A Phase Ib/II Study of Ramucirumab in Combination with Emibetuzumab in Patients with Advanced Cancer

This article is featured in Highlights of This Issue, p. 5177

VEGF and its receptor, VEGFR-2, have critical roles in the physiologic growth and maintenance of blood vessels (1). High levels of VEGF and other angiogenic factors are often associated with an increased level of tumor microvessel density, aggressive pathologic features, and poor clinical outcomes in a variety of cancer histology. In preclinical models, blockade of the VEGF/VEGFR-2 axis leads to tumor vascular remodeling and cancer cell death. Pathologic dysregulation of angiogenesis is an established hallmark of malignancy, and both VEGFR-2 and its ligand are bona fide therapeutic targets either with selective monoclonal antibodies (mAbs) or via multi-targeted tyrosine kinase inhibitors (TKI), in a variety of tumor types, including non–small lung cancer (NSCLC), ovarian cancer, renal cell carcinoma (RCC), and several gastrointestinal malignancies such as colorectal cancer (CRC), gastric or gastroesophageal junction adenocarcinoma (GEJ), and hepatocellular carcinoma (HCC; refs. 2–5).

Innate and acquired resistance to antiangiogenic therapy is frequent and multi-factorial, but is in part mediated by the hepatocyte growth factor (HGF) receptor MET (6–8). Both ligand-dependent and -independent MET activation is implicated in tumor cell motility, proliferation, survival, invasion, metastasis, and angiogenesis. Indeed, the MET receptor is overexpressed, activated, amplified, or mutated in a wide variety of solid tumors (8–10), and contributes to an aggressive tumor biology (11, 12). Importantly, VEGF blockade leads to a reciprocal increase in MET concentration and activation (7), which in turn leads to MET-mediated antiangiogenic escape (13). Simultaneous pharmacologic inhibition of MET and VEGF signaling mitigates MET-mediated escape, thereby slowing tumor growth and reducing invasion and metastasis (13). Thus, there is compelling preclinical evidence to support cotargeting VEGF and MET pathways in malignances to enhance the activity of VEGF/VEGFR-2 monotherapy. Clinically, the concept MET as a mechanism of antiangiogenic resistance is also suggested in several correlative programs across multiple tumor types indicating that unopposed VEGF inhibition, either with mAbs or TKIs, results in HGF/MET overexpression on postprogression biopsies (13–16).

Ramucirumab, a human IgG1 VEGFR-2–targeting antibody, exhibits antitumor activity and when administered as monotherapy, or in combination, improves both progression-free and overall survival over placebo in patients with advanced gastric/GEJ cancers, alpha-fetoprotein (AFP)-high HCC, and NSCLC (17–20). Given the above data on the importance of MET as a bypass pathway circumventing VEGF/VEGFR-2 inhibition, it is reasonable to test ramucirumab in combination with MET-targeting agents in solid tumors.

Emibetuzumab (LY2875358) is a humanized IgG4 bivalent mAb that prevents HGF from binding to the extracellular domain of MET, thereby inhibiting ligand-dependent MET signaling. Emibetuzumab triggers MET receptor internalization and the degradation of total membrane MET expression, leading to inhibition of ligand-independent activation of MET signaling (21). The agent does not elicit antibody-dependent cellular cytotoxicity. In preclinical models, emibetuzumab inhibits growth of MET-dependent tumors and when combined with a murine surrogate antibody of ramucirumab in a xenograft model leads to additive antitumor activity (Supplementary Fig. S1). Clinically, emibetuzumab has shown to be safe and tolerable in patients with cancer with limited antitumor activity as monotherapy in clinical studies for unselected patients with cancer (22,23). In prior studies, a MTD was not established due to a favorable safety profile and the recommended phase II dose (RP2D) range of emibetuzumab was established at 700–2,000 mg every 2 weeks (22).

