Purpose: CD19, a B-cell lineage-specific marker, is highly represented in B-cell malignancies and an attractive target for therapeutic interventions. MGD011 is a CD19 x CD3 DART bispecific protein designed to redirect T lymphocytes to eliminate CD19-expressing cells. MGD011 has been engineered with a modified human Fc domain for improved pharmacokinetic (PK) properties and designed to cross-react with the corresponding antigens in cynomolgus monkeys. Here, we report on the preclinical activity, safety and PK properties of MGD011.
Experimental Design: The activity of MGD011 was evaluated in several in vitro and in vivo models. PK, safety and pharmacodynamic activity was also assessed in dose-escalation and repeat-dose studies of MGD011 administered once weekly in cynomolgus monkeys.
Results: MGD011 mediated killing of human B-cell lymphoma lines by human or cynomolgus monkey PBMCs as well as autologous B-cell depletion in PBMCs from both species. MGD011-mediated killing was accompanied by target-dependent T-cell activation and expansion, cytokine release and upregulation of perforin and granzyme B. MGD011 demonstrated antitumor activity against localized and disseminated lymphoma xenografts reconstituted with human PBMCs. In cynomolgus monkeys, MGD011 displayed a terminal half-life of 6.7 days; once weekly intravenous infusion of MGD011 at doses up to 100 μg/kg, the highest dose tested, was well tolerated and resulted in dose-dependent, durable decreases in circulating B cells accompanied by profound reductions of B lymphocytes in lymphoid organs.
Conclusions: The preclinical activity, safety and PK profile support clinical investigation of MGD011 as a therapeutic candidate for the treatment of B-cell malignancies. Clin Cancer Res; 23(6); 1506–18. ©2016 AACR.
Progress has been made in the clinical management of B-cell neoplasms, although most remain ultimately incurable with current modalities. Redirecting T lymphocytes to lyse lymphoma/leukemia cells via bispecific molecules that simultaneously engage CD3 on T cells with a B-cell antigen, such as CD19, has emerged as a powerful novel concept, highlighted by the clinical success of blinatumomab (Blincyto). Blinatumomab, however, requires continuous infusion, owing to its short circulating half-life. We report here on the preclinical development of MGD011, a bispecific DART molecule with increased in vitro cytolytic activity compared to blinatumomab and engineered for improved circulating half-life. MGD011 showed potent antitumor activity in mouse leukemia/lymphoma models and displayed prolonged pharmacokinetic properties in cynomolgus monkeys, a cross-reacting species. MGD011 was well tolerated in monkeys, with durable and profound B-cell depletion following weekly administrations. MGD011's potent activity and pharmacokinetic properties may offer therapeutic convenience and applicability in the treatment of B-cell malignancies.
B-cell malignancies represent a heterogeneous group of disorders with varying characteristics and clinical behaviors (1). Although systemic chemotherapy is still the mainstay of treatment for B-cell malignancies, kinase inhibitors that selectively target molecules at the core of the transformation process and antibody therapy are now well established tools (2). Among the latter category, rituximab (Rituxan), a monoclonal antibody (mAb) that targets the B-cell antigen CD20, induces direct tumor cell apoptosis as well as complement- and antibody-dependent cytotoxicity (2, 3). These orthogonal mechanisms of action form the basis for therapeutic combinations that have improved outcome in advanced B-cell malignancies. Yet, approximately 20,000 patients die of lymphoma every year in the United States alone.
Providing T lymphocytes (CTL) with the ability to recognize and destroy tumor cells has shown promise in advanced forms of leukemia and lymphoma. In the form of chimeric antigen receptor (CAR) T-cell therapy, such an approach requires ex vivo isolation, transduction, and reinfusion of the patient's T cells. This complexity can be overcome with bispecific antibodies that bind simultaneously to an antigen expressed by malignant B cells and an activation molecule on T lymphocytes (2). The pan B-cell marker, CD19, has emerged as a promising antigen for targeting B-cell malignancies because of its broader expression profile and lower rate of downregulation compared with other B-cell antigens (3). Its expression is highly conserved in the majority of B-cell tumors (4), with normal to high levels of expression in 80% of acute lymphoblastic leukemia (ALL), 88% of B-cell lymphomas, and all chronic lymphocytic leukemias (CLL; refs. 5, 6).
