Population pharmacokinetic modeling for dose setting of nonacog beta pegol (N9-GP), a glycoPEGylated recombinant factor IX

Authors


Peter W. Collins, Arthur Bloom Haemophilia Centre, School of Medicine, Cardiff University, Cardiff CF14 4XW, UK.
Tel.: +44 2920 742 155; fax: +44 2920 748 266.
E-mail: peter.collins@wales.nhs.uk.

Abstract

Summary.  Background: nonacog beta pegol (N9-GP) is a glycoPEGylated recombinant factor IX (rFIX) molecule with a prolonged half-life. Objectives: To provide information on potential dose regimens for N9-GP for phase 3 pivotal and surgery trials. Methods: A population pharmacokinetic model was developed from single-dose data derived from the first human-dose trial with N9-GP in hemophilia B patients, and was used to extrapolate to steady-state conditions for different N9-GP dose regimens for prophylaxis. The model was also used to compare prophylaxis using N9-GP with standard prophylactic regimens using rFIX or plasma-derived (pd) FIX (40 IU kg−1 every third day). Plasma activity following dosing with N9-GP, rFIX and pdFIX for surgery and on-demand treatment of bleeds was also simulated. Results: A linear two-compartmental model best described the pharmacokinetic profiles of N9-GP, rFIX and pdFIX. A prophylactic regimen of 10 U kg−1 N9-GP once weekly predicted mean peak and trough levels of 18 and 4.2 U dL−1, while 40 U kg−1 once weekly predicted values of 72 and 17 U dL−1, respectively. Standard prophylactic regimens with rFIX and pdFIX predicted mean peak and trough levels of 34 and 3.9 IU dL−1 for rFIX, and mean values of 43 and 2.1 IU dL−1 for pdFIX. Additional simulations predicted significantly reduced dosing frequency and factor concentrate consumption for N9-GP vs. rFIX and pdFIX for surgery and the treatment of bleeds. Conclusions: N9-GP may allow prophylaxis, surgical dosing regimens and on-demand treatment of bleeding episodes with less frequent injections and lower factor concentrate consumption; this possibility is being investigated in prospective clinical trials.

Introduction

Primary prophylaxis in patients with severe hemophilia significantly improves patient outcomes and quality of life by decreasing the number of joint bleeds and lowering the subsequent risk of developing hemophilic arthropathy and potentially reducing the risk of life-threatening bleeds [1–4]. Therefore, prophylaxis with replacement coagulation factors is the recommended treatment for patients with severe hemophilia A or B [5–7]. Some hemophilia B patients use on-demand treatment, and with currently available factor IX (FIX) products, this may require more than one injection to maintain adequate hemostasis, an observation also reported in the setting of a clinical trial [8]. FIX is also used to maintain hemostasis in hemophilia B patients undergoing surgery. However, to maintain the FIX levels necessary to provide adequate hemostatic coverage for surgery, currently available FIX products require repeated bolus injections or continuous infusion [9–11]. Hence, in multiple clinical situations, the development of FIX replacement factors with prolonged half-lives that aim to reduce dosing frequency and maintain adequate hemostasis with fewer injections would be beneficial to patients, their families and healthcare professionals [12].

The compound nonacog beta pegol (N9-GP) is a recombinantFIX (rFIX) derivative, produced without animal-derived materials, in which site-directed glycoPEGylation has been utilized to attach a 40 kDa polyethylene glycol (PEG) molecule to the activation peptide. Upon N9-GP activation, the activation peptide with the attached PEG molecule is cleaved to leave the wild-type activated FIX (FIXa) [13]. Pharmacokinetic data from the first human-dose (FHD) phase 1, open-label, dose-escalation trial (25, 50 and 100 U kg−1) with N9-GP in hemophilia B patients in a non-bleeding state (FIX activity ≤ 2 IU dL−1) demonstrated that the mean terminal half-life of N9-GP (92.7 h) was prolonged 5-fold compared with the patients’ previous FIX products (plasma-derived [pd] FIX 18 h and rFIX 19 h) [14].

