Phase I Dose-Escalation Study of Onartuzumab As a Single Agent and in Combination with Bevacizumab in Patients with Advanced Solid Malignancies

Published OnlineFirst February 3, 2014; DOI: 10.1158/1078-0432.CCR-13-2070

Clinical Cancer Cancer Therapy: Clinical Research

Phase I Dose-Escalation Study of Onartuzumab as a Single Agent and in Combination with Bevacizumab in Patients with Advanced Solid Malignancies

Ravi Salgia1, Premal Patel2, John Bothos2, Wei Yu2, Steve Eppler2, Priti Hegde2, Shuang Bai2, Surinder Kaur2, Ihsan Nijem2, Daniel V.T. Catenacci1, Amy Peterson2, Mark J. Ratain1, Blase Polite1, Janice M. Mehnert3, and Rebecca A. Moss3

Abstract Purpose: This ﬁrst-in-human study evaluated the safety, immunogenicity, pharmacokinetics, and antitumor activity of onartuzumab, a monovalent antibody against the receptor tyrosine kinase MET. Experimental Design: This 3þ3 dose-escalation study comprised three stages: (i) phase Ia dose escalation of onartuzumab at doses of 1, 4, 10, 20, and 30 mg/kg intravenously every 3 weeks; (ii) phase Ia cohort expansion at the recommended phase II dose (RP2D) of 15 mg/kg; and (iii) phase Ib dose escalation of onartuzumab at 10 and 15 mg/kg in combination with bevacizumab (15 mg/kg intravenously every 3 weeks). Serum samples were collected for evaluation of pharmacokinetics, potential pharmacodynamic markers, and antitherapeutic antibodies. Results: Thirty-four patients with solid tumors were treated in phase Ia and 9 in phase Ib. Onartuzumab was generally well tolerated at all dose levels evaluated; the maximum tolerated dose was not reached. The most frequent drug-related adverse events included fatigue, peripheral edema, nausea, and hypoalbuminemia. In the phase Ib cohort, onartuzumab at the RP2D was combined with bevacizumab and no dose- limiting toxicities were seen. Onartuzumab showed linear pharmacokinetics in the dose range from 4 to 30 mg/kg. The half-life was approximately 8 to 12 days. There were no apparent pharmacokinetic interactions between onartuzumab and bevacizumab, and antitherapeutic antibodies did not seem to affect the safety or pharmacokinetics of onartuzumab. A patient with gastric carcinoma in the 20-mg/kg dose cohort achieved a durable complete response for nearly 2 years. Conclusions: Onartuzumab was generally well tolerated as a single agent and in combination with bevacizumab in patients with solid tumors. Clin Cancer Res; 20(6); 1–10. 2014 AACR.

Introduction poor prognosis in several cancers, including non–small cell MET, a receptor tyrosine kinase, initiates branching mor- lung cancer (NSCLC), breast cancer, gastric cancer, head phogenesis in normal cells upon binding of its only known and neck cancer, and glioblastoma (3). Components of the ligand, hepatocyte growth factor (HGF; also known as MET signaling pathway have emerged as attractive thera- scatter factor). Branching morphogenesis involves cell pro- peutic targets (5–7). liferation, transformation, and migration (1, 2). Dysregula- Onartuzumab (formerly called MetMAb and PRO143966) tion of HGF and MET plays a key role in tumorigenesis (3, is a humanized, monovalent (one-armed) monoclonal anti- 4). Clinically, MET receptor expression is associated with body that blocks HGF binding and prevents downstream cellularsignaling (8). Onartuzumab was designed asa monovalent antibody to avoid the agonistic activity that may occur when a bivalent antibody binds two MET molecules (9). Authors' Afﬁliations: 1Section of Hematology/Oncology, Department of Medicine, The University of Chicago, Chicago, Illinois; 2Genentech, Inc., Onartuzumab demonstrated antitumor activity in various South San Francisco, California; and 3Rutgers Cancer Institute of New animal tumor models (6, 8, 10). MET and its downstream Jersey, Rutgers, The State University of New Jersey, New Brunswick, New Jersey effectors interact with other tumorigenic pathways; MET activation functions as a potent regulator of the angiogenic Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/). switch by increasing expression of angiogenic molecules such as vascular endothelial growth factor (VEGF) (11). Hypoxia Corresponding Author: Rebecca A. Moss, Rutgers Cancer Institute of New Jersey, 195 Little Albany Street, New Brunswick, NJ 08903-2681. can induce expression of MET on tumor cells, which may Phone: 1-732-235-8663; Fax: 1-732-235-8808; E-mail: enable improved tumor cell survival and/or migration [email protected] toward regions of normoxia (12). Enhanced antitumor activ- doi: 10.1158/1078-0432.CCR-13-2070 ity has been observed when combining onartuzumab with 2014 American Association for Cancer Research. an anti-VEGF antibody in preclinical models (10). These

www.aacrjournals.org OF1

Salgia et al.

