Final Report Summary - PRIAT (Profiling Responders In Antibody Therapies)
Executive Summary:
Therapeutic antibodies represent the largest and fastest growing class of Pharmaceutical Biotechnology. They are expensive treatments, which can provide a considerable benefit to some (but not all!) patients in various therapeutic areas. At present, the main pharmaceutical applications of antibody therapeutics are in the field of oncology and of chronic inflammation (e.g. rheumatoid arthritis). However, antibody products are also often used to prevent graft rejection in transplantation and increasing development activities are being performed in the cardiovascular field (e.g. PCSK9 blockade).
The PRIAT Project aimed at the development and implementation of advanced methodologies for the profiling of responses in antibody therapies. This goal has been achieved by three main lines of activities:
- HLA Peptidome analysis of biological specimens (including plasma specimens). A characterization of the T cell response to disease and treatment is crucial for the understanding of immunomodulatory pharmaceutical strategies. As T cells recognize peptides presented on MHC molecules (HLA in human), a characterization of the HLA peptidome is an indispensible step, towards the profiling of T cells (e.g. by means of multiplex tetramer technologies)
- Innovative imaging techniques (e.g. PET and near-infrared fluorescence imaging). Not all therapeutic antibodies efficiently localize at the site of disease, which often represents a requirement for pharmacological activity. It is therefore crucially important to integrate non-invasive imaging techniques in the antibody development process.
- Multiplex analysis of specimens (e.g. sera, biopsies, leukocytes), in order to gain functional information about the status of immune activation and/or cytokine levels. A detailed characterization of the mechanism of action of immunomodulatory agents facilitates product development and a profiling of responses in patients.
All three activities have been successfully completed.
We have developed innovative methodologies, which allow the identification of more than 1’000 peptides bound to MHC-I and MHC-II molecules. Importantly, hundreds of HLA-I-bound peptides could be identified from the plasma of patients with melanoma (one of the most immunogenic tumors), thus paving the way to a non-invasive profiling of peptides, which could act as tumor rejection antigens in immunotherapeutic strategies.
We have performed the first Phase 0 immuno-PET clinical trial worldwide in patients with rheumatoid arthritis, thus defining a pathway for the image-based development of antibody products in the future. In addition, we have performed a large amount of preclinical imaging applications, which have facilitated the selection of clinical development candidates.
Finally, we have extensively studied pathological specimens in man and in rodents, using multiplex biomarker analysis and microscopic techniques, shedding light on mechanism of action.
Project Context and Objectives:
In the PRIAT Project, we focused on three main indications for antibody-based therapeutics (cancer, rheumatoid arthritis and graft rejection), with the aim to experimentally address three issues, which are crucial for the understanding of antibody activity and for the design of the drugs of the future:
- How efficiently do antibody drugs accumulate at the site of disease?
- How efficiently can antibody drugs modulate the activity of the immune system at the site of disease?
- How efficiently can we profile the HLA-I and HLA-II peptidomes in health and disease, in order to be able to identify the targets of T-cell recognition (e.g. tumor rejection antigens, graft rejection antigens, autoantigens for autoimmune conditions)?
The PRIAT Project has allowed our Consortium to make considerable progress in three key areas (HLA peptidome analysis, imaging and multiplex analysis of patient-derived specimens), which are of fundamental importance for the characterization of the activity of therapeutic antibodies in patients.
We have developed innovative methodologies, which allow the identification of more than 1’000 peptides bound to MHC-I and MHC-II molecules. Importantly, hundreds of HLA-I-bound peptides could be identified from the plasma of patients with melanoma (one of the most immunogenic tumors), thus paving the way to a non-invasive profiling of peptides, which could act as tumor rejection antigens in immunotherapeutic strategies.
We have performed the first Phase 0 immuno-PET clinical trial worldwide in patients with rheumatoid arthritis, thus defining a pathway for the image-based development of antibody products in the future. In addition, we have performed a large amount of preclinical imaging applications, which have facilitated the selection of clinical development candidates.
Finally, we have extensively studied pathological specimens in man and in rodents, using multiplex biomarker analysis and microscopic techniques, shedding light on mechanism of action.
Collectively, the Project allowed not only the development and the practical implementation of novel methodologies in patients or in patient-derived material, but also provided tools for the characterization of novel antibody-based products, some of which are now being studied in clinical trials or are about to begin clinical testing.
In animal models of disease, we have performed extensive quantitative biodistribution studies, using radiolabeled preparations of therapeutic antibody products. This work has led to numerous publications and, collectively, has provided us with a quantitative understanding of the selectivity which antibody products can reach in vivo in rodent models of cancer, rheumatoid arthritis and graft rejection.
In addition, we have published a dosimetric estimate of radiation doses in organs and neoplastic lesions for a pan-tumoral antibody (L19, specific to the alternatively-spliced EDB domain of fibronectin), which has been studied in patients with solid tumors or with lymphomas. These studies have shown that the L19 antibody preferentially localizes on neosplastic lesions, but also that the tumor-homing selectivity varies greatly (i.e. up to 40-fold) from lesion to lesion and from patient to patient. Moreover, we have run an immuno-PET clinical trial with Dekavil (F8-IL10), an armed antibody developed by Philogen, labeled with iodine-124, in patients with rheumatoid arthritis.
We have implemented the routine analysis of leukocyte infiltration at the site of disease and the multiplex analysis of cytokine levels (in serum and at the site of disease) for many preclinical experiments, aimed at the characterization of the activity and mechanism of action of various antibody-based therapeutics. These studies have been published in several articles and have facilitated the identification of product candidates for clinical and industrial development programs (see below). Furthermore, we have studied in substantial detail the infiltration of lymphocytes into melanoma lesions in patients receiving antibody drugs and the specificities of cytotoxic T-lymphocytes which account for long-term survival, following pharmacological treatments.
We have developed, implemented and perfected methodologies for the determination of “atlases” of peptides bound to MHC-I or MHC-II (HLA-I and HLA-II in man). Thanks to the use of state-of-the-art mass spectrometers and to the implementation of efficient antibody-based pulldown procedures, these methodologies have allowed the determination of complex peptidomes, comprising hundreds of MHC-bound peptides, in complex biological samples, such as cell lines or patients’ sera. For the first time, it has been possible to identify hundreds of HLA-I-bound peptides from sera of melanoma patients. The identified peptides included previously reported melanoma rejection antigens, thus providing confidence about the robustness of the procedure. In mice with arthritis, collagen-derived peptides were identified as MHC-II ligands, thus providing a direct analytical access to antigens which may trigger autoimmune conditions.
We believe that the PRIAT Project has produced important technologies and publications, which facilitate the development and mechanistic study of antibody drugs. In time, this will lead to the generation of better antibody products and to improved selection procedures for patients, which are more likely to benefit from therapeutic intervention. In addition to these rather indirect (but important!) results, we believe that the PRIAT Projects has also produced results, with an immediate socio-economic impact.
We believe that the PRIAT Project has been truly successful, not only in terms of methodology development but also for the boosting of European Biotech competitiveness, with novel antibody products being brought to the clinic and improved analytical procedures for the study of response to antibody treatment.
Project Results:
Introduction
The PRIAT Project aims at developing and applying innovative technologies, for the profiling of responses in antibody therapies.
Therapeutic antibodies represent the largest and fastest growing class of Pharmaceutical Biotechnology. They are expensive treatments, which can provide a considerable benefit to some (but not all!) patients in various therapeutic areas, including oncology, chronic inflammatory conditions and transplantation.
In the Project, we have developed and applied three classes of innovative technologies, in order to better understand response to therapy both in mouse models of disease and in patients:
- HLA Peptidome analysis of biological specimens (including plasma specimens). A characterization of the T cell response to disease and treatment is crucial for the understanding of immunomodulatory pharmaceutical strategies. As T cells recognize peptides presented on MHC molecules (HLA in human), a characterization of the HLA peptidome is an indispensible step, towards the profiling of T cells by means of multiplex tetramer technologies
- Imaging techniques (e.g. PET and near-infrared fluorescence imaging), to reveal the ability of antibody-based therapeutics to localize at the site of disease in rodent models and in patients
- Multiplex analysis of specimens (e.g. sera, biopsies, leukocytes), in order to gain functional information about the status of immune activation and/or cytokine levels.
The therapeutic focus of the PRIAT Project has been in three disease areas [(i) oncology; (ii) rheumatoid arthritis; (iii) transplantation], for which antibodies represent important therapeutic agents. Indeed, antibody sales in Oncology and Inflammation were greater than 30 billions $ in 2013.
The participants of the PRIAT Project were ETH Zürich (Coordinator, Switzerland), Philogen (Siena, Italy), VUMC (Amsterdam, The Netherlands), Philochem (Zurich, Switzerland), Kennedy Institute of Rheumatology (Oxford, UK), University Hospital in Tübingen (Germany), University Hospital in Graz (Austria), University Hospital in Jena (Germany).
WP1 – HLA Class I Peptidome Methodologies
The detailed characterization of peptides bound to HLA-I molecules in man (MHC-I in rodents) is of tremendous importance, if we wish to characterize (at a molecular level!) how immunostimulatory anticancer strategies work (i.e. how T cells recognize tumor cells). This research topic has become very important in oncology, after the recent approval of immunostimulatory antibodies, directed against CTLA-4 (Ipilimumab) or against PD1 (pembrolizumab). These therapeutic agents are revolutionizing the therapy of melanoma and have the potential to have a substantial impact for the therapy of other malignancies.
Production of previously described monoclonal antibodies
HLA-class I peptidome analyses are typically performed by first purifying HLA-class I/peptide complexes with specific antibodies followed by release of peptides by denaturing the complexes. For this purpose the widely applied anti-HLA-class I antibody clone W6/32 was purified from the supernatant of hybridoma cells (HB-95, ATCC). The W6/32 antibody was produced in serum free hybridoma medium for five days followed by purification over protein A resin. Large productions of the W6/32 monoclonal antibody from the corresponding hybridoma have been performed, in order to provide material for pull-down and precipitation experiments. In total, over 1.2 g of W6/32 antibody were purified from Wave bioreactor at Philogen and shipped to Zurich. In addition, approximately 0.9 g have been produced at Philochem.
Generation of novel antibodies
A good quality monoclonal antibody clone W6/32 has been successfully used for HLA-I/peptide complex purification, as indicated in the previous section. However, affinity reagents capable of purifying different MHC-I variants in mouse and rat were also needed at the beginning of the Project. Further, specificity and cross-reactivity between different MHC sub-types is very difficult to assess for less characterized antibodies against MHC-I complexes.
For this reasons, we hypothesized that if we were able to recombinantly express b2-microglobulin from different species, we should be able to generate good quality affinity reagents using antibody phage display libraries. Ideally, these reagents would react not only with b2-microglobulin, but also with the intact MHC-I/peptide complex and as b2-microglobulin is part of all MHC-I complexes, antibodies would not be limited to certain sub-types.
We initially tested this approach on b2-microglobulin of human origin. In particular, one clone was isolated, which bound both b2-microglobulin and the MHC-I/peptide complex with high affinity and with an extremely low kinetic dissociation constant (Figure 1).
The antibody was then reformatted from scFv to IgG format for expression in mammalian cells. However, we preferred to use the W6/32 clone for pull-down experiments.
The W6/32 antibody did not cross-react with mouse and rat MHC-I molecules. For this reason, using a similar strategy, antibody phage selections were also performed against murine and rat b2-microglobulin. In addition, we evaluated hybridoma-derived monoclonal antibodies, which resulted in the use of OX-18 for rat pull-down experiments and M1/42 for mouse pull-down experiments.
Comparison of proteomic platforms and mass spectrometer selection
The next step was to purify larger quantities of HLA-class I complexes, release the peptides and analyze the eluate by mass spectrometry. Until today, four different setups were tested:
- Single spot MALDI with an AB SCIEX 4800 MALDI TOF/TOF
- LC-MS on an Orbitrap classic
- LC-MS on an Orbitrap Velos
- LC-MS on a Q Exactive; Philochem has purchased the latest generation of Orbitrap analyzers, the Q Exactive, which significantly boosted peptide identifications. Now we routinely identify hundreds of peptides from cell lysates with very high confidence.
Development of innovative methodologies for HLA-I peptidome analysis in cell lines
The optimized conditions for the purification of HLA class I complexes from the surface of cancer cell lines were applied to the profiling of the HLA peptidome of the AML cell line HL-60, the multiple myeloma cell line RPMI8226 and the human embryonic kidney cell line HEK. Six samples originating from 100 million cells of each cell line were analyzed with optimized parameters leading to the identification of 423, 313, and 1105 peptides with 1% false discovery rate (FDR) from RPMI8226, HL-60 and HEK respectively (Figure 2). In single experiments of the HEK cell line, more than 700 peptides were identified. The majority of identified peptides was 8-10 amino acids in length, which is compatible with HLA peptides, but which is not a typical length for e.g. tryptic digests of unspecific cleavage (see Deliverables report for details). In such cases we expect much larger peptides and a much larger standard deviation of sizes.
Besides the appropriate length distribution, we were evaluating if peptides can bind to the respective HLA molecules expressed on the surface of the cells. Prediction tools ANN and netMHCpan accessible through (http://tools.immuneepitope.org/mhci/) were found to provide best results. 60 and 80 % of 9mers are predicted to bind to the HLA A alleles on the surface of RPMI8226 and HEK respectively. However, 60 % of 9mers identified from HL-60 cells are predicted to bind to HLA B alleles.
Development of innovative methodologies for HLA-I peptidome analysis in plasma samples
For the conclusion of the WP1, HLA class I peptidomics methodology, we evaluated the possibility to perform HLA class I purifications from plasma from the blood bank. We therefore obtained pooled blood plasma from Blutspende Zurich. We have performed four experiments with this material a total of 14 purifications with anti HLA class I antibody W6/32 and 5 with BSA-conjugated beads (serving as negative control), with the aim to better understand the level of unspecific peptide contaminations. In four independent experiments, we identified up to 154 peptides with expected HLA class I characteristics. We also learned about preferred procedures for the storage of plasma samples, which allow the reliable identification of a large number of HLA class I derived peptides (see WP4).
Deliverables
D1.1) Technical Feasibility of HLA-I peptidome analysis in cell lines and plasma samples:
In view of what is described above, this Deliverable has been successfully completed. Importantly, the large number of identified peptides and the peptide length distribution (mainly nonamers) provided confidence that the methodology can be used for practical biomedical applications (see next sections).
WP2 – HLA Class II Peptidome Methodologies
The study of the HLA-II peptidome in man (MHC-II in rodents) is extremely important, in order to profile which antigens are being recognized by CD4+ T cells. A detailed characterization of the HLA-II peptidome promises to be particularly important for the understanding of immunological responses in autoimmune conditions such as rheumatoid arthritis and multiple sclerosis, two indications for which antibody products are frequently used in the clinical practice.
The study of the HLA-II peptidome is complicated by the polymorphism of HLA-II molecules and by the absence of a common chain (such as b2-microglobulin in HLA-I). Furthermore, no mass spectrometry-based studies have been reported so far for the study of peptides bound to soluble HLA-II molecules. Indeed, two transmembrane peptides need to be cleaved, in order to release HLA-II/peptide complexes in circulation.
Production of previously described monoclonal antibodies
Commercial anti-murine MHC class II antibody M5/114 could be obtained from a commercial source in large quantities. This antibody recognizes a panel of murine I-A and I-E alleles and is therefore applicable to a variety of model systems.
To purify human HLA class II complexes, we have purchased the hybridoma cell line producing pan-HLA DR antibody L243 (ATCC HB-55, recognizing the alpha chain of HLA DR). The antibody was produced in low IgG serum and purified with Protein A resin.
Generation of novel monoclonal antibodies
Unlike polymorphic MHC-I molecules, which all share the b2-microglobulin domain and which can be purified by suitable monoclonal antibodies directed against this portion of the MHC-I molecule, the purification of MHC-II molecules is much more difficult, because they are more polygenic and polymorphic.
We decided to focus on DR HLA molecules, as they are less polymorphic than DP and DQ and have been associated with increased incidence of certain autoimmune conditions (e.g. rheumatoid arthritis). We obtained recombinant HLA-II molecules of different haplotypes in complex with CLIP peptide from collaborators in Berlin (Prof. Dr. C. Freund) and we successfully raised >20 monoclonal antibodies from phage display libraries, which were able to recognize their cognate antigen in ELISA and on BIAcore (Figure 3). The affinity of a commercial antibody, L243, however was still superior on Biacore and we decided at this step to perform experiments with this antibody.
Murine MHC preparations have been even more difficult to get hold of. Also in this case we decided to go with a commercial antibody, M5/114, with which we obtained very good results.
Development of innovative methodologies for HLA-II peptidome analysis in cell lines
Starting from 100 million A20 lymphoma cells we could repeatedly identify up to over 1000 peptides with 1% false discovery rate. These numbers are unprecedented in the scientific literature especially as we only rely on high confidence identifications. Many publications in this field rely on low confidence peptide identifications (worse than 5% FDR), which boosts the number of peptides significantly, however, the number of false identifications increases as well. Bozzacco and colleagues [Bozzacco et al. (2011) J. Prot. Res., 10, 5016) for example, used 500 million cells and accepted peptide identifications with a FDR below 10%, ending up with around 100 peptides. We are currently preparing a manuscript on these results as we believe they represent a milestone in the analysis of the MHC class II peptidome, allowing us to answer fundamental questions in immunology (e.g. identity and frequency of antigens associated with T cell recognition, leading to autoimmunity) for the first time.
In one analysis, in which we identified 1110 peptide sequences, we obtained for the first time a very detailed understanding of the size distribution of MHC class II peptides (Figure 4) The length distribution follows a Gaussian centering around 15 amino acids in length with the majority of peptides being 12 to 19 amino acids long. No preference for certain amino acids either at the N- or at the C-terminus of the identified peptides was observed. Certain central amino acids were found to be significantly enriched after alignment of the sequences with the multi sequence alignment tool Clustal X (Figure 4). To the best of our knowledge, this is the first time that the binding motif for the MHC class II complex H2 I-Ad has been described to such a level of detail.
We have also investigated on the purification of HLA class II complexes from human cell lines (B-cell lines Ramos, Burkitt's lymphoma, ATCC® CRL-1596, and DoHH2, follicular B-cell lymphoma, DMSZ ACC 47). Purification of HLA class II from DoHH2 cells could be successfully demonstrated by Western blotting. In a large scale purification from 100 mio cells each, 165 and 97 unique peptides were identified with high confidence from DoHH2 and Ramos cells respectively.
Feasibility of HLA-II peptidome analysis in human plasma samples
After optimization of elution conditions, 5 ml of human plasma from the blood bank were subjected to the purification of HLA class II complexes with L243 resin. Three samples were purified with L243 resin and one control sample with BSA conjugated resin was included. Interestingly, several HLA class II derived peptides were identified in the analysis of the supernatant after filtration of eluted complexes through 10 kDa filters. Eight peptides were identified from HLA DRA, 7 from HLA DRB5, and 5 from HLA DRB1 indicating the successful purification of HLA class II complexes. However, 149 peptides were also identified in the BSA sample and only 108, 109, and 69 peptides after HLA class II purification, raising concerns about the efficiency of peptide recovery and the stability of isolated peptides in acidic conditions. Peptides identified after purification of HLA class II were short (on average 10 amino acids): a feature which is incompatible with the knowledge which has been gained regarding the structure of HLA class II peptides.
We are currently still working on establishing the purification of HLA class II complexes from human blood. We believe that this challenging goal may be reached one day, but there are many obstacles to overcome.
Feasibility of MHC-II peptidome analysis in mouse serum samples
Since purification of murine MHC class II complexes was yielding significantly higher amounts of peptides as compared to human HLA class II, we investigated if purifications from mouse serum would similarly lead to better results. We have obtained commercial mouse serum to get an initial idea of the volumes of mouse serum necessary. 4, 2, and 1 ml of the mouse was subjected to MHC class II purification with M5/114 antibody. 12 peptides derived from MHC class II complexes were identified in supernatant digests demonstrating successful purification, 7 of these matched to I-Ab, indicating that this MHC class II subtype is present. Moreover, peptides were derived from alpha and beta chain, indicating the presence of intact complexes. We should be able to elute peptides from these intact complexes and indeed 66, 94, and 117 peptides were identified from 1, 2, and 4 ml samples, respectively. Among the identified peptides were three MHC class II protein derived peptides known to be ligands of I-Ab. One of these, RHNYEGPETHTSLR, has been identified in murine spleen samples with the same haplotype (Ab). This indicates that some of the identified peptides may represent MHC class II eluted peptides.
Deliverables
D2.1) Technical Feasibility of HLA-II peptidome analysis in cell lines and plasma samples
In view of what is described above, this Deliverable has been successfully completed. Importantly, we were able to identify at least 10-times more murine MHC class II peptides (starting from a lower number of cells), compared to previous reports. A careful analysis of peptide consensus sequences and of peptide length distribution has shed light on the types of molecules which are presented onto MHC-II molecules in rodents. We could further prove the successful purification of MHC class II complexes from murine serum where a number of known MHC-II ligends have been identified.
WP3 – Immuno-PET clinical trial preparation
It is becoming increasingly clear that many of the standard chemotherapeutic agents used for cancer treatment do not preferentially localized at the tumor site (Figure 5).
Unfortunately, a selective uptake in the tumor is often not observed also for antibody-based therapeutics, which would be expected to display a tumor-targeting potential (Figure 6).
We and others have previously recognized the importance of integrating molecular imaging procedures for a “Precision Medicine” approach to drug development [Tijink et al. (2009) Eur. J. Nucl. Med. Mol. Imaging, 36, 1235].
