Background Despite significant progress in the development of T cell-engaging therapies for various B-cell malignancies, a high medical need remains for the refractory disease setting, often characterized by suboptimal target levels.
Methods To address this issue, we have developed a 65-kDa multispecific antibody construct, CLN-978, with affinities tuned to optimize the killing of low-CD19 expressing tumor cells. CLN-978 bound to CD19 on B cells with picomolar affinity, and to CD3ε on T cells with nanomolar affinity. A serum albumin binding domain was incorporated to extend serum half-life. In this setting, we biophysically characterize and report the activities of CLN-978 in cell co-culture assays, multiple mouse models and non-human primates.
Results Human T cells redirected by CLN-978 could eliminate target cells expressing less than 300 copies of CD19 on their surface. The half-life extension and high affinity for CD19 led to significant antitumor activity in murine lymphoma models at very low doses of CLN-978. In primates, we observed a long serum half-life, deep and sustained depletion of normal B cells, and remarkable tolerability, in particular, reduced cytokine release when CLN-978 was administered subcutaneously.
Conclusions CLN-978 warrants further exploration. An ongoing clinical phase 1 trial is investigating safety, pharmacokinetics, pharmacodynamics, and the initial therapeutic potential of subcutaneously administered CLN-978 in patients with non-Hodgkin’s lymphoma.
- CD4-Positive T-Lymphocytes
- CD8-Positive T-Lymphocytes
- Cytotoxicity, Immunologic
- Drug Evaluation, Preclinical
Data availability statement
No data are available.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
- CD4-Positive T-Lymphocytes
- CD8-Positive T-Lymphocytes
- Cytotoxicity, Immunologic
- Drug Evaluation, Preclinical
WHAT IS ALREADY KNOWN ON THIS TOPIC
CD19 T-cell engagers (TCEs) show clinical activity in B-cell malignancies, yet complete response (CR) rates and duration of response are not as favorable as for CD19 CAR-T cell therapies.
WHAT THIS STUDY ADDS
CLN-978 was developed to address limitations of CD19 TCEs to offer the potential for CAR-T-like CR rates, but off-the-shelf therapy.
CLN-978 redirects T cells to lyse B-cell lymphomas expressing very low levels of CD19.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
CLN-978 has the potential to treat patients having relapsed from other T cell-based therapies due to reduced CD19 target expression.
T-cell engagers (TCEs) have shown robust single-agent activity in the treatment of relapsed or refractory (r/r) B-lineage acute lymphocytic leukemia (ALL),1 non-Hodgkin’s lymphoma (NHL)2 3 and multiple myeloma.4 5 The first bispecific T cell-engaging antibody construct to reach the market was blinatumomab (Blincyto) for the treatment of Philadelphia chromosome-negative r/r ALL in adult patients.6 7 Blinatumomab targets CD19 on ALL cells and the CD3ε subunit of the T-cell receptor and thereby mediates the redirected lysis of ALL cells by endogenous T cells. Pivotal trials showed a 40% complete response (CR) rate in patients with adult ALL with overt disease,8–10 and an 80% CR rate in patients with minimal residual disease.11 12 An alternative strategy to redirect lysis of CD19+ cells is by engineering a patient’s autologous T cells with a chimeric antigen receptor specific for CD19 (CAR-T cells). While more logistically intensive to produce, CD19 CAR-T cell therapies such as tisagenlecleucel have shown superior CR rates in patients with pediatric ALL in the range of 70–90%.13 14 Likewise, the duration of response is more favorable for CAR-T cell therapies. Another limitation of blinatumomab includes its administration by continuous intravenous infusion via a port as required due to a very short serum half-life. Frequent bag changes are inconvenient for the patient and use of ports incurs risk of infection. Blinatumomab is also currently limited to use in ALL, with challenges in expanding to NHL due to the need for a much higher dose in NHL.15
Relapses due to the downregulation or apparent loss of CD19 have been observed in both patients treated with CD19 CAR-T cells16 and with blinatumomab,17 18 where they account for ~30% of treatment failures17 and are presumably selected in response to CD19-directed T-cell pressure. While CD3-based bispecific antibodies and CAR-T therapies have been shown preclinically to mediate a bystander effect even against local antigen-negative tumor cells,19 20 the failure of blinatumomab in the clinical setting of CD19 loss suggests the dominance of CD19-null escape mutants, rendering the bystander effect insufficient to facilitate complete responses. Recent data suggests that blasts that appear to have lost CD19, as measured by immunohistochemistry (IHC), may still exhibit low levels of CD19 expression (ie, <3000 copies of CD19 per cell) when measured by more sensitive techniques like flow cytometry.21 One approach to deal with these challenges is the development of therapies targeting an alternative B-cell antigen, such as CD22, which is also expressed in ALL and NHL.22 We chose to take an alternative approach that capitalizes on the clinical success of CD19-targeted therapies.
