The PSMA-targeting Half-life Extended BiTE Therapy AMG 160 has Potent Antitumor Activity in Preclinical Models of Metastatic Castration-resistant Prostate Cancer

Prostate cancer remains a leading cause of cancer-related deaths, especially for patients who progress to metastatic castration-resistant prostate cancer (mCRPC) following androgen deprivation therapy (ADT; refs. 1, 2). mCRPC is characterized by molecular and clinical heterogeneity that may include alterations in androgen receptor (AR) pathway signaling and differentiation of tumor cells to a neuroendocrine-like state (3, 4). Typical mCRPC treatment includes novel hormonal therapy, such as enzalutamide and/or abiraterone, and taxane treatment. Additional therapies approved to treat mCRPC include radioligand therapy (radium-223), targeted therapy (rucaparib), and immunotherapy (sipuleucel-T). Although these therapies decrease levels of prostate serum antigen (PSA) and increase overall survival, their effects are generally not durable and may be associated with limited tolerability (5–8). Tumor heterogeneity and therapy resistance remain key challenges to achieving durable response and long-term survival.

Emerging therapies target prostate-specific membrane antigen (PSMA), a type II transmembrane protein that is expressed on the surface of nearly all prostate cancer cells. PSMA is highly upregulated in prostate cancer relative to normal tissues (100- to 1,000-fold), and its expression increases with progression to mCRPC (9–12). PSMA undergoes constitutive internalization that is increased by binding of anti-PSMA antibodies targeting the PSMA extracellular region (13). These characteristics make PSMA a compelling target for radioligand tracers that detect PSMA-positive tumor cells, such as gallium 68–labeled PSMA-11 (⁶⁸Ga-PSMA-11; refs. 14, 15). ⁶⁸Ga-PSMA-11 PET/CT is increasingly being used to identify PSMA-positive tumors (16, 17) due to high accuracy and reproducibility in labeling prostate cancer (18). Other radioligand molecules currently under evaluation as therapeutic agents, such as lutetium-177 [¹⁷⁷Lu]-PSMA-617 and actinium-225 [²²⁵Ac]-PSMA-617, have been shown to decrease PSA levels by >50% in up to 57% and 63% of patients, respectively, with objective responses observed in 82% (14/17) of patients treated with [¹⁷⁷Lu]-PSMA-617 and mean duration of response of 9 months for patients treated with [²²⁵Ac]-PSMA-617 as last-line therapy (19, 20). However, PSMA-targeted radioligand therapy has not enabled durable responses, and treatment is associated with xerostomia (21). New therapies that significantly improve overall survival with better tolerability are needed.

BiTE (bispecific T-cell engager) therapy is a targeted immunotherapy approach that engages patients' own T cells to kill tumor cells. BiTE molecule binding to a tumor-associated antigen on target cells and cluster of differentiation 3 (CD3) on T cells induces T-cell activation, cytokine production, and T-cell–mediated lysis of target cells (22). This mechanism can bypass the conventional pathway of T-cell receptor activation and may be active independent of the tumor's genetic background. These characteristics may enable BiTE therapy to be effective against tumors that are resistant to chemotherapy, radiotherapy, or immune checkpoint blockade (23, 24). We previously developed AMG 212 [MT112 (Micromet, Inc.)/BAY2010112 (Bayer AG)], a first-generation BiTE molecule targeting PSMA that has potent, specific activity in vitro and in mouse models of prostate cancer (25). In phase I studies, AMG 212 demonstrated early evidence of clinical activity, including a near-complete response in a patient who had received six prior therapies and had failed to respond to ¹⁷⁷Lu-PSMA-617 as his last therapy before beginning AMG 212 treatment (26). However, AMG 212 has a short half-life that requires administration by continuous intravenous infusion. We developed AMG 160, a PSMA-targeted BiTE molecule with an increased serum half-life, to enable administration at longer dosing intervals. Here, we describe the activity and safety of AMG 160 in preclinical models. These data support the ongoing evaluation of AMG 160 monotherapy and combinations in patients with mCRPC (NCT03792841).

