Clinical Predictors and Prediction Models for rFVIII-Fc Half Life in Real-World People with Severe Hemophilia A

Abstract

The half life of recombinant factor VIII-Fc (rFVIII-Fc) for people with hemophilia A (PwHA) varies greatly. Understanding the factors influencing the variation and assessment of rFVIII-Fc half life is important for personalized treatment. Eighty-five severe-type PwHA with rFVIII-Fc treatment receiving an evaluation of half life by the Web-Accessible Population Pharmacokinetic (PK) Service-Hemophilia during 2019-2021 were retrospectively enrolled. The 50-patient PK profiles before 2021 were used for analysis and developing prediction models of half life, and the 35-patient PK profiles in 2021 were used for external validation. The patients in the development cohort were aged 8-64, with a median rFVIII-Fc half life of 20.75 h (range, 8.25-41.5 h). By multivariate linear regression analysis, we found two, four, and five predictors of rFVIII-Fc half life for the blood groups non-O, O patients, and overall patients, respectively, including baseline VWF:Ag, BMI, VWF:activity/VWF:Ag ratio, body weight, O blood group, inhibitor history, HCV infection, and hematocrit. The three prediction equations of rFVIII-Fc half life (T) were respectively developed as T for non-O group patients = -0.81 + 0.63 × (BMI, kg/m²) + 6.07 × (baseline VWF:Ag, IU/mL), T for O group patients = -0.68 + 13.30 × (baseline VWF:Ag, IU/mL) + 0.27 × (BW, kg) - 1.17 × (BMI, kg/m²) + 16.02 × (VWF:activity/VWF:Ag ratio), and T for overall patients = -1.76 + 7.24 × (baseline VWF:Ag, IU/mL) - 3.84 × (Inhibitor history) + 2.99 × (HCV infection) - 2.83 × (O blood group) + 0.30 × (Hct, %), which explained 51.97%, 75.17%, and 66.38% of the half life variability, respectively. For external validation, there was a significant correlation between the predicted and observed half lives in the validation cohort. The median half life deviation was +1.53 h, +1.28 h, and +1.79 h for the equations of non-O group, O group, and overall group patients, respectively. In total, eight predictors influencing rFVIII-Fc half life were identified. Prediction equations of rFVIII-Fc half life were developed for the non-O and O blood groups and overall PwHA with a good degree of external validation. The equations could be applied to patients aged 8-64 without the need for PK blood sampling and clinically valuable for personalized therapy.

Keywords: external validation; half life; hemophilia A; pharmacokinetics; prediction model; predictor; rFVIII-Fc.

Figure 1

Distribution of rFVIII-Fc half lives marked by the O and non-O blood groups of the development cohort of the 50 people with hemophilia A (PwHA). The numbers above the bars denote the individual rFVIII-Fc half life (h). The gray-colored bars indicate PwHA with blood group O and appear more on the left side of the figure. The black-colored bars represent PwHA with blood group non-O and appear more on the right side.

Figure 2

Clinical variables, including body weight (BW), body mass index (BMI), baseline VWF:Ag, and baseline VWF:activity, related to the observed rFVIII-Fc half life in the development cohort (n = 46). Both BW and BMI had a weak or moderate positive correlation to half life (p-value < 0.01). Both baseline VWF:Ag and baseline VWF:activity had a strong positive correlation to half life (p-value < 0.0001). The drawn lines show the best fit by univariate linear regression.

Figure 3

Correlation between the observed rFVIII-Fc half life and baseline VWF:Ag and baseline VWF:activity, respectively, in blood group non-O and blood group O patients of the development cohort. For non-O group PwHA (n = 26), baseline VWF:Ag had a moderate positive correlation to the observed rFVIII-Fc half life (p-value < 0.001) and baseline VWF:activity had a strong positive correlation to the observed rFVIII-Fc half life (p-value < 0.001). For O group PwHA (n = 20), both baseline VWF:Ag and baseline VWF:activity had a strong positive correlation to the observed rFVIII-Fc half life (p-value < 0.001). The drawn lines show the best fit by univariate linear regression.

Figure 3

Correlation between the observed rFVIII-Fc half life and baseline VWF:Ag and baseline VWF:activity, respectively, in blood group non-O and blood group O patients of the development cohort. For non-O group PwHA (n = 26), baseline VWF:Ag had a moderate positive correlation to the observed rFVIII-Fc half life (p-value < 0.001) and baseline VWF:activity had a strong positive correlation to the observed rFVIII-Fc half life (p-value < 0.001). For O group PwHA (n = 20), both baseline VWF:Ag and baseline VWF:activity had a strong positive correlation to the observed rFVIII-Fc half life (p-value < 0.001). The drawn lines show the best fit by univariate linear regression.

Figure 4

Correlation between the observed and predicted rFVIII-Fc half life values in the development cohort: (A) for non-O group PwHA (n = 26) (p-rank = 0.74683, p-value < 0.0001); (B) for O group PwHA (n = 20) (P-rank = 0.93213, p-value < 0.0001); (C) for the entire cohort of PwHA (n = 46) (P-rank = 0.82892, p-value < 0.0001). The drawn lines show the best fit by univariate linear regression.

Figure 5

External validation. Correlation between the observed and predicted rFVIII-Fc half life values of the validation cohort: (A) for non-O group PwHA (n = 23) (P-rank = 0.5510, p-value < 0.01); (B) for O group PwHA (n = 12) (P-rank = 0.6157, p-value < 0.05); (C) for the entire cohort of PwHA (n = 35) (P-rank = 0.7815, p-value < 0.0001). The drawn lines show the best fit by univariate linear regression.