PK-PD modeling of doxycycline, azithromycin, and their combination in treating severe scrub typhus using a general pharmacodynamic interaction (GPDI) model.

Objectives: Severe scrub typhus, a rickettsial disease with multi-organ failure, is a highly fatal infectious disease. IV doxycycline, azithromycin, or their combination are commonly used in the treatment, but their comparative efficacy needs to be better established. This study aims to describe the PK of these two drugs and their PD as monotherapy and, in combination, using GPDI, which could help optimize drug regimens. Methods: 744 patients over 15 years old with severe scrub typhus were recruited. Plasma concentrations of azithromycin and doxycycline, as well as bacterial copy numbers, were measured. Blood samples for azithromycin and doxycycline levels were taken to match the drugs' long half-life. Bacterial DNA markers were on days 1, 3, 7, 10, and 14. A population PK model was developed to describe the effect of individual monotherapies and combination therapy in causing microbiological cures. The general pharmacodynamic interaction (GPDI) model was employed to describe the combinatorial pharmacodynamics of both these drugs1. Results: There were 2462, 2420, and 2029 measurements of doxycycline, azithromycin, and bacterial copy numbers, respectively, from 744 participants. One and two-compartment models with first-order elimination were used to describe the PK of doxycycline and azithromycin, respectively. Inter-individual variability (IIV) was incorporated into all PK parameters using an exponential relationship. Residual unexplained variability (RUV) was described using a proportional model for both drugs. Clearances (CL) and volumes of distributions (Vd) were allometrically scaled with a fixed exponent of 0.75 and 1, respectively. eGFR was added as a covariate on clearance for both the drugs, and sex was added as a covariate on the Vd of doxycycline. The pharmacodynamic models using the measured bacterial DNA copy numbers were developed initially for the Doxycycline only and Azithromycin groups individually. Various PD models, such as the direct effect, effect compartment, and turnover models, were explored, with a direct response model accurately predicting data for both azithromycin and doxycycline-only groups. A generalized pharmacodynamic interaction relationship was employed following individual model development to create the combined PD model. A simple additive model among Bliss independence and the highest single agent methods provided the best visual predictive checks with the least RSE on the estimated interaction parameters. Azithromycin positively influences the Emax of doxycycline by a factor of 1.15, whereas doxycycline diminishes the Emax of azithromycin by a factor of 0.346. Neutrophil %, eGFR, and serum albumin concentration were significant covariates affecting baseline bacterial quantity.Conclusion: The PD interaction of doxycycline and azithromycin could be well described by a GPDI model. Along with the relevant covariates, this could help with simulations to optimise treatment in severe scrub typhus.Citations: [1] A general pharmacodynamic interaction model identifies perpetrators and victims in drug interactions Nat Commun. 2017 Dec 14;8(1):2129. doi: 10.1038/s41467-017-01929-y. Sebastian G Wicha, Chunli Chen, Oskar Clewe, Ulrika S H Simonsson