LnCAP, 22Rv-1, and CHO and PC-3 cells were from the DSMZ-German Collection of Microorganisms and Cell Cultures. MDA-PCa-2b, VCAP, and C4-2B cells were from the ATCC. C4-2 cells were from ViroMed Laboratories. 22Rv-1, C4-2, C4-2B, LnCAP, VCAP, and PC-3 cells were maintained in RPMI media that contained 10% FBS, and CHO cells were grown in HyClone media with 10% FBS. MDA-PCa-2b cells were grown in F-12 media with 20% FBS, 1% insulin-transferrin-selenium solution, 10 ng/mL EGF, and 100 pg/mL hydrocortisone. Penicillin/streptomycin solution (1%) was also added to the media (Gibco). The cell lines were analyzed for authenticity upon receipt and tested negative for Mycoplasma contamination. The cell line research resource identifiers are listed in Supplementary Table S1. Human T cells were obtained from AllCells. Human peripheral blood mononuclear cells (PBMC) were isolated from buffy coats collected at the Institute of Clinical Transfusion Medicine and Immunogenetics (Ulm, Germany), and nonhuman primate (NHP) PBMC were obtained from R.C. Hartelust, B.V. The anti-programmed death-1 (anti-PD-1) antibody was generated at Amgen. Reagent sources and catalog numbers are listed in Supplementary Table S2.

AMG 160 activity in vitro was assessed in T-cell–dependent cellular cytotoxicity (TDCC) assays. Cell lines were incubated with AMG 160 in a concentration range and effector cells for 48 hours, and cell viability was determined by flow cytometry or a CellTiter-Glo Luminescent Cell Viability Assay (Promega). Activation of T cells in the cytotoxicity assays was evaluated by flow cytometry, using fluorochrome-conjugated antibodies directed against T-cell surface molecules, CD25, CD69, 4-1BB, and PD-1 (Beckton Dickinson). After incubation, cells were washed with FACS buffer (1 × PBS with 1% FBS) and T cells were incubated with the antibodies in a volume of 30 μL at 1:30 to 1:100 dilutions for 30 minutes at 4°C, protected from light. Cells were analyzed with a FACS Canto II or LSRFortessa flow cytometer, using FACSDiva Software (Becton Dickinson). Dose–response curves were analyzed with Prism (GraphPad).

Cytokine concentrations were analyzed from supernatants of cytotoxicity assays using a Cytometric Bead Array for Th1/Th2 cytokines (Becton Dickinson), including IL2, IL4, IL6, IL10, IFNγ, and TNFα. After incubation, samples were washed with PBS and analyzed by flow cytometry using a FACSCanto II or LSRFortessa (Becton Dickinson) flow cytometer and analyzed with FACSDiva and FCAP Array Software (Becton Dickinson).

All studies were approved by an Institutional Animal Care and Use Committee (IACUC) or other local institutional review board. NOD.CB17-Prkdc^scid/J (NOD/SCID) or NOD-scid IL2Rgamma^null (NSG) mice were obtained from Charles River Laboratories. Studies performed at Amgen were in accordance with the German Animal Welfare Law with permission from the responsible local authorities, in accordance with the standards of the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC). Studies performed at inviCRO were carried out in accordance with the UK Animals (Scientific Procedures) Act of 1986 and British Home Office regulations governing animal experimentation, with all protocols approved by the Queen Mary University of London Project Assessment Committee (London, United Kingdom). Studies performed at Champions Oncology were carried out according to the local IACUC. Mice were cared for in accordance with the Guide for the Care and Use of Laboratory Animals, 8th Edition (27). Mice, 6–8 weeks of age, were housed in individual ventilated caging systems on irradiated poplar bedding. Animal rooms were set to maintain a temperature of 22°C ± 2°C, relative humidity of 55% ± 20%, and intermittent light and dark cycles of 12 hours. Sterile food and sterile water were made freely available throughout the studies. Cages were changed weekly inside an engineered cage changing station.

As AMG 160 does not cross-react to mouse CD3, human T cells were administered to mice bearing xenograft tumors before treatment. The human T cells were isolated from PBMC using a pan-T Cell Isolation Kit (Miltenyi Biotec) and treated with IL2 and CD3/CD28/CD2 to induce T-cell activation and expansion, to obtain sufficient human T cells from a single donor for use in 1 mouse xenograft study. T-cell activation was assessed by flow cytometric analysis of CD25 cell surface expression, and T-cell expansion was measured by the T-cell count over time.

