A comparison of methods for prediction of pharmacokinetics when switching to extended half-life products in hemophilia A patients

Abstract

IntroductionHemophilia A is a genetic bleeding disorder resulting from a lack of clotting factor VIII. Where extended half-life products are available, people with hemophilia may stop their current drug regimen and switch to a EHL product providing a more convenient dosing regimen. While most factor VIII concentrate regimens are started prophylactically based on international units per weight, this “one-size-fits-all” approach does not account for the large pharmacokinetic variability between individuals. AimsWe explored methods to predict individual PK of an EHL product by using population pharmacokinetic models and eta-values (η), a value that quantifies how individuals deviate from a population for any PK parameter, derived from a prior product. In addition, we wanted to investigate which individuals would benefit from this method compared to using a PopPK model alone. MethodsPK data from subjects (n = 39) who have taken both Adynovate and Eloctate was collected from clinical trial data and from the Web-Accessible Population Pharmacokinetic Service – Hemophilia (WAPPS-Hemo) database. In addition, PK data from subjects (n = 200) who switched from a standard half-life product to Eloctate was also extracted from the WAPPS-Hemo database. Two methods to estimate individual PK outcomes of the second product were compared. The PopPK method used the Eloctate PopPK model published from WAPPS-Hemo, while the η-method incorporated individually scaled η from the prior product's PopPK model. Both methods were assessed for its performance in predicting PK outcomes. Absolute percent differences were calculated between the predicted and observed PK outcomes. Infusions were parsed into subgroups based on number of samples and individual η-percentiles for analysis. ResultsFor the three switching protocols (Adynovate to Eloctate, Eloctate to Adynovate, and SHL FVIII to Eloctate), the η-method resulted in a relative difference reduction in mean absolute percent difference of 27.8% (range 1–59%), 4.9% (range 0–129%), and 18.0% (0–79%) in half-life compared to the PopPK method respectively. With some exceptions (in particular central volume), the η-method produced relative difference reduction in mean absolute percent differences up to 33% lower compared to the PopPK method. When individuals were parsed based on their η-values (either CL or V1), the two methods differentiate up to 64% in terms of half-life and time to 0.02 IU/mL predictions for individuals with a low (0th to 20th percentile) ηCL or ηV1 on the first product. Individuals with higher number of observations per infusion on the first product resulted in better predictions in PK parameter estimates when using the η-method. ConclusionThe use of prior knowledge by implementing η-values into PopPK models may provide clinicians with a safer and more effective method to choose a dosing regimen for patients with hemophilia A switching from one factor concentrate to another. However, the η-method was unable to better predict an increase or decrease in half-life of a future product compared to the PopPK method, and thus supports the conclusion that most individuals would still benefit from a trial on the EHL and subsequent estimation of their individual PK profile from sparse measurements on the EHL.