Background
Interleukin-2 (IL2) is one of the three drugs currently approved by the U.S. Food and Drug Administration (FDA) for the treatment of metastatic melanoma. High dose IL2 induces tumor response rates of approximately 15% in patients with metastatic melanoma, with nearly half of these responses being extremely durable and leading to a seemingly cured subset of patients [
1,
2]. The main drawback of IL2 therapy is its toxicity, especially when administered at high doses that require hospitalization for therapy. Most patients receiving the FDA-approved high dose IL2 experience reversible grade 3 and 4 toxicities including hypotension, renal insufficiency, pulmonary edema, and cardiac arrhythmias with frequent need for continuous cardiac monitoring and administration of vasopressors such as dopamine and phenylephrine.
We hypothesized that targeted delivery of IL2 to the tumor microenvironment using immunocytokines would limit toxicity and increase efficacy of IL2-based therapies. Immunocytokines are genetically engineered fusion proteins consisting of a monoclonal antibody directed against a cancer cell surface antigen and a cytokine such as IL2 [
3]. The immunocytokine EMD 273063 (hu14.18-IL2) consists of two molecules of human recombinant IL2 genetically linked to a humanized monoclonal antibody, which is directed against the diasiologanglioside GD2 (hu14.18). GD2 is a carbohydrate antigen found on the surface of human neuroectodermally-derived tumors including melanomas, neuroblastomas and some sarcomas [
4]. Therefore, GD2 represents a target for the potential delivery of IL2 to the tumor site [
3]. The immunocytokine is expected to maintain the activities of the monoclonal antibody that include target cell binding, effector functions such as complement-dependent cytotoxicity (CDC) and antibody-dependent cellular cytotoxicity (ADCC), while possessing cytokine function. The locally delivered IL2 may activate T and natural killer (NK) cells, which could release a secondary wave of cytokines, and activate immune effector cells.
In animal models, EMD 273063 was able to completely eradicate established lung, liver, subcutaneous, and bone marrow metastases of melanoma and neuroblastoma in immunocompetent mice bearing syngeneic tumor cells transfected to express the GD2 molecule (melanoma model), and in SCID mice reconstituted with human lymphokine-activated killer (LAK) cells and bearing human tumor xenografts (neuroblastoma) [
5]. Interestingly, CD8+ T cells were required for activity of this immunocytokine in melanoma (but not in neuroblastoma), although the melanoma antigens recognized by these CD8+ T cells were not identified. Furthermore, the antitumor activity was dependent on the intact immunocytokine, since it could not be replicated by the administration of equivalent mixtures of antibody and IL2 [
6].
EMD 273063 was tested in a phase I clinical trial aimed at evaluating its safety, toxicity and
in vivo immunological effects in 33 patients with metastatic melanoma [
7]. This immunocytokine was given as a 4-h intravenous infusion on days 1, 2 and 3 of week 1 at dose levels of 0.8–7.5 mg/m
2 per day every 4 weeks (one cycle). The best response on study was stable disease for at least 2 cycles of therapy in 8 patients. Dose-limiting toxicities defining the maximum tolerable dose (MTD) of 7.5 mg/m
2 per day included hypoxia, hypotension, and elevations in liver function tests. Immune activation was induced, as measured by rebound lymphocytosis, increased peripheral-blood NK cell number and activity, and increased serum levels of the soluble alpha chain of the IL2 receptor complex (sIL2R), which was observed at doses both higher (4.8 mg/m
2 per day) and lower (3.2 mg/m
2 per day) than the dose selected for evaluation in the current study. These results were replicated in a separate phase I clinical trial in a pediatric population of patients with neuroblastoma (27 subjects) and melanoma (one subject) treated with EMD 273063 [
8]. Evidence of immune activation was based on increases in serum levels of sIL2R and rebound lymphocytosis. There were no major objective tumor responses, but some patients with chemotherapy-refractory neuroblastoma had periods of durable disease stabilization. In this population, the MTD of EMD 273063 was determined to be 12 mg/m
2 per day.