Within PRIAT, we have focused on the preparation and implementation of a Phase 0 immuno-PET clinical trial, to assist the development of F8-IL10 (an antibody-cytokine fusion protein, which has recently moved to Phase IIb clinical trials for the therapy of patients with rheumatoid arthritis). In addition, we have performed and published dosimetric evaluations on immuno-PET trials previously performed by the Philogen group, and we have executed a large number of imaging and quantitative biodistribution studies in rodent models of cancer, rheumatoid arthritis and chronic graft rejection.
Optimization of GMP radiolabeling of F8-IL10, chosen as representative innovative antibody-based therapeutic agent for clinical trials
The GMP procedure for the radiolabeling of F8-IL10 has been optimized by the fine-tuning of a set of critical parameters such as reaction time, concentration of the iodination reagent and initial formulation of F8-IL10. Furthermore, different radiolabeling modalities (i.e. Chloramine-T vs Iodogen method) have been compared and the use of Philogen's automated radiolabeling module “Easy Iodine Machine” has been implemented.
The quality and behavior of F8-IL10, iodinated under different conditions, have been characterized according to standard biochemical methods as well as with biodistribution studies performed in mice.
Stability of the radiolabelled drug was also verified upon storage at both room temperature and +4°C. A biodistribution study in mice was also performed to ensure retention of biodistribution properties after storage.
Optimization experiments showed that both Chloramine-T and Iodogen methods provide material of similar quality and that the presence of Tween-20 in the formulation buffer of F8-IL10 does not interfere with the radiolabeling process. Furthermore, to support stability study of the iodinated material, a biodistribution study in normal mice demonstrated full retention of biodistribution properties of iodinated material stored at +4°C up to 24h
Preparation work for a Phase 0 Immuno-PET Imaging Clinical Trial at VUMC in Amsterdam
A GMP Manufacturing procedure has been established at VUMC (Amsterdam) for the radiolabeling of F8-IL10, an anti-inflammatory immunocytokine which is currently being investigated by Philogen for the therapy of rheumatoid arthritis in a double-blind multicenter Phase IIb clinical trial. The product can now be reproducibly labeled with iodine-124, a PET radionuclide.
Quality data for the radiolabeled product, together with a Study Protocol, and Investigator Brochure and a Patient Information Form, have been submitted to Clinical Authorities, who have approved a Phase 0 immuno-PET imaging study in 5 patients with rheumatoid arthritis. The study has been performed, as documented in WP8.
Dosimetric Evaluation in PET
ImmunoPET combines the high resolution and sensitivity of PET and the selective localization of antibodies at their target in vivo. The technique has profited from technical advances and the increased availability of GMP-grade radionuclides and has therefore begun to play an important role in both cancer staging and tailoring of therapy. Besides pure diagnostic purposes, ImmunoPET can also be envisaged as a dosimetric tool to predict the expected doses to target lesions and healthy organs of a subsequent radioimmunotherapy using the same antibody coupled to a therapeutic isotope.
Indeed, we have performed and published a dosimetric evaluation of the radiation dose delivered to neoplastic lesions and to normal organs by a clinical-stage human monoclonal antibody (124I-L19, developed by Philogen) in patients with brain-metastases and extracranial lesions. This study nicely complements similar dosimetric findings on the targeting of radioiodinated L19 in patients with lymphoma [Erba et al. (2012) J. Nucl. Med., 53, 922].
Collectively, these studies indicated that antibody uptake can be highly variable even in different lesions of the same patient and that immunoPET procedures may guide product development with armed antibodies (Figure 7).
Deliverables
D3.1) Preparation activities for Immuno-PET clinical studies:
The Deliverable has been successfully completed, as documented by the performance of the clinical trial.
Dosimetric analyses on tumor targeting of cancer patients with radioiodinated preparations of the L19 antibody have also been published. Moreover, the implementation of reliable radiolabeling techniques has been important for the execution of many biodistribution and imaging studies in rodent models of disease (see subsequent WPs).
WP4 – Management of clinical samples
Procedures and experimental findings related to the use of patients’ sera for HLA peptidome analysis have been described in WP1.
For the analysis of biopsies, it is well established that biopsies and surgically-resected material should preferably be stored as freshly stored material, embedded in OCT.
We have evaluated the feasibility of multiplex analysis of leukocyte infiltration and of intratumoral cytokine levels using a combination of immunofluorescence microscopy techniques, FACS analysis of resuspended cells and parallel measurement of cytokine concentrations. The results of these investigations have now been described in publications and are now routinely implementing these methods in our laboratories (see also WP9).
Immune system–based prognostic markers may be useful to monitor response to immunotherapies and may serve to allow the identification of patients with different prognosis.
In a previous study, the group of Prof. Garbe at University of Tübingen demonstrated the prognostic relevance of functional circulating T cells responding to the tumor-associated antigens NY-ESO-1, Melan-A, MAGE-3, and survivin on overall survival of patients with melanoma with distant metastasis. For PRIAT activities, cryopreserved peripheral blood mononuclear cells (PBMCs) from 84 patients with follow-up after analysis (cohort A), 18 long-term survivors with an extraordinarily favorable course of disease before analysis (≥ 24 months survival after first occurrence of distant metastases; cohort B), and 14 healthy controls were collected and analyzed. Circulating antigen-reactive T cells were characterized by intracellular cytokine staining after in vitro stimulation.
The presence of circulating T cells responding to Melan-A or NY-ESO-1 have a strong independent prognostic impact on survival in advanced melanoma. The collection of samples described above represented the basis for the analysis of T cell reactivities in melanoma (see WP9).
Collectively, the partners collected and exchanged samples of plasma, sera, blood and biopsies, which were essential for the execution of the analysis described in the various Workpackages.
A standard operation procedures for collection and storage of rat serum/plasma to detect soluble MHC I/ II was established by UHJ.
Deliverables
D4.1) Implementation of Standard Operating Procedures for the collection, storage and shipment of clinical samples (biopsies and plasma samples):
The Deliverable has been successfully completed, as documented by the successful execution of the experiments described in other WPs of this report.
WP5 – HLA Class I peptidome analysis in melanoma
A recent analysis of over 7000 human cancer specimens has recently provided a solid foundation for the mutation rate which is observed in human malignancies. Typically, a prevalence greater than 1 mutation in 1 megabase is found for most tumor types. Melanoma appears to be the tumor with the highest mutation rate, which is typically in the range between 10-100 mutations per megabase [Alexandrov et al. (2013) Nature, 500, 415].
In this WP, we have focused on the profiling of HLA-I peptidome in melanoma and on the characterization of T cell reactivies (and their clinical significance).
HLA-I peptidome analysis in melanoma
At the University Hospital Tübingen, sera from cancer patients and healthy volunteers were collected according to the defined standard operating procedure for further HLA class I peptidome analysis. Patient characteristics are presented in Table 1. In total 15 sera were shipped, nine samples from cancer patients, and six from healthy volunteers. The inclusion criterion was that at least one of the HLA A haplotypes was A02, as most of the known tumor rejection antigens are described for this allele. Tumor burden was variable, ranging between none (only local tumor) and high, two sera were provided before Ipilimumab treatment, three samples after Ipilimumab.
Five ml of serum per patient were subjected to the HLA purification procedure using W6/32 antibody resin purifying all HLA class I alleles. After excessive washing, complexes were eluted with 0.1M acetic acid, heated to 95 °C, filtered through 10 kDa cutoff filters and purified over C18 resin for mass spectrometric analysis. Peptides were resuspended in 8 l 3% acetonitrile 0.1% formic acid in ultrapure water and 6 l loaded on a 15 cm C18 column. A linear gradient from 100 % water 0.1% formic acid to 20 % acetonitrile 0.1% formic acid over 45 minutes was used to elute the HLA peptides. Peptides were fragmented in the HCD cell and spectra recorded in the Orbitrap. Between 57 and 567 peptides (on average of 184) were identified with 1% FDR. Most identified peptides were 9 amino acids in length (Figure 8A). The same is true for the c-terminal anchor residues: Certain amino acids were enriched, demonstrating patterns of anchor residues depending on the individual HLA subtype (Figure 8B). Most prominent C-terminal anchor residues were K and R as well as hydrophobic residues L, V, Y, S. Patients 3 and 6 for example demonstrated a strong enrichment of C-terminal K, which is a known anchor residue for the HLA A03 allele. Indeed, almost 60% of identified peptides with 8-11 amino acids per patient were predicted to bind to the patient’s cognate HLA alleles.
Identified peptides were compared with a previously published list of melanoma related tumor rejection antigens [Andersen et al. (2012) Cancer Res., 72, 1642]. One peptide (VVQNFAKEFV) from the catalog of tumor associated antigens was identified in melanoma patient with the sample number 6. The identification however is of low confidence and will be validated with a synthetic peptide. In addition, a number of melanoma-associated antigen (MAGE) derived peptides have been identified (most of them with low confidence) and we are validating these identifications by comparing their spectra with the ones of synthetic peptides. A complementary validation activity relies on the stimulation of patient-derived lymphocytes with synthetic peptides.
At this stage, we believe that we have established a robust technology for the purification of sHLA complexes from patients’ sera. To the best of our knowledge, this is the first study on sHLA complexes from patients with solid tumors. We have selected individuals with the same haplotype and we have included a number of healthy individuals as control. In the coming months we will validate the MAGE peptides with synthetic peptides and we will try to validate some of the potential tumor rejection antigens with our collaborators at the University Hospital Tübingen.
We believe that the identification of HLA-I peptides from patients’ sera represents a convenient and efficient avenue for the experimental determination of an Atlas of peptides, some of which may represent tumor rejection antigens. The technology is ideally complemented by the analysis of T lymphocytes, using peptide stimulation or multiplex tetramer analysis, pioneered by the group of Ton Schumacher in Amsterdam.
T cell response against melanoma antigens
Philogen, the University Hospital Tübingen and the University Hospital Graz have investigated the therapeutic effect of intralesional administrations of L19-IL2 to patients with Stage IIIC melanoma, following a methodology pioneered by the group of Prof. C. Garbe in Tübingen. A first Phase II study in 25 patients has been completed and published [Weide et al. (2014) Cancer Immunol. Res., 2, 668].
Patients with stage IIIB/IIIC melanoma and cutaneous/subcutaneous injectable metastases received weekly intratumoral injections of L19–IL2 at a maximum dose of 10 MIU/week for 4 consecutive weeks. Tumor response was evaluated 12 weeks after the first treatment. Twenty-four of 25 patients were evaluable for therapy-induced responses. A complete response (CR) by modified immune-related response criteria (irRC) of all treated metastases was achieved in 6 patients (25%), with long-lasting responses in most cases (5 patients for >24 months). Objective responses were documented in 53.9% of all index lesions [44.4% CR and 9.5% partial responses (by irRC)], and 36.5% of these remained stable, while 9.5% progressed. Toxicity was comparable with that of free IL2, and no serious adverse events were recorded. A significant temporary increase of peripheral regulatory T cells and natural killer cells, sustained increase of absolute CD4+ T lymphocytes, and decrease of myeloid-derived suppressor cells were observed upon treatment. Finally, we recorded encouraging data about the progression time to distant metastases and overall survival.
The systemic immune responses induced by the immunocytokine L19-IL2 upon intratumoral injection of cutaneous/subcutaneous metastases in Stage IIIB/IIIC melanoma patients was evaluated as follows. We collected 45 blood samples from 17 patients at baseline and two different time points after injection of L19-IL2, PBMC were purified and stored at -80°C. After thawing, different populations of immune cells including CD4+ and CD8+ T cells, Myeloid-derived suppressor cells (MDSCs), Regulatory T cells (Tregs) and Natural Killer (NK cells) were evaluated using multicolor flow cytometry. MDSCs were characterized by the CD14+CD11b+HLA-DR–/low phenotype and NK cells as CD16+CD56low. The percentage of MDSCs and NK cells was calculated as the proportion of CD14+CD11b+HLA-DR–/low cells and CD16+CD56low cells within the total population of viable PBMCs.
For Tregs, which were characterized by the CD25+FoxP3+ phenotype, cells were intracellularly stained with FoxP3/Alexa647 (BD Biosciences). Frequencies of CD4+ and CD8+ T cells were determined within the total number of lymphocytes. The percentage of Tregs was calculated as the proportion of CD25+FoxP3+ cells within all CD4+ cells.
For NY-ESO-1- and Melan-A-specific T cell response detection, cells were stimulated with protein-spanning overlapping peptides (1 µg/mL). After culture for 12 days, T cells were restimulated at a ratio of 1:2 with autologous, fluorescent-labelled PBMCs either unpulsed (negative control) or peptide-pulsed with one of the antigens in the presence of Golgi-Plug for 12 hours. After blocking with Gamunex and labeling with ethidium monoazide, cells were fixed and permeabilized with CytoFix/CytoPerm and stained with the following antibodies: CD3/eFluor605, CD4/PerCP, CD8/APC-H7, IL-4/APC, IFN-γ/PE-Cy7, TNF/FITC, IL-10/PE, IL-17/PerCP-Cy5.5 and IL2/Alexa700. Samples were measured immediately using an LSR II and FACSDiva software and data were analyzed using FlowJo software.
The results showed a significant, transient increase in the frequency of CD4+CD25+Fox-P3+ Tregs in 15 of 16 evaluable patients after four weeks of treatment compared to baseline (mean frequency 12.1% vs. 8.7%; p<0.001). Frequencies returned to baseline values at day 85 (Figure 9A). There was a decrease of CD14+CD11b+HLA-DR–/low MDSCs from 11.5% at baseline to 9.0% at day 28 and 7.6% at day 85 (Figure 9B). Both the differences between baseline and end of treatment and between baseline and day 85 were statistically significant with p-values of 0.001 and 0.002 respectively. No changes in frequency were observed for NK-cells, CD4+ or CD8+ T cells.
The decrease of frequency of MDSCs in our patients might serve as an explanation for the beneficial mode of action of intratumoral IL-2, as suggested by a favorable distant metastasis-free survival as compared to historical controls.
Deliverables
D5.1) Implementation of HLA-I peptidome analysis in melanoma:
The Deliverable has been successfully completed, as documented by the successful identification of melanoma rejection antigens in soluble HLA-I/peptide complexes from patients’ sera.
Break-through methodologies, based on the work of the Tübingen group to study the reactivity of patient-derived T-cells against tumor-rejection antigens, revealed a complementary approach to study the relevance of HLA-I peptides in relation to anti-cancer immunity. We integrated both approaches in the PRIAT Project, thus complementing the MS-based analysis of soluble HLA-I peptidome.
WP6 – HLA Peptidome analysis in arthritis
The possibility to perform HLA-II analysis in the collagen-induced model of arthritis (CIA) in the mouse depends crucially on the availability of the animal model and on the ability to purify MHC-II peptide complexes (see WP2).
The CIA model has been established in Zurich and in Oxford.
HLA peptidome analysis in arthritis
After initial studies using lower amounts of serum, 5ml serum each from healthy DBA1 mice and mice with established collagen-induced arthritis (CIA) were subjected to the purification of MHC class II complexes following the methodology established on A20 lymphoma cells. Purification was performed with the M5/114 antibody and after excessive washing, complexes were eluted with 10% acetic acid. After purification of peptides, they were analyzed by mass spectrometry. The sample from heathy mouse serum clogged the HPLC column and could therefore not be analyzed. However, the sample from arthritic mice could be analyzed and a total of 147 peptides were identified with 5% false discovery rate. Moreover, the analysis of the supernatant the 10 kDa filtration confirmed that H2 I-Aq from healthy as well as from arthritic mice could be purified. 15 peptides from the alpha chain and 10 peptides from beta chain were identified indicating purification of intact MHC class II complexes. A selection of possibly MHC class II peptides is presented in Table 2. Several peptides originating from collagen were identified. Collagens are known to be degraded during CIA progression and are very promising candidate molecules for further validation. Other peptides are derived from intracellular proteins and membrane proteins, e.g. Lysosome-associated membrane glycoprotein 2, H-2 class II histocompatibility antigen gamma chain, Ras-related protein Rab-5C. These proteins have been either eluted from MHC class II of A20 cells, or represent very likely proteins to be presented by MHC class II complexes.
Emerging antibody therapeutics for the treatment of rheumatoid arthritis: a changing paradigm
Intact monoclonal antibodies (e.g. TNF blockers, pioneered by the colleagues at the Kennedy Institute of Rheumatology) represent some of the most successful therapeutic agents for the treatment of Rheumatoid Arthritis.
Within the framework of the PRIAT Project, we have extensively studied the therapeutic activity of “armed” antibody products (i.e anti-inflammatory immunocytokines, such as F8-IL10 and F8-IL4) in the murine CIA model of arthritis. Among the many interesting observations made in the course of these studies, we have found that F8-IL4, in combination with dexamethasone, is able to induce complete and lasting remissions in mice with arthritis (Figure 10). To the best of our knowledge, this is the first report of arthritis cures in rodent models of the disease. The new findings allowed a set of mechanistic investigations (see also WP9).
The immunosuppressant drug, dexamethasone and F8-IL-4 were previously shown to induce lasting remissions in arthritic mice (Hemmerle et al. (2014) Proc. Natl. Acad. Sci. U.S.A. 111, 12008). In contrast, the treatment of arthritic mice with either F8-IL-4 or dexamethasone alone did not lead to lasting remissions. To fully understand how this novel combination therapy works we examined the immune cell subpopulations in arthritic paws and lymph nodes of CIA mice treated with dexamethasone and F8-IL-4 monotherapy versus combination therapy. We found that arthritic mice that received F8-IL-4 monotherapy for 10 days had a significant increase in toleragenic TH2 cells and a decrease in the number of pathogenic TH17 cells in their paws. Interestingly F8-IL-4 monotherapy was also associated with a decrease in the percentage of anti-inflammatory Tregs in arthritic mice. On the other hand combination therapy not only lead to the reduction of pro-inflammatory TH17 cells but also a significant induction of Tregs and TH2 cell numbers which was associated with a significant decrease in arthritic score (Figure 11). In summary, combination therapy leads to lasting remissions in arthritic mice due to the effect on the balance between pathogenic, pro-inflammatory (TH17) and toleragenic, anti-inflammatory cells (TH2 and Tregs, Figure 12)
Deliverables
D6.1) Implementation of HLA-I and/or HLA-II peptidome analysis in arthritis:
The Deliverable has been successfully completed, as documented by the successful identification of MHC-II bound peptides from mice with collagen-induced arthritis.
Moreover, we have developed innovative approaches for the therapy of rheumatoid arthritis, which have resulted in complete cures in mice treated with F8-IL4 plus dexamethasone.
WP7 – HLA Class I peptidome analysis in transplant rejection
To investigate antibody therapies and to monitor their performance in the prevention of transplant rejection, a heterotopic rat heart transplantation model has been developed by the University Hospital in Jena.In this model, hearts are explanted from donor rats, subjected to cold ischemia at 4°C for 30 min, transplanted into recipient rats and harvested after 70 or 140 days respectively. For the first 14 post-transplant days, animals receive a sub-therapeutic dose of Cyclosporin A (2 mg/kg per day) to generate chronic rejection. For heterotopic heart transplantation two inbred strains were used: LEW as donors and F344 as recipients. Both strains are completely similar in the MHC RT1 genotype with only minor differences located on the RT1l gene locus for non-MHC allele. For this reason the rat combination LEW-F344 develops nearly no signs of acute rejection following heart transplantation but is a valuable model to investigate chronic rejection.
Plasma samples from rats have been collected by UHJ and have been used to determine the quantities of soluble MHC-I and II and ED-A+ fibronectin molecules in rats with chronic heart rejection in various antibody therapy conditions (see below).
Levels of soluble MHC proteins in a rat model of chronic heart transplant rejection
With respect to the planned HLA I and II peptidome analysis in serum after heterotopic rat heart transplantation, MHC I and II expression and shedding was investigated in relation to severity of rejection by immunohistochemistry and ELISA technology.
First, tissue expression of MHC I and MHC II was detected and quantified in shock frozen tissue samples of cardiac allografts by fluorescence immunohistochemistry using two commercially available antibodies specific to the rat MHC class I (OX-18) and II antigens (OX-6) provided by BD-Pharmingen. In a pre-therapeutic experiment a comparable increase in MHC I as well as MHC II expression in heart tissue in association to the severity of cardiac allograft rejection was evidenced. To answer the question if these histologically detected differences in MHC expression are also reflected by a concordant liberation of soluble MHC proteins into the serum (sMHC), commercially available ELISAs (Bio-Medical Assay). A comparative increased serum concentration of sMHCII could be evidenced. Comparable results were observed in the preventive therapeutic setting using F8-IL10 immunocytokines.
Levels of soluble ED-A+ fibronectin in a rat model of chronic heart transplant rejection
Besides its tissue expression and protein deposition into the cardiac extracellular matrix, EDA(+)-fibronectin is also released to circulating blood. Thus, the serum levels of the molecule might be a valuable biomarker for non-invasive surveillance of tissue remodeling in patients suffering from cardiovascular diseases. In the frame of the PRIAT project, 2 different ELISA protocols to detect EDA(+) fibronectin have been newly established: 1), ED A(+) fibronectin was captured with gelatin and specifically detected using the antibody IST-9 and a highly absorbed biotinylated secondary antibody and 2), the antibody IST-2 specific to the domain Fn12-14 was used as capturing antibody and a biotinylated IST-9 as detection antibody. In both methods, specificity was tested using purified ED A(+) fibronectin Fn as well as plasmatic fibronectin (pFn). Purified melanoma cell line derived ED A+ Fn was used to generate a standard curve. ELISA methods were applied for quantification of serum samples from patients and from rats after heterotopic heart transplantation. Both of the newly established ELISAs were specific to ED A(+) fibronectin whereas pFn could not be detected. The gelatin based ELISA worked in humans and rats. Compared to healthy humans, in patients with cardiovascular diseases and left ventricular dysfunction (n=133) significantly elevated serum concentrations of ED A(+) fibronectin could be measured. There were interesting correlations to clinical findings in the patient group. Using the rat heart transplantation model, serum concentrations of ED A+ Fn were decreased in transplanted rats exhibiting chronic rejection compared to non-transplanted controls. The explanation for this unexpected finding will be the object of further studies.
