Population pharmacokinetics of KAF156/LUM-SDF and exposure-response analysis for the probability of cure for patients with uncomplicated malaria
Giulia Lestini, Ivan Demin, Jing Yu.
Pharmacometrics, Analytics GDD, Novartis
Objectives: To describe the population pharmacokinetics (PK) of KAF156 (KAF) and the solid dispersion formulation of lumefantrine (LUM-SDF) when given in combination in fasted adult/adolescent and pediatric malaria patients (≥ 2 years). To describe exposure-response (ER) relationship for probability of PCR-corrected (PCR-cor.) ACPR29 (Adequate Clinical and Parasitological Response measured at day 29) and to evaluate the effect of different covariates on PK and on probability of PCR-cor. ACPR29. To predict PK of both KAF and lumefantrine as well as possible trial outcomes for PCR-cor. ACPR29 for candidate phase 3 dosing regimens according to the planned phase 3 design.
Methods: Separate population PK (popPK) models were built for KAF and LUM-SDF based on data from the combination phase 2 study NCT03167242 using the nonlinear mixed-effects approach and were implemented in MonolixSuite 2019R2. Only body weight and age were explored as covariates in the popPK modeling.
A logistic regression model describing the ER relationship between KAF, lumefantrine exposures and the probability of cure (PCR-cor. ACPR29) was built using simulated exposure metrics and efficacy data from the above-mentioned phase 2 study, as well as from two monotherapy studies of KAF and conventional formulation of lumefantrine (Coartem).
For each candidate phase 3 dosing regimen, N=1000 trial replicates were simulated according to the planned phase 3 design. Uncertainty in popPK parameters (estimated) and ER model parameters (bootstrapped) was used for simulations. Baseline covariates were re-sampled using available patient demographics from the combination phase 2 study. Individual simulated PK metrics were used for further ER simulations to derive individual probabilities of cure. ER simulations were performed in R.
The individual probabilities of cure were used to simulate individual outcome for each patient (cure or failure) and then were summarized by trial replicate to calculate an outcome of each simulated trial. The simulated trial outcomes were summarized and the mean together with 5th and 95th percentiles of the responder rate were calculated for each simulated dosing regimen.
Results: The final popPK models shared the same structure of a two-compartment model with first-order absorption, lag time to absorption and first-order elimination. For lumefantrine, due to observed nonlinearity of PK between 480 mg and 960 mg doses and higher than expected exposure after day 1, bioavailability (F) was modeled as a function that allows estimation of those differences in F between doses and days (Day 2, 3 versus Day 1). Body weight was retained as a covariate for apparent clearance and volume parameters, indicating increasing apparent clearances and volumes with increasing body weight.
The final ER model for PCR-cor. ACPR29 was a logistic regression model with an intercept, log transformed KAF day 7 concentration (Cday7), log-transformed lumefantrine Cday7, age category and log-transformed baseline parasitemia. Both KAF Cday7 (p-value <0.05) and lumefantrine Cday7 (p-value <0.001) were statistically significant, which indicated that higher drug exposures provide a higher probability of response. Age effect of children of 2 to 11 years was also statistically significant indicating lower probability of response in the younger age category.
Predictions were performed for several candidate phase 3 dosing regimens. Among considered dosing regimens, KAF156 400 mg and LUM-SDF 480 mg 3-day under fed condition when fully or partially fed (assuming 2-fold increase of lumefantrine exposure due to food intake) matched best the target dosing regimen KAF156 400 mg and LUM 960 mg once daily for 3 days in fasted patients (including a subset of children 2 to <12 years old) which showed desirable efficacy in phase 2 study.
Conclusions: PopPK and ER models can be used to simulate PK and efficacy of untested dosing regimens to further strengthen KAF/LUM phase 3 dose selection. Based on ER analysis both KAF and LUM-SDF exposures contributed to the probability of cure based on PCR cor. ACPR29. When assuming 2-fold increase of lumefantrine exposure due to food intake, KAF156 400 mg and LUM-SDF 480 mg once daily for 3 days in fed patients matched efficacy of the target dosing regimen KAF156 400 mg and LUM-SDF 960 mg once daily for 3 days in fasted patients.
