Targeting B-cell malignancies with the beta-emitting anti-CD37 radioimmunoconjugate 177Lu-NNV003

Purpose The aim of this study was to explore the β-emitting lutetium-177 labelled anti-CD37 antibody NNV003 (177Lu-NNV003, Humalutin®) for the treatment of non-Hodgkin’s lymphoma in in vitro studies and in animal models. Methods Cytotoxicity of 177Lu-NNV003 was measured in REC-1 (mantle cell lymphoma) and DOHH-2 (diffuse large B cell lymphoma) cell lines. Biodistribution was studied in mice bearing subcutaneous DOHH-2 or MEC-2 (chronic lymphocytic leukaemia) xenografts. The therapeutic effect of a single injection of 177Lu-NNV003 was measured in mice intravenously or subcutaneously injected with REC-1 cells. Haematological and histopathological assessments were used to evaluate the toxic effect of 177Lu-NNV003. The immunotherapeutic effect of NNV003 was assessed by measuring binding to Fcγ receptors, activation of ADCC and ADCP. NNV003’s immunogenicity potential was assessed using in silico immunogenicity prediction tools. Results 177Lu-NNV003 showed an activity dependent antiproliferative effect in all cell lines. Maximum tumour uptake in vivo was 45% of injected activity/g in MEC-2 tumours and 15% injected activity/g in DOHH-2 tumours. In mice injected intravenously with REC-1 cells, 177Lu-NNV003 (50–100 MBq/kg) improved survival compared to control groups (p < 0.02). In mice with subcutaneous REC-1 xenografts, 500 MBq/kg 177Lu-NNV003 extended survival compared to the control treatments (p < 0.005). Transient haematological toxicity was observed in all mice treated with radioactivity. NNV003 induced ADCC and ADCP and was predicted to have a lower immunogenicity potential than its murine counterpart. Conclusion 177Lu-NNV003 had a significant anti-tumour effect and a favourable toxicity profile. These results warrant further clinical testing in patients with CD37-expressing B cell malignancies. Electronic supplementary material The online version of this article (10.1007/s00259-019-04417-1) contains supplementary material, which is available to authorized users.


Detailed Materials and Methods
Labelling and Quality Control of antibodies with 17 7 Lu NNV003 and cetuximab (Merck KGaA, used as isotype control), both chimeric IgG1 antibodies, were first conjugated with p-SCN-Bn-DOTA (Macrocyclics, Texas, USA) dissolved in 5 mM HCl. p-SCN-Bn-DOTA was added in carbonate buffer pH 8.8, at a molar ratio of 10:1. The solution was incubated at 37°C while shaking for 2 hours, before it was stopped by addition of 50 µl of 0.2 M glycine solution (Merck Millipore, Massachusetts, USA) per mg of antibody. A buffer exchange with 50 ml 0.9% NaCl was performed by centrifugation (Centrifuge 5804R, Eppendorf, Germany) in Vivaspin 15R centrifuge tubes (Sartorius, Goettingen, Germany).
Instant thin layer chromatography (Tec-Control ITLC strips, Biodex Medical, New York, USA) was used to measure the radiochemical purity (RCP) of the conjugates. If the RCP was below 95%, the conjugate was purified using Sephadex G-25 PD-10 columns (GE Healthcare Life Sciences, Chicago, USA) and RCP >95% was verified. The immunoreactivity was measured using a modified Lindmo model [1] with one cell concentration of 75 x 10 6 cells/ml. The immunoreactivities of 177 Lu-NNV003 used in the experiments were between 70% and 88%.
ADCP and ADCC assays A FcγRIIa-H ADCP-and a FcγRIIIa ADCC Reporter Bioassay kit (Promega, USA) were used for measuring ADCP and ADCC induction in MEC-2, REC-1 and DOHH-2. 2.5 µg/ml NNV003, lilotomab or rituximab (Roche, Switzerland, used as positive control) was added to the target cells (1 x 10 6 cells/ml), using three replicates per antibody per cell line. Effector cells were added at a 2:1 effector:target cell ratio and the cells incubated for 21 hours. After bioluminescence measurements, the background signal was subtracted from each well, and the wells were normalized against the control containing only target and effector cells.