There is another interesting point to mention: In the rat model, the determination of ED A(+) fibronectin in serum after a single administration of the antibody F8 revealed significantly decreased levels compared to controls that have received an unspecific control antibody. Thus, a possible interference between the specific antibody and circulating ED-A(+) fibronectin should be taken into account when setting up therapeutic studies. This must be considered in therapeutic approaches.
Therapeutic intervention in a rat model of chronic heart transplant rejection
Philogen has provided clinical-grade F8-humanIL10 for the immunosuppressive therapy of chronic heart transplant rejection in rats. In addition to what originally described in the proposal, also F8-ratIL10 (a chimeric immunocytokine) and KSF-IL10 (a negative control immunocytokine, based on an antibody of irrelevant specificity in the rat) have been cloned, produced and used in therapy experiments.
UHJ has performed a first therapy experiment with the immunocytokines F8-humanIL10 (DEKAVIL), F8-ratIL10 as well as KSF-humanIL10 (irrelevant antigen-specificity). Treatment was performed weekly for 10 weeks starting at day 7 after transplantation (1mg/animal). Four experimental groups (PBS, KSF-IL10, DEKAVIL, F8-ratIL10; n=5 each) were compared with non-transplanted controls treated similarly. After completion of the experiment, organs and plasma were collected and subjected to further analyses.
In the non-transplanted control animals, macroscopic as well as histological analysis did not reveal any signs of tissue damage irrespective of treatment. In the cardiac allografts, treatment with DEKAVIL or F8-ratIL10 was associated by increased organ weights, a higher grade of chronic rejection, increased CIF, higher protein expression levels of alpha-smooth muscle actin (-SMA) as well as an augmented infiltration with inflammatory cells (CD4+, CD8+ and CD68+ cells) compared to the groups treated with neither PBS nor KSF-IL10. There were significant differences for the following parameters: the histological rejection score, CIF, alpha-SMA and the extent of CD4+, CD8+ and CD68+ cell infiltration levels. Compared to the DEKAVIL group, rejection-associated tissue damage was further augmented in the F8-ratIL10 group. There were no relevant differences between the KSF-IL10 and the PBS group. Serum brain natriuretic (BNP) levels, displaying heart failure grade, were higher in the groups treated with DEKAVIL and F8-ratIL10 compared to the groups that received PBS or KSF-IL10 respectively.
In another experiment, treatment was performed weekly for 10 weeks starting at day 70 after transplantation (1mg/animal), at a time point when chronic rejection has fully developed. Thus, the question should be addressed, if treatment was capable to reduce rejection. Three experimental groups (PBS, KSF-IL10, F8-ratIL10; n=4 each) were compared with non-transplanted controls treated similarly. After completion of the experiment, organs and plasma were collected and subjected to further analyses as described above. Taken together, when applying this treatment protocol, no relevant effects of administration of F8-ratIL10 could be observed compared to PBS, KSF-IL10 or untreated animals. These findings are speaking well for the fact that the time point of treatment initiation is of importance in this rejection model and should considered as a crucial experimental condition when planning further studies using F8 based immunocytokines.
MHC-I peptidome analysis in a rat model of chronic heart transplant rejection
After an in depth evaluation of existing literature on purification of MHC class I complexes from rat and antibodies against certain rat alleles and haplotypes, we decided that our best option is to obtain the OX-18 hybridoma cell line producing the antibody OX-18 directed against the MHC class I A equivalent in the mouse, RT-1A.
The hybridoma culture was established and the OX-18 antibody produced using low IgG serum. OX-18 could be purified from hybridoma cell supernatant after 6 days of culture with a final yield of 13 mg/L.
MHC class I complexes were purified from the lysate of 50 mg and 200 mg spleen tissue respectively following the procedure established for the purification of HLA class I complexes with W6/32 antibody. Mass spectrometric analysis led to the identification of a total of 125 peptides, 67 from 50 mg and 94 from 200 mg of tissue respectively. Moreover, the peptide length distribution shows a strong enrichment of 9mers and the evaluation of the c-terminal amino acid demonstrates an enrichment of histidine indicating the presence of a c-terminal anchor residue often observed for MHC class I molecules.
We therefore moved on to the purification of MHC class I from rat sera. Two small test experiments indicated, that we could also purify MHC class I from rat serum. A larger experiment was therefore performed purifying MHC class I from 3 ml of rat serum with different grade of rejection. After subtracting peptides identified in samples where BSA conjugated resin was used, 19, 33, 14, and 8 peptides were identified in healthy control, moderate, severe, and weak rejection respectively. Peptide length distribution demonstrated an enrichment of 9mers and at the peptide c-terminus histidine and leucine were significantly enriched (Figure 13).
After these promising experiments, we collected serum from a therapy study performed in the frame of the PRIAT project (see above). 144 peptides were identified in the six samples demonstrating the same pattern of enriched 9mers and mostly C-terminal histidine residue. Interestingly, the number of peptides identified from respective samples after subtracting known contaminants increases with severity of rejection. This is in line with previous findings indicating that the concentration of eluted MHC class I complexes from OX-18 resin is higher in samples of severe rejection as compared to healthy control sera. An increased concentration of soluble MHC class I complexes in serum has been described for other diseases, but it is until now unknown if this is also true for transplant rejection.
Deliverables
D7.1) Implementation of HLA-I peptidome analysis to rodent models of heart transplantation:
The Deliverable has been successfully completed, as documented by the successful identification of MHC-I-bound peptides in complex biological specimens, and by the rich series of experimental investigation in the rat model of chronic graft rejection.
WP8 – Assessment of targeting by Imaging
Immuno-PET is a break-through methodology, which is indispensible for the quantitative characterization of the localization properties of antibody-based therapeutics in normal tissues and at sites of disease.
The realization of immuno-PET studies with therapeutic antibodies has been made possible, thanks to recent work of VUMC, in collaboration with ETH and Philogen, for the radioiodination of antibodies with the PET radionuclide iodine-124 [Tijink et al. (2009) Eur. J. Nucl. Med. Mol. Imaging, 36, 1235-1244; Heuveling et al. (2013) J. Nucl. Med., 54, 397-401].
Phase 0 immuno-PET clinical trial in patients with rheumatoid arthritis
A novel Immuno-PET study with the 124I-F8-IL10 immunocytokine been launched at VUMC, in collaboration with Philogen. After receiving authorization from competent authorities, three patients with rheumatoid arthritis (RA) have been injected with the radioiodinated product and imaged by PET.
The Primary Objective of this study was to investigate the targeting performance (including pharmacokinetics/-dynamics) of 124I-F8-IL10 in patients with rheumatoid arthritis, in a phase 0 single dose study. The Secondary Objective was to investigate dosimetric parameters of 124I-F8-IL10 in arthritic joints and internal organs.
The recruited RA patients all had active disease (disease activity score of 28 joints > 3.2 at least two clinically swollen evaluable joints). Their anti-rheumatic treatment was stable for at least two weeks. After a single i.v. injection of 2 mCi (74 MBq) 124I-F8-IL10 (with an adjusted dose of 30 MBq in third patient), 3 whole body PET-CT scans and detailed scans of arthritic joints were performed at baseline, 24 and 72 hr p.i. Blood samples analysis for pharmacokinetics were withdrawn at several time points up to 72 hr p.i.
124I-F8-IL10 cleared very rapidly from the circulation. Within 30 min post-injection, less than 1% injected dose (ID)/kg was left in circulation, both in whole blood and plasma. Clear targeting of inflamed joints was obtained, with maximum uptake at 24 hr p.i. (Figure 14). Clinically affected joints showed higher 124I-F8-IL10 uptake than clinically unaffected joints.
Besides uptake in inflamed joints, dosimetric analysis revealed high 124I-F8-IL10 uptake in liver and spleen. Within 1 hour p.i. maximum liver/spleen uptake was found, decreasing with time.
Significantly increased uptake in liver and spleen was not previously found in tumor bearing mice [Schwager et al., (2009) Arthritis Res. Ther., 11, R142]. The increased liver and spleen uptake in RA patients resulted in a three times higher radiation burden of 0.39-0.46 mSv/MBq than predicted from animal data (0.14 mSv/MBq). To keep the radiation burden within acceptable limits according safety guidelines, the injection dose of 124I-F8-IL10 was lowered to 30 MBq in the third RA patient. Even with a three times lower injection dose, accumulation of 124I-F8-IL10 in inflamed joints could be clearly depicted on the images by prolongation of the scanning time.
Possible explanations for observed in vivo biodistribution differences of 124I-F8-IL10 between RA patients and animals are:
1. Radiolabeling differences (slightly different in animals and RA patients)
2. Species differences
3. F8-IL10 protein dose effects (microdose of PET in men versus higher dose in animals).
4. RA specific effects.
Irrespective of these considerations, a dosimetric analysis of arthritic and non-arthritic lesions nicely revealed a preferential accumulation of F8-IL10 in the affected lesions of patients (Figure 14).
Fluorescence imaging and immuno-PET studies in a rat model of chronic heart rejection with 124I-F8
To study the targeting properties of the F8 antibody in the heterotopic rat heart transplantation model to induce chronic cardiac rejection, imaging studies were performed by applying two different methods: 1), ex vivo near infrared fluorescence imaging (NIRF) and 2), in vivo PET detection.
1) Imaging studies using the fluorescently (DY-682) labelled SIP-F8 antibody and NIRF methodology were performed in transplanted (n=5) compared to non-transplanted rats (n=3) at day 69 after transplantation. Furthermore, a group of transplanted rats (n=5) were injected with the F16 antibody specific to the A1 domain of human tenascin-C, which is not reactive in the rats and therefore suitable as an isotype control. It could be proven that the F8 antibody selectively accumulates in chronically rejected hearts while sparing recipient hearts (heterotopic approach) [Franz et al.. (2013) J. Heart Lung Transplant., 32, 641-50].
2) To validate these findings and to test an imaging method that is also applicable in humans, we performed immuno-PET CT imaging in accordance. Therefore, transplanted animals received 124I-labeled SIP-F8 antibody and an isotype control as well (SIP-KSF). At an early time point after injection (1 hour), 124I-labeled SIP-F8 as well as 124I-labeled SIP-KSF were detectable with an accentuation in the circulation (blood pool), the liver, the kidneys and the urinary bladder. The cardiac allografts could be clearly confined in all rats. At a later time point, 24 hours after injection, the accumulation of 124I-labeled SIP-F8 was much more intensive compared to 124I-labeled SIP-KSF in the transplanted heart region (Figure 15). The findings are preliminary and have to be carried out including more animals.
Ancillary imaging and biodistribution studies with 124I-F8-IL10
In order to better understand the observed rapid blood clearance and high liver/spleen uptake of I-124I-F8-IL10 in RA patients, additional experiments were performed at VUMC, Philogen, Philochem and ETH.
Iodination method: addressing a potential effect of applied radioiodination method, additional studies in healthy mice have indicated that different modalities of F8-IL10 radioiodination (e.g. Iodogen vs. Chloramine-T and addition vs. elimination of tween20 in the formulation buffer of F8-IL10) do not have a substantial impact on the results of biodistribution studies. With either radioiodination method, there was no relevant uptake of 124I-F8-IL10 in liver and spleen.
Test species: the role of species differences on in vivo biodistribution was further investigated by injection of 124I-F8-IL10 in rats with antigen-induced arthritis. Interestingly, these experiments showed a higher uptake in liver and spleen of 124I-F8-IL10 as compared to a control 124I labeled cytokine-antibody (KSF) (Figure 16).
Other imaging and biodistribution studies
Interaction with serum protein: the interaction of F8-IL10 with RA serum was investigated ex vivo on HPLC. These data revealed a small shoulder on each side of the F8-IL10 peak which may point at some interaction, but this needs to be further explored in additional experiments.
Interaction with blood cells: the interaction with RA blood cells was studied ex-vivo by centrifugation assay as described by Doll et al. (Doll et al. Arthritis Res.Ther. 2013, 15 R:138). After incubation of 125I-F8-IL10 with fresh RA blood, cells and plasma were separated by centrifugation and counted separately. Data from this experiment show that F8-IL10 was retained at high percentage in the plasma (> 70%).
To further address RA specific contribution to uptake of F8-IL10 in liver and spleen, immunoPET treatment of patients with other chronic inflammatory disease(s) will also be planned.
Finally, dose effects of F8-IL10 on in vivo biodistribution, will also be investigated in future studies.
The data collected by the imaging and biodistribution experiments will aid in further understanding and development of F8-IL10 as promising novel therapeutic agent for RA.
In addition to the above-described clinical studies, a number of preclinical studies have been performed, including the following ones:
- radioiodinated preparations of several armed antibody products have been characterized by quantitative biodistribution studies in animal models of cancer
- radioiodinated preparations of various monoclonal antibodies have been compared in terms of their targeting performance in the mouse CIA model of arthritis, revealing that the F8 antibody has the best targeting potential for arthritic lesions
- similar arthritis targeting experiments have been performed with antibody preparations labeled with near-infrared fluorophores
Deliverables
D8.1) Imaging of patients and of animal models:
The Deliverable has been successfully completed, as documented by the execution of the Phase 0 clinical trial (in extremely rapid timelines: less than 2 years from submission to execution) and by the large number of complex preclinical imaging and biodistribution studies.
The WP ended up featuring many more experiments, in addition to the ones described in the Technical Annex. This WP is particularly important for the PRIAT Project, as it outlines how Precision Medicine approaches help develop next-generation antibody therapeutics in a controlled and predictive manner. It is one of the Flagship WPs and success stories of the PRIAT Project, which will have considerable visibility in the future.
WP9 – Analysis of biopsies
In this WP, we have analyzed material from patients and from animal models (e.g. biopsies, blood samples, plasma specimens) using complementary multiplex technologies.
In cancer (i.e. patients in clinical trials or mouse models treated with antibody therapeutics), we have aimed at:
- evaluating variability of target antigen expression and leukocyte infiltration in freshly-frozen tumor sections, using several antibody reagents
- analyzing infiltrating leukocytes by FACS analysis of resuspended cell preparations
- analyzing cytokine levels
- analyzing the T cell reactivity against tumor-rejection antigens
In mouse models of arthritis (following antibody-based treatment), we have aimed:
- evaluating CytoF technology for joint and lymphnode cells
- analyzing infiltrating cells by FACS
- analyzing cytokine levels by MesoScale Discovery
In the rat model of chronic heart transplant rejection, we have assessed the rejection grade using a defined scoring system adapted to the rat model including the extent of lymphocyte infiltration, myocyte degeneration, fibrosis and vasculopathy as well [Franz et al. (2011) J. Heart Lung Transpl., 30, 86]. Moreover, lymphocyte infiltration have been characterized with regard to the different subtypes of immune cells as well as their spatial tissue distribution (e.g. association to histological signs of chronic rejection: fibrosis and vasculopathy).
While a more complete list of all performed activities can best be derived from the data described in the publications (listed below), here we would like to outline some of the most significant findings, which have emerged from studies in cancer, arthritis and chronic heart rejection.
Use of the immunocytokine L19-IL2 for the intratumoral injection of cutaneous/subcutaneous metastases in Stage III melanoma patients.
As mentioned in WP3, clinical trials in patients with Stage IIIC and Stage IV melanoma have been performed, featuring intralesional administration of either L19-IL2 alone (25 patients), or of a combination of L19-IL2 and L19-TNF (22 patients). The latter Phase II clinical trial was motivated by the observation of strikingly potent activity in immunocompetent mouse models of cancer [Schwager et al. (2012) J. Invest. Dermatol., 133, 751; Pretto et al. (2014) Cancer Immunol. Immunother., 63, 901].
In a first study, we addressed the use of L19IL2 for the intralesional treatment of Stage IIIB/C melanoma patients [Weide et al. (2014), Cancer Immunol. Res. 2, 668; Figure 17]. We reasoned that a targeted form of IL2 injected intralesionally into melanoma metastases would exhibit prolonged residence time in the lesions [Pretto et al. (2014) Cancer Immunol. Immunother., 63, 901], as compared to the untargeted form, therefore allowing an extended immunological action, reduction in the frequency of administrations and a shorter duration of the treatment.
A complete response of all treated metastases was achieved in 6/24 evaluable patients (25%), long-lasting in most cases (5 patients ≥ 24 months). Objective responses were documented in 53.9% of all index lesions.
We also recorded a slower progression to distant metastases in this cohort of patients as compared to historical controls [Weide et al. (2014), Cancer Immunol. Res. 2, 668].
The second study describes, for the first time, the clinical use of a combination of two immunocytokines for the intralesional treatment of injectable lesions in Stage IIIB/C and Stage IVM1a metastatic melanoma patients. Twenty-two patients were treated with a mixture of L19-IL2 and L19-TNF once weekly for up to 4 weeks. The dose was shared among the lesions via multiple intralesional injections and newly occurring melanoma lesions in the regional field (in-transit metastases) within the 4 weeks treatment period were also treated as described. Treatment was in general well tolerated, with a limited number of drug-related adverse events, mostly of grade ≤ 2, which were normally rapidly resolved. In terms of efficacy, although no CR was observed according to modified RECIST criteria, PRs were recorded in 50 % of patients. A locoregional assessment of individual lesions indicated a better outcome of the study, with 5% of CR and 50% of PR. Importantly, according to these criteria, no patient was considered to be in PD at the time of tumor assessment. Interestingly, responses in some patients manifested themselves or continued to improve several weeks after the end of treatment. Furthermore, the observation of disappearance or reduction in size of 8 out of 12 lesions, which were not injected, suggests that intralesional injection of the L19-IL2 in combination with L19-TNF mediates a systemic anti-cancer activity (Figure 18A).
Histological and immunohistochemical findings in two different patients showed presence of extensive necrosis and melanophages in treated lesions while in untreated lesions only small areas of necrosis were apparent (not shown). Infiltration by CD8+ T cells was markedly increased in treated lesions as compared to untreated (Figure 18B), not responding ones, while CD4+, Tregs and NK cells infiltration of treated lesions was not significantly changed as compared to untreated, not responding lesions (manuscript submitted).
Analysis of melanoma biopsies in other types of antibody-based treatments
The University of Graz has collected biopsies from patients before and after anti-melanoma therapy. Staining protocols for immune modulating molecules expressed on cancer cells; i.e. PD-L1, MICA/B and HLA have been established for FFPE samples. Methods to detect markers creating an immune-promoting or immune-suppressive tumor microenvironment (e.g. CD68, CD163, iNOS, Arginase, IDO), as well as possible markers creating a respective microenvironment in the tumor host (e.g. immune suppressive cytokines such as IL4 or IL10, circulating MDSCs, Tregs, Th17 cells) have also been established (IHC, pPCR, FCM).
These analyses demonstrated that presence of an immune suppressive microenvironment characterized by low MICA/B, MHC class I and a high Arginase, IDO, iNOS, CD163 expression was associated with a weak to absent intratumoral CD8+ T cell infiltrate and vice versa. Interestingly, expression of the immune-inhibiting PD-L1protein, or presence of FOXP3+ Tregs, was associated with a pronounced CD8+ T-cell infiltrate. Unfortunately, the number of available sequential lesions, i.e. before and after an immune modulating therapy (which was mostly ipilimumab) was limited. Consequently, a statistical analysis with respect to whether these results can be used as predictive biomarker was not feasible. However, there was a clear trend that tumors characterized by an immune promoting microenvironment or were characterized by an intratumoral T-cell infiltrate showed a more pronounced response to ipilimumab therapy, e.g. an even further increased T-cell infiltration.
Analysis of the circulating immune regulatory cells revealed that only the frequency of Th17 cells correlated with an immune promoting microenvironemt. It should be noted, however, that this was only a trend, possibly to the limited sample size. The quantification of immune suppressive cytokines were characterized by strong intra-individual variations (e.g. when several samples obtained prior to therapy from the same patient); thus, these were not used for further analysis.
Analysis of biopsies, leukocyte infiltration and cytokine levels in preclinical models of cancer
ETH and Philochem have performed an extensive analysis of bioptic specimens and of cytokine levels in therapy experiments with immunocytokines, executed in mouse models of cancer. Products which have been tested in vivo and which have been subjected to the above mentioned analysis include immunocytokines based on IL2, IL4, IL12, IL13, IFN-g and TNF. A list of publication is provided at the end of this section, in which leukocyte infiltration into tumor lesions and cytokine levels were measured. Figure 19 illustrates a typical analysis of leukocyte infiltration in therapy experiments, performed in mouse models of cancer.
The measurement of lymphocyte infiltration into the tumor mass, alone, does not provide a functional readout. However, antibody therapy studies in tumor-bearing mice, for which a selective depletion of CD4+, CD8+ or NK cells had been performed, have shed light on the classes of lymphocytes, which contribute to an anti-cancer protective immunity, which can be boosted by antibody therapeutics.
Analysis of biopsies, leukocyte infiltration and cytokine levels in mouse models of rheumatoid arthritis and psoriasis
To gain some insight about the mechanism of action of F8-IL4 in the murine collagen-induced model of rheumatoid arthritis and three different skin inflammation models of psoriasis, samples were analyzed for change in cytokine levels due to the treatment with targeted IL4, a strong TH2 cytokine. A panel of 13 different pro- and anti-inflammatory cytokines that play a role in TH1-, TH2- and TH17-responses were analyzed in serum and at the site of disease (e.g. in paw tissue for arthritis and in ear tissue for psoriasis). The infiltration of immune cells due to therapeutic treatment was evaluated by immunofluorescence procedures.