Patients were generally excluded if they had received Translational Relevance antitumor therapies or undergone a major surgical proce- Aberrant upregulation of the MET pathway is frequent dure within 4 weeks of the start of the study. Patients were in human malignancies, and leads to branching mor- excluded from the bevacizumab cohort in phase Ib if they phogenesis and invasion. MET activation may also occur had inadequately controlled hypertension as defined by during states of hypoxia, which may enable migration/ systolic blood pressure >150 mmHg and/or diastolic blood invasion toward regions of normoxia. This phase Ia and pressure >100 mmHg, a history of cardiovascular disease or Ib study shows onartuzumab—a humanized, monova- significant vascular disease within 6 months before enroll- lent (one-armed) monoclonal antibody that blocks ment, or major surgery within 28 days before start of dosing. hepatocyte growth factor binding—is well tolerated as single agent and in combination with bevacizumab. In Study design and treatments this setting of good safety, we show how pharmacoki- The study involved three stages: (1) the phase Ia dose- netic/pharmacodynamic (PK-PD) modeling may be escalation phase was designed to determine the safety, used to derive a recommended phase II dose. One tolerability, and pharmacokinetics of single-agent onartu- patient with clinically relevant exposure and with auto- zumab (supplied by Genentech, Inc.) at dose levels of 1, 4, crine MET gastric cancer achieved a complete response. 10, 20, and 30 mg/kg administered by i.v. Q3W, and to identify the RP2D; (2) in the phase Ia expansion, additional patients were enrolled to study safety and pharmacokinetics of single-agent onartuzumab at the RP2D; and (3) the phase observations justify evaluation of combination therapies that Ib stage evaluated the combination of onartuzumab [10 target both angiogenesis and MET signaling. mg/kg (one dose level below the RP2D) or 15 mg/kg i.v. We conducted a phase I dose-escalation study of onartu- Q3W (RP2D)] with bevacizumab (15 mg/kg i.v. Q3W; zumab to identify the maximum tolerated dose (MTD) and supplied by Genentech, Inc.). Patients without progressive recommended phase II dose (RP2D), and to characterize disease or significant toxicity could continue treatment with the safety, tolerability, and pharmacokinetics of onartuzu- study drug(s) for a maximum of 16 cycles or 1 year. mab, alone and in combination with bevacizumab, in Patients were discontinued from the study for any of the patients with advanced solid malignancies. following reasons: dose-limiting toxicity (DLT), disease progression, noncompliance, change in eligibility, or if the Materials and Methods investigator felt it was in their best interest. Patients in phase Onartuzumab was administered by intravenous (i.v.) Ib who experienced bevacizumab-related adverse events infusion every 3 weeks (Q3W), either alone or in combi- (AE) discontinued bevacizumab, but could continue with nation with bevacizumab (15 mg/kg i.v. Q3W) in patients onartuzumab treatment. with advanced solid malignancies refractory to standard of Astandard3þ3 dose-escalation design applied for care therapy or for which standard of care therapy did not onartuzumab until the maximum administered dose of exist. This study was conducted at two U.S. sites (The 30 mg/kg was reached in the phase Ia escalation and until University of Chicago, IL, and Cancer Institute of New the RP2D of 15 mg/kg was reached in phase Ib with Jersey, NJ). The protocol was approved by the Institutional bevacizumab. The dose of onartuzumab for each patient Review Boards before patient recruitment and was con- was dependent on dose level assignment and patient ducted in accordance with the International Conference on weight on or within 14 days of day 1 of cycle 1. Onartu- Harmonisation E6 Guideline for Good Clinical Practice. zumab was administered over 90 10 minutes for the first two doses. Subsequent doses were administered over Eligibility criteria 30 10 minutes (for dose levels <10 mg/kg) or 60 10 Key inclusion criteria were adults with histologically minutes (for dose levels 10 mg/kg). In phase Ib, onar- confirmed solid malignancy that was incurable, locally tuzumab was given before bevacizumab. advanced, or metastatic and that had failed to respond to at least one prior regimen or for which standard of Safety, tolerability, and response assessments care therapy did not exist; measurable or evaluable dis- The primary outcome measure was the safety and toler- ease defined by Response Evaluation Criteria in Solid ability of onartuzumab alone or in combination with Tumors (RECIST; ref. 13); Eastern Cooperative Oncology bevacizumab assessed based on the frequency and nature Group performance status (ECOG PS) 0 to 2; adequate of DLTs, AEs (graded according to NCI CTCAE version 3.0), renal (creatinine 1.5 mg/dL or creatinine clearance 50 vital signs, and clinical laboratory parameters. Medical mL/min), hematologic (granulocytes 1,500/mL, plate- Dictionary for Regulatory Activities preferred terms were lets 100,000/mL, hemoglobin 10 g/dL), and hepatic used. A DLT was defined as grade 3 hematologic toxicity, function [aspartate aminotransferase, alanine amino- grade 3 nonhepatic major organ toxicity, or grade 3 transferase, and total bilirubin within normal limits or hepatic toxicity considered to be related to the study drug. National Cancer Institute Common Terminology Criteria Patients were assessed for AEs on days 1, 8, and 15 of cycles 1 for Adverse Events (NCI CTCAE) grade 1 or less if liver and 2, and on days 1 and 8 of each subsequent cycle. All AEs had tumor involvement]. and serious AEs were summarized by term and grade.