We constructed CLN-978, a novel CD19/CD3/HSA-specific TCE that is designed to engage T cells to target malignant B cells, including those that may have decreased or non-detectable CD19 expression (by IHC). We attempted to uniquely achieve this by using an affinity-matured single-chain variable fragment (scFv) with picomolar affinity for CD19, which is driven by a slow off-rate, and with cross-reactivity to cynomolgus monkey for initial safety assessment. We further added to CLN-978 a human serum albumin (HSA)-binding domain for half-life extension, eliminating the need for continuous infusion. Here, we characterized CLN-978 in cell co-cultures of T cells and target cells, in multiple murine efficacy models and in cynomolgus monkeys for tolerability, pharmacodynamic and pharmacokinetic (PK) properties. Our data suggest that cytotoxic T cells can be redirected to lyse target cells expressing less than 300 copies of CD19, a level that would be considered negative for CD19 by standard detection procedures. Our preclinical results support further exploration of CLN-978 in clinical studies with B-cell malignancy patients for safety and preliminary efficacy in a first-in-human study.
Materials and methods
The individual components of CLN-978 were optimized using Adimab’s antibody engineering yeast platform. Full-length CLN-978 was assembled by fusing a humanized anti-serum albumin single-domain antibody (sdAb), a humanized anti-CD19 scFv and a humanized anti-CD3 epsilon scFv into a single polypeptide chain using glycine/serine linkers. CLN-978-HIS consists of the same sequence as CLN-978 with a 10x Histidine tag at the C-terminal end and was produced in HEK-293 cells solely for initial characterization studies.
Biacore evaluation of CLN-978 binding activities
To determine affinity for a CD3εδ dimer and albumin, each protein was immobilized on a CAPture chip sensor surface via biotin capture. A multicycle kinetic assessment of CLN-978 interactions was performed by injection of an appropriate concentration series of test samples. To determine affinity for CD19, a human CD19-Fc fusion protein was immobilized on a Series S CM5 sensor chip via anti-human Fc antibody immobilized on the surface. A single-cycle kinetic assessment of the protein-sample interaction was performed by injection of an appropriate concentration series of test samples. Each protein-sample interaction was analyzed in triplicate and a Langmuir 1:1 binding model was fit to the data. To determine the impact of albumin binding on CLN-978 affinity, CLN-978 was first precomplexed with HSA or left unbound, then passed over either chip-immobilized, recombinant CD3εδ and the binding measured throughout. A global fit algorithm (Biacore Insight Evaluation) was used to calculate KD, ka, and kd values.
Cell binding of CLN-978
Cryopreserved peripheral blood mononuclear cell (PBMC) from human, cynomolgus monkey or mouse donors were thawed prior to addition of live/dead viability dye, Fc block, and staining solution. Briefly 0.3–1×106 cells were stained with FVS700 for viability according to manufacturer’s recommendations and blocked with mouse or human Fc block antibody. Next, cells were incubated with CLN-978 conjugated to APC for 60 min at 37°C in the dark and stained with fluorescence-conjugated antibodies. Mouse and non-human primate (NHP) cells were diluted in Brilliant Stain Buffer and FBS Stain Buffer (BD Biosciences), whereas human cells were diluted in Brilliant Stain Buffer and phosphate-buffered saline (Thermo Fisher). After staining, cells were fixed with FluoroFix Buffer (BioLegend) and stored at 2–8°C in the dark until acquisition by flow cytometry. CD4+, CD8+ T cells and CD20+ B cells were identified and gated.
