF levels were significantly dependent on time on treatment (P < 0.0001) but were not dependent on AMG 102 dose level (P = 0.6848). There was no significant effect of treatment (P = 0.874) or time (P = 0.6658) on circulating soluble c-Met levels.

Postbaseline imaging for the evaluation of tumor response per modified RECIST was available for 10 of 14 patients (25). No complete or partial responses were observed. Nine patients achieved a best result of stable disease with PFS ranging from 8.4 to 121.9 weeks. Among these, eight patients received AMG 102 plus bevacizumab (cohort 1, n = 1; cohort 2, n = 3; cohort 3, n = 4), and one patient received AMG 102 plus motesanib. Stable disease with a duration of ≥8, ≥16, and ≥24 weeks occurred in 9, 7, and 4 patients, respectively. One patient (small intestine cancer) had immediate progressive disease with PFS of 7.9 weeks (Fig. 2). Prolonged stable disease for >80 weeks was observed in a patient with a carcinoid tumor in the lung and for >121 weeks in a patient with slowly progressive endometrial cancer previously treated with a variety of modalities. Two additional patients (pancreatic cancer, breast cancer) had stable disease lasting ≥24 weeks. Changes in the best-result sum of longest diameters are presented in Fig. 2. Eight of 10 evaluable patients had best-result reductions in tumor size that did not qualify as partial responses per RECIST. These patients had the following primary tumor types: non–small cell lung (n = 2) and pancreatic, carcinoid in lung, breast, endometrial, esophageal, and rectal cancer (n = 1 each). At least 9 weeks before receiving study drug, three of the patients with best-result reductions in tumor size had received prior therapy with bevacizumab in combination with either chemotherapy (breast cancer, rectal cancer), chemotherapy and erlotinib (pancreatic cancer), or chemotherapy and cetuximab (rectal cancer). While receiving these regimens, the patient with rectal cancer had a partial response, and the patients with breast and pancreatic cancer had stable disease as a best result. All three patients had progressive disease before entering the study. Of the four patients without a postbaseline scan, one discontinued owing to undocumented progressive disease; the remainder discontinued owing to adverse events.

In this study, AMG 102, a fully human monoclonal antibody to HGF/SF, was combined with two antiangiogenic agents (either bevacizumab or motesanib) based on the hypothesis that interruption of both the c-Met and VEGR signaling pathways might improve antitumor efficacy while maintaining an acceptable toxicity profile. AMG 102 had shown antitumor effects in preclinical models and was shown to be tolerable in a first-in-human study with some patients displaying stable disease (12, 13). In this study, the combination of AMG 102 and bevacizumab seemed to be well tolerated in patients with refractory advanced solid tumors. Adverse events were generally mild to moderate in severity and were consistent with the known toxic effects of each agent. The incidence of some adverse events, such as fatigue, edema, and nausea, seemed somewhat more common in patients treated with AMG 102 and bevacizumab than in the AMG 102 first-in-human study (13). Because of the small sample size, no generalizations can be made about the effect of motesanib on the incidence or types of adverse events.

In this study, two patients experienced serious treatment-related thromboembolic events. One patient (cohort 2) with non–small cell lung cancer developed grade 4 pulmonary embolism on day 43 and grade 2 superficial venous thrombosis on day 44 attributed to both AMG 102 and bevacizumab. A second patient (cohort 4) with bladder cancer had grade 4 pulmonary embolism on day 21 attributed to motesanib and grade 3 deep vein thrombosis on day 27 attributed to motesanib. The relationship of these events to AMG 102 cannot be excluded. Neither of these patients received prior bevacizumab before entering the study. One other patient (cohort 3) with breast cancer who received prior bevacizumab before entering the study developed grade 3 deep vein thrombosis on day 197 attributed to bevacizumab. In the first-in-human study of AMG 102, one patient discontinued treatment as a result of deep vein thrombosis not related to AMG 102 (13). In a phase II study of AMG 102 in glioblastoma multiforme, one patient developed a treatment-related deep vein thrombosis leading to discontinuation of AMG 102 (26).

Thromboembolic events and hemorrhage are class effects of agents inhibiting the VEGF/VEGFR pathways (27). Both bevacizumab and motesanib have been associated with thromboembolic events or hemorrhagic complications when administered to patients with solid tumors as single agents or in combination with chemotherapy (19, 21–23, 27–30). Because of the small number of patients treated with AMG 102, it is not possible to ascribe a definite role for AMG 102 in the development of thromboembolic events in this study. The potential contribution of prior treatment with bevacizumab to the development of venous thromboembolism in this study could have possibly been underestimated by the exclusion of patients with a history of thrombotic events within one year of enrollment. On the other hand, repetitive courses of bevacizumab might have been contributory to an episode of grade 3 deep venous thrombosis occurring in one patient.