Cell lysis by ADCC was measured by chromium-51 release assay. Target cells, REC-1, DOHH-2, Ramos (Burkitt's lymphoma) and Daudi (Burkitt's lymphoma) were labelled with 51 Cr and incubated with NNV003 or lilotomab (20 µg/ml) or rituximab as positive control (10 µg/ml) for 10 min. IL-2 stimulated human PBMC was added at a 10:1 effector:target cell ratio. After 4 hours of incubation the cells were washed, and the cytotoxic effect was measured by Cr 51 release into the supernatant by a gamma counter (Cobra II, Packard, Perkin Elmer, USA). Controls: target cells were cultivated in medium alone (spontaneous lysis) and in medium containing 1% Triton X-100 (maximal lysis). Specific lysis was calculated by subtracting the spontaneous release from the experimental release and dividing by the maximum release minus the spontaneous release.
Cell cytotoxicity of 1 7 7 Lu-NNV003 Cell proliferation after treatment with unlabelled or 177 Lu labelled NNV003, or unspecific isotype antibody ( 177 Lu-cetuximab) was measured in REC-1 and DOHH-2. Cells were seeded in deep well plates at a density of 1.9-2.7 x 10 6 cells/ml. Antibodies were added to the cells at final concentrations of 10 ng/ml to 20 µg/ml, using two replicates per cell line per treatment. The cells were incubated while shaking for 20 hours and then washed three times in PBS containing 0.5% bovine serum albumin (BSA). After resuspension in cell culture medium to 0.4 x 10 6 cells/ml, the cells were re-seeded in 96-microwell plates for a further six days of growth. Cells treated with unlabelled NNV003 were not washed, only diluted to 0.4 x 10 6 cells/ml. CyQUANT™ NF Cell Proliferation Assay Kit (Thermo Fisher Scientific, USA) was used to measure cell proliferation. The experiments were performed on 2 or 3 separate occasions. The cell lines had a doubling time of approximately 2 days.
Radiation dosimetry of 1 77 Lu-NNV003 The biodistribution data from the DOHH-2 s.c. model was used to calculate the absorbed radiation doses from 177 Lu-NNV003 in different organs. The area under the activity versus time curves (AUCs) were normalized to an injection of 100 MBq/kg, the activity used in the REC-1 i.v. therapy study. AUCs, calculated by the trapezoidal rule, were multiplied by the mean energy of the β-particles, Auger-and conversion electrons of 0.147 MeV [2]. At t=0 100% of the injected activity was estimated to be in the blood. The absorbed radiation doses were adjusted for self-and cross-radiation by multiplying by factors developed by Miller et al., 2005 [3]. The self-radiation of blood and brain was set to 1. For lymph nodes, factors developed in [3] for a tumour of 25 mg was used. The factor used for selfradiation for the tumours was 0.95, according to the average tumour mass in the study (0.3 g).

Additional data
In silico Immunogenicity prediction In silico Major Histocompatibility Complex (MHC) class II-peptide binding predictive analysis of NNV003 and lilotomab was performed by EIR Science A/S (Copenhagen, Denmark) using the NetMHCIIpan 3.1 algorithm [4]. Only peptides predicted with a rank score of < 10 were considered as potential MHC class II binders. Immunogenicity risk scores (IRS) were calculated based on MHC class II binding profiles of protein derived 15-mer peptides to a "world average" population of MHC class II alleles consisting of more than 300 different MHC Class II alleles and 11 geographical defined populations. The IRS reflects the number of MHC class II-binding peptides weighted by the frequency in the investigated population. Self-adjusted IRS were obtained after subtraction of the epitopes also found in known antibody germline sequences and antibody conserved regions.