T cell-dependent cellular cytotoxicity and T-cell activation in vitro
In brief, PBMCs (stained with eFluor 450) and target cells (RAMOS, Raji, and A20-CD19) stained with eFluor 670 were assessed at a 10:1 E:T ratio with a target cell seeding density of ~3x 105 cells/well in the presence or absence of 0.15 mg/mL human serum albumin. An 8-point dilution of CLN-978 was performed starting at 2 nM with sixfold dilution steps. Cytotoxicity was assessed using a fixable viability dye (eFluor 780) by flow cytometry. T-cell activation was determined by flow cytometry (CD4, CD8, CD25 and CD69). Cytokine production (interferon (IFN)-γ and tumor necrosis factor (TNF)-α) was quantified using Luminex Multiplex Analysis where appropriate. CD19 expression was quantified using Quantibrite beads (BD Biosciences).
CHO-CD19 inducible cell line generation and in vitro studies
CHO-K1 cells were transduced with an induction cassette that expresses a chemical-induced proximity activation cassette that allows titratable expression of CD19 on treatment with the chemical inducer, ABA, enabling expression levels of CD19 on CHO cells ranging from 78 to 185,223 copies of CD19 per cell after 24, 48 or 96 hours. CD19 expression was quantified by staining with PE anti-CD19 and plotted on a curve generated using Quantibrite PE receptor quantification kit.
For T cell-dependent cellular cytotoxicity (TDCC) studies with inducible CHO CD19 cells, CHO-K1 ABA CD19 cells were seeded in xCELLigence RTCA plates and treated with a 7-point concentration range (2.4–10 µM) of ABA for 24 hours to induce CD19 expression. Isolated T cells and target cells were assessed at a 10:1 E:T ratio with two fixed CLN-978 concentrations (225 pM and 1770 pM), and four individual T cell donors. The comparison to blinatumomab used isolated T cells from 10 individual donors at an E:T ratio of 10:1, with 225 pM of either CLN-978 or blinatumomab. CD8+T-cell proliferation was measured via Ki-67 staining and flow cytometry analysis. IFN-γ production by the effector cells was measured via ELISA from supernatant.
In vivo efficacy studies
For the A20-CD19 study, hCD3ε-expressing BALB/c mice were inoculated with A20 cells expressing hCD19 which were produced by stable transfection, and individual clones selected by limiting dilution. Mice were treated once intravenously with CLN-978-HIS or blinatumomab when tumor volume reached ~100 mm3. For Raji B.luc studies, immunodeficient NCG mice were intravenously engrafted with 1×105 Raji B.luc cells. On the following day (Day 1), 2×107 PBMC from a normal human donor were implanted intraperitoneally (IP). Starting on Day 1, mice were treated intravenously weekly with CLN-978-HIS or blinatumomab.
For the comparison of intravenous and subcutaneous (SC) administration, mice that received PBMC were then placed into seven groups: PBS vehicle control (n=9) or PBMC+CLN-978 dosed at 0.3, 3 or 30 µg/kg (n=9) administered either intravenously or SC. CLN-978 was administered once weekly to the end of study in a dosing volume of 10 mL/kg, adjusted for body weight. On day 12, the number of Raji cells and B cells in peripheral blood was determined by flow cytometry.
Non-human primate pharmacokinetics and pharmacodynamics
Female cynomolgus monkeys (n=2/group) received a single intravenous or SC injection of 0.1 or 1 mg/kg CLN-978. Serum and whole blood were collected at defined time points throughout the 28 days. Pharmacokinetic analysis, cytokine sample analysis, and flow cytometric analysis were performed via standard methods. All work was performed under review of CR-MA Institutional Animal Care and Use Committee IACUC ASP #: 990301.
