Magnus Åstrand, Mats Någård, David W. Boulton, Bengt Hamrén
NULL
Objectives: Sodium zirconium cyclosilicate (SZC, LOKELMA™) is an orally administered non-absorbed treatment for hyperkalaemia, acting locally in the gastrointestinal tract by exchange of potassium for hydrogen and sodium. An exploratory modelling and simulation exercise has been undertaken to evaluate the serum potassium concentration (S-K) lowering effect of SZC in patients with hyperkalemia (S-K value >5 mmol/L). The objective was to characterize the relationship between S-K lowering response and SZC dose for the correction and maintenance treatment phases across multiple studies. The modelling included data from 3 Phase 2/3 studies and comprised of the following main steps:
i) Build a longitudinal pharmacodynamic model to describe the time course of S-K using all available data on S-K and administered doses of SZC
ii) Describe the S-K lowering dose response of SZC for correction (three time daily [TID]) and maintenance phases (once daily [OD])
iii) Identify and quantify the influence of baseline covariates on both the SZC-induced S-K lowering effect and placebo response
Methods: A longitudinal model describing the S-K vs SZC dose and time was built using data from a total of 1101 patients with hyperkalemia. The mean(SD) S-K at baseline was 5.38(0.39) mmol/L. The S-K over time was described by an indirect response model. A virtual pharmacokinetic-pharmacodynamic model (PK-PD) modelling approach was used by introducing a drug exposure component although there was no data on drug exposure measured. The level and duration of this virtual exposure was determined by the amount and frequency of SZC doses and an elimination rate parameter in line with a more standard PK-PD model. Introducing the virtual exposure enabled describing the dose response for both TID and OD dosing during the correction and maintenance phase respectively. Thus, the modelling included a virtual SZC exposure component and the S-K lowering effect was then described using the virtual exposure via a sigmoid exposure-response model. The influence of baseline covariates was explored for both the SZC-induced S-K lowering (concentration at half maximum effect [EC50]), the placebo response and the rate of change of S-K.
Results: The developed model was overall consistent with the included S-K data, as demonstrated by model diagnostics, and adequate for the purpose of predicting S-K changes from baseline. A sigmoid Emax exposure response was used in the modelling. The modelling found a high value for the EC50 parameter and a hill coefficient of 1.3. Within the range of SZC doses studied, the predicted dose-response was therefore close to linear. The predicted (with 95% confidence interval [CI]) placebo-adjusted S-K change from baseline after 48 hours TID correction phase treatment was -0.26 (-0.29 to -0.23) mmol/L for 5 g SZC, and -0.53 (-0.59 to -0.48) mmol/L for the 10 g SZC dose. The predicted (with 95% CI) placebo adjusted S-K change from baseline for 28 days OD maintenance treatment was -0.25 (-0.28 to -0.22) mmol/L for 5 g SZC, -0.52 (-0.58 to -0.45) mmol/L for 10 g SZC, and -0.75 (-0.85 to -0.65) mmol/L for the 15 g SZC dose. The forward and backward covariate selection step identified a total of 9 covariates, 5 for EC50, 3 for the placebo effect and 1 for the S-K dynamics (Kout). Greater treatment response was associated with high S-K baseline, older patients, lower body weight and lower eGFR. Greater treatment response was also predicted for Black or African American compared to White.
Conclusions: The dose response for both TID correction at doses 0 to 10g, and OD maintenance treatment at doses 0 to 15g is close to linear. A total of 9 covariates were identified. However, the placebo corrected change from baseline following 10g TID dosing for 48 hours for all covariates were within the range -0.75 to -0.39 mmol/L, supporting using the same SZC dose recommendation for all patients in the pooled study population.
Reference: PAGE 27 (2018) Abstr 8674 [www.page-meeting.org/?abstract=8674]
Poster: Drug/Disease Modelling - Other Topics