Redirection of CTL to CD19+ leukemia cells via the bispecific T-cell engager (BiTE) blinatumomab (Blincyto; ref. 7) is effective in patients with B-cell malignancies whose disease did not respond to standard chemo-immunotherapies and has been approved by the FDA for the treatment of patients with Philadelphia chromosome-negative relapsed or refractory B-cell precursor ALL. DART proteins are bispecific, antibody-based molecules with favorable stability, manufacturability, and potency; furthermore, a CD19 x CD3 DART protein compared favorably with a BiTE of the same pair of VH and VL sequences in redirected cytolysis assays (8, 9). MGD011 (also known as JNJ-64052781) is another CD19 x CD3 DART protein designed to simultaneously target CD19+ cells for recognition and elimination by CD3-expressing T lymphocytes as effector cells. MGD011 was engineered with a human immunoglobulin G1 (IgG1) Fc domain to bind the neonatal Fc receptor (FcRn) and engage the IgG salvage pathway, thus conferring prolonged circulating half-life and the resultant dosing convenience. To avoid unintended (target independent) CD3-mediated T-cell activation via interaction with Fc gamma receptors (FcγR), the Fc domain was mutated to greatly reduce or eliminate binding to these receptors as well as complement. Unlike blinatumomab, which reacts only with human or chimpanzee's antigens (10), MGD011 cross-reacts with both CD19 and CD3 molecules in macaques, enabling preclinical evaluation in a relevant model. Here, we report on MGD011 engineering and characterization as well as on its preclinical safety assessment, pharmacokinetics (PK), and dose optimization in cynomolgus monkeys. The results show robust activity in multiple models and predict effective dosing in humans at once weekly or longer intervals.
Materials and Methods
DART protein engineering, production, and purification
MGD011, an Fc-bearing CD19 x CD3 DART protein, was constructed as described (11) using VL and VH sequences from humanized anti-CD19 mAb BU12 (12) and humanized anti-CD3 mAb XR32 (13). The IgG1-derived Fc segment was modified to encode the L234A/L235A mutation to greatly reduce or eliminate FcγR and C1q binding (14). Control molecules in which the variable domain sequences of an anti-fluorescein mAb 4-4-20 (15) replaced either of the DART protein arms (Fluo x CD3 or CD19 x Fluo) were engineered in a similar manner. The DART proteins were expressed transiently in CHO-S cells (8) and purified to greater than 99% purity using protein A and either size-exclusion chromatography (SEC) or other polishing steps. The purified DARTs have very low levels (less than 1%) of high molecular weight (HMW) protein present and have an apparent molecular weight of approximately 110 kDa (Supplementary Fig. S1A and B).
Other reagents, cell lines, and tissue samples
Recombinant soluble human and cynomolgus CD3ε/δ chimeric proteins, as well as human and cynomolgus CD19 and CD19-His proteins, were expressed in CHO-S cells. MOLM-13 and JIMT-1 were obtained from DSMZ (Braunschweig, Germany); Jeko-1 cells were from the ATCC and HBL-2 from the National Institutes of Health (Bethesda, MD; ref. 16). Raji/GF, a CD19+ Burkitt's B-cell lymphoma expressing luciferase and green fluorescent protein by stable transfection, was established at MacroGenics. All cell lines were passaged for less than 3 months after thawing; all lines were confirmed to be free of mycoplasma by PCR (Taconic, 2013) and were authenticated on the basis of morphology, growth characteristics and CD19 expression. Heparinized human whole blood was from Biological Specialty Corporation. Cryopreserved purified primary CLL patient samples were from AllCells, LLC. Heparinized whole blood from cynomolgus monkeys was from Worldwide Primates, Inc.
MGD011 binding to human or cynomolgus monkey CD3 or CD19 proteins was analyzed by ELISA or surface plasmon resonance (SPR) as previously described (13); binding to primary human or cynomolgus monkey CD20+, CD4+ or CD8+ cells was analyzed by flow cytometry.
Cell killing assay
For CTL assays, DART protein mediated killing of target cell lines in the presence of human or cynomolgus PBMCs or purified T cells was determined by lactate dehydrogenase (LDH) release or luminescence assays as previously described (17).
B-cell depletion assay
For autologous B-cell depletion assays, PBMCs from normal human donors, cryopreserved primary CLL specimens, or cynomolgus monkey were incubated with DART proteins overnight and analyzed by flow cytometry using FSC/SSC gated lymphoid populations to determine B-cell depletion via CD20 staining.