Design and production of CLN-978
CLN-978 is a recombinant fusion protein comprised of a single polypeptide chain containing a humanized anti-serum albumin sdAb, a humanized anti-CD19 scFv and a human anti-CD3 epsilon scFv (figure 1A). Each binding domain was optimized for affinity, selectivity, species cross-reactivity, thermal stability, and developability using Adimab’s yeast-based antibody engineering platform.23 CLN-978 had an apparent molecular size of 65 kDa and was produced from a clonal Chinese hamster ovary (CHO) cell line. The expressed protein was affinity-purified via protein A chromatography, followed by additional polishing steps to ensure high product purity.
Binding characteristics of CLN-978
Binding of CLN-978 to human CD19, CD3 and serum albumin was determined by surface plasmon resonance (SPR) (figure 1B). The equilibrium binding constants (KD) of CLN-978 for CD19, CD3, and albumin were 0.0738 nM, 7.997 nM, and 0.6 nM, respectively. Serum albumin binding across species was confirmed by SPR to ensure physiologically relevant half-life extension in all species used for in vivo evaluation (online supplemental table S1). The affinity of CLN-978 for cynomolgus serum albumin was very similar to human albumin (0.7 nM), while the affinity for mouse albumin was approximately sevenfold lower than for human albumin (4.4 nM). Given the binding affinity for albumin and typical serum concentrations of albumin in the range of 500 μM,24 it is assumed that all CLN-978 is bound by albumin in vivo across species. SPR analyses performed in the presence of albumin resulted in a twofold to threefold decrease in the affinity and kinetics of the interaction between CLN-978 and human CD3εδ and CD19, which was primarily driven by a reduction in the on-rates (online supplemental table S1).
The cross-reactivity of CLN-978 for CD19 and CD3 across species was evaluated by cell surface binding to CD20+ B cells and CD4+ and CD8+ T cells in human, NHP and murine PBMC across multiple donors using flow cytometry. Cell surface binding of CLN-978 to B and T cells correlated with the SPR binding. Very low half-maximal effective concentration (EC50) values were calculated for human CD20+ B cells, with a narrow range between 0.33 and 0.42 µg/mL. Binding to human T cells was more moderate. EC50 values were 7.93 µg/mL (6.90–8.67 µg/mL) for CD4+ T cells. Binding to CD8+ T cells was variable, with an EC50 value of 99.5 µg/mL for one donor, whereas two donor T cells were in the range at 13.55 and 19.52 µg/mL. With PBMC from NHPs, the EC50 value for CLN-978 binding to B cells was 1.72 µg/mL (1–2.59 µg/mL). In the T-cell subpopulations, similar EC50 values were observed for CD4+ (28.86 µg/mL) and CD8+ T cells (28.66 µg/mL). Minimal binding of CLN-978 was observed to mouse B and T cells with EC50 values in excess of 100 µg/mL for B cells and CD4+ and CD8+ T-cell populations (figure 1C), and with no saturation achieved. Overall, CLN-978 exhibited more potent binding to B cells than to T cells, consistent with a higher affinity for CD19 than for CD3. Cross-reactivity with NHP PBMC was also confirmed, with similar EC50 values for cell binding to CD8+ T cells, although ~fivefold and ~threefold lower for binding to human B cells and CD4+ T cells, respectively.
Activity of CLN-978 in co-cultures of human T cells with CD19-expressing target cells
We evaluated TDCC, T-cell activation, and the release of selected cytokines as pharmacodynamic markers. These studies were performed in co-culture experiments with unstimulated human PBMC as effector cells (E) and the human B lymphoma RAMOS cell line as target cells (T) at an E:T ratio of 10:1. RAMOS cells expressed an average of 13,108 CD19 molecules on the cell surface, which is within the range of normal B cells and many primary lymphoma cells.25 26 First, the ability of CLN-978 to redirect lysis and activate T cells was assessed across multiple donors in the absence of albumin. Figure 2A shows the variation of TDCC among six human PBMC donors. While EC50 values for lysis ranged from 0.26 to 5 pM CLN-978, all donors mediated ~80% lysis or greater at 48 hours at a concentration of 2 nM. The expression of the T-cell activation markers CD69 and CD25 on CD4+ and CD8+ T cells were simultaneously evaluated. Consistent with TDCC, EC50 values for T-cell activation were found to be in the single-digit pM range or lower (0.2–1 pM and 1.6–4.7 pM for CD69 and CD25, respectively), with the majority of CD8+ and CD4+ T cells (80–100%) expressing CD69 and CD25 at 2 nM CLN-978 (figure 2B). Only two out of six donor CD8+ T cells showed a <50% induction of CD25, and one donor for CD25 on CD4+ T cells. Collectively, these data demonstrate potent activation of the majority of T cells in the co-culture.
