Leishmania blood parasite dynamics during and after treatment of visceral leishmaniasis in Eastern Africa: A pharmacokinetic-pharmacodynamic model

Abstract

Background: With the current treatment options for visceral leishmaniasis (VL), recrudescence of the parasite is seen in a proportion of patients. Understanding parasite dynamics is crucial to improving treatment efficacy and predicting patient relapse in cases of VL. This study aimed to characterize the kinetics of circulating Leishmania parasites in the blood, during and after different antileishmanial therapies, and to find predictors for clinical relapse of disease.

Methods: Data from three clinical trials, in which Eastern African VL patients received various antileishmanial regimens, were combined in this study. Leishmania kinetoplast DNA was quantified in whole blood with real-time quantitative PCR (qPCR) before, during, and up to six months after treatment. An integrated population pharmacokinetic-pharmacodynamic model was developed using non-linear mixed effects modelling.

Results: Parasite proliferation was best described by an exponential growth model, with an in vivo parasite doubling time of 7.8 days (RSE 12%). Parasite killing by fexinidazole, liposomal amphotericin B, sodium stibogluconate, and miltefosine was best described by linear models directly relating drug concentrations to the parasite elimination rate. After treatment, parasite growth was assumed to be suppressed by the host immune system, described by an Emax model driven by the time after treatment. No predictors for the high variability in onset and magnitude of the immune response could be identified. Model-based individual predictions of blood parasite load on Day 28 and Day 56 after start of treatment were predictive for clinical relapse of disease.

Conclusion: This semi-mechanistic pharmacokinetic-pharmacodynamic model adequately captured the blood parasite dynamics during and after treatment, and revealed that high blood parasite loads on Day 28 and Day 56 after start of treatment are an early indication for VL relapse, which could be a useful biomarker to assess treatment efficacy of a treatment regimen in a clinical trial setting.

Fig 1. Blood parasite loads in Eastern African visceral leishmaniasis patients during treatment and early follow-up.

Depicted are median (IQR) observed blood parasite loads coloured by treatment outcome (cured patients in red and relapsed patients in blue), stratified by treatment regimen. AmB+SSG10D: 10 mg/kg amphotericin B (day 1) + 20 mg/kg/day SSG (day 2 to 11); AmB+MF10D: 10 mg/kg amphotericin B (day 1) + 100 mg/day miltefosine (day 2 to 11); MFC28D: miltefosine conventional dose (100 mg/day) (28 days); MFA28D: miltefosine allometric dose (28 days); Fexi10D: 1800 mg/day fexinidazole (4 days) + 1200 mg/day fexinidazole (6 days). Gray dashed lines represent the end of treatment.

Fig 2. Schematic overview of the final pharmacokinetic-pharmacodynamic model, exemplified with the pharmacokinetic model for fexinidazole and its active metabolites M1 and M2.

In the parasite model, k_GR is the parasite replication rate, k_drug is the drug-driven parasite clearance rate, λ_drug the drug-specific linear effect, and C_drug the drug concentration of either miltefosine, the sum of M1 and M2 for fexinidazole, amphotericin B, or SSG. k_IMM is the immune-driven parasite clearance, I_max the maximum inhibition by the immune response, IT₅₀ the time at half-maximum inhibition, and γ the steepness of the time-effect relationship, which was empirically fixed to 5. In the fexinidazole pharmacokinetic model, CL is the clearance of fexinidazole, M1 or M2, and V₂, V₃, and V₄ the volume of distribution of fexinidazole, M1, and M2, respectively.

Fig 3. Prediction-corrected visual predictive checks for the final pharmacokinetic-pharmacodynamic blood Leishmania parasite load model until Day 56 after start of treatment.

Solid lines: median of the observed values; dashed lines: the 10^th and 90^th percentiles of the observed values; dark and light blue areas: the 90% confidence intervals of the simulated median and percentiles, based on 1000 simulations.

Fig 4

Simulations of typical pharmacokinetic profiles of patients receiving A) 1800 mg/day fexinidazole (4 days) + 1200 mg/day fexinidazole; B) 10 mg/kg amphotericin B (day 1) + 20 mg/kg/day SSG (day 2 to 11); C) 10 mg/kg amphotericin B (day 1) + 100 mg/day miltefosine (day 2 to 11); or D) 100 mg/day miltefosine (28 days).

Fig 5

A (left plot): Simulation of drug effects of different VL therapies of typical patients receiving 1) 10 mg/kg amphotericin B (day 1) + 20 mg/kg/day SSG (day 2 to 11) (blue curve), 2) 10 mg/kg amphotericin B (day 1) + 100 mg/day miltefosine (day 2 to 11) (grey curve), 3) 100 mg/day miltefosine (28 days) (red curve), or 4) 1800 mg/day fexinidazole (4 days) + 1200 mg/day fexinidazole (6 days) (yellow curve). No immune response after the end of treatment was observed. Other parameters were fixed to the population values. B (right plot): Simulation of typical patients receiving 150 mg/day miltefosine for 28 days. Patients have an IT₅₀ of 1000 h (blue curve), 5000 h (grey curve), and 100,000 h (red curve). Other parameters were fixed to the population values.

Fig 6. ROC curves of blood parasite load as predictor of clinical relapse on Day 10, 28, and 56 after start of treatment.

AUC represents the integrated area under the ROC curve. Green line: day 10 (AUC 0.64), red line: day 28 (AUC 0.82), blue line: day 56 (AUC 0.87). Abbreviations: AUC, area under the curve; ROC, receiver operating characteristic.

Fig 7. Parasite AUC_d0-28 and AUC_d0-56 and parasite load on Day 28 and Day 56 versus clinical outcome.