Cell activation studies
T-cell activation was determined by flow cytometry after staining with CD8-FITC, CD4-APC, CD25-PE, and CD69-PE-Cy5 antibodies (BD Biosciences). For intracellular granzyme B and perforin determinations, T cells were stained with anti-CD4-PerCP.Cy5.5 and anti–CD8-APC antibodies (BD Biosciences), fixed, permeabilized (Cytofix/Cytoperm solution; BD Biosciences), and incubated with anti-granzyme B-FITC or anti-perforin-PE antibody (BD BioSciences). T-cell proliferation was determined by flow cytometry as dilution of carboxyfluorescein diacetate, succinimidyl ester (CFSE; Invitrogen) in labeled cells. IL-2, IFN-γ, and TNF-α levels in culture supernatants were quantitated by ELISA (R&D Systems).
In vivo studies in murine models
All studies were reviewed and approved by MacroGenics' Institutional Animal Care and Use Committee (IACUC). MGD011 in vivo activity was evaluated in an established HBL-2 lymphoma model and an established, disseminated Raji/GF leukemia/lymphoma model in human PBMC-reconstituted NSG beta 2 microglobulin (B2m)-deficient (NSG B2m−/−) female mice aged 7 to 8 weeks (Jackson Laboratory). Tumor volume in the HBL-2 lymphoma model was calculated as follows: (length x width2)/2. Tumor burden in the Raji/GF leukemia/lymphoma model was evaluated by whole-body imaging on an IVIS Spectrum imaging system (PerkinElmer).
Cynomolgus monkey studies
The studies were performed at Charles River Laboratories (Reno, NV) under IACUC guidelines. Naive cynomolgus monkeys (Macaca fascicularis) of Chinese origin (2.5–6.3-years-old, 2.3–3.9 kg body weight) were randomized by weight and sex and received vehicle or MGD011 via 2-hour intravenous (IV) infusion (Supplementary Table S1). Serum samples were collected to examine the PK profile of MGD011 by ELISA using immobilized goat anti-hXR32 antibody (recognizes CD3 arm of MGD011) for capture and goat anti-human IgG Fc-biotin together with streptavidin-horseradish peroxidase (SA-HRP) for detection. Whole blood samples were collected for flow cytometry and the peripheral blood cell surface phenotype analyzed by using the following antibodies: CD4-V450, CD8-FITC, CD20-V500-C, CD45-PerCP, CD16-PE, CD159a-APC, CD14-APC-H7, CD69-FITC, PD-1-PE, TIM3-APC, CD4-APC-H7, CD8-V500, CD3-Pacific Blue, and Ki-67 Alexa647 (BD Biosciences). Absolute numbers of cells were determined by Trucount™ (BD Biosciences). Additional serum samples were collected to examine cytokine levels measured by the MILLIPLEX® MAP Non-Human Primate Cytokine Magnetic Bead Panel 7-Plex (EMD Millipore). Formalin-fixed paraffin-embedded sections of bone marrow, lymph node and spleen were evaluated by immunohistochemistry (IHC) for CD20. In brief, tissues were deparaffinized, rehydrated, antigen retrieved and incubated with a mouse anti-CD20 antibody (Dako), followed by the Envision+ System-HRP Labeled Polymer Anti-Mouse antibody (Dako) and developed with Diaminobenzadine (DAB).
Flow cytometric analyses were carried out on a FACS Calibur flow cytometer (4-color analyses) or BD LSRFortessa cell analyzer (8-color analyses) equipped with CellQuest Pro Version 5.2.1 (BD Biosciences) and FlowJo v9.3.3 (Treestar).
In vitro assays were repeated at least three times. Nonlinear regression analyses were used to fit curves using GraphPad Prism. For in vivo studies, intergroup differences were assessed by two-way ANOVA with a Bonferroni correction and survival curve compared by the log-rank test. All analyses were performed using GraphPad Prism software (version 5.02). P values of ≤0.05 were considered statistically significant.
Engineering, physicochemical characterization, and binding properties of MGD011
MGD011 is an Fc-bearing DART protein composed of three polypeptide chains covalently linked by disulfide bonds (Fig. 1A). The Fc domain was engineered to greatly reduce or eliminate FcγR and complement C1q binding (14), while retaining binding to FcRn to prolong MGD011 circulating half-life. MGD011 binds human and cynomolgus monkey CD3 with nearly identical affinity (KD = 21.2 and 21.9 nmol/L, respectively), while binding to CD19 is 10-fold lower in monkeys compared with humans (KD = 20.3 and 2.0 nmol/L, respectively) due to a faster dissociation rate (Fig. 1B; Supplementary Fig. S1C). Importantly, MGD011 binds to both antigens simultaneously, as shown by bispecific ELISA and SPR analyses that employ human CD3 protein for capture and human CD19 protein for detection (Fig. 1C and D); moreover, it recognizes the native antigens on both human and cynomolgus monkey B and T lymphocytes (Supplementary Fig. S2).