Next, the effect of albumin on CLN-978 bioactivity was evaluated. Figure 2C shows half-maximal lysis of RAMOS cells at 0.44 pM (±0.07 SEM; n=10) and 2.68 pM CLN-978 (±0.43 SEM: n=10) at 48 hours, in the absence and presence of HSA, respectively. As anticipated from the observed shift in binding affinity in the presence of HSA, we observed a minor shift in half-maximal CLN-978 concentration in the presence of HSA (<10-fold), with the EC50 remaining in the single digit pM range. Complete lysis of RAMOS cells was achieved both in the absence and presence of HSA at CLN-978 concentrations of 55 pM and higher, suggesting that the presence of HSA is unlikely to impact antitumor activity of CLN-978 under physiological conditions. Similar results were observed using the Raji cell line (online supplemental figure S1).
As a pharmacodynamic measure of activated cytotoxic T cells, we studied the release of TNF-α and IFN-γ in response to a CLN-978 titration in co-cultures of RAMOS human lymphoma cells with human PBMC in the presence of albumin (figure 2D). A dose-dependent release was observed for TNF-α and IFN-γ, although with EC50 values higher than those observed for TDCC and T-cell activation. The EC50 values were 23.3 pM (±9.74 SEM; n=2) for TNF-α release and 17.6 pM (±2.01 SEM; n=2) for IFN-γ. Maximal cytokine release was observed at 333 pM CLN-978 for both cytokines. In conclusion, CLN-978 is a very potent inducer of redirected lysis/TDCC, T-cell activation and cytokine release by previously unstimulated T cells.
Activity of CLN-978 in lymphoma cell lines with low CD19 expression
To determine the minimal copy number of CD19 on target cells necessary for redirected lysis by CLN-978, we used a chemically inducible system to tune the level of expression of a CD19 transgene in CHO cells. As an epithelial cell line of hamster origin, CHO cells do not express the B cell lineage-specific human CD19 protein. CHO cells transfected with the inducible transgene were incubated with a titration of the chemical inducer, ABA, ranging from 2.4 nM to 10 µM, enabling expression levels of CD19 on CHO cells ranging from 78 to 185,223 copies of CD19 per cell after 24 hours. At that time point, ABA was washed out of the media, and purified CD8+ T cells from four human donors were added at an E:T ratio of 10:1 along with CLN-978 at either 225 or 1770 pM for 68 hours. For all four T-cell donors, signals for TDCC, T-cell activation (CD25 expression), proliferation (Ki67), and release of IFN-γ were observed above the background of wild-type CHO cells (figure 3A–D, representative donor). Observed cytotoxicity was CD19-density dependent at 225 pM CLN-978, ranging from 38% lysis (78 CD19 copies) to 80% lysis (185,000 CD19 copies). At the higher CLN-978 concentration of 1770 pM, cytolysis was independent of CD19 levels. CLN-978 at 1770 pM mediated near-maximal lysis of induced CHO cells even at a level of 78 CD19 molecules per cell (figure 3A). CD25 and Ki67 expression and IFN-γ release increased with higher levels of CD19 expression at both 225 M and 1770 pM CLN-978 (figure 3B–D). In spite of this exquisite sensitivity to low levels of CD19, CLN-978 was also highly specific for CD19 expression, with no activity observed in the wild-type control cells lacking CD19 expression.