Herein we report results from a phase Ib/II dose-escalation trial of ramucirumab in combination with emibetuzumab in patients with advanced or metastatic cancers, followed by tumor-specific expansion cohorts in patients with gastric or GEJ adenocarcinoma, RCC, HCC, or NSCLC. The primary objectives of this study were to identify a RP2D range for emibetuzumab when given in combination with ramucirumab, and to explore the antitumor activity of this combination treatment in tumor types with established dependence on VEGF/VEGFR signaling.

This was a multicenter, nonrandomized, open-label phase Ib/II study of ramucirumab in combination with emibetuzumab in two parts (clinicaltrials.gov NCT02082210; Fig. 1). In part A, a standard 3+3 dose escalation design was employed with an increasing dose of emibetuzumab in combination with a fixed FDA-approved dose of ramucirumab in patients with advanced and/or metastatic cancer. After completion of dose escalation, additional patients were enrolled in part B in tumor-specific expansion cohorts for gastric or GEJ adenocarcinoma, HCC, RCC, or NSCLC for dose confirmation and exploration of clinical activity.

The primary objectives of the study were to determine the RP2D range for emibetuzumab that can be safely administered in combination with ramucirumab and to evaluate the preliminary antitumor activity of the combination in tumor-specific expansion cohorts. Secondary objectives included safety, tolerability, pharmacokinetics, and immunogenicity. Exploratory objectives included evaluation of tumoral MET expression and exploration of potential associations with observed antitumor activity.

The protocol was approved by institutional review boards before patient recruitment, and each patient provided written informed consent before enrollment. The study was conducted in accordance with the International Conference on Harmonization E6 Guidelines for Good Clinical Practice.

Eligible patients were ≥18 years of age with a confirmed diagnosis of advanced and/or metastatic cancer after failure of standard-of-care therapy(s), or for whom there was no standard therapy, or for whom standard therapy would not be appropriate (i.e., in the event of patient refusal). Additional key eligibility criteria included adequate hematologic, renal and hepatic function (Child Pugh A), as well as an Eastern Cooperative Oncology Group (ECOG) performance status of ≤2 in part A and ≤1 in part B. Patients had measurable disease as defined by RECIST1.1 (24). All patients were required to have a pretreatment tumoral sample from the primary tumor or a metastasis obtained prior to study treatment and after progression on (or discontinuation from) the most recent line of systemic tumor therapy and within at least 6 months prior to initiation of study treatment.

Patients were excluded if they had received any previous cancer therapy within 21 days or 5 half-lives prior to study enrollment (whichever was shorter) and had not recovered from the acute effects of prior treatment-related toxicities. Additional exclusion criteria included history of hypertensive crisis, poorly controlled hypertension (i.e., a history of hypertensive crisis or hypertensive encephalopathy or blood pressure systolic ≥ 150 and/or diastolic ≥ 95 despite medical management), significant venous or arterial thromboembolic events (i.e., any arterial thromboembolic event, including myocardial infarction, unstable angina pectoris, cerebrovascular accident, or transient ischemic attack within 6 months of study drug; patients with a history of deep venous thrombosis, pulmonary embolism, or any other significant venous thromboembolic event during a 3-month period prior to enrollment, chronic tumor portal venous thrombosis associated with HCC was not an exclusion criterion), and uncontrolled central nervous system metastasis. Major blood vessel invasion and evidence of intratumoral cavitation were exclusionary for patients with NSCLC.

Patients participating in part A (dose escalation) received emibetuzumab at doses of 750 mg (dose level 1) and 2,000 mg (dose level 2) as an intravenous infusion, in combination with an 8 mg/kg i.v. dose of ramucirumab, every 2 weeks over a 28-day cycle. The starting dose and the dose range of emibetuzumab were selected on the basis of clinical data and pharmacokinetic analyses from the phase I study, which indicated saturation of receptor-mediated clearance at dose levels at and above 700 mg every 2 weeks (22). No intrapatient dose escalations were allowed. In part B, all eligible patients received emibetuzumab 750 mg and ramucirumab 8 mg/kg i.v. every 2 weeks.