MGD011 redirects T-lymphocytes to kill CD19+ B lymphocytes
MGD011 activity was first evaluated by assessing its ability to mediate redirected cytolysis by using purified human primary T lymphocytes as effector cells and several CD19+ human B-lymphoma lines as target cells. Although MGD011 mediated no killing activity against the CD19− cell line, MOLM-13 (Fig. 1E), potent killing was observed against all CD19+ lines tested, including Raji/GF (Fig. 1F and G) as well as HBL-2 and Jeko-1 cells (Supplementary Fig. S3A–S3B). A control DART protein, in which the CD19 arm was replaced by one with an irrelevant specificity, did not mediate killing. Furthermore, compared to blinatumomab, MGD011 demonstrated enhanced potency, with EC50 values approximately 10-fold lower for killing of Raji/GF cells (0.02–0.03 ng/mL for MGD011 vs. 0.38–0.41 ng/mL for blinatumomab). MGD011 also induced B-cell depletion in human PBMCs from normal donors, with endogenous T cells serving as effector cells, demonstrating an approximately 10-fold greater potency than blinatumomab in this assay as well (Fig. 1H).
Incubation of MGD011 with CD19+ target cells (Raji/GF cells) was associated with concomitant CD4+ and CD8+ T-cell activation as evidenced by upregulation of CD69, but no activation occurred with CD19-negative cells or the control DART protein (Fig. 2A). CFSE dilution showed induction of T-cell proliferation by MGD011, but not with the control DART protein (Fig. 2B). A dose-dependent upregulation of granzyme B and perforin levels in both CD8+ and CD4+ T cells was observed following treatment with MGD011 (Fig. 2C). The upregulation of granzyme B and perforin was higher in CD8+ T cells compared with CD4+ T cells, consistent with the expected higher CTL potency of CD8+ T cells. Incubation with MGD011 in the presence of CD19+ target cells also resulted in the release of cytokines, exemplified in Fig. 2D by the release of IL-2 and IFN-γ. No T-cell activation, proliferation, or cytokine release was observed in the absence of co-engagement of CD19+ target cells, attesting to the strict dependence on CD19 expression for MGD011-mediated T-lymphocyte activation.
To further evaluate the cytolytic activity of MGD011 against primary malignant B cells, MGD011 was incubated with PBMCs from patients with CLL, thus relying on the endogenous residual T lymphocytes as effectors. Malignant B cells (CD20+/CD5+) were depleted in the presence of MGD011, but not control DART protein, in a time-dependent manner (Fig. 3A; Supplementary Fig. S4A), accompanied by concomitant expansion (Fig. 3B; Supplementary Fig. S4B) and activation (CD25 induction) of both CD4+ and CD8+ T cells (Fig. 3C and D; Supplementary Fig. S4C—S4D). Thus, MGD011 mediates potent tumor cell killing not only against model cell lines but also primary leukemia samples.
Antitumor activity of MGD011 in human PBMC-reconstituted, xenograft tumor-bearing mice
Redirected T-cell killing by DART molecules can be recapitulated in mouse models bearing human tumor xenografts as targets and human PBMCs as effector cells (9). To minimize graft-versus-host disease (GVHD) associated with human T-cell engraftment, mice deficient in the expression of MHC class I via knock-out of its B2m component were used. Lack of B2m expression, however, is associated with impaired FcRn expression; hence, serum half-life of MGD011 is short (∼4 hours) in these mice compared to that in wild-type mice (∼135 hours; data not shown). Frequent dosing (every 2–4 days) was therefore employed to achieve adequate exposure in this model.
