Pharmacodynamic Effects and Immune Correlates of Response to the CD20/CD3 Bispecific Antibody Mosunetuzumab in Relapsed or Refractory Non-Hodgkin Lymphoma

Introduction: Mosunetuzumab (M; RG7828) is a full-length, fully humanized immunoglobulin G1 (IgG1) bispecific antibody targeting both CD3 (on the surface of T cells) and CD20 (on the surface of B cells). It drives tumor B-cell killing independent of T-cell specificity and has been shown clinically to have single-agent activity in relapsed or refractory non-Hodgkin lymphoma, with durable complete responses (CR) at doses as low as 1.2mg (Budde et al. ASH 2018). We present biomarker data from the ongoing first-in-human multicenter Phase I/Ib study GO29781 (NCT02500407), highlighting acute pharmacodynamic effects that confirm mechanism of action (MOA), support Cycle 1 step-up dosing to mitigate cytokine-driven toxicities, and potentially predict response.

Methods: Patients (pts) with biomarker data in Group A (0.05 to 2.8mg q3w dosing, n=30) and Group B (0.4/1/2.8 to 1/2/40.5mg Cycle 1 Day 1/8/15 step-up dosing, followed by q3w dosing) were included in the analyses. Peripheral biomarkers were evaluated using whole blood flow cytometry, plasma cytokine ELISA, and ex-vivo single-cell cytokine assay of cryopreserved peripheral blood mononuclear cells. Tumor biomarkers were assessed through CD20 immunohistochemistry and CD8/Ki67 immunofluorescence staining. Average CD20 receptor occupancy (RO%), which takes into account residual levels of rituximab that may compete with M for target binding, was used to characterize the exposure-pharmacodynamic response relationship (Li et al. ASH 2019).

Results: Pharmacodynamic changes were observed in peripheral blood by 4 hours (hrs) after infusion of sub-efficacious doses of M in Group A. Transient reduction of circulating T cells was evident starting at the lowest doses administered of 0.05-0.2mg, with decreased cell numbers by 4 hrs after infusion and recovery by 72 hrs. At doses between 0.4-0.8mg, T-cell activation was observed in some pts, as evidenced by upregulation of CD69 on both CD8+ and CD4+ T cells and elevation of IFN-γ in plasma. At 1.2mg (minimal efficacious dose) to 2.8mg, all of these changes were detected in the majority of pts. M induced depletion of circulating B cells immediately after the first dose, which was sustained for the duration of treatment, even in pts refractory to prior anti-CD20 therapy. IL-6 levels peaked within 24 hrs of the Cycle 1 Day 1 dose and the kinetics of IL-6 increase were associated with the onset of cytokine release syndrome (CRS). Hence, step-up dosing at Cycle 1 was implemented in Group B to maintain the IL-6 peak at reasonably low levels and reduce the risk of toxicity. With step-up dosing, pharmacodynamic changes, including transient T-cell decrease, T-cell activation, and cytokine production, remained strongest within 4 hrs of the Cycle 1 Day 1 dose, despite higher subsequent and total cumulative doses. IL-6 peaks were similar in Group A and Group B and were in line with the low grades of CRS observed to date (Schuster et al. ASH 2019). In pts for whom the Cycle 1 Day 1 dose was fixed at 1mg, the acute pharmacodynamic changes were found to be associated with RO% at Day 1.

In a subset of diffuse large B-cell lymphoma pts, functional assays showed that peripheral T cells taken at baseline from pts who achieved CR secreted multiple effector cytokines, such as IFN-γ, TNF-α, and Granzyme B, upon M stimulation ex vivo, whereas T cells from pts with progressive disease did not show this capacity. Analysis of tumor biopsies showed similar baseline levels of CD8+ tumor infiltrating T-cells (TILs) between responders and non-responders. M induced increases in CD8+ TILs, which were mostly Ki67-, as early as Cycle 1. The median on-treatment CD8+ TIL level was significantly higher in responders than non-responders.

Conclusions: T-cell activation, cytokine elevation, and B-cell depletion in the periphery are among the earliest biomarkers of M activity and precede clinical responses. Acute pharmacodynamic changes confirm the MOA and inform dose/schedule selection. In particular, the 1mg Cycle 1 Day 1 dose, which was chosen to mitigate CRS and to dissociate toxicity and efficacy at higher subsequent doses, still induced potent pharmacodynamic changes. Finally, the data suggest a model whereby clinical efficacy may be critically dependent on activation and migration of functional T cells from the periphery into the tumor bed.