Safety and tolerability were assessed through clinical and laboratory evaluations at weekly intervals for the first two cycles and every 2 weeks thereafter. Adverse events (AE) were graded according to the Common Terminology Criteria for Adverse Events (CTCAE v.4.0) and were recorded for all patients who received at least one dose of study treatment. Dose-limiting toxicities (DLT) were defined as possibly drug-related AEs during cycle 1 if they met one of the following criteria: ≥Grade 3 nonhematologic toxicity (except for Grade 3 nausea, vomiting, diarrhea, constipation, fatigue, or anorexia for ≤3 days or Grade 3 hypertension which is controlled within ≤7 days), Grade 4 hematologic toxicity of ≥7 days duration, febrile neutropenia, or any other significant toxicity.

Tumor response was assessed by CT scans or MRI according to RECIST1.1 (24) at baseline and thereafter every 6 weeks until radiographic documentation of progressive disease for the first nine cycles. Thereafter, patients had tumor assessment every 8–12 weeks as clinically indicated. All patients receiving at least one dose of study drugs were included in the evaluation of antitumor activity.

Serial serum samples for bioanalysis of ramucirumab and emibetuzumab were obtained at scheduled times around their sequential infusions: prior to infusion, at the end of infusion, and at 3, 5, 8, 21, and 168 hours post-emibetuzumab (latter) infusion. Additional samples were collected on day 15, pre- and post-ramucirumab infusion and pre-emibetuzumab infusion (nominally 334, 335, and 336 hours after the first emibetuzumab dose on day 1). Serum concentrations of both ramucirumab and emibetuzumab were measured using validated ELISAs as described previously (22, 25). In patients with complete pharmacokinetic sampling after the first infusion, pharmacokinetic parameters for emibetuzumab were calculated by standard noncompartmental methods using Phoenix WinNonlin Version 6.4 (Certara L.P). Ramucirumab concentrations were descriptively overlaid with simulations from a previously established ramucirumab population pharmacokinetic model using the nonlinear mixed effects program NONMEM Version 7.4.2 (ICON Development Solutions) to determine the potential effect of emibetuzumab coadministration on ramucirumab pharmacokinetics.

Patient samples for immunogenicity assessment drawn during this study were analyzed for the presence of antidrug antibodies (ADA). The formation of ADA was assessed using a validated ELISA, following a 4-tier approach (26). The ADA screening assay was validated in accordance with the FDA Guidance for Industry: Assay Development for Immunogenicity Testing of Therapeutic Proteins (FDA 2016; ref. 27).

Tumor biopsies were tested for MET expression by IHC using the Dako MET 2 pharmDxTM kit, an exploratory kit employing the A2H2-3 MET antibody clone (28). A composite scoring system was devised to determine the status of MET by IHC, enumerating the percentage of tumor cells with immunoreactivity of 0, 1+, 2+, or 3+ staining intensity in the cell membrane as described previously (28). The analysis was performed by a trained pathologist in a blinded fashion.

Locally generated next-generation sequencing (NGS) data of tumor samples from patients with RECIST responses using platforms compliant with Clinical Laboratory Improvement Amendments regulations were provided as available (29).

The coprimary objective antitumor activity was measured by the overall response rate (ORR), which was defined as the proportion of patients with confirmed complete response (CR) and partial response (PR). The disease control rate (DCR) was calculated as the proportion of patients with confirmed CR, PR, and stable disease (SD). Progression-free survival (PFS) is defined as the time from the date of first dose to the date of objective disease progression or death, whichever was earlier. Median PFS was estimated using the Kaplan–Meier method. Efficacy and safety analyses were performed on patients who had received at least one dose of study treatment.