OF2 Clin Cancer Res; 20(6) March 15, 2014 Clinical Cancer Research

Phase I Dose-Escalation Study of Onartuzumab

Safety and tolerability were assessed through clinical and tumor inhibition PK/PD model was developed. The PK laboratory evaluations at weekly intervals for the first 6 driver of antitumor activity was directly determined from weeks then at a minimum of every 3 weeks throughout the time–dose fractionation studies designed to identify con- remainder of a patient’s study treatment. centration and time effects on tumor response. Human Disease response and progression, based on investiga- onartuzumab pharmacokinetics were predicted by alo- tor assessments and using RECIST 1.0 (13) were per- metric scaling based on PK data from cynomolgus monkey formed at baseline, and after cycles 2 and 4, then every and used in the integrated PK/PD model. The theoretical third cycle thereafter until study termination. The same onartuzumab treatment efficacy in humans was then pre- radiographic procedure used at baseline was used dicted using drug-mediated antitumor activity derived from throughout the study. Bone and brain scans were per- the tumor cell growth inhibition model and predicted formed as clinically indicated. clinical onartuzumab serum concentrations. Subsequently, All patients who received any amount of onartuzumab 500 simulations based on the PK/PD modeling results were alone or in combination with bevacizumab were included performed using an onartuzumab population pharmaco- in the safety analysis. Patients who withdrew before day 2 of kinetic model to determine the dose and regimen required cycle 2 for reasons other than DLT were not evaluable for to achieve steady-state trough concentrations above the TSC DLT and MTD assessments. The MTD was defined as the in 90% of patients. highest dose level at which less than one third of the group (comprising a minimum of 6 patients) experienced a DLT. Antitherapeutic antibody assessments Serum samples for evaluating antitherapeutic antibo- Pharmacokinetic assessments and analysis plan dies (ATA) to onartuzumab were collected before each Serum samples for establishing onartuzumab and bev- treatment cycle and at study termination. The samples acizumab concentrations were collected during cycle 1 were assayed in a validated bridging electrochemilumi- before dosing, 0.5, 2, 4, and 7 hours, and 1, 2, 3, 7, 10, nescence assay. The relative sensitivity of the assay was 14, and 21 days after dosing. Samples were collected before estimated to be 143 ng/mL using affinity-purified poly- and 0.5 hours after dosing at every cycle thereafter and at clonal cynomolgus monkey antionartuzumab antibodies treatment termination. and the assay could detect 500 ng/mL of the antibodies in Diluted serum samples were assayed for onartuzumab thepresenceof50mg/mL onartuzumab. Samples that and bevacizumab concentrations using validated sandwich screened positive were further evaluated by competitive ELISA with a limit of quantification of 200 and 78 ng/mL, binding experiments to understand whether the response respectively (Supplementary Material). Shed MET levels was directed against the unique engineered part of the were measured by electrochemiluminescence assay (Sup- framework. The strength of positive responses was eval- plementary Material) at baseline, 24 hours after first dose uated by titering (Supplementary Material). and at the end of the first cycle. The onartuzumab pharmacokinetic data collected in Pharmacodynamic assessments cycle 1 were analyzed by noncompartmental analysis with Blood was drawn at baseline and through cycle 3 of WinNonlin (version 5.2; Pharsight Corporation) and were study treatment for measurements of serum HGF, shed summarized by mean, standard deviation, and range. The MET, and interleukin (IL)-8. Methods for serum HGF and data were compared across phase Ia escalation, phase Ia IL-8 are reported elsewhere (16). Serum-shed MET levels expansion, and phase Ib for onartuzumab. Measured bev- were evaluated 24 hours after first dose and at the end of acizumab data in phase Ib were compared with the pre- cycle 1 (day 21) using an electrochemiluminescence assay dicted concentrations (based on published bevacizumab (Supplementary Material). Serum levels of HGF and cir- population pharmacokinetics model; ref. 14) to explore the culating shed MET were measured at baseline, cycle 1 day potential impact of onartuzumab administration on bev- 1 (predose), cycle 1 day 2; and at cycle 2 and cycle 3 day 1 acizumab exposure. (predose). Levels of HGF and shed MET were compared Pharmacokinetic/pharmacodynamic modeling and sim- among dose cohorts, and with concomitant serum con- ulation was used to estimate the onartuzumab tumoristatic centrations of onartuzumab. To address potential vari- concentration (TSC), in which there is no net tumor growth. ability due to platelet activation in serum, a comparison Estimated TSC was used as an efficacious target concentra- study was conducted in patients with NSCLC with tion to optimize the clinical dose and regimen. matched collections of plasma and serum. An assessment This integrated analysis was based on the determined of circulating HGF in this cohort showed a strong corre- relationship between onartuzumab exposure in preclinical lation of r ¼ 0.974, thus justifying the use of serum in this KP4 athymic xenograft mouse studies and the resulting phase I study (16). antitumor activity in this model, and predicted human pharmacokinetics (15). The KP4 model was selected for Additional statistical analyses this analysis, but other similar models, such as H596 and Demographic and baseline characteristics were summa- U87, had similar antitumor activity at comparable onartu- rized using mean, standard deviation, median, and range zumab doses. Onartuzumab exposure and tumor response for continuous variables, and proportion for categorical data were collected in KP4 athymic xenograft mice, and a variables. Safety data were summarized by cohort.

www.aacrjournals.org Clin Cancer Res; 20(6) March 15, 2014 OF3

Salgia et al.