In arthritis, F8-IL4 elevates certain anti-inflammatory cytokine levels such as IL10 and IL13, while other pro-inflammatory cytokines such as IL6 are down regulated in serum. In the paw tissue, F8-IL4 shifts the balance of many cytokine and therefore modulates the immune environment at the site of disease (Figure 20).
The combination of targeted IL4 and dexamethasone was able to cure mice from established arthritis. The analysis of cured paws compared to healthy paws and active arthritic paws showed that the treatment normalized cytokine levels to the levels of healthy mice (Figure 21).
In murine models of psoriasis, treatment with F8-IL4 led to increased levels of key regulatory cytokines (including IL5, IL10, IL13, and IL27), while other cytokine levels remained unchanged (e.g. IL2) or are down regulated (e.g. IL1alpha), when compared to buffer control treated mice (Figure 22). F8-IL4 also mediated an influx of immune cells such as macrophages and NK cells to the site of inflammation as assed by immunofluorescence methods (see above).
The measurement of cytokines and identification of immune cell populations at the site of disease remains an important read-out when testing new therapeutic strategies. Unfortunately, when using animal models we are limited by cell numbers found at the site of disease as well as the amount of serum taken from individual mice. In addition when examining cells in arthritic paws, antigen markers can be altered during the paw digestion process. In our lab we have optimized a DNase and Liberase digestion strategy flowed by a flow cytometric based technique which reliably examined immune cell populations present in arthritic paws (Figure 23). In addition we established the Meso Scale Discovery (MSD) platform to analyse multiple pro- and anti-inflammatory cytokines present in a small volume (10 cytokines in 25 μL). These two techniques have allowed us to examine the effect of F8-IL-4 on cytokines and immune cells in the paws of arthritic mice (WP6).
Analysis of biopsies, leukocyte infiltration and cytokine levels in a rat model of chronic graft rejection
To characterize the chronic rejection process in the heterotopic rat heart transplantation model under qualitative and quantitative aspects, besides the histological and morphometric analysis of surrogate markers of chronic rejection like cardiac allograft vascolupathy or cardiac interstitial fibrosis, also immunohistochemical determination of important immune cells (CD4+ - T helper cells, CD8+ - cytotoxic T cells; CD68+ - natural killer cells; CD161+ - macrophages etc.) rejection markers (i.e. C4d) as well as key molecules of stromal activation (i.e. alpha-SMA) were performed in detail in the chronically rejected cardiac tissue compared to healthy controls. In general, a significant up-regulation of all investigated immune cell markers as well as the rejection marker C4d and also alpha-SMA, as a marker of activated myofibroblasts or vascular smooth muscle cells, could be observed compared to controls. Moreover, using RNA extracted form rat cardiac tissue was subjected to RT-PCR based gene expression analysis of key molecules related to cardiac extracellular matrix and fibrosis. The complex results cannot be described in this but will be analyzed in detail in the future with respect to the identification of novel key genes related to the rejection process.
Analysis of predictive biomarkers and surrogate markers for outcome of melanoma patients treated with the CTLA-4 antibody ipilimumab
Only insufficient biopsies were available for systematic analysis of biomarkers in melanoma patients treated with antibody-based treatments. Therefore, we focused to identify predictive biomarkers and surrogate markers for outcome of melanoma patients receiving the CTLA-4 antibody ipilimumab in the peripheral blood.
The spectrum of biomarkers to be tested comprised frequencies of immune cell subsets in the peripheral blood, differential blood count parameters (e.g. the absolute and relative lymphocyte count), serum level of lactate dehydrogenasis (LDH) and clinical factors (stage according to the AJCC classification, age, gender). LDH-ratio was calculated by the measured value divided by the upper limit of normal in the respective test system. Immune cell subsets were analyzed using flow cytometry after staining of peripheral blood mononuclear cells (PBMC) with up to four different panels of fluorescence-labeled antibodies (s. Table 3) to cover a broad spectrum of phenotypes to be investigated.
Overall survival according to biomarker result was tested by Kaplan-Maier analysis. Different groups regarding a single factor were compared using the log-rank test. The best clinical response according to immune-related response criteria (irRC) upon CTLA-4 therapy is presented by descriptive statistics and Chi-square tests. For all patients, blood taken at least 28 days before start of antibody therapy was used for this study.
An initial cohort of 25 patients (cohort 1) had already been analyzed prior to start of PRIAT. For all continuous factors patients were dichotomized into two groups based on the median value observed in this cohort (group 1: <= median, group 2: > median) to analyze the impact of each biomarker. Significant associations with favorable survival were detected for high relative frequencies of eosinophils and for low frequencies of Lin-CD14+HLA-DRlow myeloid-derived suppressor cells (Table 4 left).
Based on these promising results, we set up a large collaborative study to confirm the initial findings identified in cohort 1. Altogether 342 PBMC samples were collected for this cohort from 8 international clinical sites (New York, Amsterdam, Siena, Napoli, Lausanne Nantes, Essen and Tübingen) comprising 184 baseline PBMCs and 158 PBMC specimen collected during and after antibody treatment.
The first part of the PRIAT project focused on the analysis of 184 baseline samples to identify predictive markers. The entire cohort was divided into an identification cohort (cohort 2: n=85 baseline samples from Amsterdam, Lausanne, Nantes, Essen and Tübingen) and a validation cohort (cohort 3: n=99 samples from New York, Siena and Napoli).
First, 22 continuous factors were analyzed using two cut-off points for each biomarker. The first cut-off point was defined by the median observed value in cohort 2 (dichotomization method). The second cut off point was that which resulted in the most significant differences in survival comparing the high vs. low group of the respective biomarker (optimization method). Gender, the clinical substage were additionally assessed. Altogether 24 candidates were significantly associated with survival in cohort 2. Next, all variables using the cut-off points established in cohort 2 patients were tested in cohort 3. Among the 24 candidates 7 were confirmed in cohort 3 (Table 4).
Multivariate analysis was applied for 134 patients of cohorts 2 and 3 with available results for all seven confirmed biomarkers to investigate their degree of independency and their relative impact if analyzed in combination. Cox regression analysis resulted in three independent biomarkers predictive for increased risk to die from melanoma: LDH-ratio (hazard ratio [HR] 2,364, p=0,00012), the relative eosinophil count (HR 1,962; p=0,00673) and Lin-CD14+HLA-DRlow myeloid-derived suppressor cells (HR 2,652; p=0,00002).
The differences in overall survival for those three biomarkers and their combined consideration is shown for the entire cohort (cohorts 2 & 3) in Figure 24. An overview over univariate survival data according to analyzed biomarkers is presented in Table 5.
The best clinical and radiological response upon treatment is demonstrated in Figure 25. A response according to irRC was significantly more frequent in patients with low LDH or low frequencies of MDSCs. No significant difference could be established for relative eosinophils. The more out of these three favorable factors (low LDH-ratio, high eosinophils, low MDSCs) were observed in patients the higher was the rate of responders (Figure 26). The manuscript of this part is in preparation and will be submitted in Q1 2015.
In the second part of this PRIAT project we are analyzing changes of the same spectrum of biomarker upon treatment by comparing the baseline findings acquired in part A to the results from later time points during and after ipilimumab. Post baseline samples were analyzed in 85 patients. Data acquisition is completed. Data analysis and interpretation is still ongoing. We intend to submit a manuscript focusing on surrogate markers by Q2 2015.
In the third part of this PRIAT project we focus on the predictive impact of γ/delta T cell subsets in patients treated with ipilimumab. Data acquisition is completed. Data analysis and interpretation is still ongoing. We intend to submit a manuscript focusing on γ/delta T cell subsets by Q2 2015.
Deliverables
D9.1) Analysis of biopsies and of biological specimens:
The Deliverable has been successfully completed, as documented by the extensive number of publications and by the analyses, performed both in patient-derived specimens and in rodent models of disease.
WP10 – Project Management
A kick-off meeting has been organized at the beginning of the Project, in order to coordinate the activities and the distribution of resources, reminding all participants of the rights and duties associated with the participation in the PRIAT Project (Figure 27).
After 12 months (i.e. 11+12 November 2013), a second meeting has taken place in Zurich, with active presentations and discussions by the groups, as well as a review of the current spending and of the status of deliverables (Figure 28).
At the end of the 24 months of the Project (i.e. 27 + 28 October 2014), a final meeting has taken place in Zurich (Figure 29), with active presentations by the individual groups, as well of experts in the field of cancer immunology and pharmacotherapy [Prof. Dr. Raffaella Giavazzi (Mario Negri Institute of Pharmacological Research, Milano, Italy), PD Dr. Christoph Schliemann (University Hospital Münster, Germany), Prof. Dr. Cornelia Halin-Winter (ETH Zürich, Switzerland) and Prof. Dr. Sine Reker-Hadrup (Technical University of Denmark)].
A visit of GMP production facilities of Philogen and other training activities have been organized during the 24 months of the Project (Figure 30).
Furthermore, various in-person meetings (including several internships) have been organized among the laboratories. These personal contacts have been fundamental for the successful completion of the Project.
At each Meeting, a General Assembly took place, as foreseen by the Consortium Agreement. There were no major deviations from the original scientific plan, as the Project ran smoothly all deliverables could be produced within the expected time period. There were small reallocations of the original budget (7.5%). Specifically, MUG used 119771.00 € less and VUMC 105242.00 € less of their budget and these funds were re-allocated to other participants, as decided in the General Assembly and as communicated to all partners in writing.
Clinical trials and the use of clinical samples has been approved by the competent Authorities and Ethical committees, with suitable Patient information forms
An Ethics Review Committee has followed our Project and participated at the yearly Meetings
WP11 - Dissemination Activities
A Web site for the PRIAT Project was set up at the beginning of the Project [www.priat.eu].
Some of the most significant milestones reached by the PRIAT Project have been communicated to the broader public.
A detailed list of Publications and of Presentations at national and international conferences is provided in a separate document.
Potential Impact:
Potential Impact
The aim of Project in relation to the Problem
The vast majority of pharmaceutical agents can be divided into two categories: small organic drugs (typically < 500 Da) and biotech drugs, mainly consisting of therapeutic proteins. About 50% of product sales in this latter category is represented by antibody-based products. Seven out of ten of top-selling therapeutic proteins are antibodies or antibody derivatives. Furthermore, Humira™ (a monoclonal antibody directed against TNF and approved for the treatment of various inflammatory conditions) is the best-selling pharmaceutical agents in all categories [Walsh (2014) Nat. Biot., 32, 992].
In spite of their commercial success, antibody products are not always as active as one would like them to be. For example, Avastin™ and Herceptin™, used in combination with chemotherapy, have been approved for the treatment of metastatic colorectal cancer and of metastatic breast cancer, respectively, on the basis of a prolongation of median overall survival for only few months. However, patients treated with these antibody drugs are typically not cured and the few extra months of life come at a very high cost for the healthcare system. TNF blockers (such as Humira™, Enbrel™ and Remicade™) provide a substantial benefit to about 50% of patients with rheumatoid arthritis. However, these drugs do not cure the disease. Thus, there is a need to develop better antibody-based drugs and to understand which patients are likely to benefit from antibody therapy most.
In the PRIAT Project, we focused on three main indications for antibody-based therapeutics (cancer, rheumatoid arthritis and graft rejection), with the aim to experimentally address three issues, which are crucial for the understanding of antibody activity and for the design of the drugs of the future:
- How efficiently do antibody drugs accumulate at the site of disease?
- How efficiently can antibody drugs modulate the activity of the immune system at the
site of disease?
- How efficiently can we profile the HLA-I and HLA-II peptidomes in health and disease, in order to be able to identify the targets of T-cell recognition (e.g. tumor rejection antigens, graft rejection antigens, autoantigens for autoimmune conditions)?
Within its two years of activity, the PRIAT Project has made a considerable impact in all three areas mentioned above. Here, we summarize the main findings and the main results.
Characterization of the disease-targeting performance of antibody products
In animal models of disease, we have performed extensive quantitative biodistribution studies, using radiolabeled preparations of therapeutic antibody products. This work has led to numerous publications and, collectively, has provided us with a quantitative understanding of the selectivity which antibody products can reach in vivo in rodent models of cancer, rheumatoid arthritis and graft rejection.
A priori, there is no guarantee that biodistribution and imaging studies in animal models may faithfully reflect the performance of the same products in patients. In order to bridge this gap of knowledge, we have made two contributions to the field, which we believe to be important:
(i) we have published a dosimetric estimate of radiation doses in organs and neoplastic lesions for a pan-tumoral antibody (L19, specific to the alternatively-spliced EDB domain of fibronectin), which has been studied in patients with solid tumors or with lymphomas. These studies have shown that the L19 antibody preferentially localizes on neosplastic lesions, but also that the tumor-homing selectivity varies greatly (i.e. up to 40-fold) from lesion to lesion and from patient to patient
(ii) we have developed and clinically implemented a methodology for the radiolabeling of antibody products in GMP conditions with iodine-124, a PET radionuclide which is particularly suitable for the study of antibody products in PET procedures. The methodology has been used for the characterization of Dekavil (F8-IL10), an armed antibody developed by Philogen, currently being investigated in randomized doubleblind Phase IIb clinical trials in patients with rheumatoid arthritis. In addition to the specific information gained in this study about the arthritis homing properties of the product, the Project has demonstrated the feasibility of mechanistic immuno-PET studies, which should facilitate product development in the future
Multiplex analysis of biomarkers and of leukocyte infiltration
We have implemented the routine analysis of leukocyte infiltration at the site of disease and the multiplex analysis of cytokine levels (in serum and at the site of disease) for many preclinical experiments, aimed at the characterization of the activity and mechanism of action of various antibody-based therapeutics. These studies have been published in several articles and have facilitated the identification of product candidates for clinical and industrial development programs (see below).
Furthermore, we have studied in substantial detail the infiltration of lymphocytes into melanoma lesions in patients receiving antibody drugs and the specificities of cytotoxic T-lymphocytes which account for long-term survival, following pharmacological treatments. These investigations have led to the identification of certain tumor rejection antigens (e.g. NY-ESO-1), which are efficiently recognized by T cells in patients that respond well to treatment. Furthermore, the demonstration that certain antibody-based therapeutic interventions (e.g. treatment with IL2-based immunocytokines) led to a dramatic increase of certain leukocytes (e.g. T cells and NK cells) in the neoplastic mass has prompted the design of novel combination therapies, which are about to be studied in clinical trials (see Sections below).
HLA peptidome analysis
We have developed, implemented and perfected methodologies for the determination of “atlases” of peptides bound to MHC-I or MHC-II (HLA-I and HLA-II in man). Thanks to the use of state-of-the-art mass spectrometers and to the implementation of efficient antibody-based pulldown procedures, these methodologies have allowed the determination of complex peptidomes, comprising hundreds of MHC-bound peptides, in complex biological samples, such as cell lines or patients’ sera. For the first time, it has been possible to identify hundreds of HLA-I-bound peptides from sera of melanoma patients. The identified peptides included previously reported melanoma rejection antigens, thus providing confidence about the robustness of the procedure. In mice with arthritis, collagen-derived peptides were identified as MHC-II ligands, thus providing a direct analytical access to antigens which may trigger autoimmune conditions. The applications opened by the successful implementation of methods for the determination of HLA peptidomes are numerous and, in our opinion, important. For example, on the basis of the knowledge of HLA-I peptidome in melanoma patients, it now becomes possible to synthesize the corresponding peptides and probe T-cell specificities in the same patients, using either peptide based stimulation or multiplex tetramer technology, pioneered by Ton Schumacher in Amsterdam.
Furthermore, we now have the opportunity to go back to experiments, in which we could cure tumor-bearing mice with innovative antibody-based therapeutics (e.g. immunocytokines based on IL2 or TNF payloads) and determine which are the tumor rejection antigens recognized by cytotoxic T-cells, providing a protective immunity against the tumor.
The importance of HLA peptidome analysis and of multiplex tetramer technology analysis for the study of the response of melanoma patients to antibody drug treatment [Kvistborg et al. (2014) Sci. Transl. Med., 6, 254ra128] has become a particularly important and timely topic, with the approval of anti-CTLA4 and anti-PD1 immunostimulatory antibody products.
Socio-economic impact
We believe that the PRIAT Project has produced important technologies and publications, which facilitate the development and mechanistic study of antibody drugs. In time, this will lead to the generation of better antibody products and to improved selection procedures for patients, which are more likely to benefit from therapeutic intervention.
In addition to these rather indirect (but important!) results, we believe that the PRIAT Projects has also produced results, with an immediate socio-economic impact. First of all, patients with melanoma, leukemia, sarcoma and rheumatoid arthritis have been studied in clinical trials. Their response to antibody-treatment has been analyzed with improved methodologies, which were not available so far. A direct clinical benefit has been demonstrated, at least for patients with oncological conditions which have responded to antibody treatment.
Secondly, the information gained on the ability of certain armed antibody products (e.g. those containing IL2 or TNF payloads) to stimulate a massive increase of T cells and NK cells in the neoplastic mass, has stimulated the design of novel combination clinical trials (e.g. L19-IL2 + Rituximab in last-line diffuse large B-cell lymphoma), which are now about to start.
Thirdly, the knowledge gained for the understanding of the disease-homing and immunoregulatory properties of armed antibodies has facilitated the development of novel potent therapeutic agents. For example, we now have armed antibody products which induce complete responses in mouse models of soft-tissue sarcoma and in patients (L19-TNF). Moreover, we have recently published the first example of a therapeutic agent, capable of curing 100% of treated mice with established rheumatoid arthritis [Hemmerle et al. (2014) Proc. Natl. Acad. Sci. U.S.A. 111, 12008]. This product is now completing safety toxicological testing in Cynomolgus monkeys and is scheduled to begin clinical trials in 2015 in patients with rheumatoid arthritis and with endometriosis.
Finally, by supporting a collaboration between academic centers, hospitals and biotech companies, the Project has reinforced the competitiveness of European biomedical research in terms of knowledge, product development and clinical translation. The Philogen and Philochem groups have grown, both in terms of revenues and in terms of number of employees, since the beginning of the PRIAT Project.
Wider societal implications
There are many direct benefits, which have stemmed from the PRIAT Project, including the training of European students at our centers and exchange programs. For example, we have had numerous exchanges of visiting scientists and we have organized visits to the GMP Production Facility of Philogen, thus providing a “non-standard” training to students on activities of profound pharmaceutical importance.
For the broader society, the main impact of the Project is related to the knowledge generated in these two years of experimental work (see next section on Dissemination Activities) and to the industrial and clinical activities, which have been facilitated by the PRIAT Project (see sections below).
In general, our impression is that the European Pharmaceutical and Biotech Industry is losing competitiveness towards other companies located in the United States or in emerging countries. However, projects like PRIAT, which allow a close interaction between experts in academia, hospitals and biotech companies, can stimulate nice synergies, resulting in the development of new products and in the creation of new jobs.
In the medium and long term, we believe that the main impact of the PRIAT Project for the broader society (in Europe and elsewhere) will be related to the implementation of better patient selection procedures (based on imaging, multiplex analysis and HLA-peptidome analysis) and to the development of better antibody products.
Main dissemination activities
The main avenues for the broad dissemination of scientific results have been:
- publications in peer-reviewed scientific Journals
- presentations at national and international conferences
In total, 18 publications on topics directly related to the PRIAT Project have been published by the Consortium, while additional manuscripts are currently in preparation.
In total, 64 presentations at national and international conferences have been given by scientists of the PRIAT consortium between November 2012 and October 2014.
At the end of the Project, a mini-Symposium on the specific topics of the PRIAT Project has been organized in Zürich. Furthermore, the European Union has covered the interim results of the PRIAT Project in a dedicated Inteview, which has appeared on the Horizon 2020 Website:
https://ec.europa.eu/programmes/horizon2020/en/news/priat-probes-how-antibody-drugs-help-body-cure-itself
Exploitation of the results
As outlined above, the main objective of the PRIAT Project was the development of innovative methodologies and their implementation in preclinical and clinical studies. The methodologies per se were not patentable, but the information gained by the application of the methodologies has led to the discovery (and patenting) novel antibody products and of novel combination strategies.
We believe that the PRIAT Project has facilitated a number of tangible and important development activities, some of which are listed below:
i) clinical implementation of immuno-PET procedures, as integral component in the development of armed antibody drugs
ii) knowledge stemming from the dosimetric evaluation of disease-targeting properties of antibody drugs in preclinical studies and in immuno-PET clinical trials
iii) characterization of the mechanism of action of novel anticancer antibody-based therapeutics, which has led to the improved use of antibody products which are already in the clinic (e.g. L19-IL2, F16-IL2, L19-TNF, as well as immune checkpoint inhibitors), or to the launch of new clinical trials (e.g. L19-IL2 + rituximab in DLBCL, L19-IL2 + L19-TNF for the locoregional treatment of Stage IIIC melanoma and of other malignancies)
iv) discovery of the first agent (an armed antibody: F8-IL4) capable of eradicating established arthritis in the collagen-induced model of the disease. The corresponding fully-human immunocytokine will begin clinical trials in 2015
v) development of antibody-based procedures, for the imaging of failing heart in a rat model of chronic graft rejection. This methodology has the potential to provide a non-invasive detection of rejection processes by SPECT or PET methodologies, perhaps avoiding the need for serial biopsies (the current methodology for the monitoring of patients, following heart transplantation)
vi) the development of predictive methods for long-term survival of melanoma patients, following pharmacological intervention (typically with antibody drugs, but also with other agents)
In summary, we believe that the PRIAT Project has been truly successful, not only in terms of methodology development but also for the boosting of European Biotech competitiveness, with novel antibody products being brought to the clinic and improved analytical procedures for the study of response to antibody treatment.
