Population Pharmacokinetic and Exposure-Response Modelling Analyses to Support Valemetostat Dose Recommendations in Patients With Relapsed/Refractory (R/R) Peripheral T-cell Lymphoma (PTCL)

Objectives: Valemetostat tosylate (valemetostat) is an oral, dual inhibitor of enhancer of zeste homolog (EZH)2 and EZH1 in clinical development for the treatment of non-Hodgkin lymphomas (NHLs). The objectives of this study were to (1) develop a population pharmacokinetic (PopPK) model, (2) characterize exposure-response (ER) relationships for efficacy and safety, and (3) assess the benefit–risk profile of valemetostat 200 mg once daily (QD) dose in patients with R/R PTCL. Methods: Simultaneous PopPK analysis of total and unbound valemetostat was performed using pooled data from patients with NHL and healthy participants in six studies: DS3201-A-J101, DS3201-A-J107, DS3201-A-J109, DS3201-A-J201, DS3201-A-U106, and DS3201-A-U202. Covariate effects were assessed with a full model approach and prespecified parameter–covariate relations were estimated. Post hoc model-estimated pharmacokinetics (PK) metrics were used in subsequent ER analyses. ER relationships were assessed using logistic regression models in a Bayesian framework for efficacy (objective response rate [ORR]) in patients with PTCL from DS3201-A-U202 study, and safety endpoints (any Grade ≥ 3 treatment-emergent adverse events [TEAEs], Grade ≥ 3 hematologic TEAEs [including thrombocytopenia, neutropenia, and anemia]; dose reduction and dose interruption due to TEAEs) in patients with PTCL or adult T-cell leukemia/lymphoma from DS3201-A-J101, DS3201-A-J201, and DS3201-A-U202. Unbound average concentration up to event (Cavgtte) was utilized as an exposure metric for all endpoints. Results: The pooled PopPK population included 270 patients with NHL and 72 healthy participants. Valemetostat PK was well described by a three-compartment model with sequential zero-/first-order absorption and a saturable binding submodel in the central compartment to characterize unbound valemetostat. Valemetostat total exposure was higher in patients with NHL compared with healthy participants, likely due to valemetostat binding to alpha-1-acid glycoprotein (AAG), whose concentration was elevated in patients with NHL. In contrast, AAG had a negligible effect on unbound exposure, which is likely correlated with efficacy and safety outcomes. The efficacy and safety analysis populations included 119 and 251 patients, respectively. A flat ER efficacy relationship was observed, with the estimated odds ratio of ORR being 1.10 (95% credible interval: 0.330, 4.05) for every 12.2 ng/mL increase of Cavgtte. There were positive associations between valemetostat unbound exposure metrics and the probability of any Grade ≥ 3 TEAE, Grade ≥ 3 thrombocytopenia, Grade ≥ 3 anemia, and dose interruption due to TEAEs. There was no obvious ER trend in the probability of Grade ≥ 3 neutropenia or dose reduction due to TEAEs. Conclusions: The PopPK model well described the PK profiles of total and unbound valemetostat. The ER data and benefit–risk profiles supported the recommended dose of 200 mg QD for patients with R/R PTCL.Citations: No citations