The sample size of approximately 15 patients per expansion cohort was selected to allow adequate confirmation of safety and tolerability of emibetuzumab in combination with ramucirumab, and to identify evidence of preliminary clinical activity worthy of further investigation in phase II, analogous to the first stage of a Simon two-stage design. In case of significant clinical activity in any of the tumor-specific expansion cohorts based on response rate (i.e., at least 1 responder was observed from the initial 15 patients in gastric and RCC cohorts, and at least 2 responders were observed in HCC and NSCLC cohorts) or disease control rate relative to previously reported landmark studies, the protocol allowed for enrollment of an additional 30 patients, for a total of approximately 45 patients per cohort to further characterize clinical activity of the combination treatment. A sample size of 45 provided a weighted average power of 83% based on a two-sided type 1 error rate of 5% to detect a significant dependence between biomarker status and clinical activity. To better understand the relevance of MET expression as a potential biomarker for emibetuzumab in combination with ramucirumab, the antitumor activity endpoints were retrospectively assessed with respect to different MET expression cut-off points as a planned exploratory analysis.

A total of 97 patients were enrolled and received at least one dose of study drugs. This comprised 6 patients in the dose-escalation part (3 at each of the two dose levels) and 91 patients in the four tumor-specific expansion cohorts including 16 patients with GEJ, 45 patients with HCC, 15 patients with RCC, and 15 patients with NSCLC. The median number of cycles of emibetuzumab and ramucirumab completed was three (range 1–14 cycles) with relative median dose intensity of 100% and 98%, respectively. As of the data cut-off, all patients had discontinued from study treatment. The most common reason for study treatment discontinuation was progressive disease (85%), followed by withdrawal by subject (4%; Supplementary Table S1).

Patient baseline characteristics are summarized in Table 1. The majority of patients were Caucasian (83%), male (72%), and had a baseline ECOG performance status of 1 (61%). The median number of prior systemic therapies was 2, and ranged from 0 to 7. For the HCC cohort, all patients had a Child Pugh A score with nonvirally mediated HCC predominating (60%), and 18 patients (40%) had AFP > 400 ng/mL. In total, 37 (82%) of 45 patients with HCC failed prior sorafenib treatment, whereas 5 (11%) were treatment naïve and 3 (4%) received prior therapies other than sorafenib (Supplementary Table S2).

Combination treatment with ramucirumab and emibetuzumab was well tolerated. No DLTs or DLT-equivalent toxicities were reported up to the maximum studied dose level of 2,000 mg emibetuzumab in combination with ramucirumab. In total, 82 of the 97 (85%) patients experienced at least one AE possibly related to study drugs. Most of these AEs were of mild or moderate intensity with no observed dose relationship to emibetuzumab at the doses administered. Among AEs possibly related to treatment, the most frequently reported (≥10% of patients) included fatigue (36%), peripheral edema (29%), and nausea (14%; Table 2). Twenty patients (21%) were reported to have possibly treatment-related Grade ≥ 3 AEs. Grade ≥ 3 events reported for more than 2 patients were pulmonary embolism [4 patients (4%)] and fatigue [3 patients (3%)]. Mild or moderate infusion-related reactions (IRR) were reported for 6 patients (6%), with symptoms manageable by standard-of-care treatment or spontaneous resolution. Except for 1 patient who had an IRR following the ramucirumab infusion but prior to emibetuzumab dosing, all IRRs occurred after the administration of both ramucirumab and emibetuzumab. This precluded causality assessment for either study treatment.

There were 9 deaths reported while patients were on study treatment or within 30 days after discontinuing study treatment. These deaths were either due to disease (n = 7) or adverse events (2 cases of pneumonia). None of the deaths were related to study treatment.