Similar assays were performed in this inducible system to compare the minimal CD19 expression level required for CLN-978 or blinatumomab to induce cytotoxicity and T-cell activation, as measured by IFN-γ production. As shown in figure 3E–F, CLN-978 demonstrated more robust efficacy at lower CD19 expression levels when compared with blinatumomab.
The findings with inducible human CD19-expressing CHO cells were corroborated with mouse A20 lymphoma cell lines stably transfected with human CD19. Three A20 clones were generated that expressed on their cell surface 17,000, 3,700 or 325 copies of human CD19. In co-culture TDCC assays, complete redirected lysis was observed with all three A20 cell clones in the presence of HSA (figure 4A). No lysis of non-transfected A20 cells was observed, in accordance with the absence of cross-reactivity of CLN-978 with mouse CD19. While EC50 values for TDCC were similar for the clones expressing >3000 copies of CD19 (in the range of 14–27 pM), the lowest expressing A20 clone with 325 copies of CD19 required 683 pM of CLN-978 for half-maximal lysis. However, no differences in maximum lysis were observed among the cell lines. These results are consistent with the inducible CHO study described above, where lysis was dependent on CD19 expression density at low, but not high, yet clinically achievable, concentrations of CLN-978. All three hCD19 expressing A20 clones activated CD8+ and CD4+ T cells, as measured by expression of CD25 and CD69, and induced the release of TNF-α and IFN-γ in a CD19-dependent and dose-dependent fashion (figure 4B–C).
In vivo evaluation of CLN-978 activity
The in vivo activity of CLN-978-HIS was compared head-to-head with blinatumomab in two different mouse models (figure 5A and B). One was a human CD19-expressing A20 lymphoma model implanted in the flank of BALB/c mice expressing a human CD3ε transgene; the other was a disseminated Raji B.luc luciferase NCG mouse model using implanted human PBMC as effector cells. In both models, CLN-978 outperformed blinatumomab. In therapeutic studies with established A20 flank tumors, a single intravenous dose of 0.1 mg/kg CLN-978 completely prevented tumor outgrowth, whereby an equivalent dose of blinatumomab was ineffective. In the disseminated Raji B.luc model, weekly intravenous dosing of 0.1 mg/kg prevented an increase in the luciferase signal, while blinatumomab could not prevent leukemic spread. This is potentially reflective of the extended PK of CLN-978 via albumin binding (online supplemental figure S2).
Comparison of antitumor activity of intravenous or SC administered CLN-978
Several TCEs have recently been evaluated for SC administration to minimize systemic toxicity and improve patient convenience.27 28 Because the formulation of CLN-978 allows for both intravenous and SC administration of CLN-978, we investigated antitumor efficacy by both routes of administration in immunodeficient NCG mice engrafted with Raji B.luc cells and human PBMC. Mice were treated with PBS as vehicle control or CLN-978 dosed at 0.3, 3 or 30 µg/kg administered either intravenously or SC. Tumor load was measured by the bioluminescent flux of mice from the luciferase transgene of Raji B.luc cells. In addition, at day 14, the number of Raji lymphoma cells and human B cells in peripheral blood was determined by flow cytometry.
Once weekly intravenous or SC dosing of CLN-978 at 3 and 30 µg/kg resulted in 10/10 complete responses after 28 days, as was evident from luciferase fluorescence signals equivalent to background (figure 5C, online supplemental figure S3). Treatment with 0.3 µg/kg had a minimal effect with SC administration and a modest effect with intravenous administration. The percentage of human B cells and Raji B.luc cells in the peripheral blood of mice on day 12 was determined by flow cytometry analysis. A highly significant and complete elimination of human B cells and lymphoma cells was observed in peripheral blood samples from mice in all CLN-978 treatment groups regardless of dose level or route of administration (online supplemental figure S4). These data suggest both intravenous and SC administration of CLN-978 are highly effective in the Raji lymphoma xenograft model.