Antitumor activity of MGD011 was first evaluated in human PBMC-reconstituted mice bearing established HBL-2 lymphoma cell tumors. MGD011 treatment (500 μg/kg IV every 3–4 days for 8 doses) resulted in regression of HBL-2 tumors, with no evidence of relapse up to Day 42, the last study day (>25 days after initiation of treatment; Fig. 4A). MGD011 activity was also evaluated in a disseminated Raji/GF leukemia/lymphoma model in human PBMC-reconstituted mice. As a prophylactic treatment (Supplementary Fig. S5), MGD011, administered at the same time of Raji cell intravenously inoculation, inhibited tumor dissemination and improved survival at all doses tested (0.16–500 μg/kg), with maximal activity at ≥20 μg/kg. A modest, albeit statistically significant effect of the CD3-targeted control DART molecule (500 μg/kg) was also observed, possibly due to micro-clustering of CD3 in vivo at such a high dose level. In the therapeutic paradigm, treatment was delayed 2 weeks after inoculation of Raji/GF cells to allow for establishment of disseminated lesions. Image data through day 46 and survival curves through study completion (day 48) are shown in Fig. 4B and C. Most vehicle-treated mice met the euthanasia endpoint by the end of the second week after Raji/GF cell inoculation and all were dead by the end of week 4. A slightly longer survival was observed with 100 μg/kg of control DART protein, albeit not significantly different from the vehicle-treated group in this experiment. In contrast, the tumor burden in the animals treated with MGD011 showed a statistically significant survival at doses ≥4 μg/kg, with near complete tumor regression at 100 μg/kg. In all experiments, treatment with MGD011 at any dose level was not associated with body weight loss (data not shown).
MGD011 demonstrates prolonged circulating half-life in cynomolgus monkeys
Two cynomolgus monkey studies were performed (Supplementary Table S1). Analysis of MGD011 serum concentration-time profiles across these studies (Fig. 5A and B) showed dose-proportional increases in maximum serum concentration (Cmax) across the entire dose range evaluated, indicating linear PK. Clearance was lower than the glomerular filtration rate for cynomolgus monkeys (∼125 mL/h/kg, Supplementary Table S2), as expected for a protein of this molecular size (∼110 kDa), indicating that virtually no elimination occurs by renal filtration. Mean initial volumes of distribution (V1) are similar to or slightly higher than the plasma volume in cynomolgus monkeys (∼45 mL/kg), suggesting little or no depletion of MGD011 by binding to target cells. MGD011 demonstrated a prolonged beta-phase half-life (t1/2,β) of 161.4 ± 61.3 hours (6.7 ± 2.6 days) and mean residence time (MRT) of 190.6 ±74.0 hours (7.9 ± 3.1 days), consistent with that of an IgG Fc-bearing molecule in this species. Notwithstanding the foreign nature of the molecule in this species, only 3/40 MGD011-treated animals in the GLP study (one in the 2 μg/kg dose group and two in the 5 μg/kg dose group) showed aberrant PK profiles after the first or subsequent infusions and were confirmed positive for anti-drug antibodies (ADA, data not shown).
Sustained B-cell depletion in cynomolgus monkeys treated with weekly doses of MGD011
MGD011 was confirmed to be active with cynomolgus monkey cells as it mediated effective autologous B-cell depletion in monkey PBMCs in vitro (Supplementary Fig. S6A); it was, however, less potent than with human PBMCs (EC50 values of 0.016-0.69 ng/mL, n = 6 for human PBMCs vs. 2.44-93.38 ng/mL, n = 5 for monkey PBMCs). The CD3 arm of the DART is unlikely to contribute to this disparity, since its affinity for CD3 of both species is nearly identical (Fig. 1B); furthermore, MGD011 mediated comparable cytotoxicity against human CD19+ target cells with human or monkey PBMCs as effectors (Supplementary Fig. S6B and S6C). The differential potency in autologous B-cell depletion between species is likely contributed by a combination of factors, including an approximately 10-fold lower affinity of MGD011 for cynomolgus monkey CD19 compared with human CD19 (Fig. 1B), a lower density of CD19 on cynomolgus monkey B cells (Supplementary Fig. S2A and S2B), and a lower average T-cell-to-B-cell (T:B) ratio in PBMCs of cynomolgus monkeys (T:B ratio = 4:1, n = 5) compared with humans (T:B ratio = 7:1, n = 6). These differences make pharmacodynamic modeling in cynomolgus monkeys a conservative estimate of MGD011 potency and should be taken into consideration when projecting human doses.
Administration of MGD011 resulted in a decrease in circulating CD20+ B cells by 24 hours following the start of the first infusion (Fig. 5C and D). Although a partial reduction in circulating B cells was noted at the 0.2 μg/kg dose, nearly complete depletion was observed at dose levels ≥0.5 μg/kg and were maintained throughout the 4 weekly treatments. Upon completion of dosing, circulating B cells returned to predose levels, albeit with a dose-proportional lag (Fig. 5C). Histopathology examination at terminal necropsy showed a dose-dependent decrease in follicles and germinal centers within the spleen, lymph nodes (inguinal, mandibular, and mesenteric), and gut-associated lymphoid tissue (GALT; data not shown). CD20 immunohistochemistry (IHC) showed a reduction in CD20+ B cells to nearly undetectable levels in the lymph nodes, spleen, and bone marrow at doses ≥10 μg/kg (Fig. 5E and F). The pharmacologic effect of MGD011 on B lymphocytes was prolonged, but ultimately reversible, as indicated by the return of circulating B cells to pretreatment levels and the absence of any abnormal histological and IHC findings at the end of a 12-week recovery period (Fig. 5E). In summary, once weekly administration of MGD011 was sufficient to induce complete and durable B-cell depletion in both the circulation and lymphoid organs.