For the 22Rv-1 mCRPC xenograft model, male NOD/SCID mice were subcutaneously implanted with 22Rv-1 cells (day 0). When the tumors reached approximately 200 mm³, the mice were intraperitoneally injected with 2 × 10⁷ activated human T cells per animal. Because of its extended half-life, AMG 160 was administered once weekly by bolus intravenous injection (days 16, 23, and 31); the vehicle control was administered on days 16 and 23 as mice in this group were taken off study by day 31. Tumor volume was measured by calipers (DCS-1, Plexx B.V.) three times a week and the data were analyzed by VIVO Manager Software (Biosystemes). Blood samples were collected from the periorbital venous plexus at various timepoints after the last dose for pharmacokinetics analysis. AMG 160 concentration in mouse serum was analyzed by an electrochemiluminescence-based immunoassay developed at Amgen, using anti-CD3 antibodies for capture and detection of AMG 160.

In vivo binding competition of AMG 160 with ⁶⁸Ga-PSMA-11 was evaluated in NSG mice bearing established 22Rv-1 tumors. A total of 5 × 10⁶ 22Rv-1 cells mixed 1:1 with Matrigel (Corning) were injected subcutaneously and allowed to form tumors; the study start (day 0) was when tumors reached approximately 150 mm³. In the single-dose study, mice were administered 10 MBq of ⁶⁸Ga-PSMA-11 by intravenous injection on days −2, 1, 2, and 4 and were treated with 0.2 mg/kg AMG 160 or an EGFR variant III (EGFRvIII) HLE BiTE molecule (see Supplementary Materials and Methods) by intravenous injection on day 0. In the efficacy study, mice were administered human T cells on day −2, 0.2 mg/kg AMG 160 or EGFRvIII HLE BiTE molecule on days 0 and 8, and 10 MBq of ⁶⁸Ga-PSMA-11 on days −1, 1, 7, and 14. Biodistribution data were acquired via 10-minute static, whole body PET/CT imaging 60–70 minutes after tracer injection. Tumor volume was measured by calipers three times per week. For each image dataset, reconstructed PET images were coregistered to their corresponding CT image and resampled to uniform isotropic voxel sizes (0.2 mm³) using VivoQuant Software (inviCRO). PET images were converted to units of percent injected dose per gram (%ID/g) with the assumption of tissue density of 1 g/mL. The total activity concentration for each region of interest was calculated and expressed as percent injected dose.

Antitumor activity of AMG 160 was also evaluated in the CTG-2428 patient-derived xenograft (PDX) model of mCRPC, which was derived from recurrent metastasis to bone and expresses PSMA at 8.1 fragment reads per kilobase million (Champions Oncology TumorGraft model). This model used NOD/SCID/gamma (NSG) mice with a beta-2-microglobin (B2M) mutation, which increases clearance of fragment crystallizable (Fc)-containing molecules due to their inability to bind the neonatal Fc receptor (FcRn)/B2M heterodimer (27). The half-life of AMG 160 is shortened in this model to 16.5 ± 0.4 hours, and the half-life of the anti-PD-1 antibody is shortened to 18.55 ± 0.55 hours in this model. To maintain sufficient exposure levels in this model, AMG 160 was dosed at 10 mg/kg and the anti-PD-1 antibody was dosed at 30 mg/kg, every other day for 22 days (12 doses). CTG-2428 tumor fragments were implanted subcutaneously into the NSG-B2M mice. Once tumors reached approximately 130 mm³, mice were administered a single infusion of human PBMC that had been activated and expanded in vitro (day 0), and then were treated with AMG 160, control HLE BiTE molecule, anti-PD-1 antibody, or control antibody. Tumor volume was measured by calipers twice a week.

Male NHP, Macaca fascicularis of Mauritian origin, weighing 6–9 kg, were cared for in accordance with the Guide for the Care and Use of Laboratory Animals, 8th Edition (27). Animals were housed individually at an indoor, AAALAC-accredited facility. All studies were approved by the IACUC. Animals were fed a certified pelleted primate diet daily in amounts appropriate for their age and size and had unlimited access to water through an automatic watering system. Animals were maintained on a 12-hour light/12-hour dark cycle in rooms kept at 18°C–29°C with 30%–70% humidity. Animals were socialized where possible to provide enrichment.