We hypothesized that the augmented immune activation detectable in peripheral blood after administration of EMD 273063 would be associated with enhanced immune cell infiltrates in melanoma lesions. Therefore, we performed this study to estimate the biologic effects of EMD 273063 at 4 mg/m
2 per day for 3 days as measured by the induction of immune activation in peripheral blood and at the tumor site in a pilot group of patients. The dose of 4 mg/m
2 was chosen for further clinical evaluation because the toxicity increased with higher doses in the prior phase I/II clinical trials, whereas there was evidence of reproducible immune activation at this dose level [
7,
8].
Methods
Study design and endpoints
Study EMR 62207-005 was a phase I/II, open-label, multi-center (4 centers in the USA) clinical trial. Prior to study initiation, the protocol and informed consent documents were approved by the Institutional Review Boards at each study center, and the study was conducted in accordance with both the provisions of the Declaration of Helsinki and Good Clinical Practice. Site monitoring included review of the accuracy of the data in the case report forms. The study planned to enroll 12 eligible patients to explore the effect of EMD 273063 on the study endpoints. This number was based on previous experience with immune analyses indicating that relevant immune responses could be detected with 9–12 patients. This clinical trial was not powered to make inferential statistical analyses. The primary study objective was to estimate the biological activity of EMD 273063 as measured by induction of immune activation in peripheral blood and at the tumor site. Secondary objectives were clinical anti-tumor activity, safety, toxicity and pharmacokinetics (PK) of EMD 273063. Toxicity grades were classified according to the NCI Common Toxicity Criteria Version 2. Objective tumor responses were assessed by the investigators using Response Evaluation Criteria in Solid Tumors (RECIST) [
9].
Patient selection
Eligible patients had histopathologically confirmed stage IV cutaneous melanoma that was not amenable to surgical treatment with curative intent, had progressed after prior therapy including IL2 and/or interferon (IFN), had a Karnofsky performance status of ≥ 70%, and had adequate organ function. Patients were enrolled at least 4 weeks after their last dose of prior therapy. Patients were to have at least 4 melanoma lesions (other than a target lesion) available for outpatient biopsies. The inclusion criteria initially required that the patients be HLA-A2-positive to allow for the assessment of CD8 responses to HLA-A2-restricted melanoma peptides. This criterion was later modified to enhance enrolment. GD2 expression by tumor cells was not an eligibility criterion because assays for GD2 surface expression were not felt to be robust at the time [
10].
Study drug administration
EMD 273063 was provided as a frozen solution in 4-mL glass vials at a concentration of 1 mg/mL, and was manufactured for EMD Serono Research Center, Inc. (Billerica, MA) and EMD Serono Biotech Center, Inc. (Billerica, MA) by Draxis Pharma Inc., Canada. EMD 273063 was diluted with 0.9% sodium chloride for injection and 0.25% human serum albumin before infusion, and administered as an intravenous infusion over 4 h at 4 mg/m2 per day for 3 consecutive days every 28 days. Infusions were performed in an inpatient setting in a General Clinical Research Center. Patients were eligible for up to 4 cycles of treatment.
Pharmacokinetics
Blood samples for PK analyses were drawn during cycles 1 and 2 as pre-dose samples taken immediately before the start of infusion, and post-dose samples collected at 2, 4, 5, 6, 8, 12, and 24 h after start of infusion on day 1. The sample taken at 4-h post-infusion corresponded to the end of infusion (EOI) sample. During cycle 2, the 12-h sample was not required. Additional pre-dose and EOI samples were taken on days 2 and 3 of both cycles. Samples were processed and analyzed for the determination of EMD 273063 in serum using a validated enzyme-linked immunosorbent assay (ELISA). Descriptive PK parameters were derived by non-compartmental and compartmental analysis using the software program Kinetica™ (Thermo Electron, Philadelphia, PA).
