Introduction
The human epidermal growth factor receptor 3 (HER3/ERBB3) is a 185-kDa member of the evolutionary-conserved family of HER transmembrane receptors. Together, these four receptor tyrosine kinases form a dynamic signaling network that transduces extracellular growth signals into the cell and activates multiple cellular pathways involved in proliferation, cell survival, and differentiation. HER receptors normally exist as inactive monomers [
1] and only become activated in response to overexpression or ligand binding followed by receptor dimerization. HER3 contains an extracellular neuregulin-binding domain, but unlike the other members of the HER family, it lacks an active intracellular kinase domain. Therefore, HER3 signaling is mediated through heterodimerization with other HER members. The HER2(ERBB2)/HER3 heterodimer is among the most stable of HER dimers and is a particularly potent initiator of phosphoinositide 3-kinase (PI3K) signaling [
1]. Once activated, PI3K then triggers the translocation of AKT to the phospholipid membrane—through the second messenger PIP3—where AKT becomes phosphorylated and acts as a kinase for multiple substrates [
2].
The role of HER1 (EGFR) and HER2 in tumorigenesis is well characterized, and multiple therapeutic compounds targeting these members of the HER family have now been developed. In contrast, less attention has been paid to the role of HER3 in human cancer. Elevated expression of HER3 is seen in many solid tumor types, and upregulation of HER3 is associated with poor outcome and reduced survival [
3‐
8]. Indeed, HER3 may be required for the oncogenic transformation of normal cells by other HER members [
9,
10]. Activated HER3 can interact directly with the p85 subunit of PI3K, whereas HER2 and HER1 cannot [
11]. Furthermore, upregulation of HER3 in response to therapeutic inhibition of other HER members is a recognized mechanism by which tumor cells can escape the action of HER1- and HER2-targeted therapies [
12,
13]. Consequently, several antibodies targeting HER3 are now under investigation [
14‐
21].
RG7116 is a novel humanized anti-HER3 therapeutic monoclonal antibody (mAb) with a dual mode of action. RG7116 binds selectively to domain 1 of HER3 with high affinity and effectively prevents binding of the HER3 ligand, heregulin [
22]. Binding of RG7116 to the extracellular domain (ECD) of HER3 is characterized as a fully enthalpy-dominated key-and-lock mechanism, typical for a fully mature antibody–antigen interaction [
23,
24]. HER3 bound to RG7116 is maintained in the inactive, unliganded conformation, and the high stability of this complex effectively prevents subsequent phosphorylation of HER3 and AKT phosphorylation in vitro. This translated into a potent inhibition of tumor cell growth in vitro, and robust efficacy in a variety of murine subcutaneous xenograft models of human cancer [
22]. In addition to the therapeutic efficacy derived from HER3 signaling inhibition, RG7116 is glycoengineered for enhanced antibody-dependent cellular cytotoxicity (ADCC) [
22].
In this study, we investigated the efficacy and pharmacokinetics (PK) of RG7116 treatment using a subcutaneous mouse xenograft model of human hypopharyngeal cancer. In addition, the expression of pharmacodynamic (PD) markers of HER3 signaling in response to RG7116 therapy was also investigated.
Materials and methods
Cell lines
FaDu human hypopharyngeal squamous cell carcinoma cells (HER1 and KRAS wild type) were obtained from the American Type Culture Collection. Cell lines obtained from these suppliers are routinely authenticated by karyotyping, short-tandem repeat profiling, assessment of cell morphology, and species verification by isoenzymology. Cell lines were expanded upon receipt and aliquots frozen. Cells were not passaged for more than 6 months after resuscitation. Tumor cells were routinely cultured in MEM medium supplemented with 10 % fetal bovine serum, 2 mM l-glutamine, 1× NEAA, and 1 mM sodium pyruvate at 37 °C in a water-saturated atmosphere and 5 % CO2. Culture passage was performed with 0.05 % trypsin and 0.02 % EDTA in phosphate-buffered saline every sixth–seventh day. All reagents were obtained from PAN Biotech GmbH, Germany.