List of Websites:
website: www.priat.eu
Therapeutic antibodies represent the largest and fastest growing class of Pharmaceutical Biotechnology. They are expensive treatments, which can provide a considerable benefit to some (but not all!) patients in various therapeutic areas. At present, the main pharmaceutical applications of antibody therapeutics are in the field of oncology and of chronic inflammation (e.g. rheumatoid arthritis). However, antibody products are also often used to prevent graft rejection in transplantation and increasing development activities are being performed in the cardiovascular field (e.g. PCSK9 blockade).
The PRIAT Project aimed at the development and implementation of advanced methodologies for the profiling of responses in antibody therapies. This goal has been achieved by three main lines of activities:
- HLA Peptidome analysis of biological specimens (including plasma specimens). A characterization of the T cell response to disease and treatment is crucial for the understanding of immunomodulatory pharmaceutical strategies. As T cells recognize peptides presented on MHC molecules (HLA in human), a characterization of the HLA peptidome is an indispensible step, towards the profiling of T cells (e.g. by means of multiplex tetramer technologies)
- Innovative imaging techniques (e.g. PET and near-infrared fluorescence imaging). Not all therapeutic antibodies efficiently localize at the site of disease, which often represents a requirement for pharmacological activity. It is therefore crucially important to integrate non-invasive imaging techniques in the antibody development process.
- Multiplex analysis of specimens (e.g. sera, biopsies, leukocytes), in order to gain functional information about the status of immune activation and/or cytokine levels. A detailed characterization of the mechanism of action of immunomodulatory agents facilitates product development and a profiling of responses in patients.
All three activities have been successfully completed.
We have developed innovative methodologies, which allow the identification of more than 1’000 peptides bound to MHC-I and MHC-II molecules. Importantly, hundreds of HLA-I-bound peptides could be identified from the plasma of patients with melanoma (one of the most immunogenic tumors), thus paving the way to a non-invasive profiling of peptides, which could act as tumor rejection antigens in immunotherapeutic strategies.
We have performed the first Phase 0 immuno-PET clinical trial worldwide in patients with rheumatoid arthritis, thus defining a pathway for the image-based development of antibody products in the future. In addition, we have performed a large amount of preclinical imaging applications, which have facilitated the selection of clinical development candidates.
Finally, we have extensively studied pathological specimens in man and in rodents, using multiplex biomarker analysis and microscopic techniques, shedding light on mechanism of action.
Project Context and Objectives:
In the PRIAT Project, we focused on three main indications for antibody-based therapeutics (cancer, rheumatoid arthritis and graft rejection), with the aim to experimentally address three issues, which are crucial for the understanding of antibody activity and for the design of the drugs of the future:
- How efficiently do antibody drugs accumulate at the site of disease?
- How efficiently can antibody drugs modulate the activity of the immune system at the site of disease?
- How efficiently can we profile the HLA-I and HLA-II peptidomes in health and disease, in order to be able to identify the targets of T-cell recognition (e.g. tumor rejection antigens, graft rejection antigens, autoantigens for autoimmune conditions)?
The PRIAT Project has allowed our Consortium to make considerable progress in three key areas (HLA peptidome analysis, imaging and multiplex analysis of patient-derived specimens), which are of fundamental importance for the characterization of the activity of therapeutic antibodies in patients.
We have developed innovative methodologies, which allow the identification of more than 1’000 peptides bound to MHC-I and MHC-II molecules. Importantly, hundreds of HLA-I-bound peptides could be identified from the plasma of patients with melanoma (one of the most immunogenic tumors), thus paving the way to a non-invasive profiling of peptides, which could act as tumor rejection antigens in immunotherapeutic strategies.
We have performed the first Phase 0 immuno-PET clinical trial worldwide in patients with rheumatoid arthritis, thus defining a pathway for the image-based development of antibody products in the future. In addition, we have performed a large amount of preclinical imaging applications, which have facilitated the selection of clinical development candidates.
Finally, we have extensively studied pathological specimens in man and in rodents, using multiplex biomarker analysis and microscopic techniques, shedding light on mechanism of action.
Collectively, the Project allowed not only the development and the practical implementation of novel methodologies in patients or in patient-derived material, but also provided tools for the characterization of novel antibody-based products, some of which are now being studied in clinical trials or are about to begin clinical testing.
In animal models of disease, we have performed extensive quantitative biodistribution studies, using radiolabeled preparations of therapeutic antibody products. This work has led to numerous publications and, collectively, has provided us with a quantitative understanding of the selectivity which antibody products can reach in vivo in rodent models of cancer, rheumatoid arthritis and graft rejection.
In addition, we have published a dosimetric estimate of radiation doses in organs and neoplastic lesions for a pan-tumoral antibody (L19, specific to the alternatively-spliced EDB domain of fibronectin), which has been studied in patients with solid tumors or with lymphomas. These studies have shown that the L19 antibody preferentially localizes on neosplastic lesions, but also that the tumor-homing selectivity varies greatly (i.e. up to 40-fold) from lesion to lesion and from patient to patient. Moreover, we have run an immuno-PET clinical trial with Dekavil (F8-IL10), an armed antibody developed by Philogen, labeled with iodine-124, in patients with rheumatoid arthritis.
We have implemented the routine analysis of leukocyte infiltration at the site of disease and the multiplex analysis of cytokine levels (in serum and at the site of disease) for many preclinical experiments, aimed at the characterization of the activity and mechanism of action of various antibody-based therapeutics. These studies have been published in several articles and have facilitated the identification of product candidates for clinical and industrial development programs (see below). Furthermore, we have studied in substantial detail the infiltration of lymphocytes into melanoma lesions in patients receiving antibody drugs and the specificities of cytotoxic T-lymphocytes which account for long-term survival, following pharmacological treatments.
We have developed, implemented and perfected methodologies for the determination of “atlases” of peptides bound to MHC-I or MHC-II (HLA-I and HLA-II in man). Thanks to the use of state-of-the-art mass spectrometers and to the implementation of efficient antibody-based pulldown procedures, these methodologies have allowed the determination of complex peptidomes, comprising hundreds of MHC-bound peptides, in complex biological samples, such as cell lines or patients’ sera. For the first time, it has been possible to identify hundreds of HLA-I-bound peptides from sera of melanoma patients. The identified peptides included previously reported melanoma rejection antigens, thus providing confidence about the robustness of the procedure. In mice with arthritis, collagen-derived peptides were identified as MHC-II ligands, thus providing a direct analytical access to antigens which may trigger autoimmune conditions.
We believe that the PRIAT Project has produced important technologies and publications, which facilitate the development and mechanistic study of antibody drugs. In time, this will lead to the generation of better antibody products and to improved selection procedures for patients, which are more likely to benefit from therapeutic intervention. In addition to these rather indirect (but important!) results, we believe that the PRIAT Projects has also produced results, with an immediate socio-economic impact.
We believe that the PRIAT Project has been truly successful, not only in terms of methodology development but also for the boosting of European Biotech competitiveness, with novel antibody products being brought to the clinic and improved analytical procedures for the study of response to antibody treatment.
Project Results:
Introduction
The PRIAT Project aims at developing and applying innovative technologies, for the profiling of responses in antibody therapies.
Therapeutic antibodies represent the largest and fastest growing class of Pharmaceutical Biotechnology. They are expensive treatments, which can provide a considerable benefit to some (but not all!) patients in various therapeutic areas, including oncology, chronic inflammatory conditions and transplantation.
In the Project, we have developed and applied three classes of innovative technologies, in order to better understand response to therapy both in mouse models of disease and in patients:
- HLA Peptidome analysis of biological specimens (including plasma specimens). A characterization of the T cell response to disease and treatment is crucial for the understanding of immunomodulatory pharmaceutical strategies. As T cells recognize peptides presented on MHC molecules (HLA in human), a characterization of the HLA peptidome is an indispensible step, towards the profiling of T cells by means of multiplex tetramer technologies
- Imaging techniques (e.g. PET and near-infrared fluorescence imaging), to reveal the ability of antibody-based therapeutics to localize at the site of disease in rodent models and in patients
- Multiplex analysis of specimens (e.g. sera, biopsies, leukocytes), in order to gain functional information about the status of immune activation and/or cytokine levels.
The therapeutic focus of the PRIAT Project has been in three disease areas [(i) oncology; (ii) rheumatoid arthritis; (iii) transplantation], for which antibodies represent important therapeutic agents. Indeed, antibody sales in Oncology and Inflammation were greater than 30 billions $ in 2013.
The participants of the PRIAT Project were ETH Zürich (Coordinator, Switzerland), Philogen (Siena, Italy), VUMC (Amsterdam, The Netherlands), Philochem (Zurich, Switzerland), Kennedy Institute of Rheumatology (Oxford, UK), University Hospital in Tübingen (Germany), University Hospital in Graz (Austria), University Hospital in Jena (Germany).
WP1 – HLA Class I Peptidome Methodologies
The detailed characterization of peptides bound to HLA-I molecules in man (MHC-I in rodents) is of tremendous importance, if we wish to characterize (at a molecular level!) how immunostimulatory anticancer strategies work (i.e. how T cells recognize tumor cells). This research topic has become very important in oncology, after the recent approval of immunostimulatory antibodies, directed against CTLA-4 (Ipilimumab) or against PD1 (pembrolizumab). These therapeutic agents are revolutionizing the therapy of melanoma and have the potential to have a substantial impact for the therapy of other malignancies.
Production of previously described monoclonal antibodies
HLA-class I peptidome analyses are typically performed by first purifying HLA-class I/peptide complexes with specific antibodies followed by release of peptides by denaturing the complexes. For this purpose the widely applied anti-HLA-class I antibody clone W6/32 was purified from the supernatant of hybridoma cells (HB-95, ATCC). The W6/32 antibody was produced in serum free hybridoma medium for five days followed by purification over protein A resin. Large productions of the W6/32 monoclonal antibody from the corresponding hybridoma have been performed, in order to provide material for pull-down and precipitation experiments. In total, over 1.2 g of W6/32 antibody were purified from Wave bioreactor at Philogen and shipped to Zurich. In addition, approximately 0.9 g have been produced at Philochem.
Generation of novel antibodies
A good quality monoclonal antibody clone W6/32 has been successfully used for HLA-I/peptide complex purification, as indicated in the previous section. However, affinity reagents capable of purifying different MHC-I variants in mouse and rat were also needed at the beginning of the Project. Further, specificity and cross-reactivity between different MHC sub-types is very difficult to assess for less characterized antibodies against MHC-I complexes.
For this reasons, we hypothesized that if we were able to recombinantly express b2-microglobulin from different species, we should be able to generate good quality affinity reagents using antibody phage display libraries. Ideally, these reagents would react not only with b2-microglobulin, but also with the intact MHC-I/peptide complex and as b2-microglobulin is part of all MHC-I complexes, antibodies would not be limited to certain sub-types.
We initially tested this approach on b2-microglobulin of human origin. In particular, one clone was isolated, which bound both b2-microglobulin and the MHC-I/peptide complex with high affinity and with an extremely low kinetic dissociation constant (Figure 1).
The antibody was then reformatted from scFv to IgG format for expression in mammalian cells. However, we preferred to use the W6/32 clone for pull-down experiments.
The W6/32 antibody did not cross-react with mouse and rat MHC-I molecules. For this reason, using a similar strategy, antibody phage selections were also performed against murine and rat b2-microglobulin. In addition, we evaluated hybridoma-derived monoclonal antibodies, which resulted in the use of OX-18 for rat pull-down experiments and M1/42 for mouse pull-down experiments.
Comparison of proteomic platforms and mass spectrometer selection
The next step was to purify larger quantities of HLA-class I complexes, release the peptides and analyze the eluate by mass spectrometry. Until today, four different setups were tested:
- Single spot MALDI with an AB SCIEX 4800 MALDI TOF/TOF
- LC-MS on an Orbitrap classic
- LC-MS on an Orbitrap Velos
- LC-MS on a Q Exactive; Philochem has purchased the latest generation of Orbitrap analyzers, the Q Exactive, which significantly boosted peptide identifications. Now we routinely identify hundreds of peptides from cell lysates with very high confidence.
Development of innovative methodologies for HLA-I peptidome analysis in cell lines
The optimized conditions for the purification of HLA class I complexes from the surface of cancer cell lines were applied to the profiling of the HLA peptidome of the AML cell line HL-60, the multiple myeloma cell line RPMI8226 and the human embryonic kidney cell line HEK. Six samples originating from 100 million cells of each cell line were analyzed with optimized parameters leading to the identification of 423, 313, and 1105 peptides with 1% false discovery rate (FDR) from RPMI8226, HL-60 and HEK respectively (Figure 2). In single experiments of the HEK cell line, more than 700 peptides were identified. The majority of identified peptides was 8-10 amino acids in length, which is compatible with HLA peptides, but which is not a typical length for e.g. tryptic digests of unspecific cleavage (see Deliverables report for details). In such cases we expect much larger peptides and a much larger standard deviation of sizes.
Besides the appropriate length distribution, we were evaluating if peptides can bind to the respective HLA molecules expressed on the surface of the cells. Prediction tools ANN and netMHCpan accessible through (http://tools.immuneepitope.org/mhci/) were found to provide best results. 60 and 80 % of 9mers are predicted to bind to the HLA A alleles on the surface of RPMI8226 and HEK respectively. However, 60 % of 9mers identified from HL-60 cells are predicted to bind to HLA B alleles.
Development of innovative methodologies for HLA-I peptidome analysis in plasma samples
For the conclusion of the WP1, HLA class I peptidomics methodology, we evaluated the possibility to perform HLA class I purifications from plasma from the blood bank. We therefore obtained pooled blood plasma from Blutspende Zurich. We have performed four experiments with this material a total of 14 purifications with anti HLA class I antibody W6/32 and 5 with BSA-conjugated beads (serving as negative control), with the aim to better understand the level of unspecific peptide contaminations. In four independent experiments, we identified up to 154 peptides with expected HLA class I characteristics. We also learned about preferred procedures for the storage of plasma samples, which allow the reliable identification of a large number of HLA class I derived peptides (see WP4).
Deliverables
D1.1) Technical Feasibility of HLA-I peptidome analysis in cell lines and plasma samples:
In view of what is described above, this Deliverable has been successfully completed. Importantly, the large number of identified peptides and the peptide length distribution (mainly nonamers) provided confidence that the methodology can be used for practical biomedical applications (see next sections).
WP2 – HLA Class II Peptidome Methodologies
The study of the HLA-II peptidome in man (MHC-II in rodents) is extremely important, in order to profile which antigens are being recognized by CD4+ T cells. A detailed characterization of the HLA-II peptidome promises to be particularly important for the understanding of immunological responses in autoimmune conditions such as rheumatoid arthritis and multiple sclerosis, two indications for which antibody products are frequently used in the clinical practice.
The study of the HLA-II peptidome is complicated by the polymorphism of HLA-II molecules and by the absence of a common chain (such as b2-microglobulin in HLA-I). Furthermore, no mass spectrometry-based studies have been reported so far for the study of peptides bound to soluble HLA-II molecules. Indeed, two transmembrane peptides need to be cleaved, in order to release HLA-II/peptide complexes in circulation.
Production of previously described monoclonal antibodies
Commercial anti-murine MHC class II antibody M5/114 could be obtained from a commercial source in large quantities. This antibody recognizes a panel of murine I-A and I-E alleles and is therefore applicable to a variety of model systems.
To purify human HLA class II complexes, we have purchased the hybridoma cell line producing pan-HLA DR antibody L243 (ATCC HB-55, recognizing the alpha chain of HLA DR). The antibody was produced in low IgG serum and purified with Protein A resin.
Generation of novel monoclonal antibodies
Unlike polymorphic MHC-I molecules, which all share the b2-microglobulin domain and which can be purified by suitable monoclonal antibodies directed against this portion of the MHC-I molecule, the purification of MHC-II molecules is much more difficult, because they are more polygenic and polymorphic.
We decided to focus on DR HLA molecules, as they are less polymorphic than DP and DQ and have been associated with increased incidence of certain autoimmune conditions (e.g. rheumatoid arthritis). We obtained recombinant HLA-II molecules of different haplotypes in complex with CLIP peptide from collaborators in Berlin (Prof. Dr. C. Freund) and we successfully raised >20 monoclonal antibodies from phage display libraries, which were able to recognize their cognate antigen in ELISA and on BIAcore (Figure 3). The affinity of a commercial antibody, L243, however was still superior on Biacore and we decided at this step to perform experiments with this antibody.
Murine MHC preparations have been even more difficult to get hold of. Also in this case we decided to go with a commercial antibody, M5/114, with which we obtained very good results.
Development of innovative methodologies for HLA-II peptidome analysis in cell lines
Starting from 100 million A20 lymphoma cells we could repeatedly identify up to over 1000 peptides with 1% false discovery rate. These numbers are unprecedented in the scientific literature especially as we only rely on high confidence identifications. Many publications in this field rely on low confidence peptide identifications (worse than 5% FDR), which boosts the number of peptides significantly, however, the number of false identifications increases as well. Bozzacco and colleagues [Bozzacco et al. (2011) J. Prot. Res., 10, 5016) for example, used 500 million cells and accepted peptide identifications with a FDR below 10%, ending up with around 100 peptides. We are currently preparing a manuscript on these results as we believe they represent a milestone in the analysis of the MHC class II peptidome, allowing us to answer fundamental questions in immunology (e.g. identity and frequency of antigens associated with T cell recognition, leading to autoimmunity) for the first time.
In one analysis, in which we identified 1110 peptide sequences, we obtained for the first time a very detailed understanding of the size distribution of MHC class II peptides (Figure 4) The length distribution follows a Gaussian centering around 15 amino acids in length with the majority of peptides being 12 to 19 amino acids long. No preference for certain amino acids either at the N- or at the C-terminus of the identified peptides was observed. Certain central amino acids were found to be significantly enriched after alignment of the sequences with the multi sequence alignment tool Clustal X (Figure 4). To the best of our knowledge, this is the first time that the binding motif for the MHC class II complex H2 I-Ad has been described to such a level of detail.
We have also investigated on the purification of HLA class II complexes from human cell lines (B-cell lines Ramos, Burkitt's lymphoma, ATCC® CRL-1596, and DoHH2, follicular B-cell lymphoma, DMSZ ACC 47). Purification of HLA class II from DoHH2 cells could be successfully demonstrated by Western blotting. In a large scale purification from 100 mio cells each, 165 and 97 unique peptides were identified with high confidence from DoHH2 and Ramos cells respectively.
Feasibility of HLA-II peptidome analysis in human plasma samples
After optimization of elution conditions, 5 ml of human plasma from the blood bank were subjected to the purification of HLA class II complexes with L243 resin. Three samples were purified with L243 resin and one control sample with BSA conjugated resin was included. Interestingly, several HLA class II derived peptides were identified in the analysis of the supernatant after filtration of eluted complexes through 10 kDa filters. Eight peptides were identified from HLA DRA, 7 from HLA DRB5, and 5 from HLA DRB1 indicating the successful purification of HLA class II complexes. However, 149 peptides were also identified in the BSA sample and only 108, 109, and 69 peptides after HLA class II purification, raising concerns about the efficiency of peptide recovery and the stability of isolated peptides in acidic conditions. Peptides identified after purification of HLA class II were short (on average 10 amino acids): a feature which is incompatible with the knowledge which has been gained regarding the structure of HLA class II peptides.
We are currently still working on establishing the purification of HLA class II complexes from human blood. We believe that this challenging goal may be reached one day, but there are many obstacles to overcome.
Feasibility of MHC-II peptidome analysis in mouse serum samples
Since purification of murine MHC class II complexes was yielding significantly higher amounts of peptides as compared to human HLA class II, we investigated if purifications from mouse serum would similarly lead to better results. We have obtained commercial mouse serum to get an initial idea of the volumes of mouse serum necessary. 4, 2, and 1 ml of the mouse was subjected to MHC class II purification with M5/114 antibody. 12 peptides derived from MHC class II complexes were identified in supernatant digests demonstrating successful purification, 7 of these matched to I-Ab, indicating that this MHC class II subtype is present. Moreover, peptides were derived from alpha and beta chain, indicating the presence of intact complexes. We should be able to elute peptides from these intact complexes and indeed 66, 94, and 117 peptides were identified from 1, 2, and 4 ml samples, respectively. Among the identified peptides were three MHC class II protein derived peptides known to be ligands of I-Ab. One of these, RHNYEGPETHTSLR, has been identified in murine spleen samples with the same haplotype (Ab). This indicates that some of the identified peptides may represent MHC class II eluted peptides.
Deliverables
D2.1) Technical Feasibility of HLA-II peptidome analysis in cell lines and plasma samples
In view of what is described above, this Deliverable has been successfully completed. Importantly, we were able to identify at least 10-times more murine MHC class II peptides (starting from a lower number of cells), compared to previous reports. A careful analysis of peptide consensus sequences and of peptide length distribution has shed light on the types of molecules which are presented onto MHC-II molecules in rodents. We could further prove the successful purification of MHC class II complexes from murine serum where a number of known MHC-II ligends have been identified.
WP3 – Immuno-PET clinical trial preparation
It is becoming increasingly clear that many of the standard chemotherapeutic agents used for cancer treatment do not preferentially localized at the tumor site (Figure 5).
Unfortunately, a selective uptake in the tumor is often not observed also for antibody-based therapeutics, which would be expected to display a tumor-targeting potential (Figure 6).
We and others have previously recognized the importance of integrating molecular imaging procedures for a “Precision Medicine” approach to drug development [Tijink et al. (2009) Eur. J. Nucl. Med. Mol. Imaging, 36, 1235].
