(M-012) Population pharmacokinetics and pharmacodynamics analysis of PF-07328948, a branched-chain ketoacid dehydrogenase kinase inhibitor, in healthy participants

Spinel Karas – Pfizer Inc; Zhenhong Li – Pfizer Inc; Bimal Malhotra – Pfizer Inc; Ruolun Qiu – Pfizer Inc

Objectives: PF-07328948, a branched-chain ketoacid dehydrogenase kinase (BDK) inhibitor/degrader, is in development for the treatment of heart failure. Emerging evidence suggests that heart failure is associated with impaired branched-chain amino acid (BCAA) and branched-chain ketoacid (BCKA) catabolism.[1-4] Targeting the BDK enzyme through inhibition and/or degradation may restore BCAA/BCKA catabolism and improve heart failure outcomes. The objective of this study was to develop a population pharmacokinetics (PK)/pharmacodynamics (PD) model to characterize the PK of PF-07328948 and its relationship with BCAA and BCKA responses in a healthy population.

Methods: Population PK/PD analyses were conducted using nonlinear mixed-effect modeling in Monolix (2021R1, Lixoft, Antony, France). Data from two Phase 1 studies (NCT05654181; NCT05807490) in 63 healthy participants were used to characterize PK (plasma PF-07328948) and PD (plasma BCAA/BCKA). Participants received a single dose of PF-07328948 (10–1,500 mg) or placebo after fasting or a high-protein meal, or multiple doses of PF-07328948 (100–750 mg once- or twice-daily for 14 days) or placebo with a standard meal. PK was characterized by a two-compartment model with first-order absorption, and elimination followed a power-law relationship to account for the greater-than-dose-proportional increase in exposure at higher doses. For PD, a semi mechanistic model developed from preclinical data was used. The model described dual effects of PF-07328948: inhibiting BDK enzyme activity and accelerating BDK enzyme degradation, which enhances BCAA/BCKA catabolism and lowers BCAA/BCKA levels. Interindividual variability was estimated using a log-normal distribution model for PK/PD parameters. Potential covariates were selected based on COnditional Sampling use for Stepwise Approach based on Correlation tests (COSSAC),[5] followed by covariate-effect size and precision evaluation. The final model was assessed and validated using goodness-of-fit plots, prediction-corrected visual predictive checks (VPCs), and nonparametric bootstrap approach.

Results: The final PK/PD model adequately described PF-07328948 PK/PD. Interindividual variability was included on clearance, volume of distribution of central compartment, baseline BCAA and BCKA (BCAA0 and BCKA0), and BCKA degradation rate constant, with shrinkage < 15%. Final parameters were generally well estimated (relative standard error < 34%). Sex was a significant covariate on BCAA0 and BCKA0, with males showing 35% higher BCAA0 and 42% higher BCKA0 than females. Goodness-of-fit plots and dose-stratified VPCs showed no major misspecification or bias. Nonparametric bootstrap confirmed model robustness, with final estimates consistent with bootstrap medians.

Conclusions: The final model offers a robust and well-characterized framework for PF-07328948 PK/PD, demonstrating its suitability for predicting exposure-BCAA/BCKA response relationships to guide dose selection in future clinical trials.

Citations: 1. Uddin GM et al. Cardiovasc Diabetol 2019; 18: 86.
2. Sun H et al. Circulation 2016; 133: 2038–2049.
3. Previs MJ et al. Circ Heart Fail 2022; 15: e009521.
4. Hahn VS et al. Circulation 2023; 147: 1147–1161.
5. Ayral G et al. CPT Pharmacometrics Syst Pharmacol 2021; 10: 318–329.

Pfizer’s generative artificial intelligence tool MAIA was used to assist production of the abstract first draft. Authors reviewed/edited and take responsibility for the content.

Keywords: BDK inhibitor, pharmacokinetics, pharmacodynamics