The planned sample size was 21 to 36 patients in the In phase Ib, a single DLT of grade 1 hemoptysis in a phase Ia dose-escalation cohort, 6 to 12 patients in the patient with a prior history of pulmonary hemorrhage was phase Ia expansion cohort, and up to 9 patients in phase Ib, observed at the onartuzumab dose 15 mg/kg Q3W in depending on the number of DLTs observed. combination with bevacizumab (see below). Based on this, the RP2D of onartuzumab in combination with bevacizu- Results mab was also determined to be 15 mg/kg. Patient disposition and demographics In total, 43 patients with solid malignancies (including Safety and tolerability colorectal, gastric, head and neck, liver/biliary, lung, For the 34 patients treated with onartuzumab alone, the melanoma, ovarian, pancreatic, and sarcoma tumors) were median number of doses received was 2 (range, 1–10). In included in this study: 21 patients were enrolled in phase Ia phase Ib, the median number of onartuzumab and beva- dose escalation, 13 in phase Ia expansion, and 9 in phase Ib. cizumab doses received by the 9 patients was 4 (range, 1–6). Baseline demographic characteristics are summarized Most AEs reported were of grade 1 or 2 severity and were in Table 1. reversible. Table 2 lists all onartuzumab-related grade 1 or 2 AEs experienced by >5% of patients, and all onartuzumab- Dose-limiting toxicity related grade 3 AEs experienced in either phase Ia or Ib In the phase Ia dose escalation, a single DLT of grade 3 (there were no onartuzumab-related grade 4/5 events). pyrexia was observed in the 4 mg/kg cohort. There was no During phase Ia, the most frequently reported onartuzu- > MTD identiﬁed, the maximum administered dose was 30 mab-related AEs occurring in 10% of patients comprised mg/kg, and the RP2D, based on preclinical pharmacoki- fatigue (38%), peripheral edema (27%), nausea (12%), and netic/pharmacodynamic modeling (15), was determined to hypoalbuminemia (12%). During phase Ia, 7 patients be 15 mg/kg; a total of 13 patients were treated at this dose (21%) experienced grade 3 onartuzumab-related AEs. Two level in the dose-expansion phase. patients experienced serious AEs [fever (also a DLT) and peripheral edema] judged possibly related to study drug. With the exception of peripheral edema (severity reported n ¼ n ¼ n ¼ Table 1. Patient demographics and baseline as grade 1, 6; grade 2, 0; grade 3, 3), no other obvious dose relationship associated with AEs was demon- characteristics strated (Supplementary Table S1). During phase Ib, 3 of the 9 treated patients (33%) expe- Phase Ia rienced onartuzumab-related AEs (Table 2); no AEs grade escalation Phase Ib 3 were observed. One serious AE of grade 1 hemoptysis (also and expansion escalation a DLT) was reported as possibly related to bevacizumab. Characteristic (n ¼ 34) (n ¼ 9) Pulmonary bleeding (any grade) was a protocol-deﬁned Age, y DLT. This patient with gastric cancer and pulmonary metas- Median 63 46 tases had a history of hemoptysis before entering the study; Range 29–85 42–80 computed tomography scan of chest showed evidence of Gender, n (%) pulmonary lesions with central necrosis. Female 15 (44) 7 (78) For the 34 patients treated in phase Ia, the most com- Male 19 (56) 2 (22) monly reported AEs (>10% frequency) regardless of rela- n Primary tumor, (%) tionship to study drug were fatigue (56%), peripheral Colorectal 6 (18) 4 (44) edema (35%), and decreased appetite (32%; Supplemen- Gastric 5 (15) 1 (11) tary Table S2). For 9 patients in phase Ib, the most com- — Head and neck 1 (3) monly (>10% frequency) reported AE regardless of rela- — Kidney 1 (11) tionship to study drug was fatigue (56%), followed by Liver/biliary 1 (3) — increased weight (33%; Supplementary Table S2). Lung 6 (18) 2 (22) In summary, an MTD for onartuzumab was not reached Melanoma 2 (6) — when administered alone and the RP2D of onartuzumab in Ovarian 2 (6) — combination with bevacizumab had no apparent impact on Pancreatic 3 (9) — the severity or frequency of known bevacizumab-related Sarcoma 2 (6) — Other 6 (18) 1 (11) AEs. ECOG PS, n (%) 0 19 (56) 3 (33) Pharmacokinetics/pharmacodynamics 1 14 (41) 5 (56) Onartuzumab concentration–time data (Fig. 1a) were 2 1 (3) 1 (11) obtained from all 43 patients. Table 3 summarizes the main pharmacokinetic parameters estimated using noncompart- Abbreviation: ECOG PS, Eastern Cooperative Oncology mental analysis. Onartuzumab seemed to manifest linear Group performance status. pharmacokinetics in the dose range from 4 to 30 mg/kg, although clearance of the 1 mg/kg dose was faster than the

OF4 Clin Cancer Res; 20(6) March 15, 2014 Clinical Cancer Research

Phase I Dose-Escalation Study of Onartuzumab

Table 2. Treatment-related adverse events for onartuzumab administered by intravenous infusion once every 3 weeks, alone or in combination with bevacizumab, in Phase Ia escalation (onartuzumab 1 to 30 mg/kg) and expansion (onartuzumab 15 mg/kg), and phase Ib (bevacizumab 15 mg/kg plus onartuzumab 10 and 15 mg/kg)

Phase Ia escalation Phase Ib and expansion (n ¼ 34) escalation (n ¼ 9) Grade 1 or 2 Grade 3a Grade 1 or 2 Grade 3a Any adverse event, n (%) 15 (44) 7 (21) 3 (33) 0 Fatigue 13 (38) 0 1 (11) 0 Edema, peripheral 6 (18) 3 (9) 1 (11) 0 Hypoalbuminemia 4 (12) 0 1 (11) 0 Nausea 4 (12) 0 1 (11) 0 Vomiting 3 (9) 0 —— Weight increased 1 (3) 0 2 (22) 0 Anorexia 3 (9) 0 1 (11) 0 Muscle spasms 3 (9) 0 —— Abdominal pain 0 1 (3) —— AST increased 0 1 (3) —— Pyrexia 0 1 (3) —— Hyponatremia 0 1 (3) ——

