Model-Based Characterization of the Bidirectional Interaction Between Pharmacokinetics and Tumor Growth Dynamics in Patients with Metastatic Merkel Cell Carcinoma Treated with Avelumab

Abstract

Purpose: Empirical time-varying clearance models have been reported for several immune checkpoint inhibitors, including avelumab (anti-programmed death ligand 1). To investigate the exposure-response relationship for avelumab, we explored semimechanistic pharmacokinetic (PK)-tumor growth dynamics (TGD) models.

Patients and methods: Plasma PK data were pooled from three phase I and II trials (JAVELIN Merkel 200, JAVELIN Solid Tumor, and JAVELIN Solid Tumor JPN); tumor size (TS) data were collected from patients with metastatic Merkel cell carcinoma (mMCC) enrolled in JAVELIN Merkel 200. A PK model was developed first, followed by TGD modeling to investigate interactions between avelumab exposure and TGD. A PK-TGD feedback loop was evaluated with simultaneous fitting of the PK and TGD models.

Results: In total, 1,835 PK observations and 338 TS observations were collected from 147 patients. In the final PK-TGD model, which included the bidirectional relationship between PK and TGD, avelumab PK was described by a two-compartment model with a positive association between clearance and longitudinal TS, with no additional empirical time-varying clearance identified. TGD was described by first-order tumor growth/shrinkage rates, with the tumor shrinkage rate decreasing exponentially over time; the exponential time-decay constant decreased with increasing drug concentration, representing the treatment effect through tumor shrinkage inhibition.

Conclusions: We developed a TGD model that mechanistically captures the prevention of loss of antitumor immunity (i.e., T-cell suppression in the tumor microenvironment) by avelumab, and a bidirectional interaction between PK and TGD in patients with mMCC treated with avelumab, thus mechanistically describing previously reported time variance of avelumab elimination.

Trial registration: ClinicalTrials.gov NCT02155647.

©2021 The Authors; Published by the American Association for Cancer Research.

Figures

Figure 1.

Schematic representation of the final pharmacokinetics–tumor size model. Positive (+) and negative (−) associations are indicated. Ac, avelumab concentration in the central compartment; Ap, avelumab concentration in peripheral compartment; CL, avelumab clearance; KD, tumor shrinkage rate; KD,0, baseline tumor shrinkage rate; KG, tumor growth rate; Q, intercompartmental exchange; t, time; Vc, avelumab central volume of distribution; Vp, avelumab peripheral volume of distribution; λ, exponential effect decay constant.

Figure 2.

Individual predictions (purple lines) and observations (yellow dots) of TS over time for five patients representing different TS profiles.

Figure 3.

Spider plots of percentage tumor change from baseline versus time after the first tumor assessment in patient subgroups according to FcγRII/III genotype (FcγRIIA131H and FcγRIIIA158V) in patients with mMCC. Blue lines show polynomial regression curves.

Figure 4.

Boxplots of random parameters according to FcγRII/III genotype (FcγRIIA131H and FcγRIIIA158V) in patients with mMCC. CL, avelumab clearance; KD, tumor shrinkage rate; KG, tumor growth rate; λ, decay constant of KD.

Figure 5.

Illustration of drug exposure–TS bidirectional effects via simulation of drug concentration (top panels) and percentage change of tumor size from baseline (bottom panels) over 12 weeks since first dose in 9 simulated individuals differing only in initial CL and TS. Initial CL and TS values correspond to median, and 5th and 95th percentiles of individual Bayes estimates from the final model. CL, avelumab clearance; TS, tumor size.