Within PRIAT, we have focused on the preparation and implementation of a Phase 0 immuno-PET clinical trial, to assist the development of F8-IL10 (an antibody-cytokine fusion protein, which has recently moved to Phase IIb clinical trials for the therapy of patients with rheumatoid arthritis). In addition, we have performed and published dosimetric evaluations on immuno-PET trials previously performed by the Philogen group, and we have executed a large number of imaging and quantitative biodistribution studies in rodent models of cancer, rheumatoid arthritis and chronic graft rejection.
Optimization of GMP radiolabeling of F8-IL10, chosen as representative innovative antibody-based therapeutic agent for clinical trials
The GMP procedure for the radiolabeling of F8-IL10 has been optimized by the fine-tuning of a set of critical parameters such as reaction time, concentration of the iodination reagent and initial formulation of F8-IL10. Furthermore, different radiolabeling modalities (i.e. Chloramine-T vs Iodogen method) have been compared and the use of Philogen's automated radiolabeling module “Easy Iodine Machine” has been implemented.
The quality and behavior of F8-IL10, iodinated under different conditions, have been characterized according to standard biochemical methods as well as with biodistribution studies performed in mice.
Stability of the radiolabelled drug was also verified upon storage at both room temperature and +4°C. A biodistribution study in mice was also performed to ensure retention of biodistribution properties after storage.
Optimization experiments showed that both Chloramine-T and Iodogen methods provide material of similar quality and that the presence of Tween-20 in the formulation buffer of F8-IL10 does not interfere with the radiolabeling process. Furthermore, to support stability study of the iodinated material, a biodistribution study in normal mice demonstrated full retention of biodistribution properties of iodinated material stored at +4°C up to 24h
Preparation work for a Phase 0 Immuno-PET Imaging Clinical Trial at VUMC in Amsterdam
A GMP Manufacturing procedure has been established at VUMC (Amsterdam) for the radiolabeling of F8-IL10, an anti-inflammatory immunocytokine which is currently being investigated by Philogen for the therapy of rheumatoid arthritis in a double-blind multicenter Phase IIb clinical trial. The product can now be reproducibly labeled with iodine-124, a PET radionuclide.
Quality data for the radiolabeled product, together with a Study Protocol, and Investigator Brochure and a Patient Information Form, have been submitted to Clinical Authorities, who have approved a Phase 0 immuno-PET imaging study in 5 patients with rheumatoid arthritis. The study has been performed, as documented in WP8.
Dosimetric Evaluation in PET
ImmunoPET combines the high resolution and sensitivity of PET and the selective localization of antibodies at their target in vivo. The technique has profited from technical advances and the increased availability of GMP-grade radionuclides and has therefore begun to play an important role in both cancer staging and tailoring of therapy. Besides pure diagnostic purposes, ImmunoPET can also be envisaged as a dosimetric tool to predict the expected doses to target lesions and healthy organs of a subsequent radioimmunotherapy using the same antibody coupled to a therapeutic isotope.
Indeed, we have performed and published a dosimetric evaluation of the radiation dose delivered to neoplastic lesions and to normal organs by a clinical-stage human monoclonal antibody (124I-L19, developed by Philogen) in patients with brain-metastases and extracranial lesions. This study nicely complements similar dosimetric findings on the targeting of radioiodinated L19 in patients with lymphoma [Erba et al. (2012) J. Nucl. Med., 53, 922].
Collectively, these studies indicated that antibody uptake can be highly variable even in different lesions of the same patient and that immunoPET procedures may guide product development with armed antibodies (Figure 7).
Deliverables
D3.1) Preparation activities for Immuno-PET clinical studies:
The Deliverable has been successfully completed, as documented by the performance of the clinical trial.
Dosimetric analyses on tumor targeting of cancer patients with radioiodinated preparations of the L19 antibody have also been published. Moreover, the implementation of reliable radiolabeling techniques has been important for the execution of many biodistribution and imaging studies in rodent models of disease (see subsequent WPs).
WP4 – Management of clinical samples
Procedures and experimental findings related to the use of patients’ sera for HLA peptidome analysis have been described in WP1.
For the analysis of biopsies, it is well established that biopsies and surgically-resected material should preferably be stored as freshly stored material, embedded in OCT.
We have evaluated the feasibility of multiplex analysis of leukocyte infiltration and of intratumoral cytokine levels using a combination of immunofluorescence microscopy techniques, FACS analysis of resuspended cells and parallel measurement of cytokine concentrations. The results of these investigations have now been described in publications and are now routinely implementing these methods in our laboratories (see also WP9).
Immune system–based prognostic markers may be useful to monitor response to immunotherapies and may serve to allow the identification of patients with different prognosis.
In a previous study, the group of Prof. Garbe at University of Tübingen demonstrated the prognostic relevance of functional circulating T cells responding to the tumor-associated antigens NY-ESO-1, Melan-A, MAGE-3, and survivin on overall survival of patients with melanoma with distant metastasis. For PRIAT activities, cryopreserved peripheral blood mononuclear cells (PBMCs) from 84 patients with follow-up after analysis (cohort A), 18 long-term survivors with an extraordinarily favorable course of disease before analysis (≥ 24 months survival after first occurrence of distant metastases; cohort B), and 14 healthy controls were collected and analyzed. Circulating antigen-reactive T cells were characterized by intracellular cytokine staining after in vitro stimulation.
The presence of circulating T cells responding to Melan-A or NY-ESO-1 have a strong independent prognostic impact on survival in advanced melanoma. The collection of samples described above represented the basis for the analysis of T cell reactivities in melanoma (see WP9).
Collectively, the partners collected and exchanged samples of plasma, sera, blood and biopsies, which were essential for the execution of the analysis described in the various Workpackages.
A standard operation procedures for collection and storage of rat serum/plasma to detect soluble MHC I/ II was established by UHJ.
Deliverables
D4.1) Implementation of Standard Operating Procedures for the collection, storage and shipment of clinical samples (biopsies and plasma samples):
The Deliverable has been successfully completed, as documented by the successful execution of the experiments described in other WPs of this report.
WP5 – HLA Class I peptidome analysis in melanoma
A recent analysis of over 7000 human cancer specimens has recently provided a solid foundation for the mutation rate which is observed in human malignancies. Typically, a prevalence greater than 1 mutation in 1 megabase is found for most tumor types. Melanoma appears to be the tumor with the highest mutation rate, which is typically in the range between 10-100 mutations per megabase [Alexandrov et al. (2013) Nature, 500, 415].
In this WP, we have focused on the profiling of HLA-I peptidome in melanoma and on the characterization of T cell reactivies (and their clinical significance).
HLA-I peptidome analysis in melanoma
At the University Hospital Tübingen, sera from cancer patients and healthy volunteers were collected according to the defined standard operating procedure for further HLA class I peptidome analysis. Patient characteristics are presented in Table 1. In total 15 sera were shipped, nine samples from cancer patients, and six from healthy volunteers. The inclusion criterion was that at least one of the HLA A haplotypes was A02, as most of the known tumor rejection antigens are described for this allele. Tumor burden was variable, ranging between none (only local tumor) and high, two sera were provided before Ipilimumab treatment, three samples after Ipilimumab.
Five ml of serum per patient were subjected to the HLA purification procedure using W6/32 antibody resin purifying all HLA class I alleles. After excessive washing, complexes were eluted with 0.1M acetic acid, heated to 95 °C, filtered through 10 kDa cutoff filters and purified over C18 resin for mass spectrometric analysis. Peptides were resuspended in 8 l 3% acetonitrile 0.1% formic acid in ultrapure water and 6 l loaded on a 15 cm C18 column. A linear gradient from 100 % water 0.1% formic acid to 20 % acetonitrile 0.1% formic acid over 45 minutes was used to elute the HLA peptides. Peptides were fragmented in the HCD cell and spectra recorded in the Orbitrap. Between 57 and 567 peptides (on average of 184) were identified with 1% FDR. Most identified peptides were 9 amino acids in length (Figure 8A). The same is true for the c-terminal anchor residues: Certain amino acids were enriched, demonstrating patterns of anchor residues depending on the individual HLA subtype (Figure 8B). Most prominent C-terminal anchor residues were K and R as well as hydrophobic residues L, V, Y, S. Patients 3 and 6 for example demonstrated a strong enrichment of C-terminal K, which is a known anchor residue for the HLA A03 allele. Indeed, almost 60% of identified peptides with 8-11 amino acids per patient were predicted to bind to the patient’s cognate HLA alleles.
Identified peptides were compared with a previously published list of melanoma related tumor rejection antigens [Andersen et al. (2012) Cancer Res., 72, 1642]. One peptide (VVQNFAKEFV) from the catalog of tumor associated antigens was identified in melanoma patient with the sample number 6. The identification however is of low confidence and will be validated with a synthetic peptide. In addition, a number of melanoma-associated antigen (MAGE) derived peptides have been identified (most of them with low confidence) and we are validating these identifications by comparing their spectra with the ones of synthetic peptides. A complementary validation activity relies on the stimulation of patient-derived lymphocytes with synthetic peptides.
At this stage, we believe that we have established a robust technology for the purification of sHLA complexes from patients’ sera. To the best of our knowledge, this is the first study on sHLA complexes from patients with solid tumors. We have selected individuals with the same haplotype and we have included a number of healthy individuals as control. In the coming months we will validate the MAGE peptides with synthetic peptides and we will try to validate some of the potential tumor rejection antigens with our collaborators at the University Hospital Tübingen.
We believe that the identification of HLA-I peptides from patients’ sera represents a convenient and efficient avenue for the experimental determination of an Atlas of peptides, some of which may represent tumor rejection antigens. The technology is ideally complemented by the analysis of T lymphocytes, using peptide stimulation or multiplex tetramer analysis, pioneered by the group of Ton Schumacher in Amsterdam.
T cell response against melanoma antigens
Philogen, the University Hospital Tübingen and the University Hospital Graz have investigated the therapeutic effect of intralesional administrations of L19-IL2 to patients with Stage IIIC melanoma, following a methodology pioneered by the group of Prof. C. Garbe in Tübingen. A first Phase II study in 25 patients has been completed and published [Weide et al. (2014) Cancer Immunol. Res., 2, 668].
Patients with stage IIIB/IIIC melanoma and cutaneous/subcutaneous injectable metastases received weekly intratumoral injections of L19–IL2 at a maximum dose of 10 MIU/week for 4 consecutive weeks. Tumor response was evaluated 12 weeks after the first treatment. Twenty-four of 25 patients were evaluable for therapy-induced responses. A complete response (CR) by modified immune-related response criteria (irRC) of all treated metastases was achieved in 6 patients (25%), with long-lasting responses in most cases (5 patients for >24 months). Objective responses were documented in 53.9% of all index lesions [44.4% CR and 9.5% partial responses (by irRC)], and 36.5% of these remained stable, while 9.5% progressed. Toxicity was comparable with that of free IL2, and no serious adverse events were recorded. A significant temporary increase of peripheral regulatory T cells and natural killer cells, sustained increase of absolute CD4+ T lymphocytes, and decrease of myeloid-derived suppressor cells were observed upon treatment. Finally, we recorded encouraging data about the progression time to distant metastases and overall survival.
The systemic immune responses induced by the immunocytokine L19-IL2 upon intratumoral injection of cutaneous/subcutaneous metastases in Stage IIIB/IIIC melanoma patients was evaluated as follows. We collected 45 blood samples from 17 patients at baseline and two different time points after injection of L19-IL2, PBMC were purified and stored at -80°C. After thawing, different populations of immune cells including CD4+ and CD8+ T cells, Myeloid-derived suppressor cells (MDSCs), Regulatory T cells (Tregs) and Natural Killer (NK cells) were evaluated using multicolor flow cytometry. MDSCs were characterized by the CD14+CD11b+HLA-DR–/low phenotype and NK cells as CD16+CD56low. The percentage of MDSCs and NK cells was calculated as the proportion of CD14+CD11b+HLA-DR–/low cells and CD16+CD56low cells within the total population of viable PBMCs.
For Tregs, which were characterized by the CD25+FoxP3+ phenotype, cells were intracellularly stained with FoxP3/Alexa647 (BD Biosciences). Frequencies of CD4+ and CD8+ T cells were determined within the total number of lymphocytes. The percentage of Tregs was calculated as the proportion of CD25+FoxP3+ cells within all CD4+ cells.
For NY-ESO-1- and Melan-A-specific T cell response detection, cells were stimulated with protein-spanning overlapping peptides (1 µg/mL). After culture for 12 days, T cells were restimulated at a ratio of 1:2 with autologous, fluorescent-labelled PBMCs either unpulsed (negative control) or peptide-pulsed with one of the antigens in the presence of Golgi-Plug for 12 hours. After blocking with Gamunex and labeling with ethidium monoazide, cells were fixed and permeabilized with CytoFix/CytoPerm and stained with the following antibodies: CD3/eFluor605, CD4/PerCP, CD8/APC-H7, IL-4/APC, IFN-γ/PE-Cy7, TNF/FITC, IL-10/PE, IL-17/PerCP-Cy5.5 and IL2/Alexa700. Samples were measured immediately using an LSR II and FACSDiva software and data were analyzed using FlowJo software.
The results showed a significant, transient increase in the frequency of CD4+CD25+Fox-P3+ Tregs in 15 of 16 evaluable patients after four weeks of treatment compared to baseline (mean frequency 12.1% vs. 8.7%; p<0.001). Frequencies returned to baseline values at day 85 (Figure 9A). There was a decrease of CD14+CD11b+HLA-DR–/low MDSCs from 11.5% at baseline to 9.0% at day 28 and 7.6% at day 85 (Figure 9B). Both the differences between baseline and end of treatment and between baseline and day 85 were statistically significant with p-values of 0.001 and 0.002 respectively. No changes in frequency were observed for NK-cells, CD4+ or CD8+ T cells.
The decrease of frequency of MDSCs in our patients might serve as an explanation for the beneficial mode of action of intratumoral IL-2, as suggested by a favorable distant metastasis-free survival as compared to historical controls.
Deliverables
D5.1) Implementation of HLA-I peptidome analysis in melanoma:
The Deliverable has been successfully completed, as documented by the successful identification of melanoma rejection antigens in soluble HLA-I/peptide complexes from patients’ sera.
Break-through methodologies, based on the work of the Tübingen group to study the reactivity of patient-derived T-cells against tumor-rejection antigens, revealed a complementary approach to study the relevance of HLA-I peptides in relation to anti-cancer immunity. We integrated both approaches in the PRIAT Project, thus complementing the MS-based analysis of soluble HLA-I peptidome.
WP6 – HLA Peptidome analysis in arthritis
The possibility to perform HLA-II analysis in the collagen-induced model of arthritis (CIA) in the mouse depends crucially on the availability of the animal model and on the ability to purify MHC-II peptide complexes (see WP2).
The CIA model has been established in Zurich and in Oxford.
HLA peptidome analysis in arthritis
After initial studies using lower amounts of serum, 5ml serum each from healthy DBA1 mice and mice with established collagen-induced arthritis (CIA) were subjected to the purification of MHC class II complexes following the methodology established on A20 lymphoma cells. Purification was performed with the M5/114 antibody and after excessive washing, complexes were eluted with 10% acetic acid. After purification of peptides, they were analyzed by mass spectrometry. The sample from heathy mouse serum clogged the HPLC column and could therefore not be analyzed. However, the sample from arthritic mice could be analyzed and a total of 147 peptides were identified with 5% false discovery rate. Moreover, the analysis of the supernatant the 10 kDa filtration confirmed that H2 I-Aq from healthy as well as from arthritic mice could be purified. 15 peptides from the alpha chain and 10 peptides from beta chain were identified indicating purification of intact MHC class II complexes. A selection of possibly MHC class II peptides is presented in Table 2. Several peptides originating from collagen were identified. Collagens are known to be degraded during CIA progression and are very promising candidate molecules for further validation. Other peptides are derived from intracellular proteins and membrane proteins, e.g. Lysosome-associated membrane glycoprotein 2, H-2 class II histocompatibility antigen gamma chain, Ras-related protein Rab-5C. These proteins have been either eluted from MHC class II of A20 cells, or represent very likely proteins to be presented by MHC class II complexes.
Emerging antibody therapeutics for the treatment of rheumatoid arthritis: a changing paradigm
Intact monoclonal antibodies (e.g. TNF blockers, pioneered by the colleagues at the Kennedy Institute of Rheumatology) represent some of the most successful therapeutic agents for the treatment of Rheumatoid Arthritis.
Within the framework of the PRIAT Project, we have extensively studied the therapeutic activity of “armed” antibody products (i.e anti-inflammatory immunocytokines, such as F8-IL10 and F8-IL4) in the murine CIA model of arthritis. Among the many interesting observations made in the course of these studies, we have found that F8-IL4, in combination with dexamethasone, is able to induce complete and lasting remissions in mice with arthritis (Figure 10). To the best of our knowledge, this is the first report of arthritis cures in rodent models of the disease. The new findings allowed a set of mechanistic investigations (see also WP9).
The immunosuppressant drug, dexamethasone and F8-IL-4 were previously shown to induce lasting remissions in arthritic mice (Hemmerle et al. (2014) Proc. Natl. Acad. Sci. U.S.A. 111, 12008). In contrast, the treatment of arthritic mice with either F8-IL-4 or dexamethasone alone did not lead to lasting remissions. To fully understand how this novel combination therapy works we examined the immune cell subpopulations in arthritic paws and lymph nodes of CIA mice treated with dexamethasone and F8-IL-4 monotherapy versus combination therapy. We found that arthritic mice that received F8-IL-4 monotherapy for 10 days had a significant increase in toleragenic TH2 cells and a decrease in the number of pathogenic TH17 cells in their paws. Interestingly F8-IL-4 monotherapy was also associated with a decrease in the percentage of anti-inflammatory Tregs in arthritic mice. On the other hand combination therapy not only lead to the reduction of pro-inflammatory TH17 cells but also a significant induction of Tregs and TH2 cell numbers which was associated with a significant decrease in arthritic score (Figure 11). In summary, combination therapy leads to lasting remissions in arthritic mice due to the effect on the balance between pathogenic, pro-inflammatory (TH17) and toleragenic, anti-inflammatory cells (TH2 and Tregs, Figure 12)
Deliverables
D6.1) Implementation of HLA-I and/or HLA-II peptidome analysis in arthritis:
The Deliverable has been successfully completed, as documented by the successful identification of MHC-II bound peptides from mice with collagen-induced arthritis.
Moreover, we have developed innovative approaches for the therapy of rheumatoid arthritis, which have resulted in complete cures in mice treated with F8-IL4 plus dexamethasone.
WP7 – HLA Class I peptidome analysis in transplant rejection
To investigate antibody therapies and to monitor their performance in the prevention of transplant rejection, a heterotopic rat heart transplantation model has been developed by the University Hospital in Jena.In this model, hearts are explanted from donor rats, subjected to cold ischemia at 4°C for 30 min, transplanted into recipient rats and harvested after 70 or 140 days respectively. For the first 14 post-transplant days, animals receive a sub-therapeutic dose of Cyclosporin A (2 mg/kg per day) to generate chronic rejection. For heterotopic heart transplantation two inbred strains were used: LEW as donors and F344 as recipients. Both strains are completely similar in the MHC RT1 genotype with only minor differences located on the RT1l gene locus for non-MHC allele. For this reason the rat combination LEW-F344 develops nearly no signs of acute rejection following heart transplantation but is a valuable model to investigate chronic rejection.
Plasma samples from rats have been collected by UHJ and have been used to determine the quantities of soluble MHC-I and II and ED-A+ fibronectin molecules in rats with chronic heart rejection in various antibody therapy conditions (see below).
Levels of soluble MHC proteins in a rat model of chronic heart transplant rejection
With respect to the planned HLA I and II peptidome analysis in serum after heterotopic rat heart transplantation, MHC I and II expression and shedding was investigated in relation to severity of rejection by immunohistochemistry and ELISA technology.
First, tissue expression of MHC I and MHC II was detected and quantified in shock frozen tissue samples of cardiac allografts by fluorescence immunohistochemistry using two commercially available antibodies specific to the rat MHC class I (OX-18) and II antigens (OX-6) provided by BD-Pharmingen. In a pre-therapeutic experiment a comparable increase in MHC I as well as MHC II expression in heart tissue in association to the severity of cardiac allograft rejection was evidenced. To answer the question if these histologically detected differences in MHC expression are also reflected by a concordant liberation of soluble MHC proteins into the serum (sMHC), commercially available ELISAs (Bio-Medical Assay). A comparative increased serum concentration of sMHCII could be evidenced. Comparable results were observed in the preventive therapeutic setting using F8-IL10 immunocytokines.
Levels of soluble ED-A+ fibronectin in a rat model of chronic heart transplant rejection
Besides its tissue expression and protein deposition into the cardiac extracellular matrix, EDA(+)-fibronectin is also released to circulating blood. Thus, the serum levels of the molecule might be a valuable biomarker for non-invasive surveillance of tissue remodeling in patients suffering from cardiovascular diseases. In the frame of the PRIAT project, 2 different ELISA protocols to detect EDA(+) fibronectin have been newly established: 1), ED A(+) fibronectin was captured with gelatin and specifically detected using the antibody IST-9 and a highly absorbed biotinylated secondary antibody and 2), the antibody IST-2 specific to the domain Fn12-14 was used as capturing antibody and a biotinylated IST-9 as detection antibody. In both methods, specificity was tested using purified ED A(+) fibronectin Fn as well as plasmatic fibronectin (pFn). Purified melanoma cell line derived ED A+ Fn was used to generate a standard curve. ELISA methods were applied for quantification of serum samples from patients and from rats after heterotopic heart transplantation. Both of the newly established ELISAs were specific to ED A(+) fibronectin whereas pFn could not be detected. The gelatin based ELISA worked in humans and rats. Compared to healthy humans, in patients with cardiovascular diseases and left ventricular dysfunction (n=133) significantly elevated serum concentrations of ED A(+) fibronectin could be measured. There were interesting correlations to clinical findings in the patient group. Using the rat heart transplantation model, serum concentrations of ED A+ Fn were decreased in transplanted rats exhibiting chronic rejection compared to non-transplanted controls. The explanation for this unexpected finding will be the object of further studies.