Comparison of pharmacokinetics, pharmacodynamics and tolerability of intravenous-administered or SC-administered CLN-978 in non-human primates
We hypothesized that SC administration may reduce systemic toxicities commonly associated with intravenous-delivered TCEs due to a reduction in maximal concentration in serum (Cmax),25 and therefore directly compared CLN-978 at single doses of 0.1 and 1 mg/kg by intravenous or SC administration in cynomolgus monkeys (n=2). Serum CLN-978 dose levels increased in a dose-dependent manner for both routes of administration (figure 6A). Serum half-lives ranged from 5 to 7 days, suggesting that half-life extension will support infrequent dosing. Exposure and half-life were similar for both routes of administration. As expected, Cmax values were at least five times lower by the SC route than seen for the intravenous route.
A single dose of CLN-978 reduced the number of circulating B cells to background levels for the entire duration of the 28-day experiment with no apparent recovery of B-cell counts (figure 6B). T cells also transiently disappeared from the periphery 2 days post-dose but quickly recovered to predose levels by day 8 post infusion, consistent with the observations of other TCEs (figure 6C).29
As expected, administration of CLN-978 induced the transient release of various cytokines and chemokines into peripheral blood (figure 6D). Following intravenous infusion of CLN-978, blood levels of TNF-α, interleukin (IL)-2, IFN-γ, and IL-8 were highest at 2 hours post infusion and trended towards baseline by 6 hours, while levels of IL-6 and IL-10 peaked at 6 hours. Notably, the release of the majority of the cytokines was lower after SC delivery compared with intravenous infusion, and the kinetics of cytokine release were altered as well. Twenty-four hours post infusion, all cytokine levels were back to baseline irrespective of the route of delivery.
While CLN-978 was well tolerated by cynomolgus monkeys at the 0.1 mg/kg dose level by both intravenous and SC routes, the 1 mg/kg dose level was tolerated only after SC administration. In contrast, intravenous administration of 1 mg/kg CLN-978 caused one of two cynomolgus monkeys to experience symptoms of cytokine release, which required euthanasia on Day 1 of study. A dose of 1 mg/kg administered SC exposed the animals to serum concentrations of ~1–5 µg/mL (~15–75 nM) CLN-978 over a time period of 300 hours (12.5 days). This was >10-fold greater than the concentrations of CLN-978 needed for complete target cell lysis in vitro, even for low CD19 expressing cells.
In this study, we describe a novel CD19/CD3/HSA-specific T cell-engaging antibody construct for the treatment of diverse B-cell malignancies expressing the B-cell lineage marker CD19. Target expression levels on cancer cells typically show a Gaussian distribution. Depending on the potency of a cytotoxic TCE, it is likely that very low target-expressing cells cannot be effectively eliminated, and in this way get enriched by Darwinian selection for repopulation with target-dim cancer cells that can no longer be detected by IHC. Next-generation TCEs should therefore aim at binding a target antigen with as high affinity as possible to increase the number of bound TCE molecules on the surface of cancer cells as is critically required for formation of functional cytolytic synapses. With efficient cytotoxic T cell synapse formation on target-dim cancer cells, overall target cell elimination can become deeper, which may translate into better response rates and longer duration of responses.
In view of the profound clinical activity of blinatumomab, we opted for a tandem scFv format where anti-CD19 and anti-CD3 scFvs are similarly arranged as in blinatumomab, but selected scFvs derived from novel, optimized and humanized antibody fragments. A small single-domain anti-HSA antibody of 12 kDa recombinantly fused to the N-terminus was incorporated for serum half-life extension to avoid the requirement for continuous intravenous infusion.
In co-culture experiments of T cells with target cells, CLN-978 demonstrated TDCC and T-cell activation at single-digit picomolar concentrations. EC50 values for cytokine release were approximately 10-fold higher, in the double-digit picomolar range, which may positively contribute to the therapeutic window of CLN-978. Our data demonstrate that CLN-978 efficiently binds and eliminates low CD19 expressing lymphomas, with exquisite sensitivity to CD19 levels significantly lower than the threshold required for blinatumomab. At concentrations of CLN-978 that were safely tolerated by cynomolgus monkeys, CLN-978 could effectively mediate the lysis of lymphoma cells expressing <300 CD19 molecules per cell, levels that are typically below the limit of detection by IHC. These data suggest that CLN-978 may have the potential to deepen overall response rates and improve the sustained duration of response, and may be effective post-CD19 CAR-T therapy. The high potency of redirected lysis by CLN-978 is supported by results from multiple mouse models and cynomolgus monkey studies. A dose of 3 µg/kg of CLN-978 given by either intravenously or SC administration was sufficient to effectively prevent leukemia in the mouse models. In monkeys, a single intravenous or SC dose of 0.1 mg/kg led to depletion of normal B cells for the entire observation period of 28 days.