Supplementary Table S3 summarizes emibetuzumab noncompartmental pharmacokinetic parameters after the first dose, calculated from patients with complete sampling after the first infusions. While no formal dose proportionality assessments were performed in this study, emibetuzumab clearance is known to be linear in the dose range investigated, and was approximately dose proportional between the 750 and 2,000-mg dose levels. Estimated half-life was approximately 7 days in both treatment groups. These results were also comparable with those from previous studies in which patients were treated with emibetuzumab alone (22, 23), suggesting that coadministration of ramucirumab does not appreciably impact the pharmacokinetic of emibetuzumab. Likewise, an overlay plot comparing observed ramucirumab concentrations in this study with simulated values obtained from a ramucirumab population pharmacokinetic model fit to patients treated with ramucirumab alone (not shown) suggests that coadministration of emibetuzumab has no appreciable effect on the concentration–time profile of ramucirumab.

Immunogenicity samples available for evaluation of ADA were analyzed for the presence of anti-emibetuzumab antibodies from 93 patients and anti-ramucirumab antibodies from 96 patients. None of these patients developed treatment-emergent ADA against ramucirumab or emibetuzumab. Ten (11%) and 12 (13%) patients had detectable ADA against emibetuzumab or ramucirumab, respectively, detected prior to dosing or at any time on treatment. For those detected while on treatment, titers did not reach the prespecified criteria for treatment-emergent ADA.

Across all cohorts, 87 of 97 patients (90%) receiving ≥1 dose of study drugs were evaluable for tumor response assessment by RECIST1.1. Overall, there were 5 confirmed PRs (1 gastric/GEJ, 1 NSCLC, and 3 HCC), and no CR observed, for an ORR of 5.2% [95% confidence interval (CI), 0–10] across the entire population (Table 3). Four of the 5 responders were previously exposed to anti-VEGF treatment (3 HCC: sorafenib; gastric: ramucirumab). An additional 55 patients (57%) exhibited SD as best response to therapy for a DCR of 62% (95% CI, 50–70). Overall, a decrease in sum of target lesions relative to baseline was observed in a total of 31 patients (32%; Fig. 2A). The median PFS for patients enrolled in the tumor-specific expansion cohorts ranged between 1.6 months for gastric cancer (95% CI, 1.4–4.6) and 6.6 months for patients with NSCLC (95% CI, 2.9–9.7). Tumor response and PFS for each of the tumor-specific expansion cohorts are presented in Table 3.

Neither the RCC cohort nor the NSCLC cohorts met criteria to expand based on a prior assumption of antitumor activity. Although 1 PR was observed in the gastric/GEJ cohort, and therefore met the predetermined boundary to expand, the principal investigators and the sponsor chose not to explore this given the short lived nature of this PR (PFS of 1.6 months) and emerging contemporary clinical data indicating that MET inhibition might not be effective in this disease type (30). Among the initial 15 patients in the HCC expansion cohort, 2 of 15 patients (13%) had a confirmed PR and 9 of 11 evaluable patients showed a reduction in AFP. The HCC cohort was therefore expanded to enroll an additional 30 patients to further characterize the antitumor activity of emibetuzumab and ramucirumab. The ORR across the 45 total patients with HCC finally enrolled was 6.7%, with a DCR of 60% (95% CI, 0.4–0.7). The median PFS in these patients was 5.4 months (95% CI, 1.6–8.1; Fig. 2B).

Of the 97 patients enrolled, 73 had sufficient tumor material for MET expression analysis. In the overall patient population, patient with MET expression of ≥2+ staining intensity in ≥50% of their tumor cells had a PFS of 7.4 months (n = 41) versus a PFS of 2.8 months in patients with MET expression below this cut-off (n = 32; HR, 0.48; 90% CI, 0.29−0.81; Fig. 3).