There is another interesting point to mention: In the rat model, the determination of ED A(+) fibronectin in serum after a single administration of the antibody F8 revealed significantly decreased levels compared to controls that have received an unspecific control antibody. Thus, a possible interference between the specific antibody and circulating ED-A(+) fibronectin should be taken into account when setting up therapeutic studies. This must be considered in therapeutic approaches.
Therapeutic intervention in a rat model of chronic heart transplant rejection
Philogen has provided clinical-grade F8-humanIL10 for the immunosuppressive therapy of chronic heart transplant rejection in rats. In addition to what originally described in the proposal, also F8-ratIL10 (a chimeric immunocytokine) and KSF-IL10 (a negative control immunocytokine, based on an antibody of irrelevant specificity in the rat) have been cloned, produced and used in therapy experiments.
UHJ has performed a first therapy experiment with the immunocytokines F8-humanIL10 (DEKAVIL), F8-ratIL10 as well as KSF-humanIL10 (irrelevant antigen-specificity). Treatment was performed weekly for 10 weeks starting at day 7 after transplantation (1mg/animal). Four experimental groups (PBS, KSF-IL10, DEKAVIL, F8-ratIL10; n=5 each) were compared with non-transplanted controls treated similarly. After completion of the experiment, organs and plasma were collected and subjected to further analyses.
In the non-transplanted control animals, macroscopic as well as histological analysis did not reveal any signs of tissue damage irrespective of treatment. In the cardiac allografts, treatment with DEKAVIL or F8-ratIL10 was associated by increased organ weights, a higher grade of chronic rejection, increased CIF, higher protein expression levels of alpha-smooth muscle actin (-SMA) as well as an augmented infiltration with inflammatory cells (CD4+, CD8+ and CD68+ cells) compared to the groups treated with neither PBS nor KSF-IL10. There were significant differences for the following parameters: the histological rejection score, CIF, alpha-SMA and the extent of CD4+, CD8+ and CD68+ cell infiltration levels. Compared to the DEKAVIL group, rejection-associated tissue damage was further augmented in the F8-ratIL10 group. There were no relevant differences between the KSF-IL10 and the PBS group. Serum brain natriuretic (BNP) levels, displaying heart failure grade, were higher in the groups treated with DEKAVIL and F8-ratIL10 compared to the groups that received PBS or KSF-IL10 respectively.
In another experiment, treatment was performed weekly for 10 weeks starting at day 70 after transplantation (1mg/animal), at a time point when chronic rejection has fully developed. Thus, the question should be addressed, if treatment was capable to reduce rejection. Three experimental groups (PBS, KSF-IL10, F8-ratIL10; n=4 each) were compared with non-transplanted controls treated similarly. After completion of the experiment, organs and plasma were collected and subjected to further analyses as described above. Taken together, when applying this treatment protocol, no relevant effects of administration of F8-ratIL10 could be observed compared to PBS, KSF-IL10 or untreated animals. These findings are speaking well for the fact that the time point of treatment initiation is of importance in this rejection model and should considered as a crucial experimental condition when planning further studies using F8 based immunocytokines.
MHC-I peptidome analysis in a rat model of chronic heart transplant rejection
After an in depth evaluation of existing literature on purification of MHC class I complexes from rat and antibodies against certain rat alleles and haplotypes, we decided that our best option is to obtain the OX-18 hybridoma cell line producing the antibody OX-18 directed against the MHC class I A equivalent in the mouse, RT-1A.
The hybridoma culture was established and the OX-18 antibody produced using low IgG serum. OX-18 could be purified from hybridoma cell supernatant after 6 days of culture with a final yield of 13 mg/L.
MHC class I complexes were purified from the lysate of 50 mg and 200 mg spleen tissue respectively following the procedure established for the purification of HLA class I complexes with W6/32 antibody. Mass spectrometric analysis led to the identification of a total of 125 peptides, 67 from 50 mg and 94 from 200 mg of tissue respectively. Moreover, the peptide length distribution shows a strong enrichment of 9mers and the evaluation of the c-terminal amino acid demonstrates an enrichment of histidine indicating the presence of a c-terminal anchor residue often observed for MHC class I molecules.
We therefore moved on to the purification of MHC class I from rat sera. Two small test experiments indicated, that we could also purify MHC class I from rat serum. A larger experiment was therefore performed purifying MHC class I from 3 ml of rat serum with different grade of rejection. After subtracting peptides identified in samples where BSA conjugated resin was used, 19, 33, 14, and 8 peptides were identified in healthy control, moderate, severe, and weak rejection respectively. Peptide length distribution demonstrated an enrichment of 9mers and at the peptide c-terminus histidine and leucine were significantly enriched (Figure 13).
After these promising experiments, we collected serum from a therapy study performed in the frame of the PRIAT project (see above). 144 peptides were identified in the six samples demonstrating the same pattern of enriched 9mers and mostly C-terminal histidine residue. Interestingly, the number of peptides identified from respective samples after subtracting known contaminants increases with severity of rejection. This is in line with previous findings indicating that the concentration of eluted MHC class I complexes from OX-18 resin is higher in samples of severe rejection as compared to healthy control sera. An increased concentration of soluble MHC class I complexes in serum has been described for other diseases, but it is until now unknown if this is also true for transplant rejection.
Deliverables
D7.1) Implementation of HLA-I peptidome analysis to rodent models of heart transplantation:
The Deliverable has been successfully completed, as documented by the successful identification of MHC-I-bound peptides in complex biological specimens, and by the rich series of experimental investigation in the rat model of chronic graft rejection.
WP8 – Assessment of targeting by Imaging
Immuno-PET is a break-through methodology, which is indispensible for the quantitative characterization of the localization properties of antibody-based therapeutics in normal tissues and at sites of disease.
The realization of immuno-PET studies with therapeutic antibodies has been made possible, thanks to recent work of VUMC, in collaboration with ETH and Philogen, for the radioiodination of antibodies with the PET radionuclide iodine-124 [Tijink et al. (2009) Eur. J. Nucl. Med. Mol. Imaging, 36, 1235-1244; Heuveling et al. (2013) J. Nucl. Med., 54, 397-401].
Phase 0 immuno-PET clinical trial in patients with rheumatoid arthritis
A novel Immuno-PET study with the 124I-F8-IL10 immunocytokine been launched at VUMC, in collaboration with Philogen. After receiving authorization from competent authorities, three patients with rheumatoid arthritis (RA) have been injected with the radioiodinated product and imaged by PET.
The Primary Objective of this study was to investigate the targeting performance (including pharmacokinetics/-dynamics) of 124I-F8-IL10 in patients with rheumatoid arthritis, in a phase 0 single dose study. The Secondary Objective was to investigate dosimetric parameters of 124I-F8-IL10 in arthritic joints and internal organs.
The recruited RA patients all had active disease (disease activity score of 28 joints > 3.2 at least two clinically swollen evaluable joints). Their anti-rheumatic treatment was stable for at least two weeks. After a single i.v. injection of 2 mCi (74 MBq) 124I-F8-IL10 (with an adjusted dose of 30 MBq in third patient), 3 whole body PET-CT scans and detailed scans of arthritic joints were performed at baseline, 24 and 72 hr p.i. Blood samples analysis for pharmacokinetics were withdrawn at several time points up to 72 hr p.i.
124I-F8-IL10 cleared very rapidly from the circulation. Within 30 min post-injection, less than 1% injected dose (ID)/kg was left in circulation, both in whole blood and plasma. Clear targeting of inflamed joints was obtained, with maximum uptake at 24 hr p.i. (Figure 14). Clinically affected joints showed higher 124I-F8-IL10 uptake than clinically unaffected joints.
Besides uptake in inflamed joints, dosimetric analysis revealed high 124I-F8-IL10 uptake in liver and spleen. Within 1 hour p.i. maximum liver/spleen uptake was found, decreasing with time.
Significantly increased uptake in liver and spleen was not previously found in tumor bearing mice [Schwager et al., (2009) Arthritis Res. Ther., 11, R142]. The increased liver and spleen uptake in RA patients resulted in a three times higher radiation burden of 0.39-0.46 mSv/MBq than predicted from animal data (0.14 mSv/MBq). To keep the radiation burden within acceptable limits according safety guidelines, the injection dose of 124I-F8-IL10 was lowered to 30 MBq in the third RA patient. Even with a three times lower injection dose, accumulation of 124I-F8-IL10 in inflamed joints could be clearly depicted on the images by prolongation of the scanning time.
Possible explanations for observed in vivo biodistribution differences of 124I-F8-IL10 between RA patients and animals are:
1. Radiolabeling differences (slightly different in animals and RA patients)
2. Species differences
3. F8-IL10 protein dose effects (microdose of PET in men versus higher dose in animals).
4. RA specific effects.
Irrespective of these considerations, a dosimetric analysis of arthritic and non-arthritic lesions nicely revealed a preferential accumulation of F8-IL10 in the affected lesions of patients (Figure 14).
Fluorescence imaging and immuno-PET studies in a rat model of chronic heart rejection with 124I-F8
To study the targeting properties of the F8 antibody in the heterotopic rat heart transplantation model to induce chronic cardiac rejection, imaging studies were performed by applying two different methods: 1), ex vivo near infrared fluorescence imaging (NIRF) and 2), in vivo PET detection.
1) Imaging studies using the fluorescently (DY-682) labelled SIP-F8 antibody and NIRF methodology were performed in transplanted (n=5) compared to non-transplanted rats (n=3) at day 69 after transplantation. Furthermore, a group of transplanted rats (n=5) were injected with the F16 antibody specific to the A1 domain of human tenascin-C, which is not reactive in the rats and therefore suitable as an isotype control. It could be proven that the F8 antibody selectively accumulates in chronically rejected hearts while sparing recipient hearts (heterotopic approach) [Franz et al.. (2013) J. Heart Lung Transplant., 32, 641-50].
2) To validate these findings and to test an imaging method that is also applicable in humans, we performed immuno-PET CT imaging in accordance. Therefore, transplanted animals received 124I-labeled SIP-F8 antibody and an isotype control as well (SIP-KSF). At an early time point after injection (1 hour), 124I-labeled SIP-F8 as well as 124I-labeled SIP-KSF were detectable with an accentuation in the circulation (blood pool), the liver, the kidneys and the urinary bladder. The cardiac allografts could be clearly confined in all rats. At a later time point, 24 hours after injection, the accumulation of 124I-labeled SIP-F8 was much more intensive compared to 124I-labeled SIP-KSF in the transplanted heart region (Figure 15). The findings are preliminary and have to be carried out including more animals.
Ancillary imaging and biodistribution studies with 124I-F8-IL10
In order to better understand the observed rapid blood clearance and high liver/spleen uptake of I-124I-F8-IL10 in RA patients, additional experiments were performed at VUMC, Philogen, Philochem and ETH.
Iodination method: addressing a potential effect of applied radioiodination method, additional studies in healthy mice have indicated that different modalities of F8-IL10 radioiodination (e.g. Iodogen vs. Chloramine-T and addition vs. elimination of tween20 in the formulation buffer of F8-IL10) do not have a substantial impact on the results of biodistribution studies. With either radioiodination method, there was no relevant uptake of 124I-F8-IL10 in liver and spleen.
Test species: the role of species differences on in vivo biodistribution was further investigated by injection of 124I-F8-IL10 in rats with antigen-induced arthritis. Interestingly, these experiments showed a higher uptake in liver and spleen of 124I-F8-IL10 as compared to a control 124I labeled cytokine-antibody (KSF) (Figure 16).
Other imaging and biodistribution studies
Interaction with serum protein: the interaction of F8-IL10 with RA serum was investigated ex vivo on HPLC. These data revealed a small shoulder on each side of the F8-IL10 peak which may point at some interaction, but this needs to be further explored in additional experiments.
Interaction with blood cells: the interaction with RA blood cells was studied ex-vivo by centrifugation assay as described by Doll et al. (Doll et al. Arthritis Res.Ther. 2013, 15 R:138). After incubation of 125I-F8-IL10 with fresh RA blood, cells and plasma were separated by centrifugation and counted separately. Data from this experiment show that F8-IL10 was retained at high percentage in the plasma (> 70%).
To further address RA specific contribution to uptake of F8-IL10 in liver and spleen, immunoPET treatment of patients with other chronic inflammatory disease(s) will also be planned.
Finally, dose effects of F8-IL10 on in vivo biodistribution, will also be investigated in future studies.
The data collected by the imaging and biodistribution experiments will aid in further understanding and development of F8-IL10 as promising novel therapeutic agent for RA.
In addition to the above-described clinical studies, a number of preclinical studies have been performed, including the following ones:
- radioiodinated preparations of several armed antibody products have been characterized by quantitative biodistribution studies in animal models of cancer
- radioiodinated preparations of various monoclonal antibodies have been compared in terms of their targeting performance in the mouse CIA model of arthritis, revealing that the F8 antibody has the best targeting potential for arthritic lesions
- similar arthritis targeting experiments have been performed with antibody preparations labeled with near-infrared fluorophores
Deliverables
D8.1) Imaging of patients and of animal models:
The Deliverable has been successfully completed, as documented by the execution of the Phase 0 clinical trial (in extremely rapid timelines: less than 2 years from submission to execution) and by the large number of complex preclinical imaging and biodistribution studies.
The WP ended up featuring many more experiments, in addition to the ones described in the Technical Annex. This WP is particularly important for the PRIAT Project, as it outlines how Precision Medicine approaches help develop next-generation antibody therapeutics in a controlled and predictive manner. It is one of the Flagship WPs and success stories of the PRIAT Project, which will have considerable visibility in the future.
WP9 – Analysis of biopsies
In this WP, we have analyzed material from patients and from animal models (e.g. biopsies, blood samples, plasma specimens) using complementary multiplex technologies.
In cancer (i.e. patients in clinical trials or mouse models treated with antibody therapeutics), we have aimed at:
- evaluating variability of target antigen expression and leukocyte infiltration in freshly-frozen tumor sections, using several antibody reagents
- analyzing infiltrating leukocytes by FACS analysis of resuspended cell preparations
- analyzing cytokine levels
- analyzing the T cell reactivity against tumor-rejection antigens
In mouse models of arthritis (following antibody-based treatment), we have aimed:
- evaluating CytoF technology for joint and lymphnode cells
- analyzing infiltrating cells by FACS
- analyzing cytokine levels by MesoScale Discovery
In the rat model of chronic heart transplant rejection, we have assessed the rejection grade using a defined scoring system adapted to the rat model including the extent of lymphocyte infiltration, myocyte degeneration, fibrosis and vasculopathy as well [Franz et al. (2011) J. Heart Lung Transpl., 30, 86]. Moreover, lymphocyte infiltration have been characterized with regard to the different subtypes of immune cells as well as their spatial tissue distribution (e.g. association to histological signs of chronic rejection: fibrosis and vasculopathy).
While a more complete list of all performed activities can best be derived from the data described in the publications (listed below), here we would like to outline some of the most significant findings, which have emerged from studies in cancer, arthritis and chronic heart rejection.
Use of the immunocytokine L19-IL2 for the intratumoral injection of cutaneous/subcutaneous metastases in Stage III melanoma patients.
As mentioned in WP3, clinical trials in patients with Stage IIIC and Stage IV melanoma have been performed, featuring intralesional administration of either L19-IL2 alone (25 patients), or of a combination of L19-IL2 and L19-TNF (22 patients). The latter Phase II clinical trial was motivated by the observation of strikingly potent activity in immunocompetent mouse models of cancer [Schwager et al. (2012) J. Invest. Dermatol., 133, 751; Pretto et al. (2014) Cancer Immunol. Immunother., 63, 901].
In a first study, we addressed the use of L19IL2 for the intralesional treatment of Stage IIIB/C melanoma patients [Weide et al. (2014), Cancer Immunol. Res. 2, 668; Figure 17]. We reasoned that a targeted form of IL2 injected intralesionally into melanoma metastases would exhibit prolonged residence time in the lesions [Pretto et al. (2014) Cancer Immunol. Immunother., 63, 901], as compared to the untargeted form, therefore allowing an extended immunological action, reduction in the frequency of administrations and a shorter duration of the treatment.
A complete response of all treated metastases was achieved in 6/24 evaluable patients (25%), long-lasting in most cases (5 patients ≥ 24 months). Objective responses were documented in 53.9% of all index lesions.
We also recorded a slower progression to distant metastases in this cohort of patients as compared to historical controls [Weide et al. (2014), Cancer Immunol. Res. 2, 668].
The second study describes, for the first time, the clinical use of a combination of two immunocytokines for the intralesional treatment of injectable lesions in Stage IIIB/C and Stage IVM1a metastatic melanoma patients. Twenty-two patients were treated with a mixture of L19-IL2 and L19-TNF once weekly for up to 4 weeks. The dose was shared among the lesions via multiple intralesional injections and newly occurring melanoma lesions in the regional field (in-transit metastases) within the 4 weeks treatment period were also treated as described. Treatment was in general well tolerated, with a limited number of drug-related adverse events, mostly of grade ≤ 2, which were normally rapidly resolved. In terms of efficacy, although no CR was observed according to modified RECIST criteria, PRs were recorded in 50 % of patients. A locoregional assessment of individual lesions indicated a better outcome of the study, with 5% of CR and 50% of PR. Importantly, according to these criteria, no patient was considered to be in PD at the time of tumor assessment. Interestingly, responses in some patients manifested themselves or continued to improve several weeks after the end of treatment. Furthermore, the observation of disappearance or reduction in size of 8 out of 12 lesions, which were not injected, suggests that intralesional injection of the L19-IL2 in combination with L19-TNF mediates a systemic anti-cancer activity (Figure 18A).
Histological and immunohistochemical findings in two different patients showed presence of extensive necrosis and melanophages in treated lesions while in untreated lesions only small areas of necrosis were apparent (not shown). Infiltration by CD8+ T cells was markedly increased in treated lesions as compared to untreated (Figure 18B), not responding ones, while CD4+, Tregs and NK cells infiltration of treated lesions was not significantly changed as compared to untreated, not responding lesions (manuscript submitted).
Analysis of melanoma biopsies in other types of antibody-based treatments
The University of Graz has collected biopsies from patients before and after anti-melanoma therapy. Staining protocols for immune modulating molecules expressed on cancer cells; i.e. PD-L1, MICA/B and HLA have been established for FFPE samples. Methods to detect markers creating an immune-promoting or immune-suppressive tumor microenvironment (e.g. CD68, CD163, iNOS, Arginase, IDO), as well as possible markers creating a respective microenvironment in the tumor host (e.g. immune suppressive cytokines such as IL4 or IL10, circulating MDSCs, Tregs, Th17 cells) have also been established (IHC, pPCR, FCM).
These analyses demonstrated that presence of an immune suppressive microenvironment characterized by low MICA/B, MHC class I and a high Arginase, IDO, iNOS, CD163 expression was associated with a weak to absent intratumoral CD8+ T cell infiltrate and vice versa. Interestingly, expression of the immune-inhibiting PD-L1protein, or presence of FOXP3+ Tregs, was associated with a pronounced CD8+ T-cell infiltrate. Unfortunately, the number of available sequential lesions, i.e. before and after an immune modulating therapy (which was mostly ipilimumab) was limited. Consequently, a statistical analysis with respect to whether these results can be used as predictive biomarker was not feasible. However, there was a clear trend that tumors characterized by an immune promoting microenvironment or were characterized by an intratumoral T-cell infiltrate showed a more pronounced response to ipilimumab therapy, e.g. an even further increased T-cell infiltration.
Analysis of the circulating immune regulatory cells revealed that only the frequency of Th17 cells correlated with an immune promoting microenvironemt. It should be noted, however, that this was only a trend, possibly to the limited sample size. The quantification of immune suppressive cytokines were characterized by strong intra-individual variations (e.g. when several samples obtained prior to therapy from the same patient); thus, these were not used for further analysis.
Analysis of biopsies, leukocyte infiltration and cytokine levels in preclinical models of cancer
ETH and Philochem have performed an extensive analysis of bioptic specimens and of cytokine levels in therapy experiments with immunocytokines, executed in mouse models of cancer. Products which have been tested in vivo and which have been subjected to the above mentioned analysis include immunocytokines based on IL2, IL4, IL12, IL13, IFN-g and TNF. A list of publication is provided at the end of this section, in which leukocyte infiltration into tumor lesions and cytokine levels were measured. Figure 19 illustrates a typical analysis of leukocyte infiltration in therapy experiments, performed in mouse models of cancer.
The measurement of lymphocyte infiltration into the tumor mass, alone, does not provide a functional readout. However, antibody therapy studies in tumor-bearing mice, for which a selective depletion of CD4+, CD8+ or NK cells had been performed, have shed light on the classes of lymphocytes, which contribute to an anti-cancer protective immunity, which can be boosted by antibody therapeutics.
Analysis of biopsies, leukocyte infiltration and cytokine levels in mouse models of rheumatoid arthritis and psoriasis
To gain some insight about the mechanism of action of F8-IL4 in the murine collagen-induced model of rheumatoid arthritis and three different skin inflammation models of psoriasis, samples were analyzed for change in cytokine levels due to the treatment with targeted IL4, a strong TH2 cytokine. A panel of 13 different pro- and anti-inflammatory cytokines that play a role in TH1-, TH2- and TH17-responses were analyzed in serum and at the site of disease (e.g. in paw tissue for arthritis and in ear tissue for psoriasis). The infiltration of immune cells due to therapeutic treatment was evaluated by immunofluorescence procedures.
