PB0153 - Comparative Pharmacokinetics of Coagulation Factor VIII and IX Products Between Independently Developed and Web-based Population Pharmacokinetic Models

Background: Many variables affect factor replacement therapy for hemophilia patients. Population PK (PPK) modeling can facilitate dosing estimates even with reduced sampling schedule.

Aims: To compare pharmacokinetic (PK) estimates between PPK models for Factor VIII (FVIII) and Factor IX (FIX) products.

Methods: Retrospective review of hemophilia A or B patients. Comparison of PK estimates from in-house Nonlinear Mixed Effects Modeling (NONMEM) model, online WAPPS-Hemo, and expected age-based mean product half-lives (t1/2) in FDA prescriber labels (PL).

Results: A one-compartment model with first-order elimination was best fit for both cohorts. For hemophilia A (n=51, males=47, age 0.1-84.6 years, weight 3.5-151 kg), correlation coefficient (r) for t1/2 estimates between in-house NONMEM and WAPPS-Hemo models was 0.69 (95%CI 0.51, 0.81). Agreement between PL t1/2 values and WAPPS-Hemo results was the best (mean bias 3.03; 95%CI 0.77, 5.29), while that with NONMEM was poorest (mean bias 6.48; 95%CI 3.44, 9.51). In-house model t1/2 values demonstrated two distinct clusters (Figure 1). Confounding covariates were median age (p=0.0010) and PK study type (peri-procedural vs. prospective; p=0.0222).
For hemophilia B (n=18, males=17, age 5-79 years, weight 20.8-105 kg), correlation coefficient (r) for t1/2 estimates between in-house NONMEM and WAPPS-Hemo models was 0.89 (95%CI 0.78, 0.94). Agreement between WAPPS-Hemo and NONMEM was lowest (mean bias 36.88; 95%CI 28.57, 45.20), while that between PL t1/2 and NONMEM was the best (mean bias 13.46; 95%CI 7.59, 19.33). Comparison of in-house t1/2 with PL demonstrated two distinct clusters (Figure 2). Confounding covariates were product type (standard vs extended t1/2; p< 0.0001), PK study type (peri-procedural vs. prospective; p=0.0003), and weight (p=0.0296).

Conclusion(s): Despite differential comparative results with PL and WAPPS-Hemo estimates, likely due to limited sample size and heterogeneity of factor products, an external validation of our independently built FVIII and FIX PPK models was possible and their performance characteristics were consistent with published literature.