When assessing this biomarker separately by treatment effect for each of the tumor-specific expansion cohorts, no obvious trend toward longer PFS was observed in patients with gastric cancer, RCC, or NSCLC with MET expression of ≥2+ in ≥50% of tumor cells receiving emibetuzumab in combination with ramucirumab (Supplementary Fig. S2). However, patients with HCC with MET expression of ≥2+ in ≥50% of tumor cells had a median PFS of 8.1 months (n = 19) relative to patients below this MET expression cut-off, who had a median PFS of 2.8 months (n = 14; HR, 0.22; 90% CI, 0.08–0.59; Supplementary Fig. S2B). Further exploratory analysis of different MET expression cut-offs (e.g., MET ≥2+ in ≥80% or ≥30%) did not provide any stronger association of MET status and PFS for the treatment effect of emibetuzumab and ramucirumab (Supplementary Fig. S3A and S3B). This trend for a MET biomarker by treatment effect in patients with HCC was also observed for ORR, with 2 PRs observed out of 16 evaluable patients for tumor response with MET ≥2+ in ≥50% of tumor cells, and no PRs noted in 9 patients with MET status below this cut-off (Supplementary Fig. S3C and S3D). None of the patients with a PR and NGS data available had a MET genomic amplification or alterations associated with MET sensitivity (Supplementary Table S4).

This phase Ib/II study determined the RP2D range of emibetuzumab when given in combination with ramucirumab and explored antitumor activity across select solid tumors. In the dose-escalation phase, the safety profile of emibetuzumab in patients with different solid tumors was consistent across both dose levels with no new safety signals observed for the combination treatment. The combination was tolerable with no DLTs or DLT-equivalent toxicities outside the DLT assessment period during treatment. A MTD for the combination was not established, and there was no apparent emibetuzumab dose-dependent effect on antitumor activity. Furthermore, pharmacokinetic assessments (noncompartmental/descriptive analyses and population modeling) indicated that saturation of target–receptor occupancy occurs at emibetuzumab exposures associated with dose levels > 210 mg every 2 weeks. Population pharmacokinetic model simulations demonstrated that at doses of ≥700 mg emibetuzumab every 2 weeks, 100% of the population was predicted to have a minimum plasma concentration at steady state (C_min,ss) of ≥50 mg/mL, which is the emibetuzumab C_min,ss associated with ≥90% tumor growth inhibition in xenograft models (22). Consistent with monotherapy, no remarkable differences in emibetuzumab pharmacokinetic profiles were observed among the patients with different tumor types when emibetuzumab was coadministered with ramucirumab, nor were there any signs of drug–drug interactions between the two antibodies when administered in combination. Thus, given the totality of safety, pharmacokinetic, and preliminary efficacy data, the decision was made to select emibetuzumab at 750 mg every 2 weeks as the RP2D for solid tumor cohort expansions.

The most common adverse events included fatigue, peripheral edema, and nausea, and the majority of events were of mild to moderate intensity (Grade 1 or 2). Not surprisingly, the overall safety profile of emibetuzumab and ramucirumab differed slightly based on tumor type. For example, edema, hypoalbuminemia, and thrombocytopenia predominated in the HCC cohort, whereas nausea and vomiting were more common in the GEJ cohort. Nevertheless, the rates of Grade ≥3 AEs were comparable in patients with different types of cancers.

Ramucirumab combined with emibetuzumab exhibited variable antitumor activity across a spectrum of solid tumors. Five confirmed PRs according to RECIST1.1 were reported and the disease control rate of 62% reflects the cytostatic mechanism of action expected for this combination. The clinical trial design allowed to explore activity of this combination in several tumor types that are known to be dependent on angiogenic pathways. No compelling antitumor activity was observed in NSCLC, gastric/GEJ, or RCC in the initial expansion cohorts. Regarding RCC, this was somewhat surprising, especially given the favorable data observed in patients with advanced RCC who were treated with cabozantinib, a multi-targeted TKI that blocks both VEGFR-2 and MET (31), as well as several other key signaling proteins. Our novel approach of selectively inhibiting only MET and VEGFR-2 by mAbs suggests that perhaps other cabozantinib targets, such as AXL, or the relatively nonselective mechanism of action of the agent may be required for clinical benefit in RCC. Furthermore, the METEOR study indicates that tumoral MET expression is not predictive or prognostic in cabozantinib-treated patients with RCC, suggesting that the summation of multiple inhibitor signals might be required for antitumor activity (32).