Xenograft model
FaDu cells (5.0 × 106 cells/mL) were injected subcutaneously under anesthesia into the right flank of female SCID-beige mice (CB17.Cg-PrKdcscidLystbg; age 5–6 weeks at arrival; Charles River, Germany). After inoculation, FaDu xenograft tumors displayed rapid progressive growth (take rate 100 %) with an in vivo tumor doubling time of 2–3 days. Mice were maintained under specific-pathogen-free condition with daily cycles of 12-h light/12-h darkness according to the guidelines (GV-Solas; Felasa; TierschG) with food, and water was provided ad libitum. All animal experiments were conducted according to the guidelines of the German Animal Welfare Act and were approved by local government. Animals were examined daily for clinical symptoms, detection of adverse effects, and assessment of body weight. Mice were randomized on Days 14–18 when tumor volume was approximately 200 mm3 and treatment started immediately.
Study FaDu_001: FaDu-bearing SCID-beige mice (n = 9–10/group) were randomized on Day 14 after inoculation (mean tumor volume [TV] 150 mm3), and RG7116 was administered at dose levels of 0.3, 1.0, 3.0, and 10 mg/kg weekly ip on Days 14, 21, and 28. Serum was collected before the second (Day 21) and last treatment (Day 35) together with tumor (Day 35). Study FaDu_006: Mice were randomized on Day 17 (mean TV 220 mm3), and RG7116 was administered as a single i.v. bolus dose at 0.3 and 1 mg/kg on Day 18 (n = 35 mice/dose group; n = 15/control group). Serum and tumor were collected 1, 3, 6, 24, 96, 168, and 240 h after RG7116 injection from five mice/time point/group (controls: 1, 96, 168 h). Study FaDu_008: 15 mice/group (mean TV 180 mm3) were randomized on Day 18, and RG7116 was given weekly i.v. at 0.3 and 3.0 mg/kg on Days 18, 25, 32, 39, 46, 53, and 60 (or 3.0 mg/kg q3w on Days 18 and 39). Serum and tumor samples were collected on Days 25, 32, and 39 (five mice/time point). The 0.3 mg/kg dosing group was rechallenged with RG7116 at 5 mg/kg on Days 39, 46, 53, and 60.
Assessment of anti-tumor efficacy
Tumor volume (TV) was measured by caliper twice-weekly beginning at randomization according to a standard formula (TV = ½ length × width
2); values were documented as means and standard error of the mean (SEM). Tumor growth inhibition (TGI) during the treatment period was calculated by comparing each treated group (treated) with its respective vehicle-treated control (resp. control) using the formula:
$${\text{TGI}}\; ( {\text{\%)}} = 100 - \frac{{\overline{\text{TV}} \left( {\text{treated}} \right)_{{{\text{day}}\;x}} - \overline{\text{TV}} \left( {\text{treated}} \right)_{{{\text{day}}\;y}} }}{{\overline{\text{TV}} \left( {{\text{resp}}.\;{\text{control}}} \right)_{{{\text{day}}\;x}} - \overline{\text{TV}} \left( {{\text{resp}}.\;{\text{control}}} \right)_{{{\text{day}}\;y}} }} \times 100$$
where
\(\overline{\text{TV}}_{{{\text{day}}\,x}}\) represents the average tumor volume of a study group on study day
x.
Data regarding tumor growth were analyzed using nonparametric methods, since the data showed asymmetrical behavior. Prior to this procedure, the data were baseline-corrected with the tumor volume at the of the start treatment.
Pharmacodynamic analysis of the effect of RG7116 on target modulation
Immunohistochemistry
Tissue was formalin-fixed and paraffin-embedded, and standard immunohistochemistry (IHC) was conducted using a murine mAb specific for HER3 (clone DAK-H3-IC, DAKO, #M7297) and rabbit mAb specific for phosphorylated HER3 (pHER3; clone 21D3, Cell Signaling Technologies, #4791). Staining was conducted using the DAKO autostainer (HER3) or Ventana Benchmark XT system (pHER3).