NHP received AMG 160 as a flat dose, five doses of 5 μg/kg, or as a step dose, a single dose of 5 μg/kg followed by four doses of 15 μg/kg of AMG 160, or a single dose of 5 μg/kg followed by four doses of 30 μg/kg AMG 160. All animals were dosed on days 1, 5, 12, 19, and 26 of a 33-day treatment period. NHP in a control group received vehicle only. Multiple parameters were evaluated, including clinical observations, body weight, food consumption, body temperature, cardiovascular and neurologic examinations, clinical pathology, and histopathology. Tissue sections were stained with hematoxylin and eosin (H&E) to enable evaluation by light microscopy and all observations were semiquantitatively graded in severity on the basis of a standard 5-point grading scheme [within normal limits (0), minimal (1), mild (2), moderate (3), marked (4), and severe (5); ref. 28]. Sexual maturity of animals was confirmed via light microscopic evaluation of testis. Serum samples for toxicokinetic analysis were collected at multiple timepoints to profile the AMG 160 exposure levels. Total serum concentration of AMG 160 was estimated using the Phoenix Software (Pharsight Corporation). Noncompartmental parameters, including area under the serum concentration time curve, maximum serum concentration, and half-life, were estimated.

A two-compartment model was used to characterize the toxicokinetic profile of AMG 160 from pooled NHP data. The model was parameterized using systemic clearance, volume of distribution of central compartment, intercompartmental clearance, and peripheral volume of distribution. The derived NHP model was scaled to humans using allometry. Modeling was used to simulate the human pharmacokinetics resulting from AMG 160 administration once every 2 weeks.

AMG 160 contains an anti-PSMA binding domain and an anti-CD3 binding domain connected by a short linker, with the core BiTE molecule structure fused to the amino terminus of an immunoglobulin fragment crystallizable (IgG Fc) domain (Fig. 1A). The Fc domain is effectorless and functions to increase the molecule size and enable binding to the FcRn to extend serum half-life (29, 30). AMG 160 was designed to bind human PSMA and CD3 and to cross-react to NHP PSMA and CD3. AMG 160 bound to the PSMA-positive prostate cancer cell line LNCaP and to CHO cells stably expressing NHP PSMA, but did not bind to PC-3 or untransfected CHO cells (CHO DHFR⁻) that do not express PSMA (Supplementary Fig. S1A). AMG 160 had equilibrium dissociation (K_D) values of 14.8 ± 2 nmol/L for human PSMA and 9.1 ± 2 nmol/L for NHP PSMA, as determined by flow cytometry analysis of CHO cells expressing human or NHP PSMA (Supplementary Fig. S1B). AMG 160 bound to human CD3 with a K_D value of 22.4 ± 2.77 nmol/L and to NHP CD3 with a K_D value of 18.1 ± 2.15 nmol/L, as determined by surface plasmon resonance (SPR; Supplementary Fig. S1C). SPR showed qualitative binding of AMG 160 to human and NHP FcRn (Supplementary Fig. S1D), although the K_D value was not calculated because of the fast off-rate and overall low affinity of AMG 160 to FcRn. These data demonstrate the high binding affinity and specificity of AMG 160 for human and NHP PSMA and CD3.

In cocultures of PSMA-positive C4-2 cells with human PBMC, AMG 160 induced dose-dependent increases in CD25, CD69, 4-1BB, and PD-1 expression on T cells at 48 hours, indicating T-cell activation (Fig. 1B). AMG 160–mediated T-cell activation required the presence of PSMA-positive cells, as minimal T-cell activation was observed in cocultures of PSMA-negative PC-3 cells with PBMC (Fig. 1B). AMG 160 also induced dose-dependent secretion of cytokines, including IL2, IL4, IL6, IL10, IFNγ, and TNFα when cocultured with PSMA-positive prostate cancer cells and human PBMC in vitro (Fig. 1C).

AMG 160 induced potent lysis of PSMA-positive prostate cancer cell lines in cocultures with PBMC, with EC₅₀ values of 6–42 pmol/L after 48 hours (Fig. 1D; Supplementary Table S3). AMG 160 cytotoxic activity was observed across prostate cancer cell lines with a range of PSMA expression levels (∼7,000–5 × 10⁵ receptors/cell), with the most potent activity observed against cell lines LnCAP and C4-2, with the highest levels of PSMA (Fig. 1D; Supplementary Table S3). PSMA expression is upregulated in prostate tumors after treatment with ADT (12, 31), suggesting that treatment of prostate cancer cells with enzalutamide has the potential to enhance the cell potency of AMG 160. Treatment of 22Rv-1 or C4-2B cells in vitro with clinically relevant concentrations of enzalutamide increased PSMA cell surface expression, as assessed by flow cytometry (Supplementary Fig. S2). The increased PSMA expression on prostate cancer cells was able to enhance AMG 160–mediated cytotoxicity, although differences in the mean EC₅₀ values were not statistically significant (Supplementary Fig. S2). AMG 160 also induced cytotoxicity of PSMA-positive mCRPC cell lines (Supplementary Table S3). Together, these data support the idea that AMG 160 has potent cell killing activity across cells with a range of PSMA expression and may be effective against tumor cells that have become resistant to other therapies.