Indeed, only the HCC expansion cohort demonstrated meaningful antitumor activity. Acknowledging the hazards of cross-trial comparison and the observation that 11% of the HCC cohort was sorafenib naïve, the objective response rate as well as PFS observed in unselected patients with HCC herein is comparable with that seen with multi-targeted tyrosine kinase anti-angiogenesis inhibitors in the second-line HCC setting (33, 34). Our preliminary clinical data is also in line with preclinical and emerging clinical data, supporting the notion that concurrent MET inhibition may delay antiangiogenic resistance (10, 12). Single-agent ramucirumab led to a PFS of 2.8 months in two recent phase III studies (20, 35), while the addition of MET blockade in our study yielded a favorable PFS of 5.4 months in a similar advanced HCC patient population. Furthermore, MET expression has been shown to increase in patients with HCC following treatment with antiangiogenic agents (15), and recent positive phase III data of cabozantinib further bolster the biological rationale in this study of cotargeting MET signaling and the VEGF/VEGFR-2 axis by emibetuzumab and ramucirumab, respectively (33). Our data underscore the importance of MET and VEGFR in HCC biology and highlight these two pathways as critical targets in HCC. Although further exploration of the combination of ramucirumab and emibetuzumab in HCC appears warranted on the basis of efficacy and a favorable safety profile, the potential financial cost associated with such a regimen, the logistical concerns associated with biweekly intravenous administrations, and the emerging landscape of oral TKIs and immunotherapy for patients with HCC, limit the future potential of a ramucirumab plus emibetuzumab combination for HCC (36, 37).

A critical question for the field is whether selecting for MET-high tumors will help select patients who will benefit from VEGFR-MET–targeted treatments. Acknowledging the small sample size of the study, our exploratory data indicate that determination of MET status by IHC might not enrich for clinical activity of emibetuzumab in combination with ramucirumab for tumor types such as gastric/GEJ, RCC, and NSCLC. In accordance with the available literature (23, 30, 38, 39), it may be that genomic MET alterations (e.g., MET amplification, MET exon 14 variants in NSCLC, or MET kinase domain mutations) are requisite for response to MET-targeted therapeutics in these cancer types. The observation that 4 out of the 5 patients with a PR in this study had no such genomic alterations suggests that functional methods such as RNA sequencing or improved methods of quantifying active MET may be required to help select patients with such tumor types who are apt to respond. In HCC, a randomized, prospective phase III study restricting enrollment to high MET expression (IHC > 2+, 50% cells) did not show efficacy with the low potency, nonselective MET inhibitor tivantinib (16). Our data are informative, suggesting that MET expression by IHC may have a predictive role for certain combination strategies, such as selective inhibitors of MET and VEGFR-2. Indeed, our exploratory data indicate that patients with HCC with higher tumoral MET expression at baseline had an approximately 3-fold longer PFS than patients with lower MET expression when treated with ramucirumab and emibetuzumab (8.1 vs. 2.8 months; HR, 0.22).

A limitation of this study is that the single-arm design clearly restricts the ability to evaluate the impact of prognostic patient baseline characteristics contributing to the treatment effects observed for the combination treatment. We are also unable to differentiate between the prognostic and the predictive effect of MET expression within this single-arm study. While in HCC MET expression by IHC has been shown to be a negative prognostic biomarker in phase III with a median PFS of 2.0 months using the same cut-off as that in our study (15), a randomized, controlled study stratifying for MET expression would be required to adequately differentiate between prognostic and predictive effects. Despite these shortcomings, we show that the combination of emibetuzumab and ramucirumab is safe and tolerable with favorable antitumor efficacy, particularly in patients with advanced HCC. Further evaluation of biomarkers to identify patients who may benefit from treatment with emibetuzumab in combination with ramucirumab is warranted.