Immune monitoring in peripheral blood samples
All assays on peripheral blood were performed at the Immunologic Monitoring and Cellular Products Laboratory of the University of Pittsburgh Cancer Institute Research Pavilion at the Hillman Cancer Center, Pittsburgh, PA. Patients underwent collection of peripheral blood (20–90 mL depending on the study day) pre-study, on days 1 and 10 of each cycle of therapy and at the completion of therapy. Peripheral blood mononuclear cells (PBMC) were separated by density gradient centrifugation over Ficoll gradients and cryopreserved for later analyses. The following analyses were performed as a readout of immune activation: T cell phenotyping for CD3, CD4, CD8, CD16, CD25, CD27 and CD56 by flow cytometry; intracellular granzyme B by flow cytometry as a surrogate marker of the cytotoxic potential of circulating lymphocytes; NK cytotoxic activity against the erythroleukemia cell line K562 (NK-sensitive target) as assessed by standard 51Chromium release assays; ADCC was determined by incubating NK cells with an NK-resistant melanoma cell line (FEMX) and EMD 273063; sIL2R, neopterin and the cytokines IL6, IL10, tumor necrosis factor alpha (TNF-α), and S100 were all measured in serum by commercially available ELISA kits (R&D Systems, Minneapolis, MN). The ELISA analyses of sIL2R, neopterin, IL6, IL10 and TNF-α were conducted with peripheral blood samples obtained on each of the first 3 days of the first 2 treatment cycles. The peripheral blood sample for baseline measurements was obtained by combining two pre-treatment samples (a screening sample and a sample obtained just before the first dose).
Analysis of tumor biopsies
All biopsy tissue assays were performed at Genzyme Analytical Services, Los Angeles, CA. Tumor tissue specimens were obtained at initial screening and at approximately day 10 of the first 2 cycles. Sections of biopsies were snap-frozen using liquid nitrogen, embedded in epoxy, cut and stained with hematoxylin and eosin. Additional sections were embedded in paraffin and labeled with appropriate antibodies for immunophenotyping by immunohistochemistry (IHC). Assays included the density of inflammatory and immune cells; the expression of the T and NK cell cytotoxic granule granzyme B; GD2 immunostaining to define changes in the target of EMD 273063; and major histocompatibility complex (MHC) class I antigen expression. Photographs were taken with an Olympus DP10 digital camera attachment with a C-mount adapter mounted on an Olympus BX40 compound microscope with 4×, 10×, 20× and 40× power objectives. Samples were scored as positive if there were ≥ 50% of cells with 1+ or greater staining intensity (GD2, S100, or HLA-A), or ≥ 1.0 cells per high power field (cell/HPF). In addition, the relative intensity of staining (0, 1+, 2+, and 3+) and the percentage of cells with each degree of staining were also recorded.
Statistical analysis
Exploratory analyses using descriptive statistics were performed to study the biologic activity of the study drug. For parameters in peripheral blood with 3 or more observations, the Mack-Skillings test was conducted as an omnibus test of changes over time. Mack-Skillings p values were adjusted by the step-down Bonferroni method. If an endpoint produced an adjusted p value that was less than 0.05, contrasts between specific study days were tested with the signed rank test. These included comparing days 1–10 of cycle 1 except for serum cytokines for which day 1 to day 3 comparisons were conducted for the first 2 cycles. Some immune parameters that lacked enough samples for the omnibus test were analyzed by comparing pre-treatment to cycle 1 day 10 with the signed rank test. Signed rank p values were not adjusted for multiple hypothesis tests. Semi-quantitative changes in immunohistochemical staining of tumor tissue before and after treatment were analyzed for significance with the McNemar's test.