During the conduct of this study, experience with motesanib in other studies suggested that some dose schedules of motesanib may be associated with an increased incidence of cholecystitis, as well as gallbladder enlargement (21, 22). Because of these concerns, it was decided to prematurely discontinue enrollment into the motesanib arm. It should be noted that although cholecystitis did not occur in this study, one patient developed gallbladder edema that was monitored closely and deemed to be associated with motesanib. It is important to note, however, that in a recent phase Ib trial combining motesanib with gemcitabine in patients with solid tumors, no instances of cholecystitis were observed (23).

To determine whether AMG 102 pharmacokinetic parameters were affected by either bevacizumab or motesanib, a comparison of AMG 102 CL values in the presence or absence of these agents was done. Based on the parameters estimated, the pharmacokinetics of AMG 102 in the bevacizumab cohorts seemed similar to the pharmacokinetics of AMG 102 observed in the first-in-human study (13). The effect of motesanib on AMG 102 pharmacokinetics could not be estimated because of the small sample size.

Treatment with the combination of AMG 102 plus bevacizumab or motesanib led to a best result of no tumor progression (stable disease) in 9 of 10 evaluable patients. Eight of these nine patients received bevacizumab, whereas one received motesanib. Stable disease with a duration of ≥8 and ≥16 weeks was noted in 9 and 7 patients, respectively, whereas 4 patients maintained stable disease for ≥24 weeks. Tumor reductions (as shown in Fig. 2A) occurred in 8 of 10 evaluable patients, but no partial or complete responses (as defined by RECIST) occurred. In the first-in-human study of AMG 102, 70% of evaluable patients achieved a best result of stable disease (13). Bevacizumab as a single agent is known to possess limited activity in a number of tumor types; however, partial responses do not occur in a large proportion of patients (28, 29, 31, 32). Thus, the contribution of bevacizumab to the instances of stable disease and measurable tumor regressions seen in combination with AMG 102 cannot be accurately estimated. It should be noted, however, that five patients entered the study having received prior bevacizumab, and in only one case was a prior tumor partial response documented. In any event, in this small study with varying tumor types, it is not possible to speculate further on the impact of AMG 102 on either the magnitude of tumor responses or the duration of PFS.

The HGF/SF:c-Met axis is targeted by a number of recently developed agents. Several selective and nonselective c-Met multikinase inhibitors have been developed and entered clinical trials (9). Among these, ARQ197, a selective non-ATP competitive c-Met kinase inhibitor, showed encouraging antitumor activity and a favorable safety profile in a phase I dose-escalation study of patients with metastatic solid tumors (33). Adverse events of fatigue and nausea similar to the present study were noted. Treatment with the nonselective c-Met multikinase inhibitors XL184 and XL880, which also show anti-VEGFR2 activity, have been associated with partial responses in some patients, as well as an adverse event profile consistent with VEGFR inhibition, including hypertension, fatigue, and thromboembolic events (34–36).

The expression of HGF/SF and c-Met and aberrant signaling via the HGF/SF:c-Met axis have been implicated in the development and progression of human cancers (5, 7, 37–41). For example, among patients with solid tumors, HGF/SF (42) and c-Met (43–47) expression has been associated with poor prognosis, tumor invasiveness, and/or migration. We thus attempted to correlate HGF/SF and soluble c-Met as potential biomarkers of disease progression and response to therapy with AMG 102. Similar to the first-in-human study (13), levels of plasma total HGF/SF increased upon administration of AMG 102, but the AMG 102 dose level did not significantly predict HGF/SF concentration. This lack of correlation might have been the result of small sample size. Increases from baseline in HGF/SF levels might have been due to stabilization of HGF/SF by formation of AMG 102:HGF/SF complexes because the assay is able to detect free HGF/SF, as well as HGF/SF bound to AMG 102. Soluble c-Met levels were not correlated with treatment with AMG 102. Similar results were observed with AMG 102 monotherapy in the first-in-human study (13).

In summary, the combination of AMG 102 and bevacizumab seemed to be relatively safe and tolerable in patients with refractory solid tumors, but the combination of AMG 102 and motesanib could not be adequately assessed because of the small cohort size. An encouraging proportion of patients achieved durable stable disease lasting ≥16 weeks, suggesting that further investigation of AMG 102 in combination with angiogenesis inhibitors is warranted.

D. Beaupre, H. Deng, I. Leitch, P. Shubhakar, M. Zhu, K. Oliner, A. Anderson, employees of and shareholders in Amgen Inc.; P. Rosen, member of Amgen Inc. Speakers' Bureau. This study was funded by Amgen Inc.

We thank Erik Rasmussen, Charity Scripture, Jian-Feng Lu, Mario Bejarano, Bryan Kadotani, Cindy Farrell, Nick Yeager, Jennifer Malella, Mark Ma, Mukesh Verma (Amgen Inc.), and Marie Fuerst (Tower Cancer Research Foundation). We also acknowledge Benjamin Scott, PhD, whose work was funded by Amgen Inc., for assistance in writing the manuscript.

Grant Support: Amgen Inc.

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.