The aim of the current study was to develop a population-based pharmacokinetic model using single-dose data derived from the FHD trial with N9-GP in hemophilia B patients, and to use the model to simulate different N9-GP dose levels and dosing frequencies. Population pharmacokinetic modeling has previously been utilized to translate pharmacokinetic results into suggestions for dosing with rFIX in prophylactic regimens for patients with hemophilia B [15]. Further, pharmacokinetic modeling is of special interest for clinical trials in rare bleeding disorders such as hemophilia B, in which the number of patients available for recruitment into clinical trials is limited. The results of the pharmacokinetic modeling approach utilized in the current study were used to design studies to investigate different N9-GP dosing regimens for phase 3 pivotal and surgical trials.

Methods

Data

Data from the single-dose FHD trial of hemophilia B patients in a non-bleeding state [14] were used to generate population pharmacokinetic models for N9-GP, pdFIX and rFIX based on the three dose levels administered in the trial (25, 50 and 100 U kg−1). The FIX activity was measured using a modified activated partial thromboplastin time assay and was used for both N9-GP and the patients’ previous FIX products [14]. N9-GP plasma samples were analyzed using an N9-GP calibrator. The lower limit of quantitation (LLOQ) for N9-GP was 0.009 U mL−1 whereas the LLOQ was 0.010 and 0.011 IU mL−1 for pdFIX and rFIX, respectively.

Analysis

Population pharmacokinetic analysis was performed in NONMEM VI, version 1.2 (ICON Development Solutions, Ellicott City, MD, USA). This software optimizes the combination of parameter values in a specified mathematical model to maximize the overall likelihood of the experimental data. The overall likelihood is labeled the objective function value (OFV). When calculating derivatives of the OFV with respect to each fitted model parameter, the model is linearized with a first-order Taylor approximation (the first-order conditional estimation or FOCE option). The model-building process was guided by visual inspection of goodness-of-fit and changes in the OFV obtained from NONMEM. For nested models the change in OFV is assumed to be χ2 distributed. A P value of 0.001 was used in model extension/reduction criteria throughout the analysis; hence, a drop in OFV of at least 10.83 was required for the addition of a single parameter to be a significant improvement in model fit.

The population model estimated the typical values of the model parameters and the inter-patient variability of these parameters along with the asymptotic standard errors of the estimates derived from the Fischer information matrix. The inter-patient variabilities in parameter estimates were expressed as exponential models, where between-patient variability of a given parameter η is a normal distributed random variable with mean zero and variance ω2. Residual error was also included and assessed in a proportional or a combination of additive and proportional residual error. The residual variability is denoted by ε, which is a normal distributed random variable with mean zero and variance σ2.

Modeling strategy

One of the inclusion criteria in the FHD trial was FIX activity ≤ 2 IU dL−1. The highest measured baseline level of FIX activity assessed immediately prior to dosing found in a single patient was 1.3 IU dL−1 whereas the remaining patients were below the LLOQ (0.9 IU dL−1). In the modeling, baseline FIX activity was set to zero because the endogenous contribution would be insignificant for the vast majority of observations.

Linear one-, two- and three-compartmental models were assessed, with each additional compartment providing two more model parameters to be fitted to the data. Based on balancing the improved fit to the data of the more flexible model against the increase in number of fitted parameters with likelihood ratio tests on a 0.1% level of significance [16], the two-compartmental model was selected among the three candidate models.

The final pharmacokinetic model for N9-GP was validated by simulating the N9-GP data from the FHD trial 1000 times with the parameter estimates obtained from the final model. The ability of the model to replicate the observations was evaluated by comparing the mean of the simulated concentrations to the mean of the observed data. In addition, the accuracy of the random effect estimates (i.e. inter-patient variability and residual error) was evaluated by comparing the mean 95% prediction interval (PI) across 1000 trial replications to the 95% PI interval calculated on the observed data.

