Population Pharmacokinetic/Pharmacodynamic Modelling from the Phase I/II COMPOSER Trial to Predict Lowering of Lactate Dehydrogenase in Crovalimab-Treated Patients with Paroxysmal Nocturnal Hemoglobinuria

Introduction

Paroxysmal nocturnal hemoglobinuria (PNH) is a hematopoietic stem cell disorder characterized by intravascular hemolytic anemia and hemoglobinuria. Lactate dehydrogenase (LDH) is a cytoplasmic enzyme that is abundant in red blood cells. Serum LDH level is used as a biomarker for intravascular hemolysis to monitor response to PNH treatment. Crovalimab is a novel antihuman C5 antibody engineered with Sequential Monoclonal Antibody Recycling Technology (SMART-Ig; Fukuzawa et al, Sci Rep. 2017) that is being evaluated as a therapy for PNH in the 4-part, dose-optimization Phase I/II COMPOSER trial (NCT03157635; Röth et al, Blood. 2020), which includes healthy volunteers and patients with PNH who received or did not receive prior C5 inhibition therapy with eculizumab. In this study, COMPOSER data were used to obtain a dose-concentration effect relationship between crovalimab and LDH through development of a pharmacokinetic (PK)-LDH model.

Objectives

To describe the LDH time course in patients with PNH and to confirm the adequate LDH-lowering effect of the optimized dose and regimen of crovalimab.

Study Design and Methods

At the clinical data cutoff date of September 4, 2019, data from 15 treatment-naive patients with PNH were used to develop the population PK-LDH model. Observations collected in patients switching treatment from eculizumab to crovalimab (n = 26) were not used to build the PK-LDH model but were used as an external validation tool to evaluate the predictive performance of the model at steady state.

A sequential approach was followed (Zhang et al, J Pharmacokinet Pharmacodynam. 2003;30:387-404; Zhang et al, J Pharmacokinet Pharmacodynam. 2003;30:405-416): the individual PK parameters of crovalimab were derived from a population PK model by an empirical Bayesian analysis, and subsequently, the PK-LDH response relationship was modeled with an indirect response model in which LDH synthesis was inhibited by crovalimab concentrations. The predictive performance of the population PK-LDH model was evaluated using a prediction-corrected visual predictive check (pcVPC). The population PK-LDH analysis and all simulations were performed using NONMEM 7.4.0 software. All data sets were analyzed using SAS 9.4.

Results

For a typical patient, the maximal LDH-lowering effect was estimated to be 80% and the concentration of crovalimab leading to 90% of maximal reduction of LDH synthesis was estimated to be 34.9 μg/mL (90% CI, 15.9-67.5 μg/mL). The half-life for LDH was estimated to be 3 days and steady state for the LDH-lowering effect was predicted to be reached 15 days after first drug intake. The pcVPC in treatment-naive patients showed that the PK-LDH model had good predictive qualities (Figure 1). An external pcVPC was performed in patients who had switched from eculizumab to crovalimab with good predictive qualities, supporting the use of the model for simulation.

Using the optimized crovalimab dose and regimen from Part 4, model-based simulations of the LDH time course showed that > 95% of patients were predicted to have LDH levels of ≤ 1.5 × upper limit of normal (ULN) (Figure 2). A sensitivity analysis, to assess the impact of uncertainty around the estimated concentration of crovalimab leading to 90% of maximal reduction of LDH synthesis, found a moderate impact on the predicted LDH-lowering effect of crovalimab, with a small decrease in the percentage of patients predicted to reach an LDH level of < 1.5 × ULN.

Conclusions

Simulations show a rapid and sustained decrease in LDH levels to < 1.5 × ULN in over 95% of treatment-naive patients with PNH who receive the optimized dose and regimen of crovalimab. The confirmation of the LDH-lowering effect of crovalimab supports further clinical study of the optimized regimen, which is being evaluated in the Phase III studies COMMODORE 1 (NCT04432584) and COMMODORE 2 (NCT04434092).

View largeDownload slide

View largeDownload slide

Close modal