AMG 160 antitumor activity was evaluated in NOD/SCID mice bearing established 22Rv-1 CRPC xenograft tumors. As AMG 160 does not cross-react with mouse CD3, human T cells that were activated and expanded in vitro were administered to tumor-bearing mice by intraperitoneal injection on day 12 after implantation of the 22Rv-1 cells. Starting on day 16, when tumors had reached a size of approximately 200 mm³, mice were treated with AMG 160 or a vehicle control once weekly. AMG 160 treatment resulted in significant (P < 0.001) tumor growth inhibition, reaching a maximum of 91% (at doses of 2 or 0.2 mg/kg) or 63% (at 0.02 mg/kg) on day 26 (Fig. 2A). AMG 160 exposure levels in serum samples collected after the last dose were dose dependent and above the cell EC₅₀ value at all doses tested (Fig. 2B). These data provide the first evidence that AMG 160 may promote significant antitumor activity against established CRPC tumors.

To evaluate whether AMG 160 treatment could be used following ⁶⁸Ga-PSMA-11 PET imaging to detect PSMA-positive tumors, we conducted binding competition assays on PSMA-expressing prostate cancer cells. In vitro, AMG 160 binding to PSMA-expressing C4-2B cells was similar in the absence or presence of the unlabeled precursor of ⁶⁸Ga-PSMA-11, PSMA-11 (Supplementary Fig. S3A), indicating that PSMA-11 did not affect AMG 160 binding. An anti-PSMA BiTE molecule targeting the same PSMA epitope as AMG 160 did inhibit AMG 160 binding to C4-2B cells and served as a control for the experiment (Supplementary Fig. S3B). In vivo binding competition of AMG 160 with the radiolabeled form of PSMA-11, ⁶⁸Ga-PSMA-11, was evaluated in the 22Rv-1 tumor xenograft model. NOD/SCID mice bearing established 22Rv-1 tumors were administered a single treatment of 10 MBq ⁶⁸Ga-PSMA-11 and a baseline PET/CT scan confirmed that the tumor cells expressed PSMA (day −2; Fig. 3A). Mice were then administered a single intravenous dose of 0.2 mg/kg AMG 160 or a nontargeting control HLE BiTE molecule (day 0). Additional PET/CT scans performed over 4 days following BiTE molecule treatment demonstrated that the ⁶⁸Ga-PSMA-11 signal remained detectable at levels that were equivalent or greater to those observed at baseline and was relatively similar between the two groups (Fig. 3A). These data suggest that AMG 160 does not affect ⁶⁸Ga-PSMA-11 binding to PSMA. The effects of ⁶⁸Ga-PSMA-11 treatment on AMG 160 antitumor activity were also evaluated in the 22Rv-1 xenograft model. NOD/SCID mice bearing established 22Rv-1 xenograft tumors were administered a single injection of human T cells (day 0) and were treated with 0.2 mg/kg AMG 160 or a control HLE BiTE molecule on days 0 and 8. Mice were injected with 10 MBq ⁶⁸Ga-PSMA-11 on day −1 to establish baseline levels of PSMA by PET/CT (Fig. 3B). Subsequent scans showed a decrease in signal on days 7 and 14 in mice that had been treated with AMG 160 (Fig. 3B). The decrease in ⁶⁸Ga-PSMA-11 signal was associated with a significant reduction in tumor volume (P = 0.04; Fig. 3C). In contrast, the ⁶⁸Ga-PSMA-11 signal in mice treated with the nontargeting control HLE BiTE molecule increased over time, consistent with the continued growth of the 22Rv-1 tumors in this group (Fig. 3B and C). Together, these data indicate that AMG 160 may be used following ⁶⁸Ga-PSMA-11 PET/CT imaging.