Pharmacokinetic simulations to predict different dosing scenarios

To predict the outcomes of different dosing scenarios with pdFIX, rFIX and N9-GP, respectively, a random selection of 1000 patients exhibiting mutual variability in pharmacokinetic parameters was simulated to receive each treatment. The results of these simulations were summarized with mean values and 95% PIs spanning the 2.5th and the 97.5th percentile of the predicted values.

Pharmacokinetic simulations for once-weekly dose regimens of 10, 20, 30, 40 and 50 U kg−1 N9-GP showed that steady-state will be achieved after four doses (data not shown). The standard prophylaxis dosing regimen for currently available FIX products was defined as 40 IU kg−1 given every 3 days [6]. Two dose regimens of N9-GP (10 and 40 U kg−1 once weekly) were chosen for further simulations. The 10 U kg−1 once- weekly regimen was selected to explore the scenario of a low dose of N9-GP selected to achieve similar steady-state trough activity levels to those predicted for standard FIX prophylaxis. The 40 U kg−1 N9-GP once-weekly regimen was chosen to reach and explore higher trough levels and near normal maximal concentration (Cmax) levels. The model was also used to evaluate whether prophylaxis with 40 U kg−1 N9-GP given every 2 weeks would result in trough activity levels in line with standard FIX prophylaxis.

Further simulations compared the pharmacokinetic profile of N9-GP with rFIX and pdFIX for the on-demand treatment of hypothetical knee bleeds and intracranial hemorrhage. For the treatment of a mild/moderate knee bleed, the nominal target was to maintain a FIX activity above 40 IU dL−1 for about 24 h. To treat a severe knee bleed, the nominal target was to maintain a FIX activity above 50 IU dL−1 for about 3 days. For an intracranial hemorrhage, the minimum acceptable FIX activity was set at between 60 and 80 IU dL−1 for the first 7 days and then 30 IU dL−1 until 21 days. Simulations were performed to explore the N9-GP dose and frequencies required to reach these target FIX levels.

For surgery, simulations were performed to explore and compare the dosing regimens for N9-GP, rFIX and pdFIX that were needed to reach and maintain target FIX levels as recommended by the World Federation of Hemophilia (WFH) guidelines [6]. These guidelines suggest that the FIX level should be maintained above 40 IU dL−1 for 3 days, above 30 IU dL−1 for a further 3 days, and then above 20 IU dL−1 until 14 days.

Results

Pharmacokinetic model

The pharmacokinetics of N9-GP following intravenous bolus doses was most appropriately described by a linear two-compartmental model, with first-order elimination (Table 1). A covariance matrix on the inter-patient variability of clearance (CL) and volume of distribution (V1) was also found to significantly improve the model fit. The residual error was best described by a proportional and additive error model. As shown by the schematic depiction of the model structure (Fig. 1), intravenously administered N9-GP was injected into plasma, with a volume (V1) corresponding to 0.0739 L kg−1, and distributed to a peripheral compartment with a volume (V2) corresponding to 0.0156 L kg−1. Thus, the data suggest that after injection, N9-GP primarily circulates in the blood, which is in agreement with non-clinical data [17]. Clearance from the central compartment was 0.00068 L h kg−1.

Table 1.  Population estimates of pharmacokinetic parameters for N9-GP, pdFIX and rFIX
ParameterPopulation estimates (medians)
N9-GP (±SE)rFIXpdFIX (±SE)
  1. N9-GP, nonacog beta pegol; rFIX, recombinant factor IX; pdFIX, plasma-derived factor X; CL, clearance; NA, not applicable; SE, standard error of the mean; V, volume of distribution; Q, inter-compartmental clearance.