In arthritis, F8-IL4 elevates certain anti-inflammatory cytokine levels such as IL10 and IL13, while other pro-inflammatory cytokines such as IL6 are down regulated in serum. In the paw tissue, F8-IL4 shifts the balance of many cytokine and therefore modulates the immune environment at the site of disease (Figure 20).
The combination of targeted IL4 and dexamethasone was able to cure mice from established arthritis. The analysis of cured paws compared to healthy paws and active arthritic paws showed that the treatment normalized cytokine levels to the levels of healthy mice (Figure 21).
In murine models of psoriasis, treatment with F8-IL4 led to increased levels of key regulatory cytokines (including IL5, IL10, IL13, and IL27), while other cytokine levels remained unchanged (e.g. IL2) or are down regulated (e.g. IL1alpha), when compared to buffer control treated mice (Figure 22). F8-IL4 also mediated an influx of immune cells such as macrophages and NK cells to the site of inflammation as assed by immunofluorescence methods (see above).
The measurement of cytokines and identification of immune cell populations at the site of disease remains an important read-out when testing new therapeutic strategies. Unfortunately, when using animal models we are limited by cell numbers found at the site of disease as well as the amount of serum taken from individual mice. In addition when examining cells in arthritic paws, antigen markers can be altered during the paw digestion process. In our lab we have optimized a DNase and Liberase digestion strategy flowed by a flow cytometric based technique which reliably examined immune cell populations present in arthritic paws (Figure 23). In addition we established the Meso Scale Discovery (MSD) platform to analyse multiple pro- and anti-inflammatory cytokines present in a small volume (10 cytokines in 25 μL). These two techniques have allowed us to examine the effect of F8-IL-4 on cytokines and immune cells in the paws of arthritic mice (WP6).
Analysis of biopsies, leukocyte infiltration and cytokine levels in a rat model of chronic graft rejection
To characterize the chronic rejection process in the heterotopic rat heart transplantation model under qualitative and quantitative aspects, besides the histological and morphometric analysis of surrogate markers of chronic rejection like cardiac allograft vascolupathy or cardiac interstitial fibrosis, also immunohistochemical determination of important immune cells (CD4+ - T helper cells, CD8+ - cytotoxic T cells; CD68+ - natural killer cells; CD161+ - macrophages etc.) rejection markers (i.e. C4d) as well as key molecules of stromal activation (i.e. alpha-SMA) were performed in detail in the chronically rejected cardiac tissue compared to healthy controls. In general, a significant up-regulation of all investigated immune cell markers as well as the rejection marker C4d and also alpha-SMA, as a marker of activated myofibroblasts or vascular smooth muscle cells, could be observed compared to controls. Moreover, using RNA extracted form rat cardiac tissue was subjected to RT-PCR based gene expression analysis of key molecules related to cardiac extracellular matrix and fibrosis. The complex results cannot be described in this but will be analyzed in detail in the future with respect to the identification of novel key genes related to the rejection process.
Analysis of predictive biomarkers and surrogate markers for outcome of melanoma patients treated with the CTLA-4 antibody ipilimumab
Only insufficient biopsies were available for systematic analysis of biomarkers in melanoma patients treated with antibody-based treatments. Therefore, we focused to identify predictive biomarkers and surrogate markers for outcome of melanoma patients receiving the CTLA-4 antibody ipilimumab in the peripheral blood.
The spectrum of biomarkers to be tested comprised frequencies of immune cell subsets in the peripheral blood, differential blood count parameters (e.g. the absolute and relative lymphocyte count), serum level of lactate dehydrogenasis (LDH) and clinical factors (stage according to the AJCC classification, age, gender). LDH-ratio was calculated by the measured value divided by the upper limit of normal in the respective test system. Immune cell subsets were analyzed using flow cytometry after staining of peripheral blood mononuclear cells (PBMC) with up to four different panels of fluorescence-labeled antibodies (s. Table 3) to cover a broad spectrum of phenotypes to be investigated.
Overall survival according to biomarker result was tested by Kaplan-Maier analysis. Different groups regarding a single factor were compared using the log-rank test. The best clinical response according to immune-related response criteria (irRC) upon CTLA-4 therapy is presented by descriptive statistics and Chi-square tests. For all patients, blood taken at least 28 days before start of antibody therapy was used for this study.
An initial cohort of 25 patients (cohort 1) had already been analyzed prior to start of PRIAT. For all continuous factors patients were dichotomized into two groups based on the median value observed in this cohort (group 1: <= median, group 2: > median) to analyze the impact of each biomarker. Significant associations with favorable survival were detected for high relative frequencies of eosinophils and for low frequencies of Lin-CD14+HLA-DRlow myeloid-derived suppressor cells (Table 4 left).
Based on these promising results, we set up a large collaborative study to confirm the initial findings identified in cohort 1. Altogether 342 PBMC samples were collected for this cohort from 8 international clinical sites (New York, Amsterdam, Siena, Napoli, Lausanne Nantes, Essen and Tübingen) comprising 184 baseline PBMCs and 158 PBMC specimen collected during and after antibody treatment.
The first part of the PRIAT project focused on the analysis of 184 baseline samples to identify predictive markers. The entire cohort was divided into an identification cohort (cohort 2: n=85 baseline samples from Amsterdam, Lausanne, Nantes, Essen and Tübingen) and a validation cohort (cohort 3: n=99 samples from New York, Siena and Napoli).
First, 22 continuous factors were analyzed using two cut-off points for each biomarker. The first cut-off point was defined by the median observed value in cohort 2 (dichotomization method). The second cut off point was that which resulted in the most significant differences in survival comparing the high vs. low group of the respective biomarker (optimization method). Gender, the clinical substage were additionally assessed. Altogether 24 candidates were significantly associated with survival in cohort 2. Next, all variables using the cut-off points established in cohort 2 patients were tested in cohort 3. Among the 24 candidates 7 were confirmed in cohort 3 (Table 4).
Multivariate analysis was applied for 134 patients of cohorts 2 and 3 with available results for all seven confirmed biomarkers to investigate their degree of independency and their relative impact if analyzed in combination. Cox regression analysis resulted in three independent biomarkers predictive for increased risk to die from melanoma: LDH-ratio (hazard ratio [HR] 2,364, p=0,00012), the relative eosinophil count (HR 1,962; p=0,00673) and Lin-CD14+HLA-DRlow myeloid-derived suppressor cells (HR 2,652; p=0,00002).
The differences in overall survival for those three biomarkers and their combined consideration is shown for the entire cohort (cohorts 2 & 3) in Figure 24. An overview over univariate survival data according to analyzed biomarkers is presented in Table 5.
The best clinical and radiological response upon treatment is demonstrated in Figure 25. A response according to irRC was significantly more frequent in patients with low LDH or low frequencies of MDSCs. No significant difference could be established for relative eosinophils. The more out of these three favorable factors (low LDH-ratio, high eosinophils, low MDSCs) were observed in patients the higher was the rate of responders (Figure 26). The manuscript of this part is in preparation and will be submitted in Q1 2015.
In the second part of this PRIAT project we are analyzing changes of the same spectrum of biomarker upon treatment by comparing the baseline findings acquired in part A to the results from later time points during and after ipilimumab. Post baseline samples were analyzed in 85 patients. Data acquisition is completed. Data analysis and interpretation is still ongoing. We intend to submit a manuscript focusing on surrogate markers by Q2 2015.
In the third part of this PRIAT project we focus on the predictive impact of γ/delta T cell subsets in patients treated with ipilimumab. Data acquisition is completed. Data analysis and interpretation is still ongoing. We intend to submit a manuscript focusing on γ/delta T cell subsets by Q2 2015.
Deliverables
D9.1) Analysis of biopsies and of biological specimens:
The Deliverable has been successfully completed, as documented by the extensive number of publications and by the analyses, performed both in patient-derived specimens and in rodent models of disease.
WP10 – Project Management
A kick-off meeting has been organized at the beginning of the Project, in order to coordinate the activities and the distribution of resources, reminding all participants of the rights and duties associated with the participation in the PRIAT Project (Figure 27).
After 12 months (i.e. 11+12 November 2013), a second meeting has taken place in Zurich, with active presentations and discussions by the groups, as well as a review of the current spending and of the status of deliverables (Figure 28).
At the end of the 24 months of the Project (i.e. 27 + 28 October 2014), a final meeting has taken place in Zurich (Figure 29), with active presentations by the individual groups, as well of experts in the field of cancer immunology and pharmacotherapy [Prof. Dr. Raffaella Giavazzi (Mario Negri Institute of Pharmacological Research, Milano, Italy), PD Dr. Christoph Schliemann (University Hospital Münster, Germany), Prof. Dr. Cornelia Halin-Winter (ETH Zürich, Switzerland) and Prof. Dr. Sine Reker-Hadrup (Technical University of Denmark)].
A visit of GMP production facilities of Philogen and other training activities have been organized during the 24 months of the Project (Figure 30).
Furthermore, various in-person meetings (including several internships) have been organized among the laboratories. These personal contacts have been fundamental for the successful completion of the Project.
At each Meeting, a General Assembly took place, as foreseen by the Consortium Agreement. There were no major deviations from the original scientific plan, as the Project ran smoothly all deliverables could be produced within the expected time period. There were small reallocations of the original budget (7.5%). Specifically, MUG used 119771.00 € less and VUMC 105242.00 € less of their budget and these funds were re-allocated to other participants, as decided in the General Assembly and as communicated to all partners in writing.
Clinical trials and the use of clinical samples has been approved by the competent Authorities and Ethical committees, with suitable Patient information forms
An Ethics Review Committee has followed our Project and participated at the yearly Meetings
WP11 - Dissemination Activities
A Web site for the PRIAT Project was set up at the beginning of the Project [www.priat.eu].
Some of the most significant milestones reached by the PRIAT Project have been communicated to the broader public.
A detailed list of Publications and of Presentations at national and international conferences is provided in a separate document.
Potential Impact:
Potential Impact
The aim of Project in relation to the Problem
The vast majority of pharmaceutical agents can be divided into two categories: small organic drugs (typically < 500 Da) and biotech drugs, mainly consisting of therapeutic proteins. About 50% of product sales in this latter category is represented by antibody-based products. Seven out of ten of top-selling therapeutic proteins are antibodies or antibody derivatives. Furthermore, Humira™ (a monoclonal antibody directed against TNF and approved for the treatment of various inflammatory conditions) is the best-selling pharmaceutical agents in all categories [Walsh (2014) Nat. Biot., 32, 992].
In spite of their commercial success, antibody products are not always as active as one would like them to be. For example, Avastin™ and Herceptin™, used in combination with chemotherapy, have been approved for the treatment of metastatic colorectal cancer and of metastatic breast cancer, respectively, on the basis of a prolongation of median overall survival for only few months. However, patients treated with these antibody drugs are typically not cured and the few extra months of life come at a very high cost for the healthcare system. TNF blockers (such as Humira™, Enbrel™ and Remicade™) provide a substantial benefit to about 50% of patients with rheumatoid arthritis. However, these drugs do not cure the disease. Thus, there is a need to develop better antibody-based drugs and to understand which patients are likely to benefit from antibody therapy most.
In the PRIAT Project, we focused on three main indications for antibody-based therapeutics (cancer, rheumatoid arthritis and graft rejection), with the aim to experimentally address three issues, which are crucial for the understanding of antibody activity and for the design of the drugs of the future:
- How efficiently do antibody drugs accumulate at the site of disease?
- How efficiently can antibody drugs modulate the activity of the immune system at the
site of disease?
- How efficiently can we profile the HLA-I and HLA-II peptidomes in health and disease, in order to be able to identify the targets of T-cell recognition (e.g. tumor rejection antigens, graft rejection antigens, autoantigens for autoimmune conditions)?
Within its two years of activity, the PRIAT Project has made a considerable impact in all three areas mentioned above. Here, we summarize the main findings and the main results.
Characterization of the disease-targeting performance of antibody products
In animal models of disease, we have performed extensive quantitative biodistribution studies, using radiolabeled preparations of therapeutic antibody products. This work has led to numerous publications and, collectively, has provided us with a quantitative understanding of the selectivity which antibody products can reach in vivo in rodent models of cancer, rheumatoid arthritis and graft rejection.
A priori, there is no guarantee that biodistribution and imaging studies in animal models may faithfully reflect the performance of the same products in patients. In order to bridge this gap of knowledge, we have made two contributions to the field, which we believe to be important:
(i) we have published a dosimetric estimate of radiation doses in organs and neoplastic lesions for a pan-tumoral antibody (L19, specific to the alternatively-spliced EDB domain of fibronectin), which has been studied in patients with solid tumors or with lymphomas. These studies have shown that the L19 antibody preferentially localizes on neosplastic lesions, but also that the tumor-homing selectivity varies greatly (i.e. up to 40-fold) from lesion to lesion and from patient to patient
(ii) we have developed and clinically implemented a methodology for the radiolabeling of antibody products in GMP conditions with iodine-124, a PET radionuclide which is particularly suitable for the study of antibody products in PET procedures. The methodology has been used for the characterization of Dekavil (F8-IL10), an armed antibody developed by Philogen, currently being investigated in randomized doubleblind Phase IIb clinical trials in patients with rheumatoid arthritis. In addition to the specific information gained in this study about the arthritis homing properties of the product, the Project has demonstrated the feasibility of mechanistic immuno-PET studies, which should facilitate product development in the future
Multiplex analysis of biomarkers and of leukocyte infiltration
We have implemented the routine analysis of leukocyte infiltration at the site of disease and the multiplex analysis of cytokine levels (in serum and at the site of disease) for many preclinical experiments, aimed at the characterization of the activity and mechanism of action of various antibody-based therapeutics. These studies have been published in several articles and have facilitated the identification of product candidates for clinical and industrial development programs (see below).
Furthermore, we have studied in substantial detail the infiltration of lymphocytes into melanoma lesions in patients receiving antibody drugs and the specificities of cytotoxic T-lymphocytes which account for long-term survival, following pharmacological treatments. These investigations have led to the identification of certain tumor rejection antigens (e.g. NY-ESO-1), which are efficiently recognized by T cells in patients that respond well to treatment. Furthermore, the demonstration that certain antibody-based therapeutic interventions (e.g. treatment with IL2-based immunocytokines) led to a dramatic increase of certain leukocytes (e.g. T cells and NK cells) in the neoplastic mass has prompted the design of novel combination therapies, which are about to be studied in clinical trials (see Sections below).
HLA peptidome analysis
We have developed, implemented and perfected methodologies for the determination of “atlases” of peptides bound to MHC-I or MHC-II (HLA-I and HLA-II in man). Thanks to the use of state-of-the-art mass spectrometers and to the implementation of efficient antibody-based pulldown procedures, these methodologies have allowed the determination of complex peptidomes, comprising hundreds of MHC-bound peptides, in complex biological samples, such as cell lines or patients’ sera. For the first time, it has been possible to identify hundreds of HLA-I-bound peptides from sera of melanoma patients. The identified peptides included previously reported melanoma rejection antigens, thus providing confidence about the robustness of the procedure. In mice with arthritis, collagen-derived peptides were identified as MHC-II ligands, thus providing a direct analytical access to antigens which may trigger autoimmune conditions. The applications opened by the successful implementation of methods for the determination of HLA peptidomes are numerous and, in our opinion, important. For example, on the basis of the knowledge of HLA-I peptidome in melanoma patients, it now becomes possible to synthesize the corresponding peptides and probe T-cell specificities in the same patients, using either peptide based stimulation or multiplex tetramer technology, pioneered by Ton Schumacher in Amsterdam.
Furthermore, we now have the opportunity to go back to experiments, in which we could cure tumor-bearing mice with innovative antibody-based therapeutics (e.g. immunocytokines based on IL2 or TNF payloads) and determine which are the tumor rejection antigens recognized by cytotoxic T-cells, providing a protective immunity against the tumor.
The importance of HLA peptidome analysis and of multiplex tetramer technology analysis for the study of the response of melanoma patients to antibody drug treatment [Kvistborg et al. (2014) Sci. Transl. Med., 6, 254ra128] has become a particularly important and timely topic, with the approval of anti-CTLA4 and anti-PD1 immunostimulatory antibody products.
Socio-economic impact
We believe that the PRIAT Project has produced important technologies and publications, which facilitate the development and mechanistic study of antibody drugs. In time, this will lead to the generation of better antibody products and to improved selection procedures for patients, which are more likely to benefit from therapeutic intervention.
In addition to these rather indirect (but important!) results, we believe that the PRIAT Projects has also produced results, with an immediate socio-economic impact. First of all, patients with melanoma, leukemia, sarcoma and rheumatoid arthritis have been studied in clinical trials. Their response to antibody-treatment has been analyzed with improved methodologies, which were not available so far. A direct clinical benefit has been demonstrated, at least for patients with oncological conditions which have responded to antibody treatment.
Secondly, the information gained on the ability of certain armed antibody products (e.g. those containing IL2 or TNF payloads) to stimulate a massive increase of T cells and NK cells in the neoplastic mass, has stimulated the design of novel combination clinical trials (e.g. L19-IL2 + Rituximab in last-line diffuse large B-cell lymphoma), which are now about to start.
Thirdly, the knowledge gained for the understanding of the disease-homing and immunoregulatory properties of armed antibodies has facilitated the development of novel potent therapeutic agents. For example, we now have armed antibody products which induce complete responses in mouse models of soft-tissue sarcoma and in patients (L19-TNF). Moreover, we have recently published the first example of a therapeutic agent, capable of curing 100% of treated mice with established rheumatoid arthritis [Hemmerle et al. (2014) Proc. Natl. Acad. Sci. U.S.A. 111, 12008]. This product is now completing safety toxicological testing in Cynomolgus monkeys and is scheduled to begin clinical trials in 2015 in patients with rheumatoid arthritis and with endometriosis.
Finally, by supporting a collaboration between academic centers, hospitals and biotech companies, the Project has reinforced the competitiveness of European biomedical research in terms of knowledge, product development and clinical translation. The Philogen and Philochem groups have grown, both in terms of revenues and in terms of number of employees, since the beginning of the PRIAT Project.
Wider societal implications
There are many direct benefits, which have stemmed from the PRIAT Project, including the training of European students at our centers and exchange programs. For example, we have had numerous exchanges of visiting scientists and we have organized visits to the GMP Production Facility of Philogen, thus providing a “non-standard” training to students on activities of profound pharmaceutical importance.
For the broader society, the main impact of the Project is related to the knowledge generated in these two years of experimental work (see next section on Dissemination Activities) and to the industrial and clinical activities, which have been facilitated by the PRIAT Project (see sections below).
In general, our impression is that the European Pharmaceutical and Biotech Industry is losing competitiveness towards other companies located in the United States or in emerging countries. However, projects like PRIAT, which allow a close interaction between experts in academia, hospitals and biotech companies, can stimulate nice synergies, resulting in the development of new products and in the creation of new jobs.
In the medium and long term, we believe that the main impact of the PRIAT Project for the broader society (in Europe and elsewhere) will be related to the implementation of better patient selection procedures (based on imaging, multiplex analysis and HLA-peptidome analysis) and to the development of better antibody products.
Main dissemination activities
The main avenues for the broad dissemination of scientific results have been:
- publications in peer-reviewed scientific Journals
- presentations at national and international conferences
In total, 18 publications on topics directly related to the PRIAT Project have been published by the Consortium, while additional manuscripts are currently in preparation.
In total, 64 presentations at national and international conferences have been given by scientists of the PRIAT consortium between November 2012 and October 2014.
At the end of the Project, a mini-Symposium on the specific topics of the PRIAT Project has been organized in Zürich. Furthermore, the European Union has covered the interim results of the PRIAT Project in a dedicated Inteview, which has appeared on the Horizon 2020 Website:
https://ec.europa.eu/programmes/horizon2020/en/news/priat-probes-how-antibody-drugs-help-body-cure-itself
Exploitation of the results
As outlined above, the main objective of the PRIAT Project was the development of innovative methodologies and their implementation in preclinical and clinical studies. The methodologies per se were not patentable, but the information gained by the application of the methodologies has led to the discovery (and patenting) novel antibody products and of novel combination strategies.
We believe that the PRIAT Project has facilitated a number of tangible and important development activities, some of which are listed below:
i) clinical implementation of immuno-PET procedures, as integral component in the development of armed antibody drugs
ii) knowledge stemming from the dosimetric evaluation of disease-targeting properties of antibody drugs in preclinical studies and in immuno-PET clinical trials
iii) characterization of the mechanism of action of novel anticancer antibody-based therapeutics, which has led to the improved use of antibody products which are already in the clinic (e.g. L19-IL2, F16-IL2, L19-TNF, as well as immune checkpoint inhibitors), or to the launch of new clinical trials (e.g. L19-IL2 + rituximab in DLBCL, L19-IL2 + L19-TNF for the locoregional treatment of Stage IIIC melanoma and of other malignancies)
iv) discovery of the first agent (an armed antibody: F8-IL4) capable of eradicating established arthritis in the collagen-induced model of the disease. The corresponding fully-human immunocytokine will begin clinical trials in 2015
v) development of antibody-based procedures, for the imaging of failing heart in a rat model of chronic graft rejection. This methodology has the potential to provide a non-invasive detection of rejection processes by SPECT or PET methodologies, perhaps avoiding the need for serial biopsies (the current methodology for the monitoring of patients, following heart transplantation)
vi) the development of predictive methods for long-term survival of melanoma patients, following pharmacological intervention (typically with antibody drugs, but also with other agents)
In summary, we believe that the PRIAT Project has been truly successful, not only in terms of methodology development but also for the boosting of European Biotech competitiveness, with novel antibody products being brought to the clinic and improved analytical procedures for the study of response to antibody treatment.
List of Websites:
website: www.priat.eu