AMG 160 treatment of prostate cancer cells induced upregulation of PD-1 on T cells in vitro (Fig. 1B). Increased PD-1 expression on T cells has the potential to decrease BiTE molecule activity or to be a mechanism of therapy resistance (32). The combination of AMG 160 with an anti-PD-1 antibody was evaluated in a PDX model of mCRPC, CTG-2428. In this model, immunodeficient NSG-B2M mice bearing CTG-2428 tumor xenografts were injected with human T cells, and then treated with 10 mg/kg of AMG 160 or a control HLE BiTE molecule, and 30 mg/kg of an anti-PD-1 antibody or control antibody, every 2 days for 22 days. The B2M strain background results in fast clearance of Fc-based molecules (Supplementary Fig. S4), and the dosing regimen was selected to enable similar exposure levels as the 0.2 mg/kg dose in the 22Rv-1 xenograft study (Fig. 2B). AMG 160 monotherapy induced regression of CTG-2428 tumors (P < 0.0001), with 30% tumor regression on day 22 as compared with the control HLE BiTE molecule (Fig. 4). Anti-PD-1 treatment did not inhibit CTG-2428 tumor growth (Fig. 4). The combination of AMG 160 and anti-PD-1 induced an antitumor response starting on day 11, and the effect of AMG 160 monotherapy became manifest 4 days later. By day 43 of the study, more than 3 weeks after the final treatment with AMG 160 alone or the combination of AMG 160 with an anti-PD-1 antibody, tumors had continued to regress and only small nodules of <30 mm³ remained. This study provides further evidence for AMG 160 monotherapy activity against established mCRPC tumors and indicates that the combination with an anti-PD-1 antibody may potentiate AMG 160 efficacy.

The cross-reactivity of AMG 160 to NHP PSMA and CD3 was evaluated in TDCC assays. In cocultures of CHO cells expressing NHP PSMA with NHP PBMC, AMG 160 induced upregulation of the T-cell markers, CD25 and CD69 (Supplementary Fig. S5A), cytokine production (Supplementary Fig. S5B), and target cell lysis (Supplementary Fig. S5C). AMG 160 did not induce T-cell activation or cytotoxicity in cocultures of NHP PBMC with CHO cells that did not express NHP PSMA (Supplementary Fig. S5A and S5C). These data, together with the binding affinity studies (Supplementary Fig. S1), support the selection of NHP as a relevant model for AMG 160 toxicology studies.

A repeat-dose toxicology study was conducted in mature male NHP to evaluate potential effects of AMG 160. NHP were administered AMG 160 by intravenous infusion for 30 minutes and a flat dose of 5 μg/kg was compared with a step-dose design where a 5 μg/kg dose was followed by 15 or 30 μg/kg on the subsequent days of dosing. AMG 160 was well-tolerated during the treatment period at all doses tested. There were no changes in body weight or food consumption, and there were no AMG 160–related ocular or neurologic effects. Evidence for BiTE molecule target engagement was observed in all dose groups, including transient body temperature changes (Fig. 5A), increased C-reactive protein (CRP) levels (Fig. 5B), and transient cytokine production (Fig. 5C). The elevation in CRP levels suggested an acute phase response and correlated with a decrease in albumin, an increase in globulins, and a decrease in albumin/globulin ratio. There were also increased levels of lipids that were considered part of the acute phase response, and additional changes in bilirubin, blood urea nitrogen, and creatinine that were likely related to AMG 160 treatment. Cytokine release was observed after the first dose of AMG 160 and could be attenuated with a step-dose design that allowed higher doses of AMG 160 to be tolerated subsequently (Fig. 5C). Mononuclear cell or mixed cell infiltrates were observed in several NHP tissues known to express PSMA, including brain, duodenum, kidney, lung, and prostate gland, following AMG 160 treatment (Fig. 5D). These findings were largely reversible after 1 month without AMG 160 treatment.

The toxicokinetic profile of AMG 160 was evaluated during the NHP toxicology study. Exposure levels were dose dependent, with minimum effective concentration (C_min) maintained above the in vitro cell EC₅₀ value concentration for all doses evaluated. The toxicokinetic profile following the last dose from the 5/30 μg/kg group (Fig. 5E) showed clearance of 0.47 mL/hour/kg, and a terminal half-life of 6.1 days. The NHP data were scaled to project human pharmacokinetics for a 10 μg dose of AMG 160 administered every other week. This analysis shows that AMG 160 has an extended half-life that may support intermittent dosing in humans (Fig. 5F).