Discussion
The purpose of this study was to explore the biologic and immunologic activity of the immunocytokine EMD 273063 and provide estimates for designing a future definitive study. We hypothesized that EMD 273063 would bind to GD2 on tumor cells; its IL2 moiety would then activate T and NK cells, which would release a secondary wave of cytokines, orchestrating an antitumor immune response. The main finding of this study was an increase in intratumoral CD8+ CTL with possible increased expression of CD3zeta and granzyme B after administration of EMD 273063. Since these results are based on a small sample size, they would require confirmation in a larger study.
Compared to the results from a previous phase I study with hu14.18-IL2 [
7], peak concentrations and AUC values were only 1/3 of expected values. Since the half-life was in the same range in both studies, the clearance values obtained with the current study were higher. We do not have an obvious explanation for this finding, but several possibilities exist. In the phase I study, the peak serum levels of EMD273063 and AUC during course 1 showed a significant dose-dependent increase, whereas clearance showed a dose-dependent decrease. The dose used in the current study is 4 mg/m
2 which was between dose levels in the phase I study, so that our expected values could have been inaccurate. According to the phase I study, the presence or absence of macroscopic tumor does not influence the clearance of EMD 273063 [
7]. In our study, the safety profile of EMD 273063 was consistent with the expected IL2 side effect profile as reported in the previous phase I clinical trial [
7], except for a lower incidence of hyperglycemia and hypophosphatemia. Despite the intratumoral changes observed in our study of tumor biopsies, this clinical trial demonstrated no definitive antitumor activity with EMD 273063, which may be reflective of the small number and heavily pre-treated nature of the patients enrolled in this study or the small sample size with an inherently low probability (0.60) of observing even a single clinical response with 9 patients and an underlying response rate of 15%.
To gain insight on the effects of the immunocytokine on the immune system, we measured serum levels of immune-activating cytokines over the first 3 days of each treatment cycle. Our results show an increase in serum levels of sIL2R, IL10, and neopterin post-dosing. These findings may suggest the induction of both a T
h1 response (sIL2Rα), monocyte activation (neopterin) as well as a T
h2 response (IL10). Neopterin is produced in monocytes/macrophages upon stimulation with IFNγ and is commonly elevated in inflammatory conditions. Neopterin levels have been reported to be elevated following administration of IL2 [
12], and our study demonstrates a similar increase with the administration of IL2 immunocytokines. The elevated IL10 could be evidence of monocyte stimulation or activation of Th2 cells since it is produced primarily by these cells. In contrast, we did not observe an increase in the percentage of CD16+ and CD56+ PBMC, an increase in NK lysis, or an increase in ADCC. These results should be interpreted with caution given that EMD 273063 has been previously shown to induce ADCC and NK cell-mediated lysis [
7]. This discrepancy may be due to different techniques or the smaller sample size analyzed in our study. Regarding regulatory T cells (Treg), there was no comparable difference in the frequency of CD4 with CD25 staining (the phenotype of both T
reg and activated T helper cells) comparing pre- and post-dosing samples. We did not have additional specimens for functional T
reg determination, but this would be important to assess in further studies since IL2 has been shown to expand T
reg [
13] which could have a negative impact on the effector immune response activated by this immunocytokine.
The staining characteristics of the tumor cells suggest that the EMD 273063 immunocytokine had gained access to the tumor milieu. Although the choice of biopsy site and the random pathologic sampling in small specimens is likely to introduce variability not related to the treatment effect, exposure to EMD 273063 resulted in a decrease for 4 patients in GD2 staining on melanoma cells and increases in staining for S100. Whether this decrease in GD2 staining intensity represents antigen downregulation versus steric hindrance from the EMD 273063 bound to the tumor is not known. In studies of other anti-GD2 antibodies, conflicting results regarding internalization of the antibody (and presumably the GD2) have been observed with some showing that the GD2 remains on the surface [
14] and others reporting internalization [
15]. We also observed that some patients do not have GD2 expressed on their tumor and possibly these should be excluded in future studies. The explanation for increased S100 expression is also unclear; the 2 patients with stable disease did not demonstrate major changes in S100 intensity.
