A phase 1 first‐in‐human study of GS‐0189, an anti‐signal regulatory protein alpha (SIRPα) monoclonal antibody, in patients with relapsed/refractory (R/R) non‐Hodgkin lymphoma (NHL)

Abstract Signal regulatory protein alpha (SIRPα) is the receptor for cluster of differentiation (CD)47, a potent “don't eat me” signal for macrophages. Disruption of CD47‐SIRPα signaling in the presence of prophagocytic signals can lead to enhanced phagocytosis of tumor cells, resulting in a direct antitumor effect; agents targeting this pathway have shown efficacy in non‐Hodgkin lymphoma (NHL) and other tumor types. GS‐0189 is a novel anti‐SIRPα humanized monoclonal antibody. Here we report: (1) clinical safety, preliminary activity, and pharmacokinetics of GS‐0189 as monotherapy and in combination with rituximab from a phase 1 clinical trial in patients with relapsed/refractory NHL (NCT04502706, SRP001); (2) in vitro characterization of GS‐0189 binding to SIRPα; and (3) in vitro phagocytic activity. Clinically, GS‐0189 was well tolerated in patients with relapsed/refractory NHL with evidence of clinical activity in combination with rituximab. Receptor occupancy (RO) of GS‐0189 was highly variable in NHL patients; binding affinity studies showed significantly higher affinity for SIRPα variant 1 than variant 2, consistent with RO in patient and healthy donor samples. In vitro phagocytosis induced by GS‐0189 was also SIRPα variant–dependent. Although clinical development of GS‐0189 was discontinued, the CD47‐SIRPα signaling pathway remains a promising therapeutic target and should continue to be explored.

phagocytosis induced by GS-0189 was also SIRPα variant-dependent. Although clinical development of GS-0189 was discontinued, the CD47-SIRPα signaling pathway remains a promising therapeutic target and should continue to be explored.

K E Y W O R D S
CD47, GS-0189, SIRPα, non-Hodgkin lymphoma, monoclonal antibodies INTRODUCTION Non-Hodgkin lymphoma (NHL) is one of the most common cancers in the United States, accounting for about 4% of all cancers [1]. Systemic treatment options include chemotherapy, immuno-and targeted therapy, chimeric antigen receptor (CAR) T-cell therapy, and stem cell transplant [2]. Many patients have relapsed/refractory (R/R) disease following frontline treatment for NHL [3][4][5][6]. The overall prognosis and long-term survival for R/R patients after multiple lines of therapy are often poor [7][8][9]. CAR T-cell therapy provides impressive response rates in heavily pretreated R/R follicular and other B-cell NHLs, but toxicities, complicated logistics, limited access, and delays in delivering treatment may limit use of this therapy in many patients [10][11][12]. The development of more effective and tolerable therapies for R/R NHL represents a high unmet medical need.
GS-0189 was designed as an alternative to magrolimab, an anti-CD47 monoclonal antibody (mAb) in clinical development in hematologic malignancies and solid tumors, as it had been found not to impact red blood cells in preclinical experiments [23]. Here we report: (1) clinical safety, preliminary activity, and pharmacokinetics (PK) of GS-0189 as monotherapy and combination therapy with rituximab, from a phase 1 first-in-human (FIH) clinical trial in patients with R/R NHL (NCT04502706); (2) in vitro characterization of GS-0189 binding potency to SIRPα variants; (3) and in vitro phagocytic activity of macrophages derived from donors possessing different SIRPα variants relative to the activity of a control pan-SIRP blocking antibody with an inert Fc, KWAR23.

Design and participants
This was an open-label FIH trial to evaluate GS-0189 safety, PK, and preliminary efficacy of monotherapy and in combination with rituximab in patients with select R/R NHL histologies (Supplemental Methods; Figure S1): here, we report monotherapy dose escalation

Endpoints
The primary endpoint was incidence of adverse events defined by National Cancer Institute Common Terminology Criteria for Adverse Events, Version 5.0. Secondary endpoints included GS-0189 PK, objective response rate per Lugano response criteria [24], duration of response (DOR), and progression-free survival (PFS).

Statistical analysis
Statistical analyses were performed on all patients who received ≥1 dose of GS-0189. For categorical variables, frequencies and percentages were calculated.

Nonclinical and ex vivo assays
Antibodies and reagents used in these studies are listed in Table S1

GS-0189 and KWAR23 binding to SIRPα variants
Binding experiments quantifying the affinity of GS-0189 or KWAR23 for recombinant SIRPα variants were performed on either a Biacore T100 or T200 instrument using a CM5 sensor chip. GS-0189 and KWAR23 as active control [25,26] were captured and regenerated according to manufacturer instructions. SIRPα V1 and SIRPα V2 were injected using maximum concentrations of either 0.3 μM and three-fold serial dilutions for the higher affinity interactions, or 3 μM and 4-fold serial dilutions for the lower affinity interactions. Data were fitted to a simple kinetic model to derive k on , k off , and K D , using the relationship K D = k off / k on .

2.3.3
In vitro phagocytosis assay

Clinical study results
Nine patients were enrolled and treated between December 2, 2020 and  Table 1. Reasons for GS-0189 discontinuation were patient's decision (n = 2) and progressive disease (n = 7).
Reasons for study discontinuation were death (n = 1), withdrawn consent (n = 2), and study terminated by sponsor (n = 6).    at the time of last adequate tumor assessment. Study treatment was discontinued due to patient decision, which coincided with the time of sponsor decision to discontinue the study. Two patients had stable disease (1 each, from MDE2 and CDE1) with estimated PFS of 5.55 and 3.71 months, respectively; 5 had progressive disease (Table 4); and 1 withdrew consent prior to the first response assessment.