Western Blot analyses
Tissue was immediately lysed using Triton Lysis Buffer (1 % Triton-X-100; 10 µg/mL aprotinin; 0.4 mM orthovanadate; 1 mM phenylmethylsulfonyl fluoride), and lysates were denatured in NuPAGE Sample Reducing Agent at 70 °C for 10 min. SDS–PAGE and Western blotting were conducted as described previously [
22] using 20 μg protein per lane measured using a bicinchoninic acid assay and antibodies specific for HER1 (Upstate/Millipore, #06-847), HER2 (Dako, #A485), HER3 (clone C-17, Santa Cruz, #sc-285), phosphorylated HER1 (pHER1, clone ID Y39; Epitomics/Abcam, #ab32086), or pHER3 (clone 21D3 [Tyr
1289], Cell Signaling Technologies, #4791).
ELISA
Inhibition of AKT phosphorylation was examined in FaDu tumor lysates using the anti-phospho-AKT (pSer473) enzyme immune assay kit according to the manufacturer’s instructions (Enzo Life Sciences). Briefly, cell lysates and reference standards were incubated in wells coated with antibody specific for the amino terminus of AKT. Plates were washed and further incubated in biotinylated antibody specific for AKT phosphorylated at Ser473. After another wash, signal was detected using a solution of streptavidin-HRP conjugate and tetramethylbenzidine substrate and quantified in a spectrometer at 450 nm. The amount of signal was directly proportional to the level of AKT 1/2 pSer473 in the sample.
Gene expression profiling
A portion of tumor was preserved in RNAlater (Qiagen) for gene expression analysis. 50 ng of total RNA was amplified using the NuGen Ovation kit followed by labeling with the NuGen Encore kit. Gene expression profiling for HER1 (232541_at) and ERBB2 (probe set 216836_s_at) was performed on U133 plus 2.0 Genechips (Affymetrix). Preprocessing of the mRNA data was performed using the statistics software R (version 2.13.2), running in house scripts for quality control and data consolidation via the robust multi-array average method.
Analysis of oral mucosa and skin in cynomolgus monkeys
Pharmacodynamic analyses in cynomolgus monkeys were conducted at Covance under standard operating procedures and in compliance with applicable regulations about the use of laboratory animals. Skin and buccal mucous membrane biopsies were collected once before dosing i.v. 20 mg/kg RG7116 and at approximately 2 and 6 h after dosing. A skin patch of approximately 1 g (approximately 40 × 10 mm) was excised from the back, and a mucosal patch of approximately 0.5 g (approximately 25 × 7 mm) was excised from the buccal mucosa. Biopsies were performed under ketamine and xylazine and cut into four pieces for further preparation. The wound was closed with surgical suture, and flunixin was given as antiphlogistic after biopsy. Samples were lysed, and HER3 levels were assessed by Western Blot or pHER3 IHC as described.
Pharmacokinetic analysis of serum RG7116 levels
The concentration of RG7116 in mouse serum was measured using a generic human IgG ELISA (lower limit of quantification: 8 ng/mL) [
25]. Serum concentration–time profiles were used to estimate the following PK parameters in mouse using non-compartmental analysis (WinNonlin, version 6.2; Pharsight Corporation, Mountain View, CA, USA): total drug exposure defined as area under the serum concentration–time curve extrapolated to infinity (AUC
inf), total clearance (CL
total), and observed maximum serum concentration (
C
max). A naïve pooled approach was used in mouse to provide one estimate for each dose group.
Discussion
RG7116 is a novel anti-HER3 antibody that can exert a therapeutic effect through both the inhibition of HER3 signaling and by recruiting immune effector cells for ADCC [
22]. In order to better understand the therapeutic potential of RG7116, we further assessed the anti-tumoral efficacy mediated by HER3-signaling inhibition in a FaDu cell line-based xenograft of human hypopharyngeal cancer and characterized the PK and PD changes occurring in response to treatment with RG7116 at different doses and dosing schedules.