Abbreviation: AST, aspartate aminotransferase. aThere were no grade 4 events. other dose levels, suggesting target-mediated drug disposi- nation therapy with bevacizumab (phase Ib; Table 3). The tion at lower doses, which is commonly observed for anti- majority of observed bevacizumab concentration data bodies. In the linear dose range, the group mean clearance points in the treatment cycle fell within the 90% confidence values varied from 6.8 to 9.9 mL/d/kg, which may be interval of predicted concentrations (Fig. 1c). Similar find- slightly faster than typical bivalent antibodies (approxi- ings were observed using Cmin and Cmax data points in other mately 3–5 mL/day/kg; ref. 17). The other pharmacokinetic treatment cycles (data not shown). These results suggest that parameters increased proportionally with dose. The mean coadministration of onartuzumab and bevacizumab does terminal half-life was 8 to 12 days. Based on population, PK not seem to affect the exposure to either drug. simulated concentrations at the end of cycles 1 and 3 (steady-state), Ctrough concentrations increased slightly Antitherapeutic antibodies from 29.4 to 40.4 mg/mL, and the percentage of patients ATAs were evaluated in 42 patients. Six patients tested with Ctrough above the TSC of 15 mg/mL increased from 86% positive for ATAs to onartuzumab; for all 6 patients, the to 90%. Consistent with typical monoclonal antibodies, the response was minimal, with a titer range of 1.5 to 1.9 volume of distribution was small across all dose groups, (minimum reportable titer, 1.4). For 1 of these patients, indicating limited tissue distribution. an ATA response was also observed in the pretreatment Pharmacokinetic/pharmacodynamic modeling estimat- sample, which fell within the assay’s validated false-positive ed the TSC for onartuzumab in KP4 pancreatic xenograft rate. All ATA responses were determined to be primarily tumor–bearing mice to be 15 mg/mL, suggesting that con- toward the nonengineered framework of onartuzumab; the centrations above 15 mg/mL could result in net tumor pharmacokinetic profiles of the ATA-positive patients did shrinkage (15). Therefore, 15 mg/mL was selected as the not seem to be affected by the immunogenicity. None of the Ctrough to achieve in the clinic. Population pharmacokinetic ATA responses seemed to correlate with any AEs. simulations based on data collected in this phase I trial indicated that a dose of 15 mg/kg Q3W would achieve Evaluation of on-treatment shed MET levels in serum as steady-state trough concentrations of 15 mg/mL in 90% of a marker of target engagement patients (Fig. 1b). Therefore, 15 mg/kg on a Q3W dosing Other biomarkers thought to be modulated by MET regimen was determined to be the RP2D. Other xenograft signaling, such as HGF and IL-8, were evaluated at similar models representing diverse tumor-types, such as H596 time points and are reported elsewhere (16). The extracel- (NSCLC) and U87 (glioblastoma multiforme), had similar lular domain of the MET receptor is proteolytically cleaved antitumor activity at comparable onartuzumab doses. and shed into circulation (18) and was detected at baseline Onartuzumab pharmacokinetic parameters at 15 mg/kg concentrations of 158 62 ng/mL in the serum of the phase were similar for both monotherapy (phase Ia) and combi- I patients. Serum was available from a total of 27 patients

www.aacrjournals.org Clin Cancer Res; 20(6) March 15, 2014 OF5

Salgia et al.

receiving any dose of single-agent onartuzumab. Shed MET A levels were evaluated to conﬁrm target engagement in humans. As shown in Fig. 2, drug treatment resulted in an immediate increase in shed MET levels in all patients. An approximately 2-fold increase in shed MET was observed in all patients in all dose groups at the 24-hour time point. Evaluation of shed MET at the Cmin concentration of onartuzumab at the end of cycle 1 showed that patients in the 1 mg/kg group continued to maintain a 2-fold increase in shed MET. However, dose groups of 4 mg/kg and higher showed a 3- to 7-fold increase over predose levels, suggesting saturation of the peripheral receptors at these dose levels.

Tumor assessments Figure 3 shows waterfall plots for best tumor response observed in individual patients during the study. There was one complete response (CR); no other RECIST responses were observed. B The CR was observed following 4 cycles of treatment with single-agent onartuzumab (20 mg/kg) in a 50-year-old patient with a solitary hepatic metastasis from gastric carcinoma (19) who had received prior chemotherapy and a prior investigative therapy. The CR was confirmed by mag- netic resonance imaging and after the tenth cycle of onartuzumab the patient decided to stop study drug. Analysis of the pretreatment biopsy sample produced findings consistent with autocrine HGF/MET biology: Intratumoral MET and HGF expression were identified by immunohistochemistry (Fig. 4), chromosome 7 polyploidy (but no MET focal gene amplification) was identified by fluorescence in situ hybridization, and there was no evidence of MET gene mutation (19). This patient had elevated pretreatment C serum HGF levels and was the only patient to show sus- tained decrease in HGF after onartuzumab treatment. This represents the sole patient in the trial with an objective response, and thus the only patient with tissue assessed for MET pathway markers.

Discussion In this phase I dose-escalation study, onartuzumab was well tolerated when administered intravenously as a single agent at doses up to 30 mg/kg Q3W and in combination with bevacizumab at the RP2D of 15 mg/kg Q3W. The RP2D was selected based on the overall safety, as well as pharmacokinetic/pharmacodynamic modeling (15). It is notable that no dose relationship with any AEs was evident, except for peripheral edema, which was observed as a treatment-related AE at doses exceeding 10 mg/kg. Figure 1. Serum concentration–time proﬁles for onartuzumab in cycle 1 The majority of AEs were grade 1 or 2, with grade 3 events (n ¼ 43; A), model of onartuzumab dose projected to achieve minimum (three in total) reported during the phase Ia expansion tumoristatic concentration (MTC; B), and bevacizumab for 9 patients in (Table 2). One patient in phase Ia escalation (onartuzumab phase Ib treated with onartuzumab (10 and 15 mg/kg) and bevacizumab (15 mg/kg; C). Concentrations, mean ( SD) in A. B, percentage of patients with onartuzumab steady-state trough concentration projected to have a greater than the minimum tumoristatic concentration versus measurements were completed at cycle 1, immediately before Q3W doses of onartuzumab. C, solid line, median; dashed lines, 2.5th and onartuzumab dosing, and after dose at the following time intervals: 30 97.5th percentiles of bevacizumab-predicted concentrations based on minutes, 2 hours, 4 hours, 7 hours, 24 hours, 48 hours, 72 hours, day 8, the population pharmacokinetics model. Pharmacokinetics and day 11.

