An affinity threshold for maximum efficacy in anti-PD-1 immunotherapy

Abstract

Monoclonal antibodies targeting the programmed cell death protein 1 (PD-1) remain the most prevalent cancer immunotherapy both as a monotherapy and in combination with additional therapies. Despite the extensive success of anti-PD-1 monoclonal antibodies in the clinic, the experimental relationship between binding affinity and functional potency for anti-PD-1 antibodies in vivo has not been reported. Anti-PD-1 antibodies with higher and lower affinity than nivolumab or pembrolizumab are entering the clinic and show varied preclinical efficacy. Here, we explore the role of broad-ranging affinity variation within a single lineage in a syngeneic immunocompetent mouse model. By developing a panel of murine anti-PD-1 antibodies with varying affinity (ranging from K_D = 20 pM - 15 nM), we find that there is a threshold affinity required for maximum efficacy at a given dose in the treatment of the MC38 adenocarcinoma model with anti-PD-1 immunotherapy. Physiologically based pharmacokinetic modeling complements interpretation of the experimental results and highlights the direct relationship between dose, affinity, and PD-1 target saturation in the tumor.

Keywords: PD-1; affinity; antibody; cancer immunotherapy; pharmacokinetic modeling.

Figure 1.

Engineered anti-PD-1 monovalent mutants. (a) Mutations across complementary determining regions (CDR) compared to the parental murine anti-PD-1 clone, 29 F.1A12. (b) Bivalent and monovalent antibody formats including LALA-PG mutations to silence Fc effector function. (c) Homology model generated using the ROSIE platform. Colors match the CDR as indicated in panel (a). (d) Fitted association and dissociation curves generated via Octet (FortéBio, Sartorius AG) for the monovalent antibody panel. Figure 1 shows the sequences of the antibody variants, a graphical model of the protein structures, a homology model, and the biolayer interferometry results displaying a range of affinity for mouse PD-1

Figure 2.

Internalization and clearance of antibodies. (a) Log rate of internalization of the full panel of monovalent antibodies as well as the bivalent parental antibody. (b) Log half-life of internalization of the full panel of monovalent antibodies and the bivalent parental antibody. (c) Biexponential decay curve fit of clearance rate of the full panel of monovalent antibodies and the parental bivalent antibody. Figure 2 shows the internalization rates, internalization half lives, and pharmacokinetic clearance rates of the full panel of monovalent antibodies and the bivalent parental antibody.

Figure 3.

Physiological-based pharmacokinetic model of anti-PD-1 immunotherapy. (a) Simplified diagram of a two-compartment model with relevant parameters. (b) Concentration of drug in the plasma over time for a 100-µg dose given every three days. (c) Concentration of free receptor in the tumor. (d) Concentration of free drug in the tumor. (e) Concentration of receptor bound by drug in the tumor. Figure 3 shows a graphical depiction of the pharmacokinetic model and the resulting curves produced by the model. The curves include the drug concentration in the blood, the free drug concentration in the tumor, the free receptor concentration in the tumor, and the concentration of drug bound to the receptor in the tumor.

Figure 4.

In vivo efficacy in MC38 adenocarcinoma model. (a) Dosing paradigm. (b) Model timescale prediction of PD-1 saturation in the tumor above a 99% threshold for an antibody of a given affinity for this 135 ug dose. (c) Overall survival in vivo for the full panel of monovalent antibodies using this dosing paradigm. (d) Adjusted dosing paradigm. (e) Model timescale prediction of PD-1 saturation in the tumor above a 99% threshold for an antibody of a given affinity for these adjusted doses. (f) Overall survival in vivo for the parental 29 F and L1 lower affinity monovalent antibodies using the adjusted dosing paradigm. Figure 4 shows two dosing schemes for the antibodies used to treat the MC38 adenocarcinoma model in C57BL/6 mice. The results from the model at these doses display the time of expected receptor occupancy for a range of antibody affinities. The survival results from the in vivo studies are shown.

Figure 5.

In vivo efficacy in MC38 adenocarcinoma model. (a) Dosing paradigm. (b) Model timescale prediction of PD-1 saturation in the tumor above a 99% threshold for an antibody of a given affinity for a 50 ug dose. (c) Overall survival in vivo comparing the parental 29 F and higher affinity 2.13 monovalent antibodies using this dosing paradigm. (d) Adjusted dosing paradigm. (e) Model timescale prediction of PD-1 saturation in the tumor above a 99% threshold for an antibody of a given affinity for a 100 ug dose. (f) Overall survival in vivo comparing the parental 29 F and 2.13 higher affinity monovalent antibodies using the adjusted dosing paradigm. Figure 5 shows two adjusted dosing schemes for the antibodies used to treat the MC38 adenocarcinoma model in C57BL/6 mice. The results from the model at these doses display the time of expected receptor occupancy for a range of antibody affinities. The survival results from the in vivo studies are shown.