Treatment of patients with mCRPC has proven challenging due to heterogeneity of the disease and alterations in AR and immune pathway signaling as cells develop resistance to ADT (33, 34). PSMA is a compelling target in this setting as the receptor levels increase as tumors transition to androgen independence (10–12). The most advanced PSMA-targeted therapy, ¹⁷⁷Lu-PSMA-617, decreased PSA levels >50% in 64% (32/50) of patients and enabled an objective response in 56% (15/27) of patients with measurable disease in a long-term follow-up from a phase II study (35). That study, as well as a recent phase I/II study on the lutetium-labeled anti-PSMA mAb, ¹⁷⁷Lu-J591, showed that patients with high PSMA levels as determined by ⁶⁸Ga-PSMA-11 PET imaging were more likely to respond to radioligand therapy (35, 36).

AMG 160 eradicates PSMA-expressing cells by a completely different mechanism than radioligand therapy. BiTE molecules redirect T cells to lyse tumor cells with a wide range of target expression levels, including cells that express very low levels of target (<1,000 receptors/cell; ref. 37). While AMG 160 cell potency increased with increasing PSMA expression levels, it also effectively mediated redirected lysis of the VCAP cell line (∼7,000 PSMA molecules/cell). This PSMA expression level is below the limit of detection of ⁶⁸Ga-PSMA-11 PET imaging in mouse models (20,000 PSMA molecules/cell; ref. 38), suggesting that AMG 160 may be effective even against tumors that appear to be PSMA-negative in ⁶⁸Ga-PSMA-11 PET scans. AMG 160 is also active against prostate cancer cell lines, including LNCaP and VCAP, that contain mutations in DNA repair genes that are potentially associated with resistance to radioligand therapy (39, 40). The nonclinical safety profile of AMG 160 suggests the potential for cytokine production, which is expected on the basis of the BiTE molecule mechanism of action, but was not associated with xerostomia as has been observed with PSMA-targeted radioligand therapy (21). AMG 160, therefore, may be effective in patients who do not respond to PSMA-targeted radioligand therapy with the potential for an improved safety profile. Early data support this idea, as a patient with mCRPC achieved a near-complete remission with AMG 212 treatment after failing to respond to ¹⁷⁷Lu-PSMA-617 (26).

BiTE immuno-oncology therapy also differs from immune checkpoint blockade and may be effective in tumors that do not respond to anti-PD-1 monotherapy. Despite early evidence for antitumor activity of PD-1 inhibitors in a subset of patients with mCRPC (41), immune checkpoint blockade has had limited impact on the mCRPC treatment landscape. The anti-PD-1 inhibitor, pembrolizumab, is approved for the treatment of microsatellite instability-high solid tumors (42), which represent 3% of patients with prostate cancer (43). In the KEYNOTE-028 study, which evaluated pembrolizumab monotherapy in patients with mCRPC selected on the basis of expression of the PD-1 ligand, PD-L1, 4 of 23 (17.4%) patients had a partial response (PR) and overall survival was 7.9 months (44). The KEYNOTE-199 study showed a response rate of 5% (9/166 patients), with two complete responses and five PRs in PD-L1–selected patients and two PRs in PD-L1–negative patients, and median response duration of 16.8 months (45). In our preclinical studies, AMG 160 engaged systemically administered T cells to induce regression of established CRPC tumors, including the CTG-2428 model, which showed limited response to anti-PD-1 inhibition alone. AMG 160 also induced immune cell infiltration in NHP tissues that express PSMA. Together, these preclinical data suggest that BiTE immuno-oncology therapy may induce inflammation in prostate tumors and enhance the antitumor immune response. Preliminary clinical data support this concept: in addition to the AMG 212 patient described above, AMG 212 treatment induced a dose-dependent decrease in PSA in patients with mCRPC and enabled long-term response in a second patient. These responses were observed at dose levels below the MTD (26).

Unlocking immune activation in mCRPC to promote durable response may require combination therapy. The enhanced activity of the combination of AMG 160 with an anti-PD-1 antibody may reflect the ability of immune checkpoint blockade to maintain T-cell activity or could result from increased levels of tumor-infiltrating lymphocytes driven by BiTE molecule treatment. The combination of AMG 160 with enzalutamide also increases AMG 160 cytotoxic activity. While this may be driven by increased levels of cell surface PSMA expression in our in vitro studies, enzalutamide might increase AMG 160 activity in patients by multiple mechanisms. Patients with mCRPC responded to anti-PD-1 after progression on enzalutamide, suggesting that enzalutamide increases T-cell infiltration into tumors (46).