The inclusion of the HSA binding domain effectively improved pharmacokinetic properties and preserved CLN-978 functional activity. Our non-human primate PK studies showed a prolonged serum half-life and exposure to CLN-978 after intravenous or SC administration. The therapeutic benefit of half-life extension also likely accounts for the superior efficacy observed in preclinical tumor models comparing CLN-978 with blinatumomab. In vitro data suggests that albumin had only a small inhibitory impact on the various biological activities of CLN-978 in terms of EC50, with no impact on maximal activity. This phenomenon may be driven by the binding characteristics of CLN-978, where the slow off-rates of CLN-978 to CD19 were unaffected by the presence of albumin. The pronounced activity of CLN-978 in the presence of albumin is also confirmed by efficacy in murine models and complete B-cell depletion in cynomolgus monkeys, where CLN-978 is expected to be fully bound to serum albumin.
Our preclinical data also helped inform the preferred route of administration of CLN-978 in the clinic. Collectively, our data demonstrate that similar pharmacodynamics and efficacy was achieved with both intravenous-administered and SC-administered CLN-978 in mouse models and cynomolgus monkey studies. However, SC administration resulted in better tolerability in monkeys with reduced cytokine release, likely due to the lowered Cmax achieved with SC administration. These data demonstrate SC delivery of CLN-978 further widens the therapeutic index compared with intravenous delivery.
It is challenging to compare the in vivo activity of CLN-978 with that of blinatumomab given the limited cross reactivity and very short serum half-life of the latter. When compared side-by-side, the in vitro and in vivo potency of CLN-978 surpassed that of blinatumomab. Both the higher affinity binding of CLN-978 to both CD19 and to CD3 and its extended serum half-life may have contributed to its superior antitumor activity in mouse models. A major distinction from blinatumomab will also be that CLN-978 is being developed for the treatment of patients with NHL where blinatumomab would require a four times higher dose than is registered for patients with ALL. Given the promising potency data observed thus far, CLN-978 may have the potential to generate CAR-T like response rates as an off-the-shelf therapy.
One other CD19/CD3-bispecific TCE in clinical development is TNB-486.30 It is based on an IgG4 backbone using an sdAb arm for binding CD19 and an Fab arm for very low affinity binding of CD3. The KD value for CD19 was reported to be 1.8 nM. Preclinical data showed that lowering the affinity to CD3 greatly reduced both potency and the release of IL-2 compared with a positive control. EC50 values for TDCC, T-cell activation and IL-2 release by T cells were on the order of 0.1–1 nM, that is, 2 logs higher than for CLN-978. In NALM6 and SUDHL-10 flank tumor models, doses of 10 mg/kg were required for significant tumor control. In contrast to the CD3 affinity detuning strategy used by TNB-486, our cynomolgus monkey data suggests that cytokine release syndrome (CRS) may be appropriately managed by SC administration of CLN-978. In conclusion, CLN-978 is expected to have robust antitumor activity in the clinic in patients with B-cell malignancies.
Data availability statement
No data are available.
Patient consent for publication
We thank Antibody Analytics for their collaboration in developing the low CD19-expressing cells lines, and for the in vitro evaluation of cytotoxicity and T-cell activation of endogenous CD19-expressing and low CD19-expressing cell lines.
Contributors Experiments were planned by NKM, KM, JSM, BL, and PAB. Data analysis and interpretation was performed by KM and NKM. KM, NKM, JSM, and PAB wrote the manuscript.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests NKM, KM, BL, PAB, and JSM have ownership in Cullinan Oncology, which is seeking to commercialize CLN-978.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.