NNV003 is a mouse-human chimeric IgG1 monoclonal antibody (mAb), containing the variable heavy (VH) and light (VL) regions of its murine analogue, lilotomab, a mouse IgG1, κ (mIgG1, κ) mAb, and the constant regions of the heavy (CH) and light (CL) chains of human IgG1, κ (hIgG1, κ). In silico MHC class II-binding prediction analysis of peptides derived from NNV003 light and heavy chain sequences suggests that replacing lilotomab (CH+CL) with hIgG1 (CL+CH) decreases NNV003 immunogenicity potential, by removing the predicted T cell neo-epitopes identified in lilotomab CH and CL (see Figure  S1, red profile, lilotomab only). It is of interest to note that this decrease of immunogenicity potential is mainly the consequence of T cell neo-epitopes removal in lilotomab CH, as replacing lilotomab CL with the hIgG1, κ counterparts generates a new promiscuous overlapping T cell neo-epitope spanning NNV003 VL-CL in position 102-116 (see Figure S1, NNV003 blue immunogenicity profile not overlapping lilotomab red immunogenicity profile). Of note, this T cell neo-epitope is also identified in the rituximab light chain sequence overlapping the murine VL-human CL region [5].

Figure S1) Predicted immunogenicity profile of NNV003.
Overlapping self-adjusted predicted immunogenicity profiles of NNV003 (blue) and lilotomab (red) light (top panel) and heavy (bottom panel) chains generated using sequence position-specific immunogenicity risk score obtained from an in-silico MHC class II-peptide binding prediction analysis for the world population, performed using the algorithm NetMHCIIpan version 3.1. Grey dotted lines highlight the position of the complementary-determining regions.
Binding of 17 7 Lu-NNV003 to CD37 positive cell lines The equilibrium dissociation constant, Kd, and the mean number of binding sites, Bmax, were measured for 177 Lu-NNV003 in MEC-2, REC-1 and DOHH-2 using Scatchard analysis [6]. 10 x 10 6 cells/ml in 0.2 ml cold PBS + 0.5% BSA were incubated with 177 Lu-NNV003, at concentrations between 20 ng/ml and 25 µg/ml, for one hour at 4°C while shaking. Three to four replicates were made where one or two of them were blocked with 20 µg unlabelled NNV003 prior the addition of 177 Lu-NNV003. Cells were washed three times in cold PBS + 0.5% BSA. Activity of cells was measured before and after washing using a calibrated gamma counter (Wizard 3470, PerkinElmer, USA). The experiment was performed on 2-3 separate occasions. The results are presented in Table S1.
The results from the two therapy studies are presented in Figure S3. In the MEC-2 i.v. model the median survival time of the control groups, NaCl, 0.33 mg/kg IgG1, 200MBq/kg 177 Lu-IgG1 and the NNV003 group was between 19 and 21 days. In the two 177 Lu-NNV003 groups the median survival time was extended to 32 days for the single injection group and 29 days for the repeated injection group. The survival in the 177 Lu-NNV003 groups was significantly better than the NNV003 groups and the 177 Lu-IgG1 group (p < 0.025), however they were not significantly better than the NaCl group (p > 0.05). In the DOHH-2 study, the median survival time of the control groups were 46, 47 and 49 days for the unspecific IgG1, 177 Lu-IgG1 and NaCl group, respectively. Survival of mice treated with NNV003 or 177 Lu-NNV003 at the end of the study, 219 days after cell injection, were 89-100 % ( Figure S3b). Histopathology of organs of the RAG-2 mice revealed tumour infiltration in 100% of the control groups, 1/9 mouse in both NNV003 groups, 2/9 mice in the 200 MBq/kg 177 Lu-NNV003 group, none in the 300 MBq/kg 177 Lu-NNV003 group and 1/10 mouse in the 400 MBq/kg 177 Lu-NNV003. Tumour infiltration was generally observed in a high percentage of organs, mainly related to ovaries, bone G/I tract, lymph nodes, spleen and perirenal adipose tissues. In 177 Lu-NNV003 or NNV003 treated mice the number of infiltrated organs was lower than the control groups. with MEC-2 cells intravenously and (b) CB17 SCID mice (n=9 or 10) were injected intravenously with DOHH-2 cells at day 0. Therapy injections were given at day 2 (MEC-2 study) or day 3 (DOHH-2 study) and some groups (marked 2 x) received a second injection at day 9. One death in (a), marked with an asterisk, was probably caused by radiation toxicity as no macroscopic tumours were found during autopsy