Treatment of MGD011 was associated with transient, dose-dependent fluctuations of both circulating CD4+ and CD8+ T cells, with the lowest nadir following the first infusion and with a decreasing magnitude following subsequent infusions (Fig. 6A and B). Contrary to the durable reduction in B cells, T lymphocytes recovered quickly after each infusion and reached or exceeded baseline levels before the next dose. Similar T-cell kinetics were observed in cynomolgus monkeys after the administration of MGD006, a CD123 x CD3 DART protein (13), and in humans during blinatumomab infusion (18). Fluctuations were also observed with natural killer cells and monocytes in MGD011-treated monkeys (data not shown), likely representing transient cell margination.
MGD011 treatment was also associated with upregulation of the T-cell activation markers, CD69 and PD-1 (Fig. 6C–F), together with an increase of the Ki67+ proliferating fraction of both CD4+ and CD8+ T-cell subsets during the treatment phase (Fig. 6G and H). Therefore, MGD011 is capable of mediating T-cell activation and expansion in vivo. The activation markers trended toward or returned to baseline by the end of the experiment; however, unexplained isolated increases in one group 4 animal (showing up to 10-fold higher frequency of CD4+CD69+ cells than the remaining 3 animals) and in two vehicle control animals (up to 10-fold higher frequency of CD8+PD1+ cells than the other 2 animals) during the recovery phase contributed to swelling the average for these groups (Fig. 6C and F). An asymptomatic, transient, dose-dependent increase in circulating cytokine levels, a first-dose effect, was observed for IFN-γ (Fig. 6I), IL-6 (Fig. 6J), and IL-10 (Fig. 6K) and, to a lesser extent, TNF-α levels (data not shown), all occurring 2 hours following the end of the first infusion and returning to at or near baseline within 24 hours of the start of the infusion. Smaller levels of cytokine increase were noted with subsequent infusions, which were of a magnitude comparable with those observed following vehicle infusions, except IL-10, which exceeded levels of the vehicle control group in some instances. No MGD011-related changes in circulating IL-2, IL-4, or IL-5 levels were noted.
MGD011 was well tolerated in cynomolgus monkeys at all dose levels tested, with no compound-related toxicological effects, including cage-side observations, changes in body weight or food consumption, vital parameters, ophthalmology, electrocardiograms, body temperature, blood pressure, heart rate, respiration rate, neurological examinations, coagulation, or clinical chemistry (data not shown); furthermore, with the exception of the aforementioned hematological, bone marrow, and lymphoid organ changes, no other gross or microscopic pathological findings were noted (data not shown).
MGD011 is a bispecific DART protein developed for the treatment of B-cell malignancies and designed to redirect the T lymphocytes via their CD3 molecule to kill target cells expressing CD19. MGD011 was shown to eliminate normal or pathogenic CD19+ cells in vitro and in vivo. Furthermore, weekly administration of MGD011 to cynomolgus monkeys resulted in complete and durable depletion of B cells in the circulation and in lymphoid tissues. MGD011 was well tolerated in toxicology studies under treatment regimens that exaggerated potential clinical setting scenarios in patients. Thanks to its favorable PK properties, convenient once-a-week dosing or longer interval is predicted to be clinically feasible in humans.
The structure of MGD011 represents a refinement of a previously reported format (8), where the DART chains (Chain 1 and Chain 2) preserve the original C-terminal stabilizing disulfide linkage, while opposite E/K-coil sequences were added that further improve heterodimer formation. To promote the desired heterodimerization of Chain 1 and the Fc-domain Chain 3, knob-into-hole mutations were incorporated in the CH3 region. To facilitate the removal of residual Chain 3 dimers during purification, a H435R mutation was included in this chain to abolish protein A binding. MGD011 was engineered with humanized antibody arms displaying 10-fold greater affinity for human CD19 than for human CD3, thus providing for preferential initial binding to target cells, while minimizing CD3 engagement in the absence of target cells. MGD011 exhibits binding to monkey CD3 and CD19, allowing for preclinical modeling in this species. Specifically, MGD011 cross-reacts with cynomolgus monkey CD3 with nearly identical affinity as for human CD3 and with a 10-fold lower affinity for monkey CD19. Consistent with its binding properties, MGD011 mediated similar levels of cytotoxicity against human CD19+ Raji/GF target cells by either human or cynomolgus monkey PBMCs. MGD011 mediated also a dose-dependent, autologous ex vivo B-cell depletion in both human and cynomolgus monkey PBMCs, albeit with a decreased potency in the latter, due, at least in part, to its lower affinity for monkey CD19. Because of the lower CD19 affinity and reduced potency in cynomolgus monkeys, this species may offer a conservative estimate of potency that needs to be taken into consideration in extrapolating non-human primate data to potential clinical settings.