RG7116 showed nonlinear PK in mice with rapid serum clearance at low doses (0.3–1 mg/kg) and slower clearance at higher doses (3 mg/kg). Systemic exposure (AUC) showed a greater than dose-proportional increase accompanied by a decline in total clearance with increasing dose, indicating that elimination of RG7116 is target-mediated [
27]. Maximal efficacy was seen at doses ≥3 mg/kg, consistent with the slower clearance observed at that dose. In addition, the lack of RG7116
C
min accumulation following chronic administration of 0.3 mg/kg indicates a predominant involvement of target-mediated clearance resulting in suboptimal efficacy. Accumulation of RG7116 in serum was observed with doses from 1 to 10 mg/kg in study FaDu_001 and following 3 mg/kg administration in study FaDu_008. This distinct exposure pattern was consistent with the improved efficacy and greater decrease in the pHER3/HER3 ratio seen with higher doses, and is likely due to the target saturation. In addition, the PK profile of RG7116 was correlated with the pHER3/HER3 ratio profile following single-dose administration, with the faster elimination shown in the lower dose group (0.3 mg/kg) being consistent with an earlier return of pHER3/HER3 to basal levels.
This cell line-based tumor model is characterized by a rapid and progressive tumor growth in vivo indicated by a short tumor doubling time of 2–3 days. Therefore, randomization is done between Days 14 and 18 with individual variability, and control-treated animals have to be excluded early on, as is often seen with other models. Despite these model imitations, the anti-tumor efficacy induced by RG7116 in the FaDu xenograft model was strong and dose-dependent in both multiple treatment studies and consistent with previous data [
22]. Efficacious doses of RG7116 also inhibited HER3 phosphorylation and down-modulated membrane HER3 expression in explanted tumor tissue in a similar manner as reported previously. Furthermore, dose-related downstream inhibition of pAKT was also observed. In all studies, the 3 mg/kg dose appeared to be optimal for the FaDu subcutaneous xenograft model system investigated. At this dose level, tumor stasis was achieved with weekly RG7116 in both multiple dosing studies, and target modulation was observed. No difference in efficacy was seen between the 3 and 10 mg/kg doses, suggesting that higher doses do not appear to offer further benefit. A lesser degree of tumor growth inhibition was seen with a suboptimal dose (0.3 mg/kg); however, rechallenging mice treated at this level with a higher dose (5 mg/kg) appeared to restore anti-tumor efficacy. Efficacy with 3 mg/kg given weekly was similar to that of 3 mg/kg given on a 3-week cycle. However, pHER3/HER3 was only ~50 % inhibited by the end of the q3w dosing interval (day 21). Further work is required to understand the extent of pHER3/HER3 inhibition required to achieve efficacy.
Despite initial tumor stasis, tumor regrowth was observed in mice treated once-weekly with 3 mg/kg RG7116 after approximately 46 days. This escape from RG7116-induced growth inhibition was associated with upregulation of both HER1 and HER2. Therefore, two reasons for failure to control tumor growth with RG7116 were observed: suboptimal dosing and activation of alternative signaling pathways involving HER1/2. Combined inhibition of HER3 and other members of the HER family is an attractive option for enhancing the efficacy of RG7116. The availability of potent mAb inhibitors of individual members of the HER family will allow for a personalized medicine approach in the clinic, with appropriate antibodies selected based on the HER profile of the patient’s tumor. This could also imply that a detailed molecular characterization of the tumor needs to be addressed at the time of progression to overcome bypass mechanisms. We previously demonstrated that combining RG7116 (anti-HER3 antibody) with RG7160 (anti-HER1 antibody) or pertuzumab (anti-HER2) achieved complete and long-lasting tumor growth inhibition in murine subcutaneous xenograft models in which tumor growth was driven by HER1 or HER2, respectively [
22].
