Model-informed early clinical development of BI-1808, a novel monoclonal antibody to tumor necrosis factor receptor 2
Anna M. Mc Laughlin (1), Julie M. Janssen (1), Marie Borggren, (2), Ingrid Teige (2), Johan Wallin (2)
(1) Pharmetheus AB, Uppsala, Sweden (2) BioInvent International AB, Lund, Sweden
Objectives: BI-1808 is a novel immunotherapy targeting tumor necrosis factor receptor 2 (TNFR2), which is abundantly expressed on tumor-resident regulatory T cells (Tregs) [1]. It is a monoclonal antibody interfering with the binding of tumor necrosis factor alpha (TNF-α) to TNFR2, resulting in a reduction of intratumoral Tregs and the simultaneous expansion of intratumoral effector T cells.
BI-1808 has shown promising results in preclinical in-vivo studies and is currently in early clinical development. In the ongoing Phase 1/2a dose-escalation trial KEYNOTE-D20, a dose range of 25 mg – 1000 mg BI-1808 with or without pembrolizumab every three weeks (q3w) administered via i.v. infusion has so far been explored in 39 subjects with advanced malignancies or T-cell lymphoma. Of the 39 subjects, 29 received BI-1808 alone and 10 received 225 mg BI-1808 with pembrolizumab. No grade 3/4 adverse events or dose-limiting toxicities have been observed so far.
The observed favorable safety profile across the wide dose range does not suggest obvious doses for further exploration in a dose expansion trial. Consequently, the purpose of this work was to characterize the population PK, the receptor occupancy (RO), and concentrations of the target engagement biomarker soluble TNFR2 (sTNFR2) at different BI-1808 doses, to support the dose selection for a dose expansion trial.
Methods: PK and RO data were available for 32 subjects (25 mg: n=3, 75 mg: n=5, 225 mg without pembrolizumab: n=3, 225 mg with pembrolizumab: n=9, 675 mg: n=3, 1000 mg: n=9) and sTNFR2 data was available for 20 of the 32 subjects. Model building proceeded in a sequential manner: First, a joint PK-RO model was developed to simultaneously characterize the BI-1808 concentrations and RO. The PK was described by a two-compartment model with parallel linear and non-linear CL and the RO by a direct effect model. Both models were linked through a joint KM parameter, which was incorporated into the non-linear CL equation of the PK submodel and the Emax equation of the RO submodel. Potential exploratory covariate-parameter relationships were assessed based on interindividual variability (IIV) vs. covariate plots. Next, the joint PK-RO model was extended with an indirect response model describing the sTNFR2 data. In this model, all three analytes were characterized simultaneously with individually predicted BI-1808 concentrations impacting the sTNFR2 concentrations through an inhibitory Emax model on the sTNFR2 kout.
To inform dose selection for a dose expansion trial, typical profiles were simulated for PK, RO, and sTNFR2 using the final sTNFR2 model. Profiles were simulated at different doses (25 mg, 75 mg, 150 mg, 200 mg, 250 mg, 300 mg, 350 mg, 400 mg, 450 mg, 500 mg, 675 mg, and 1000 mg) and different dosing frequencies (q3w, every fourth week (q4w) and every sixth week (q6w)).
Results: The developed joint PK-RO model described the observed BI-1808 concentration-time and RO data well. Body weight was implemented on clearance and volume parameters using allometric scaling with estimated exponents (0.437 for CL, Q, Vmax, and 0.495 for VC and VP). No other significant covariate, including co-treatment with pembrolizumab, was identified. The estimates for the linear CL, Vmax, and KM were 427 mL/d, 7.81 mg/d, and 223 ng/mL. IIV was estimated on CL (51.2% CV), VC (23.1% CV), Vmax (26.4% CV), and KM (72.1% CV). The residual unexplained variability was 42.8% CV for the BI-1808 PK and 4.05% (additive) for the RO. Upon inclusion of the sTNFR2 data into the model, the legacy parameters remained stable. The estimate for the maximum inhibitory effect of BI-1808 serum concentrations on the sTNFR2 kout was 97.7%.
In line with the observed data, steady-state simulations indicated RO close to 100% throughout a dosing interval for doses of ≥450 mg q3w. The mean simulated RO within a dosing interval was ≥80% for doses ≥250 mg, and RO exceeded 50% for a full dosing interval for doses ≥350 mg. In line with the simulated RO, maximum simulated concentrations of sTNFR2 only increased minimally from 350 mg q3w up to 1000 mg q3w.
Conclusions: A simultaneous model of the BI-1808 PK, RO, and sTNFR2 successfully explained the observed profiles across a wide dose range. The performed simulations provide insights into the expected PK, RO, and sTNFR2 levels across potential additional dose levels and dosing frequencies and will support the selection of doses for further exploration.
References:
[1] É. S. Vanamee and D. L. Faustman, “TNFR2: A Novel Target for Cancer Immunotherapy,” Trends in Molecular Medicine, vol. 23, no. 11, pp. 1037–1046, Nov. 2017, doi: 10.1016/j.molmed.2017.09.007.