To confer prolonged circulating half-life, MGD011 was also engineered with an Fc-domain; with an estimated terminal half-life of approximately 161 hours (6.7 days) in monkeys and a predicted half-life of 343-495 hours (14.3–20.6 days) in humans, which is similar to that of conventional mAbs, delivering MGD011 as an intermittent dosing regimen appears feasible. In comparison, blinatumomab, which is engineered as a single-chain Fv pair, is administered by continuous intravenous infusion for repeated 4-week courses, owing to its short approximately 2-hour half-life.
MGD011 administered to cynomolgus monkeys by a 2-hour intravenous infusion on a weekly schedule at doses ranging from 0.2 to 100 μg/kg was safe and well tolerated. Dose-dependent, on-target activity with nearly complete depletion of circulating CD20+ B cells was observed at doses ≥0.5 μg/kg. Histologic evaluation confirmed a dose-dependent decrease in B-cell zones in lymphoid tissues at the terminal necropsy. Although bone marrow showed no obvious histological changes, B-cell depletion was observed by IHC, further attesting to the specificity of the on-target effects of MGD011. The pharmacologic effect of MGD011 on B lymphocytes was reversible, as indicated by flow cytometry and IHC analysis in addition to hematology and histopathology (data not shown). Each of these parameters returned to baseline, with no differentiation between the animals receiving MGD011 or control article at the end of the 12-week recovery period.
Maximal cytolytic activity against B-cell lines and in patient samples ex vivo were observed at MGD011 concentrations of 50 to 100 ng/mL. These levels are at or below the Cmax observed in monkeys at ≥5 μg/kg or below trough levels at ≥50 μg/kg. Either dose level resulted in near complete depletion of B lymphocytes in the circulation and lymphoid tissues in monkeys, attesting to the potency of the DART molecule, particularly in account it decreased potency in this species compared with humans. In B-cell malignancies, a low effector-to-target (E:T) cell ratio could limit the effectiveness of redirected cytolysis. MGD011, however, was able to eliminate leukemic cells in CLL patient samples by engaging and expanding the patient's residual T cells in vitro from a low E:T cell ratio to one exceeding the target cell population. These data also confirm that CLL T cells can be redirected to become cytolytic against leukemic cells in the presence of MGD011, as previously reported for blinatumomab (19), even though T-cell dysfunction in CLL, such as defects in immune synapse formation, co-stimulatory/accessory molecule expression, and cytokine release have been reported (20–22).
A safety concern associated with CD3-targeting therapies is cytokine release. The monovalent nature of the binding arm of MGD011 ensures that T-cell activation and cytokine release depend exclusively on target-cell engagement. Consistent with its desired design, no T-cell activation or cytokine release were observed in vitro in the absence of CD19+ target cells or with a control DART protein that includes only the CD3-targeting arm. No cytokine storm was observed in MGD011-treated cynomolgus monkeys. Transient dose-dependent increases in IFN-γ, IL-6, and IL-10 were observed following the first dose; with subsequent doses, they were generally no greater than those observed in animals receiving vehicle. There were no MGD011-related changes in circulating IL-2, IL-4, or IL-5 levels and only small, transient increases in TNF-α, exceeding 100 pg/mL in only a few instances, were observed in animals that received 10 μg/kg MGD011. First-dose cytokine release events have been previously observed in patients treated with blinatumomab (18) or with our CD123 x CD3 DART molecule in monkeys (13); the rapid target-cell depletion following the first administration of the bispecific molecule may limit further T-cell engagement and explain the transient nature of cytokine release observed.