Pharmacokinetics
Concentration-time profiles for patients in MDE1, 2, and 3 and CDE1 are shown in Figure 1. Relevant PK parameters for MDE and CDE (Cohort 1) are listed in Table 5. The area under the curves for GS-0189 serum concentration-time profiles in the given dose range of 10 to 100 mg were lower than projected based on cyno PK-PD studies.
This could potentially be due to target-mediated drug disposition for the antibody in humans. GS-0189 RO over time from three patients in CDE1 is shown in Figure S2 for comparison with concentration-time profile. The antidrug antibody incidence rate for GS-0189 cannot be interpreted due to a small sample size.

TA B L E 5
Median pharmacokinetic parameters for GS-0189 post first dose (cycle 1, day 1).

GS-0189 and KWAR23 binding to SIRPα variants
Binding affinities of GS-0189 to recombinant SIRPα V1 and SIRPα V2 compared with those of KWAR23 are shown by kinetic and equilibrium constants (Table 6)

Antibody
Antigen values for k on ,k off , and K D varied less than 2-fold between experiments. 2 The equilibrium dissociation constant K D = k off /k on .

F I G U R E 3 GS-0189 binding to SIRP isoforms and variants shown by sensogram for GS-0189 (A and B) or KWAR23 (C and D) binding to
SIRPα V1 (A and C), and SIRPα V2 (B and D). The highest concentration injected was 3 μM for lower affinity interactions, and 0.3 μM for higher affinity interactions for SIRPα V1 and SIRPα V2 for GS-0189. The highest concentration injected was 0.3 μM for all SIRP variants for KWAR23. Black lines denote binding data; orange lines represent the kinetic fit. Abbreviations: SIRPα, signal regulatory protein alpha.

Potency of GS-0189 in phagocytosis assays is dependent on SIRPα polymorphism
Maximal phagocytosis of Raji cells by macrophages from PBMC donors with SIRPα V1/V1 and SIRPα V2/V2 variants was induced by GS-0189 combined with rituximab ( Figure S3), with only a small increase in maximal phagocytosis induced by the combination versus rituximab alone. Therefore, we identified a human colorectal cancer cell line DLD-1 for which phagocytosis was more dependent on inhibition of the CD47-SIRPα axis and for which phagocytosis was significantly induced in response to CD47-SIRPα blockade alone [25]. The ability of GS-0189 to potentiate phagocytosis of DLD-1 cells by human PBMCderived macrophages from donors with different SIRPα variants was dose dependent (Figure 4). The average half-maximal effective concentration (EC 50 ) of GS-0189 with macrophages from donors expressing SIRPα V1/V1 was 0.02 μg/mL, expressing SIRPα V1/V2 was 13.8 μg/mL, and expressing SIRPα V2/V2 was 9.5 μg/mL. KWAR23 induced phagocytosis across SIRPα variants with EC 50 ranging from 0.04 to 0.10 μg/mL ( Figure S4). Regardless of SIRPα genotype, maximal phagocytosis induced by GS-0189 was equivalent to that induced by KWAR23 at concentrations above 10 μg/mL (Figure 4).

DISCUSSION
GS-0189 up to 100 mg as monotherapy and in combination with rituximab was well tolerated by patients with R/R NHL in this phase 1 study. There were no DLTs, no grade 4 TEAEs, and no deaths related to TEAEs. Since SIRPα expression is mainly on macrophages and GS-0189 does not have a functional Fc, it was theorized that GS-0189 would result in less anemia compared to most CD47-targeting agents [23,27]. In SRP001, anemia occurred in 2 patients, 1 of whom had anemia at baseline. Lymphocyte count decreases observed with GS-0189 in combination with rituximab were transient and resolved quickly.
Although lymphopenia has been reported with rituximab monotherapy [28], the causality of this phenomenon will remain unclear unless a randomized clinical study is conducted to evaluate each drug's contribution.
GS-0189 is an anti-SIRPα antibody with an aglycosylated (inert) Fc region that was theorized to benefit from the presence of another drug that provides an "eat me" signal, such as rituximab, to induce phago- GS-0189 was intended as a low anemia-risk alternative to magrolimab. With priming-dose regimen and clinical management, acute anemia observed with magrolimab is no longer expected to limit magrolimab clinical development [29,30]. Therefore, the decision was made to terminate the SRP001 study and discontinue clinical development of GS-0189. Nevertheless, the CD47-SIRPα interaction/phagocytic mechanism remains a promising target for treatment of patients with solid tumors and hematologic malignancies and should continue to be explored.

AUTHOR CONTRIBUTIONS
MN, NLB, SI, LP, AS, JB, VG, TC, and MP conceived of or designed the study. MN, LP, JB, YL, and MP acquired and provided data.
All authors analyzed or interpreted the data, drafted, or critically reviewed the manuscript, approved the final version, and agreed to be accountable for all aspects of the work.

ACKNOWLEDGMENTS
We thank the patients who participated in the clinical study and their families, the study coordinators, and the support staff at the clinical sites. We would also like to acknowledge the contributions of the GS-