AMG 160–mediated T-cell activation and cytotoxicity of prostate cancer cells in vitro, and antitumor activity against established 22Rv-1 xenograft tumors in vivo, were similar to that of the first PSMA-targeted BiTE molecule, AMG 212 (25). The extended serum half-life of AMG 160 enables treatment at longer dosing intervals and increases convenience of administration as compared with AMG 212 (25). The nonclinical safety profile of AMG 160 is acceptable and consistent with BiTE target engagement in vivo. These attributes, together with the potent, specific antitumor activity of AMG 160 in vitro and in mouse xenograft models, make AMG 160 a compelling immuno-oncology therapy for the treatment of mCRPC. AMG 160 is currently being evaluated in a phase I clinical study as monotherapy and in combination with pembrolizumab (NCT03792841).

P. Deegen reported employment with and stock ownership in Amgen. O. Thomas reported employment with and stock ownership in Amgen. O. Nolan-Stevaux reported other from Amgen, Inc. during the conduct of the study and outside the submitted work. S. Li reported employment with and stock ownership in Amgen. J. Wahl reported employment with and stock ownership in Amgen. P. Bogner reported employment with and stock ownership in Amgen. F. Aeffner reported other from Amgen outside the submitted work. M. Friedrich reported employment with and stock ownership in Amgen. M.Z. Liao reported employment with and stockholder of Amgen. K. Matthes reported employment with and stock ownership in Amgen. D. Rau reported employment with and stock ownership in Amgen and is listed on relevant patents as inventor. B. Rattel reported employment with and stock ownership in Amgen. T. Raum reported full-time employment with and stock ownership in Amgen; moreover, as part of work at Amgen is listed as an inventor on patent applications/patents relevant to the work. P. Kufer reported a patent for PCT/EP1998/007313 licensed to Amgen, and employment with and stock ownership in Amgen. A. Coxon reported employment with and stock ownership in Amgen Inc. J.M. Bailis reported employment with and stock ownership in Amgen, Inc.

P. Deegen: Conceptualization, supervision, investigation, visualization, methodology, writing–original draft, writing–review and editing. O. Thomas: Conceptualization, resources, formal analysis, supervision, visualization, methodology, writing–original draft, writing–review and editing. O. Nolan-Stevaux: Conceptualization, supervision, visualization, methodology, writing–review and editing. S. Li: Validation, investigation, visualization, methodology, writing–review and editing. J. Wahl: Conceptualization, formal analysis, supervision, investigation, visualization, methodology, writing–original draft, writing–review and editing. P. Bogner: Conceptualization, supervision, investigation, visualization, methodology, writing–original draft, writing–review and editing. F. Aeffner: Investigation, visualization, methodology, writing–review and editing. M. Friedrich: Conceptualization, supervision, validation, writing–original draft, writing–review and editing. M.Z. Liao: Conceptualization, investigation, visualization, methodology, writing–original draft, writing–review and editing. K. Matthes: Investigation, visualization, methodology, writing–original draft, writing–review and editing. D. Rau: Conceptualization, supervision, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. B. Rattel: Supervision, writing–review and editing. T. Raum: Conceptualization, supervision, writing–review and editing. P. Kufer: Conceptualization, supervision, writing–review and editing. A. Coxon: Supervision, writing–review and editing. J.M. Bailis: Conceptualization, supervision, validation, visualization, methodology, writing–original draft, writing–review and editing.

This study was funded by Amgen Inc. We thank Ciara Finucane and Sven Macholl (InviCRO) for performing ⁶⁸Ga-PSMA-11 PET studies. Katie Hsu provided technical support for cytokine analysis of NHP serum, and Kevin Cook conducted the pharmacokinetic analysis in NSG-B2M mice. Medical writing support was provided by Sukanya Raghuraman, PhD (Cactus Life Sciences, part of Cactus Communications) and Micah Robinson, PhD. We thank Sylvie Tiso, PhD, for advice and Mark Salvati, PhD, and Hosein Kouros-Mehr, MD, PhD, for critical review of the article.

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.