Inhibition of HER3 signaling alone was insufficient to control tumor growth in the immune-deficient mouse model used here. Additional efficacy is expected in humans through RG7116-mediated ADCC. RG7116 binds with high affinity to the FcRγIIIa receptor found on ADCC-competent cells such as macrophages and natural killer (NK) cells (absent in the SCID-beige mice used in these studies) and significantly enhances in vitro ADCC compared to non-glycoengineered RG7116 [
22]. RG7116-mediated cell killing may also help suppress the development of resistance to HER3 signaling inhibition. Assessing the ADCC potential of novel humanized mAbs in animal models is challenging, as this requires the presence of immune effector cells (i.e., immunocompetent mice) and also that the human mAb interacts with the receptors on these cells. A SCID–human FcγRIIIa transgenic mouse model (which bears murine FcγRIV-positive macrophages and human FcγRIIIa-positive transgenic murine NK cells) was recently used to demonstrate the ADCC activity of imgatuzumab, which is glycoengineered in the same way as RG7116 [
28]. A NOD/SCID/gamma c(null) mouse model, which is receptive to the administration of human NK cells, has also been developed to specifically investigate NK cell-mediated ADCC [
29]. Further evaluation of RG7116 in models such as these is clearly warranted to investigate the contribution of RG7116-mediated ADCC to its efficacy.
RG7116-induced effects on HER3 and pHER3 levels were also seen in the skin and oral mucosa of a cynomolgus monkey, suggesting that these may be useful surrogate tissues for PD evaluation of RG7116 in early development clinical trials. Initial data from a first-in-human trial of RG7116 have shown that downregulation of membrane HER3 is also seen in skin samples of patients with HER3-positive epithelial tumors treated with RG7116 (at a dose of 100 mg and above), and this was associated with downregulation of HER3 in on-treatment tumor samples (at 200 mg and above) [
30]. Skin biopsies have been used as a surrogate tissue for monitoring the effects of anti-HER1 therapies [
31‐
34]. The value of this approach has been shown for erlotinib, where suppression of HER1 phosphorylation in the skin of patients with head and neck cancer was correlated with increased survival [
35].
While these studies were not designed to address differences among baseline tumor volumes, response, and dose/exposure, we analyzed different mice representing animals with the smallest and highest tumor volume at baseline from each of the multiple dosing studies in an attempt to understand the variability in anti-tumor activity observed primarily in mice treated with the lowest dose (0.3 mg/kg). This dose was efficacious in animals with smallest baseline tumor volumes, but not in animals with larger tumors. The efficacy observed was consistent with the
C
min accumulation observed in the mouse with the smallest tumor volume compared to no RG7116 accumulation observed in the mouse with the largest tumor volume. Such a finding might indicate that baseline tumor characteristics affect the efficacy of RG7116 at a given dose. Similar findings have been reported with the anti-CD20 mAb rituximab, where baseline tumor burden was correlated with survival in mice treated at the same dose [
36]. Notably, size-matched FaDu tumor-bearing mice belonging to the vehicle group displayed progressive growth independent of their baseline size. However, further work will be required to properly explore the association between tumor volume, response, and dose for the RG7116 molecule.
In summary, RG7116 demonstrated strong dose-dependent tumor growth inhibition in a rapidly growing FaDu xenograft model of human cancer. Promising preclinical efficacy was demonstrated, and target modulation of pHER3 and HER3 was observed following weekly administration and with administration once every 3 weeks. These studies highlight the value of investigating pharmacokinetic behavior and measuring total HER3 or pHER3/HER3 ratio in tumor, and the value of skin biopsies as potential surrogate markers for efficacy and guiding optimal dosing in the clinic.
Acknowledgments
The authors would like to thank the entire study team at Roche Diagnostics GmbH, Penzberg, for their assistance in the preparation of this manuscript. Support for third-party writing assistance for this article, furnished by Jamie Ashman, PhD, was provided by Prism Ideas and funded by Roche Diagnostics GmbH, Penzberg, Germany.