J.J. Harding reports receiving commercial research grants from Bristol-Myers Squibb and is a consultant/advisory board member for Bristol-Myers Squibb, Eli Lilly, CytomX, and Eisai. A.X. Zhu is a consultant/advisory board member for Eisai, Lilly, Merck, Roche-Genentech, and Bayer. T.M. Bauer reports receiving speakers bureau honoraria from Bayer and is a consultant/advisory board member for LOXO, Pfizer, Bayer, and Guardant Health. T.K. Choueiri reports receiving commercial research grants from Bristol-Myers Squibb, Exelixis, Pfizer, and Novartis and is a consultant/advisory board member for Lilly, Merck, Bristol-Myers Squibb, Exelixis, Pfizer, Up-to-Date, and Roche. A. Drilon is a consultant/advisory board member for Ignyta/Roche/Genentech, Loxo/Bayer/Lilly, TP Therapeutics, AstraZeneca, Pfizer, Blueprint, Takeda/Ariad/Millenium, Helsinn, Beigene, BergenBio, Hengrui, Exelixis, Tyra, Verastem, MORE Health, and Foundation Medicine and reports receiving other remuneration from Merck, Teva, Taiho, WoltersKluwer, GlaxoSmithKline, Puma, PharmaMar, and Medscape/OncLive/PeerVoice/PER/TargetedOnc/RTP. M.H. Voss is a consultant/advisory board member for Eisai, Exelixis, Pfizer, Novartis, and Corvus. C.S. Fuchs is an employee of CytomX Therapeutics, holds ownership interest (including patents) in CytomX Therapeutics and Entrinsic Health, and is a consultant/advisory board member for Agios, Bain Capital, Bayer, Celgene, Dicerna, Five Prime Therapeutics, Gilead Sciences, Eli Lilly, Entrinsic Health, Genentech, KEW, Merck, Merrimack Pharma, Pfizer, Sanofi, Taiho, Unum Therapeutics, and CytomX Therapeutics. G.K. Abou-Alfa is a consultant/advisory board member for Lilly. S.R. Wijayawardana is a consultant/advisory board member for Eli Lilly. X.A. Wang holds ownership interest (including patents) in the form of stock in Eli Lilly. B.A. Moser holds ownership interest (including patents) in Eli Lilly and Company. V. Wacheck holds ownership interest (including patents) in Eli Lilly. No potential conflicts of interest were disclosed by the other authors.

Conception and design: J.J. Harding, A.X. Zhu, T.K. Choueiri, C.S. Fuchs, G.K. Abou-Alfa, X.A. Wang, V. Wacheck

Development of methodology: J.J. Harding, T.K. Choueiri, G.K. Abou-Alfa, X.A. Wang, V. Wacheck

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J.J. Harding, A.X. Zhu, T.M. Bauer, T.K. Choueiri, A. Drilon, M.H. Voss, C.S. Fuchs, G.K. Abou-Alfa, X.A. Wang, J.C. Bendell

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J.J. Harding, A.X. Zhu, T.K. Choueiri, A. Drilon, M.H. Voss, G.K. Abou-Alfa, S.R. Wijayawardana, X.A. Wang, B.A. Moser, V. Wacheck

Writing, review, and/or revision of the manuscript: J.J. Harding, A.X. Zhu, T.M. Bauer, T.K. Choueiri, A. Drilon, M.H. Voss, C.S. Fuchs, G.K. Abou-Alfa, S.R. Wijayawardana, X.A. Wang, B.A. Moser, A. Uruñuela, V. Wacheck, J.C. Bendell

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J.J. Harding, C.S. Fuchs

Study supervision: J.J. Harding, A. Drilon, A. Uruñuela, V. Wacheck

We thank the study investigators and their support staff, as well as the patients and their caregivers for participating in the JTBF clinical trial. Sambasiva Kolati and Aditya Pramod of Eli Lilly and Company provided medical writing support.

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.