Page 133 - 2021_07-Haematologica-web

FVIII activity and bleeding risk in hemophilia A
In severe hemophilia A, the annualized bleeding rate (ABR) during prophylaxis was shown to correlate with time spent with FVIII activity (FVIII:C) below 1%, as pre- dicted from the patient’s individual FVIII pharmacokinet- ics (PK).6 However, actual time below 1% or any other tar- geted trough level also depends on the prescribed prophy- laxis regimen and the patient’s adherence to it. The risk of spontaneous bleeds during periods with FVIII:C ≥1% in subjects on prophylaxis is unknown, although an epidemi- ological study in individuals with mild/moderate hemo- philia suggested that maintaining FVIII:C above 1% will not protect all patients from bleeding.7,8 Evaluation of the association between FVIII:C and bleeding pattern is important and can support evidence-based tailoring of prophylaxis independently of the factor concentrate used.
We have used data from the guardian clinical trial pro- gram to examine bleeding risks according to FVIII:C with a recombinant FVIII (rFVIII) molecule during prophylaxis in patients with severe hemophilia A. The product, turoc- tocog alfa (NovoEight®, Novo Nordisk Health Care AG, Zürich, Switzerland), is a B-domain-truncated rFVIII. Data from three studies were combined: a pivotal trial in adults and adolescents (guardian 1; NCT00840086),9 a pivotal trial in children (guardian 3; NCT01138501),10 and an extension trial (guardian 2, which was an extension trial of guardian 1 and 3; NCT00984126).11 Estimated mean (95% confidence interval) spontaneous ABR in these studies were 4.32 (3.34-5.59) (pivotal trial in adults and adoles- cents), 1.69 (0.94-3.03) (pivotal trial in children), and 1.34 (1.07-1.68) (extension trial).9-11
The PK characteristics of turoctocog alfa, which have been extensively studied using standard PK assessments (based on plasma FVIII:C) and a population PK model, have been shown to be consistent over time, reproducible between lots and similar to those of other commercially available FVIII products.12
Defining the relationship between FVIII:C and bleeding pattern is important not only because bleeding frequency is an important outcome, but also to guide robust, individ- ualized prophylactic treatment schedules. Given the importance of spontaneous bleeding in hemophilia, we assessed the relationship between FVIII:C and ABR, as a measure of bleeding frequency for spontaneous bleeds, including spontaneous joint bleeds, which were recorded in dosing diaries by patients with severe hemophilia A who received prophylaxis with turoctocog alfa in the guardian clinical trials.
Methods
Trial design and patients’ eligibility
The pivotal trials (guardian 1 and 3) were multinational, phase III trials assessing prophylaxis with turoctocog alfa in adults/adolescents (≥12 years) and pediatric patients (0-11 years), respectively. All patients had severe hemophilia A (FVIII ≤1%) without inhibitors and had already been exposed to FVIII for ≥150 days (adults/adolescents) or ≥50 days (children).9,10 Patients completing the pivotal trials or phase I PK trials could continue into the open-label extension trial (guardian 2).13 Patients received turoctocog alfa as prophylaxis (20-50 IU/kg every second day or 20-60 IU/kg three times weekly, depending on age), as well as for treatment of bleeds. All three trials are now complete. Patients in the trials reported their dosing and bleeding data in diaries. Baseline joint disease at pivotal study
entry was not assessed; joint disease in the current analysis was therefore inferred based on treatment history.
Pharmacokinetic assessment in the pivotal studies
Consistent with the International Society of Thrombosis and Haemostasis FVIII product guidelines,14 the pivotal trials includ- ed a single-dose PK assessment in a subset of patients after the first dose and in adults/adolescents after 3-6 months’ treatment; this involved nine (adults/adolescents) or six (children) sampling time points up to 48 h after injection of the product.9,10,12 In addi- tion, samples were taken from all patients before and after dos- ing at the routine visits. All plasma samples were analyzed at a central laboratory using a one-stage clot assay.12,15 and post-dose FVIII:C values were entered into the PK model.
Population pharmacokinetic modeling
Population PK modeling was conducted with the nonlinear, mixed-effects modeling software NONMEM (v.7.1.2, ICON Development Solutions, Ellicott City, MD, USA). Variability between patients was quantified within the population PK model12 by including between-patient variability on PK parame- ters using a log-normal distribution with no correlation. Allometric scaling based on body weight was applied to clear- ance (CL) and volume of distribution (Vd) parameters, and age was estimated as a linear covariate on CL. A combined propor- tional and additive residual error model was used.
Exposure-response analysis: estimating the association between factor VIII activity and annualized bleeding rate
The exposure-response analysis of the association between predicted FVIII:C and spontaneous bleeds, including joint bleeds, was conducted with the statistical software R3.2.3 (Comprehensive R Archive Network [CRAN] project, University of Münster, Germany).
The population PK model estimates were applied to diary- recorded dosing from all patients receiving prophylaxis with tur- octocog alfa in the pivotal and extension trials who had appropri- ate records, regardless of whether patients reported a sponta- neous bleed or not, to produce a predicted FVIII:C time course for each patient over the entire span of prophylaxis for that patient. These predicted profiles had peaks following each recorded dose and subsequent gradual decline, taking into account the contribu- tions from the last three doses at any time point (Figure 1). Each patient’s time on prophylaxis was divided into five clinically meaningful categories of predicted FVIII:C: 0-1%, >1-5%, >5-15%, >15-50%, and >50%.
Subsequently, the diary-reported bleeding (spontaneous, including joint bleeds) data were compared to the predicted FVIII:C–time-profiles to estimate the FVIII:C–bleeding pattern relationships. Patients were excluded from the analysis if they switched from prophylaxis to on-demand treatment during the extension trial, received only on-demand treatment throughout the extension trial, had missing diary returns, or failed to docu- ment time of bleeding for >50% of recorded bleeds.
For the purposes of this analysis, we have described bleeds as those bleeds that patients reported in their diary returns. This is to acknowledge that the bleeds were subjectively evaluated by the patient.
For each patient, exposure time to turoctocog alfa, mean FVIII:C, and number of spontaneous bleeds (including joint bleeds) requiring treatment (based on the patient’s diary entries and excluding re-bleeds within 48 h) were calculated for each predicted FVIII:C category. To assess the association between FVIII:C and ABR, bleeding data were used to estimate ABR for
haematologica | 2021; 106(7)
1903