OF6 Clin Cancer Res; 20(6) March 15, 2014 Clinical Cancer Research

Phase I Dose-Escalation Study of Onartuzumab 2.80 2.09 9.92 17.8 390 95.6 þ 1.32 7.93 6.70 24.5 12.1 81.6 199 1,680 0.611 8.26 49.2 383 bevacizumab) Phase Ib (onartuzumab , trough concentration; LTR, less than min

Figure 2. Evaluation of shed MET through cycle 1 of onartuzumab therapy. 1.70 6.83 22.1 22.2 3.63 11.4 144 239 22.7 96.8 452 1,140 Serum-shed MET levels were evaluated 24 hours after ﬁrst dose and at the end of cycle 1 (day 21) using an ELISA-based assay. Data, ratio to predose levels at each dose group.

20 mg/kg, grade 1) and 1 patient in phase 1a expansion (onartuzumab 15 mg/kg, grade 3) discontinued treatment , peak concentration; C 2.63 7.86 21.6 48.2 4.99 8.65 132 631 443 3,280

2.31 85.3 due to peripheral edema. Peripheral edema has also been max reported in trials of other monoclonal antibodies targeting the MET signaling pathway, such as rilotumumab (AMG102; ref. 20) and ficlatuzumab (AV-299; ref. 21), two anti-HGF antibodies, suggesting that this toxicity may be an on-target finding. The MET/HGF axis is involved in many biologic func- 1.94 7.58 5.54 11.8 23.4 105 86.2 441 13.1 36.0 398 2,170 Dose group (mg/kg) tions, including a central role in liver regeneration (22). No significant hepatotoxicity (manifested by changes in liver enzymes or total bilirubin) was observed with onartuzumab. It is important to note that small-molecule inhibitors against MET have been reported to lead to liver AEs, which may be related to off-target mechanisms 2.39 6.85 5.07 36.7 1.99 11.5 135 1,740 12.7 348 2.52 96.8 (23). Additionally, MET activation also functions as a potent regulator of the angiogenic switch by increasing (single-agent onartuzumab) expression of angiogenic molecules such as VEGF, includ- Phase Ia escalation and expansion ing during states of hypoxia (11, 12). The ability of onartuzumab and bevacizumab to be combined relatively 1.88 9.89 17.2 166 9.56 105 2.74 13.5 2.23 8.13 99.5 877 safely allows broader testing of the combination in phase time curve from 0 to 21 days; CL, systemic clearance; C – II programs. Onartuzumab exhibits linear pharmacokinetics within the tested dose range of 4 to 30 mg/kg, indicating the saturation of target-mediated clearance. The clearance was slightly faster and the terminal half-life was slightly shorter 2.38 7.07 2.64 95.5 11.9 77.6 10.8 460 0.404 8.55 , steady-state volume of distribution. compared with typical bivalent antibodies (17). Although ss onartuzumab has a slightly faster clearance compared with 64.0 2.74 other antibodies, pharmacokinetic/pharmacodynamic modeling and simulation predicts the RP2D of 15 mg/kg Q3W will achieve steady-state trough concentrations above , area under the concentration m 21 the TSC of 15 g/mL in 90% of patients. Analyses of both SD) 1 4 10 15 20 30 10 15 0

onartuzumab and bevacizumab concentration data suggest that coadministration does not seem to affect drug exposure g/mL) 67.3 , elimination half-life; V to either. m 1/2 Pharmacokinetic parameters estimated using noncompartmental analysis for cycle 1 data (d Because of the unique monovalent structure of onartu- 21 g/mL) 27.1

g/mL) LTR 8.26 zumab, induction of ATAs was a concern. Although some m m ( ( 0 (d) (mL/kg) onartuzumab-treated patients were ATA positive at one or max min ss

1/2 more assessments, all ATA responses were low, did not seem V NOTE: The data were groupedAbbreviations: on AUC the basis of the dose level and study stage. AUC C t CL (mL/d/kg) 15.0 Table 3. Number of patientsC 3 6 3 13 3 6 3 6 Parameter (mean reportable; t to affect pharmacokinetics, were not associated with AEs,

www.aacrjournals.org Clin Cancer Res; 20(6) March 15, 2014 OF7

Salgia et al.

Figure 3. Waterfall plot showing the best response (as assessed by RECIST) to onartuzumab administered by intravenous infusion once every 3 weeks as a single agent (1 to 30 mg/kg) or in combination (10 and 15 mg/kg) with bevacizumab (15 mg/kg). SLD, sum of longest diameters. Highlighted patient (red dotted line) had a complete response. , patient discontinued study treatment early—subject's/legal guardian's decision to withdraw. , patient discontinued study treatment early—disease progression.

and were found to be primarily toward the nonengineered drug, which has slower clearance than shed MET alone framework of onartuzumab. and could thereby result in an increase in serum-shed As this phase I study was the first examination of MET concentrations. Because shed MET levels can onartuzumab administrationinhumans,evidenceof increase in circulation, levels may not necessarily reflect target engagement in humans was evaluated. On-treat- true peripheral receptor saturation; therefore, pharmaco- ment biopsies were unavailable to assess pharmacody- kinetic/pharmacodynamic modeling from preclinical namic effects in the tumor directly. However, because the xenograft models was used to support determination of extracellular domain of the MET receptor is shed into the phase II dose. We also previously examined circulat- circulation and serum was available at multiple time ing HGF levels in phase I patients as a biomarker of MET points, we evaluated soluble shed MET levels as a measure inhibition for onartuzumab, and observed an increased of target engagement and peripheral receptor occupancy level following onartuzumab treatment, which was inde- (24). It is noteworthy that shed MET levels can increase in pendent of dose, but reflected the pharmacodynamic circulation as a result of complex formation with the activity of the drug (16). One patient with likely autocrine HGF/MET-driven gastric cancer had an objective CR to onartuzumab (20 mg/kg) A B during phase Ia. The activity observed supports the pre- sumed biology and suggests there may be certain disease c-MET IHC HGF IHC states in which single-agent onartuzumab could provide clinical benefit. Further studies will be required to better characterize the efficacy of onartuzumab in populations with autocrine MET, high MET, amplified MET, or even mutated MET disease (19), as well as to elucidate possible mechanisms of resistance to onartuzumab therapy, such as KRAS or RON amplification (25, 26). In conclusion, this phase I study shows onartuzumab to be generally well tolerated when administered both alone and in combination with bevacizumab. The MTD was not reached with either single agent or combination dose escala- tions. Patients received the RP2D of 15 mg/kg Q3W without

