Prediction of PD-L1 receptor occupancy in the tumor with PBPK/RO model of PD-L1 inhibitors
Antonina Nikitich, Oleg Demin Jr
InSysBio, Moscow, Russia
Objectives: Measurements of receptor occupancy (RO) are believed to provide the necessary information on the pharmacodynamics of immune checkpoint inhibitors. Testing for target receptor occupancy in the blood can be assessed in clinical studies. However, it is difficult to sample tumor tissues in patients to assess RO. The objectives of the present study were:
- To develop physiologically based pharmacokinetic (PBPK) and RO model of anti-PD-L1 therapeutic antibodies;
- To describe clinical pharmacokinetic (PK) and RO data available for the anti-PD-L1 antibodies;
- To predict PD-L1 RO in the tumor for various regimens and doses of PD-L1 inhibitors
Methods: General structure of developed PBPK/RO model is similar to the published minimal PBPK models, but there are some differences. Key features of developed model: (1) surface binding of anti-PD-L1 antibodies with membrane bound receptor, i.e., number of target receptor per cell, number of cell expressing target receptor and the valency of therapeutic antibodies were taken into account; (2) internalization of target receptor bound with anti-PD-L1 antibody; (3) uptake of therapeutic antibodies by endothelial cells, their binding with FcRn and recycling, degradation of free anti-PD-L1 antibodies; (4) competition of anti-PD-L1 antibodies with endogenous IgG. Physiological parameters were taken from published literature; other parameters were identified on the basis of in vitro and in vivo data. Clinical data on anti-PD-L1 monoclonal antibodies were used for model validation.
Results: Developed PBPK model was tested on a set of four anti-PD-L1 monoclonal antibodies: atezolizumab, durvalumab, avelumab, and BMS-93655. To this end, clinical findings were compared to model simulations performed under similar conditions (dose and schedule).The model adequately described the PK profile of all tested drugs, as well as target RO in the blood, without any additional parameter fitting. For instance, the predicted mean PD-L1 occupancy on the T cells in the blood at the end of the cycle of 1 mg/kg of avelumab Q2W was 75%, whereas clinical data showed 76% . Simulated trough PDL1 occupancy in the tumor for atezolizumab 840 mg Q2W, durvalumab 10 mg/kg Q2W, 800 mg Q2W avelumab were 99.75%, 99.9%, 99.07%, respectively. Less frequent administration of higher doses of atezolizumab showed almost complete trough RO of PD-L1: 99.77% and 99.78% after the treatment with 1200 mg Q3W and 1640 mg Q4W, respectively. The results can be explained by similar Kd values (0.4nM for atezolizumab, 0.02 nM for durvalumab, 0.7 nM for avelumab)and applied doses of these three approved anti-PD-L1 therapeutic antibodies.
Conclusions: PBPK/RO model for anti-PDL1 monoclonal antibodies was developed and successfully validated against clinical data. Similar doses of three approved anti-PD-L1 monoclonal antibodies showed almost equal PD-L1 RO in the tumor. PD-L1 RO in the tumor is basically determined by binding properties of therapeutic antibodies due to the similar structure and PK of these antibodies. This model can be used as a tool for the prediction of PD-L1 RO in the tumor for the other therapeutic antibodies targeting PD-L1 and to optimize their doses and regimens.
 Heery et al. Lancet Oncol. 2017 May;18(5):587-598
 Tiffany K. Ricks, FDA Pharmacology review of Atezolizumab
 EMA Assessment report of Imfinzi
 FDA Pharmacology review of Avelumab