Tumor lysis syndrome, which has been observed with blinatumomab (7), should be anticipated as another potential safety concern for MGD011 and proper preventive measures be implement to limit its consequences. Blinatumomab treatment was also associated with an increased risk of serious and/or severe neurological toxicities (7); a similar safety risk of neurological toxicities for MGD011 is therefore recognized. It should be noted, however, that no signs of neurological toxicity were observed in the MGD011 toxicology studies, although the predictive value of the cynomolgus monkey with respect to neurotoxicity of CD19 x CD3 bispecific interventions is unknown. Similarly, no infection-related adverse events were observed in monkeys treated with MGD011, although 25% of patients receiving blinatumomab showed signs of infection (7). Hence, patients should be monitored for infections and treated appropriately.
Immunogenicty remains a potential concern associated with antibody therapy. MGD011 limits this through the incorporation of humanized Fv regions and the minimal use of linkers. Although immunogenicity in monkeys does not predict immunogenicity in humans, notably, only 3 of 40 monkeys treated with MGD011 showed an ADA response; given that MGD011 targets B lymphocytes, ADA development is expected to be low in all species.
Upregulation of activation markers, including PD-1, was observed amongst circulating T lymphocytes in monkeys treated with MGD011. Importantly, the frequency of regulatory T cells or TIM-3+ cells in both the CD4+ and CD8+ T-lymphocyte populations remained low (<1%) throughout the entire dosing period (data not shown), suggesting that T-cell activation induced by MGD011 was not accompanied by expansion of regulatory T cells or T-cell exhaustion.
The off-the-shelf convenience of bispecific antibodies offers substantial advantages compared with personalized CAR T cells. A point of contention, however, is whether a trade-off of bispecific molecules is a decrease in potency. The impressive response to CAR T-cell therapy is associated with a significant rate of severe adverse events, resulting in a narrow therapeutic window, with little, if any, opportunity for response fine tuning. We should further note that MGD011, on a molar basis, is more potent than blinatumomab and that complete depletion of B lymphocytes in both marrow and lymphoid tissues can be achieved in cynomolgus monkeys in the absence of adverse effects. Ultimately, only clinical data will be able to address this issue.
In summary, MGD011 demonstrated potent activity in redirecting T lymphocytes to eliminate CD19-expressing cells in vitro, induced growth inhibition and tumor regression of B-cell lymphoma/leukemia models in mice, showed prolonged circulating half-life and was safe and well tolerated in cynomolgus monkeys at doses that resulted in complete and durable B-cell depletion in the circulation, bone marrow, and lymphoid organs. The data from these nonclinical studies provided rationale for the clinical development of MGD011 as a treatment for patients with B-cell malignancies; a phase I dose-escalation study of the safety, tolerability, dose-limiting toxicity, maximal tolerated does, and recommended phase II dose of MGD011 when administrated intravenously over a 2-hour period once every 2 weeks to subjects with B-cell malignancies is currently recruiting patients (NCT02454270).
Disclosure of Potential Conflicts of Interest
All authors are or were employees of MacroGenics, Inc. and received compensation from and hold ownership interest (including patents) in MacroGenics, Inc.
Conception and design: L. Liu, C.-Y.K. Lam, R. Alderson, J.L. Nordstrom, S. Koenig, P.A. Moore, S. Johnson, E. Bonvini
Development of methodology: L. Liu, C.-Y.K. Lam, L. Widjaja, H. Li, R. Alderson
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): L. Liu, C.-Y.K. Lam, V. Long, L. Widjaja, Y. Yang, H. Li, L. Jin, J. Brown, R. Alderson
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): L. Liu, C.-Y.K. Lam, L. Widjaja, H. Li, R. Alderson, J.L. Nordstrom, P.A. Moore, S. Johnson, E. Bonvini
Writing, review, and/or revision of the manuscript: L. Liu, C.-Y.K. Lam, M.D. Lewis, J.L. Nordstrom, S. Koenig, P.A. Moore, S. Johnson, E. Bonvini
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): L. Widjaja, S. Burke, S. Gorlatov
Study supervision: L. Liu, P.A. Moore, E. Bonvini
The research funding was provided by MacroGenics.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We would like to thank Cynthia Sung (Rainbow Pharma Consulting) for data analysis, Robert Burns, Qin Tang, Nancy O'Gwin, and Xioqi Gong for technical assistance, Melinda Hanson for administrative and editorial support, Timothy Mayer and James Karrels for critical discussions and review of the manuscript. We thank the teams at the Charles River Laboratories, Inc. for the diligent conduct of the cynomolgus monkey studies.
Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/).
- Received March 11, 2016.
- Revision received August 8, 2016.
- Accepted September 1, 2016.
- ©2016 American Association for Cancer Research.