Figure 4. IHC staining for intratumoral MET (A) and HGF (B) protein evidence of DLTs. A number of phase II studies (Clinical- expression in the patient with a complete response. HGF, hepatocyte Trials.gov: NCT01632228; NCT01590719; NCT01519804; growth factor; IHC, immunohistochemistry. NCT01418222; NCT01186991; and NCT01496742) and

OF8 Clin Cancer Res; 20(6) March 15, 2014 Clinical Cancer Research

Phase I Dose-Escalation Study of Onartuzumab

phase III studies (ClinicalTrials.gov: NCT01662869 and Writing, review, and/or revision of the manuscript: P. Patel, J. Bothos, W. Yu, S. Eppler, P. Hegde, S. Bai, S. Kaur, I. Nijem, D.V.T. Catenacci, A. NCT01456325) are ongoing to evaluate the efﬁcacy and Peterson, M.J. Ratain, B. Polite, J.M. Mehnert, R.A. Moss safety of onartuzumab in patients with a variety of advanced Administrative, technical, or material support (i.e., reporting or orga- solid tumors. nizing data, constructing databases): J. Bothos, W. Yu, D.V.T. Catenacci, A. Peterson, M.J. Ratain, R.A. Moss Study supervision: P. Patel, J. Bothos, A. Peterson, J.M. Mehnert, R.A. Moss Disclosure of Potential Conﬂicts of Interest S. Eppler has ownership interest in a patent. P. Hegde is employed as a Acknowledgments Senior Scientist with Genentech Inc. I. Nijem has ownership interest in The authors thank David Geary, the study coordinator, as well as former, company stock. D.V.T. Catenacci is a consultant or advisory board member Yan Xin, and current Roche employees, Denise Jin, for their work on the and has honoraria from the speakers bureau of Genentech. M.J. Ratain is a pharmacokinetics analysis. The authors from the Rutgers Cancer Institute of consultant or advisory board member of Genentech. B. Polite has honoraria New Jersey also thank Antoinette R. Tan, MD, for guidance in conducting this from the speakers bureau of Bayer and is a consultant or advisory board study and Kiomi Harris Addo for protocol management. Support for third- member of GLC. party writing assistance for this article was provided by Genentech Inc.

Authors' Contributions Grant Support Conception and design: R. Salgia, P. Patel, J. Bothos, W. Yu, S. Eppler, The study was funded by Genentech Inc. and sponsored by F. Hoffmann- S. Kaur, A. Peterson, M.J. Ratain La Roche Ltd. Financial support for third-party writing assistance for this Development of methodology: J. Bothos, W. Yu, S. Eppler, S. Kaur, article was provided by F. Hoffmann-La Roche Ltd. I. Nijem The costs of publication of this article were defrayed in part by the payment Acquisition of data (provided animals, acquired and managed patients, of page charges. This article must therefore be hereby marked advertisement in provided facilities, etc.): R. Salgia, P. Patel, P. Hegde, S. Bai, D.V.T. accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Catenacci, A. Peterson, M.J. Ratain, B. Polite, J.M. Mehnert, R.A. Moss Analysis and interpretation of data (e.g., statistical analysis, biosta- tistics, computational analysis): R. Salgia, P. Patel, J. Bothos, W. Yu, Received July 26, 2013; revised December 5, 2013; accepted January 4, S. Eppler, P. Hegde, S. Bai, I. Nijem, A. Peterson, M.J. Ratain, R.A. Moss 2014; published OnlineFirst February 3, 2014.