In this study, staining with a pan-HLA-A antibody did not change post-dosing, which suggests that tumor escape might not have been mediated through downregulation of MHC molecules after administration of EMD 273063. There was a trend towards an increase in intratumoral cell staining with the lymphocyte markers CD3 (total T lymphocytes) and CD8 (cytotoxic T lymphocytes), with possible increased staining for CD3zeta and granzyme B, effector molecules related to cytotoxic activity. However, there was no post-dosing change in NK infiltration as detected by CD16 and CD56 IHC staining. This observation is in contrast with findings in the peripheral blood that show no change in the number of lymphocytes that display CD3 and CD8 markers (total and CD8+ T cells, respectively), and may suggest that EMD 273063 effectively targets GD2 expressing tumors, and delivery of IL2 to the tumor microenvironment, resulting in expansion of CD8+ T cells, more notably CTL. It also supports the notion that the tumor may be a more appropriate site to study the interaction between the immune system and cancer cells, as opposed to the more common analysis of immune parameters in peripheral blood [
16].
Whether directing IL2 to the tumor environment is the most appropriate way to enhance local immunity will require further study. Other approaches for introducing IL2 into the tumor environment include injection of the cytokine intratumorally [
17] and administering intratumoral injections of adenovirus encoding IL2 [
18]. In the later study, an objective response rate of 17% was observed for the injected lesions and stable disease was noted in some cases for non-injected lesions. In agreement with our study, they also noted increased intratumoral CD8+ T cells that were mainly of a cytotoxic phenotype, but minimal change in NK cell or CD4+ T cells. Similar results were observed for intratumoral injection of canarypox encoding IL2 [
19]. These data suggest that intratumoral IL2 delivered by different strategies does result in enhanced CD8+ cytotoxic T cells intratumorally. Recently, intratumoral and intravenous immunocytokine administration was compared in murine models and the IT route [
20] was more effective. Thus, future studies should evaluate the IT route in human tumors.
Acknowledgements
The authors would like to thank Roland Neugebauer from Merck KGaA, Darmstadt, Germany, for his contribution to the pharmacokinetics part of this study and EMD Serono Inc., Rockland, MA, for provision of hu14.18-IL2 and financial support for the conduct of this study. In addition, the authors would like to thank Genzyme Analytical Services, Los Angeles, CA, for biopsy tissue analysis. Editorial assistance for this article was provided by Physicians World GmbH, Mannheim, Germany. The study and the manuscript preparation were supported by Merck KGaA, Darmstadt, Germany.
Competing interests
WG and TLW declare that they have no competing interests. AR is a speaker, consultant and/or receives grant support from: Amgen, Mannkind Corporation and Pfizer. JMK has received commercial research grants from Schering, BMS and Pfizer and served as a speaker for the Schering Plough Corporation. MBA has received commercial research grants from Novartis and Bayer/Onyx, served on Advisory Boards for Novartis, Antigenics, Schering and Medarex. AK and OK are employees of Merck KGaA; OK is also an Adjunct Professor at the University of North Carolina at Chapel Hill, NC, USA. SDG is a former employee of Merck KGaA and an inventor of patents related to hu14.18-IL2. MAM is a speaker, consultant and/or receives grant support form: Amgen, BMS, Bayer, Genentech, GlobeImmune, Immunitope, Novartis, Onyx, Roche, Sanofi-Aventis, Pfizer.
Authors' contributions
AR participated in the study design and coordination and helped to draft the manuscript. JMK participated in the study design and coordination. MBA participated in the study design and coordination. TLW carried out and interpreted the immunoassay and drafted portions of the manuscript describing the assays. WG participated in the design of the study and performed the statistical analysis. AK participated in the study design and coordination. SDG participated in the study design and coordination. OK participated in the study design and coordination and helped to draft the manuscript. MAM participated in the study design and coordination and helped to draft the manuscript. All authors read and approved the final manuscript.