PII-212 - A QUANTITATIVE SYSTEMS PHARMACOLOGY (QSP) MODEL FOR PREDICTING EFFICACY FOR A COMBINATION OF AN ANTIBODY DRUG CONJUGATE (ADC) AND A CHECKPOINT INHIBITOR (CPI) IN PHASE II/III CLINICAL TRIALS

V. Prabhakar¹, R. Sing², R. Dutta², D. Mehra², A. Sareen³, B. Paleja², M. Channavazzala², S. Pallikonda Chakravarthy², K. Thiagarajan⁴; ¹Vantage Research Inc, ²Vantage Research, ³Vantage Research, Lewes, Delaware, USA, ⁴Vantage Research Inc, Lewes, Delaware, USA.

Background: Combining ADCs with Checkpoint Inhibitors could potentially be an effective strategy to improve patient outcomes. ADCs activate immunogenic cell death [1], which could lead to a synergistic combination with immunotherapies like CPIs. To identify optimal combinations of ADCs and CPIs, we have developed a QSP Model platform (calibrated to both CPI and ADC monotherapies), with the ability to predict phase II/III responses for the CPI + ADC combination therapy.

Methods: In this work, we have developed a platform model that includes an mPBPK model [2] for systemic PK and cellular-level drug disposition, as well as a module for tumor growth inhibition (TGI). The model incorporates dosing and PK for both ADC and CPI therapies. This model incorporates a combination therapy based response by simulating TGI as a function of both ADC cytotoxicity and CTL infiltration caused by the CPI.

Specifically, we chose an anti-PD 1 checkpoint inhibitor (Pembrolizumab) in combination with Nectin-4 targeting ADC (Enfortumab Vedotin (‘EV’)) against advanced/metastatic Urothelial Cancer (mUC) [4,5]. First, we calibrate to each monotherapy [4,6] for 1st line therapy. Further, we quantified the model using compiled public data to inform physiological constraints such as T cell numbers and antigen expression levels.We then validated model predictions against available combination anti-PD1 and ADC data [5].

Results: The model predicted a range of Objective Response Rate (ORR) for the combination based on estimated CTL infiltration rate induced by ADC and CPI co-administration. We also simulated alternate clinical dosing strategies to assess and identify optimal trial design. Finally, we ran population subgroups analyses based on PDL1 and Nectin-4 expression for patient enrichment.

Conclusion: Because of the ability to simulate drug MoA and disease pathways, QSP models have a strong application in combination therapy. By calibrating to available public (and if available sponsor) data, such models are able to run multiple scenarios and determine the most optimal combinations. Further applications include patient enrichment and alternative dosing strategies. Especially in therapy classes such as ADCs and checkpoints in which plentiful monotherapy data exists, QSP models provide a powerful tool to prioritize combinations and design optimal trials.