References 1. Zhang YW, Vande Woude GF. HGF/SF-met signaling in the control of tional activation of the met protooncogene. Cancer Cell 2003;3: branching morphogenesis and invasion. J Cell Biochem 2003;88: 347–61. 40–17. 13. Therasse P, Arbuck SG, Eisenhauer EA, Wanders J, Kaplan RS, 2. Lutterbach B, Zeng Q, Davis LJ, Hatch H, Hang G, Kohl NE, et al. Lung Rubinstein L, et al. New guidelines to evaluate the response to cancer cell lines harboring MET gene amplification are dependent on treatment in solid tumors. European Organization for Research and Met for growth and survival. Cancer Res 2007;67:2081–8. Treatment of Cancer, National Cancer Institute of the United States, 3. Birchmeier C, Birchmeier W, Gherardi E, Vande Woude GF. Met, National Cancer Institute of Canada. J Natl Cancer Inst 2000;92: metastasis, motility and more. Nat Rev Mol Cell Biol 2003;4: 205–16. 915–25. 14. Lu JF, Bruno R, Eppler S, Novotny W, Lum B, Gaudreault J. Clinical 4. Trusolino L, Bertotti A, Comoglio PM. MET signalling: principles and pharmacokinetics of bevacizumab in patients with solid tumors. functions in development, organ regeneration and cancer. Nat Rev Mol Cancer Chemother Pharmacol 2008;62:779–86. Cell Biol 2010;11:834–48. 15. Xiang H, Bender BC, Reyes AE, Merchant M, Jumbe NL, Romero M, 5. Cao B, Su Y, Oskarsson M, Zhao P, Kort EJ, Fisher RJ, et al. Neu- et al. Onartuzumab (MetMAb): using nonclinical pharmacokinetic and tralizing monoclonal antibodies to hepatocyte growth factor/scatter concentration-effect data to support clinical development. Clin Cancer factor (HGF/SF) display antitumor activity in animal models. Proc Natl Res 2013;19:5068–78. Acad Sci U S A 2001;98:7443–8. 16. Penuel E, Li C, Parab V, Burton L, Cowan KJ, Merchant M, et al. HGF as 6. Martens T, Schmidt NO, Eckerich C, Fillbrandt R, Merchant M, Schwall a circulating biomarker of onartuzumab treatment in patients with R, et al. A novel one-armed anti-c-Met antibody inhibits glioblastoma advanced solid tumors. Mol Cancer Ther 2013;12:1120–30. growth in vivo. Clin Cancer Res 2006;12:6144–52. 17. Dirks NL, Meibohm B. Population pharmacokinetics of therapeutic 7. Puri N, Khramtsov A, Ahmed S, Nallasura V, Hetzel JT, Jagadeeswaran monoclonal antibodies. Clin Pharmacokinet 2010;49:633–59. R, et al. A selective small molecule inhibitor of c-Met, PHA665752, 18. Prat M, Crepaldi T, Gardino L, Giordano S, Longati P, Comoglio P. C- inhibits tumorigenicity and angiogenesis in mouse lung cancer xeno- terminal truncated forms of Met, the hepatocyte growth factor recep- grafts. Cancer Res 2007;67:3529–34. tor. Mol Cell Biol 1991;11:5954–62. 8. Jin H, Yang R, Zheng Z, Romero M, Ross J, Bou-Reslan H, et al. 19. Catenacci DV, Henderson L, Xiao SY, Patel P, Yauch RL, Hegde P, MetMAb, the one-armed 5D5 anti-c-Met antibody, inhibits orthotopic et al. Durable complete response of metastatic gastric cancer with pancreatic tumor growth and improves survival. Cancer Res 2008; anti-MET therapy followed by resistance at recurrence. Cancer Dis- 68:4360–8. covery 2011;1:573–79. 9. Prat M, Crepaldi T, Pennacchietti S, Bussolino F, Comoglio PM. 20. Wen PY, Schiff D, Cloughesy TF, Raizer JJ, Laterra J, Smitt M, et al. A Agonistic monoclonal antibodies against the Met receptor dissect the phase II study evaluating the efficacy and safety of AMG 102 (rilotu- biological responses to HGF. J Cell Sci 1998;111:237–47. mumab) in patients with recurrent glioblastoma. Neuro Oncol 10. Merchant M, Zhang Y, Su Y, Romero MS, Severin C, Zheng Z, 2011;13:437–46. et al.. MetMAb significantly enhances anti-tumor activity of anti- 21. Ramanathan RK, Payumo FC, Papdopoulos KP, Weiss GJ, Chambers VEGF and/or erlotinib in several animal tumor models. Presented G, Iyengar T, et al. A phase 1, first in human, study of SCH 900105, an at AACR-NCI-EORTC International Conference 2008. Abstract antihepatocyte growth factor (HGF) monoclonal antibody, in subjects 556. with advanced solid tumors. Mol Cancer Ther 2009;8 (suppl 1; abstract 11. Zhang YW, Su Y, Volpert OV, Vande Woude GF. Hepatocyte growth A100). factor/scatter factor mediates angiogenesis through positive VEGF 22. Nakamura T, Mizuno S. The discovery of hepatocyte growth factor and negative thrombospondin 1 regulation. Proc Natl Acad Sci U S A (HGF) and its significance for cell biology, life sciences and clinical 2003;100:12718–23. medicine. Proc Jpn Acad Ser B Phys Biol Sci 2010;86:588–610. 12. Pennacchietti S, Michieli P, Galluzzo M, Mazzone M, Giordano S, 23. Kurzrock R, Sherman SI, Ball DW, Forastiere AA, Cohen RB, Mehra R, Comoglio PM. Hypoxia promotes invasive growth by transcrip- et al. Activity of XL184 (Cabozantinib), an oral tyrosine kinase inhibitor,

www.aacrjournals.org Clin Cancer Res; 20(6) March 15, 2014 OF9

Salgia et al.

in patients with medullary thyroid cancer. J Clin Oncol 2011;29: 25. Catenacci DV, Cervantes G, Yala S, Nelson EA, El-Hashani E, Kanteti 2660–6. R, et al. RON (MST1R) is a novel prognostic marker and therapeutic 24. Mai E, Zheng Z, Chen Y, Peng J, Severin C, Filvaroff E, et al. Non- target for gastroesophageal adenocarcinoma. Cancer Biol Ther clinical evaluation of the serum pharmacodynamic biomarkers HGF 2011;12:9–46. and shed MET following dosing with the anti-MET monovalent mono- 26. Cepero V, Sierra JR, Corso S, Ghiso E, Casorzo L, Perera T, et al. MET clonal antibody, onartuzumab. Mol Cancer Ther. 2013 Nov 20. [Epub and KRAS gene ampliﬁcation mediates acquired resistance to MET ahead of print]. tyrosine kinase inhibitors. Cancer Res 2010;70:7580–90.

OF10 Clin Cancer Res; 20(6) March 15, 2014 Clinical Cancer Research

Phase I Dose-Escalation Study of Onartuzumab as a Single Agent and in Combination with Bevacizumab in Patients with Advanced Solid Malignancies

Ravi Salgia, Premal Patel, John Bothos, et al.

Clin Cancer Res Published OnlineFirst February 3, 2014.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-13-2070

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2014/02/04/1078-0432.CCR-13-2070.DC1

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/early/2014/02/26/1078-0432.CCR-13-2070. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.