Purpose: Ivermectin (IVM) is a lipophilic BCS-II compound (molecular weight=875 g/mole, LogP=3.22, intrinsic solubility=700 ug/L). IVM is used as antiparasitic drug in both humans and animals. IVM is known to have a half-life of 12-56 hours in humans. Strongyloidiasis is a chronic parasitic infection of humans caused by Strongyloides stercoralis, with an estimated 30-100 million people infected worldwide. Infection may be severe and even life-threatening in cases of immunodeficiency. Patients with disseminated strongyloidiasis are usually bedridden hospitalized patients that show symptoms such as paralytic ileus and reduced plasma albumin and cholesterol. Oral IVM is the only FDA-approved treatment but may not be effective in patients with disseminated disease. Veterinary subcutaneous formulations have been used in severe infections. We hypothesized that IVM PK in patients with disseminated strongyloidiasis can be predicted using PBPK model originally built and refined in healthy human and animal species. This hypothesis was tested and shown to be valid.
Methods:A systematic method was used to build and refine different parts of the PBPK model. The process involved construction of models, parameterization of these models, evaluation of the effect of uncertainty in model parameters on model prediction via local and global sensitivity analyses and finally, refinement of model predictions.
Two disposition models that differ in the rate limiting step in drug distribution were constructed and include perfusion-limited and permeability-limited distribution models. The ability of each model to predict IVM disposition was evaluated using plasma PK data in rat after intra-arterial dosing and in dog after intravenous bolus dosing.
Then the disposition model was scaled to humans and an oral input model was constructed as a modification on the well-known ACAT model. The oral input model was coupled with the disposition model and used to predict IVM plasma concentration-time profile in healthy fasted human subject after oral dosing.
Two subcutaneous (SQ) input models were constructed and used to evaluate the effect of IVM precipitation at the injection site. Plasma PK data in dog after SQ dosing was used to refine the constructed SQ input models.
The refined disposition, oral input and SQ input physiologically-based models were used to predict IVM PK in patients with disseminated strongyloidiasis after a complex dosing regimen. The physiological parameters of the model were modified to account for the effect of the disease-induced pathophysiological changes on the body physiology and hence on the drug PK. Plasma PK data from hospitalized subjects with disseminated strongylidiasis was used in this part.
Results and conclusions:The disposition model with assumption of permeability-limited distribution was more capable of describing IVM disposition in rat after intra-arterial dosing compared to when perfusion-limited distribution was assumed. The model predicted that hepatic clearance is the most impactful parameter on model-predicted plasma concentration of the drug. Also, IVM was shown to have low hepatic extraction ratio along with high binding in plasma and large volume of distribution, which collectively may explain the long half-life in the plasma of 63 hours in rat after intra-arterial dosing.
The oral input model predicted that the oral input is limited by drug dissolution in the GI lumen and that a very small fraction of oral tablet dose (0.03) is available in the systemic circulation in healthy fasted human subjects. Both of the studied SQ input models predicted that majority of IVM absorption after SQ dosing is via the lymphatic route and that drug precipitation at the injection site can further slowdown the drug absorption after SQ administration.
The PBPK model was able achieve the main goal of this research which is to predict IVM pharmacokinetics in patients with disseminated strongyloidiasis after a complex dosing regimen of multiple oral and SQ dosing. This was achieved by modifying the most impactful physiological parameters of the model affected by the disease state and that are related to drug binding in the plasma (fraction unbound), the GI motility (gastric emptying rate) and the lymphatic flow rate. Based on our analysis, we recommend measurement of plasma IVM concentrations early after initiation of therapy to exclude treatment failure due to reduced oral and/or SQ absorption. Also, we recommend measurement of plasma lipoprotein levels and their composition in these patients to differentiate between low total plasma concentrations due to low binding plasma as opposed to low drug input. Finally, interventional procedures that enhance lymphatic flow rate to site of SQ injection are recommended to enhance SQ absorption.