CL (L h kg−1)0.000684 ± 0.0000313 (4.6%)0.005110.0054 ± 0.0005 (9%)
Inter-subject variability on CL (%)16.80.1420.32
V1 (L kg−1)0.0739 ± 0.00358 (4.8%)0.1370.103 ± 0.0103 (10%)
Inter-subject variability on V1 (%)18.7128.811.4
Correlation covariance matrix on CL and V1 (%)16.09 ± 0.8NA25.4 ± 0.7
Q (L h kg−1)0.000614 ± 0.000216 (35.2%)0.004080.003 ± 0.0025 (83%)
Inter-subject variability on Q (%)127.2818.0NA
V2 (L kg−1)0.0156 ± 0.00184 (11.8%)0.09350.0283 ± 0.007 (24.6%)
Inter-subject variability on V2 (%)NA84.6NA
Proportional residual variability (%)6.47 ± 3.639.9518.0 ± 7.7
Additive residual variability (U dL−1)0.0267 ± 0.0111 NANA
Figure 1.

 Schematic depiction of the two-compartmental model describing the pharmacokinetics of N9-GP. Comp, compartment; CL, clearance; iv, intravenous; K, first-order distribution rate constants; Q, inter-compartmental clearance; V, volumes.

The pharmacokinetics of the patients’ previous FIX products, rFIX and pdFIX following intravenous bolus doses, were also best described by a linear two-compartmental model with first-order elimination with only a proportional residual error model required (Table 1).

Model validation

The majority of data points (93%) from each of the three N9-GP doses (25, 50, 100 U kg−1) fell within the 95% PI of the model. Patients previously reported to have outlying pharmacokinetic profiles [14] were also outside the 95% PI (Fig. 2). The mean activity level predicted by the model for each dose corresponded to the mean activity level observed in the FHD trial, demonstrating the suitability of the model for pharmacokinetic predictions.

Figure 2.

 Comparison of N9-GP pharmacokinetic data from the single-dose FHD trial [14] with profiles predicted by the pharmacokinetic population-based model for (A) 25, (B) 50 and (C) 100 U kg−1 doses. Blue shading, 95% prediction intervals; blue dashed line, mean predicted values; circles, pharmacokinetic data obtained from the FHD trial; red line, 3 U dL−1 FIX activity.

Pharmacokinetic simulations to compare once-weekly dosing of N9-GP with standard prophylaxis

Pharmacokinetic simulations performed for standard FIX prophylaxis (40 IU kg−1 pdFIX or rFIX every 3 days) predicted mean steady-state peak levels of 43 IU dL−1 (95% PI, 20–81 IU dL−1) and 34 IU dL−1 (95% PI, 21–52 IU dL−1), and mean steady-state trough levels of 2.1 IU dL−1 (95% PI, 1.1–3.3 IU dL−1) and 3.9 IU dL−1 (95% PI, 2.1–5.4 IU dL−1), respectively.

The predicted steady-state profiles for once-weekly dose regimens of N9-GP (10 and 40 U kg−1) were compared with the standard FIX dose regimen (every third day with 40 U kg−1 pdFIX or rFIX) (Fig. 3). A once-weekly dose regimen of 10 U kg−1 N9-GP was predicted to provide a mean steady-state peak of 18 U dL−1 (95% PI, 13–24 U dL−1) and a mean steady-state trough level of 4.2 U dL−1 (95% PI, 2.7–6.0 U dL−1). Increasing the dose level of N9-GP to 40 U kg−1 (once weekly) was predicted to give a steady-state mean peak of 72 U dL−1 (95% PI: 52–99 U dL−1) (Fig. 3). Of note, the predicted FIX level 3 days after dosing with 40 U kg−1 N9-GP was 35 U dL−1 (95% PI, 25–48 U dL−1) and equivalent to the peak level estimated after dosing with rFIX or pdFIX at 40 IU kg−1 every 3 days. The mean steady-state trough level for prophylaxis with 40 U kg−1 N9-GP once weekly was predicted to be 17 U dL−1 (95% PI, 11–25 U dL−1).

Figure 3.

 The steady-state predicted profiles for N9-GP dose regimens of 10 (light pink) and 40 (dark pink) U kg−1 once-weekly vs. standard FIX dose regimens of 40 IU kg−1 rFIX (blue) and pdFIX (green) every 3 days. Blue horizontal dashed line, 3 IU dL−1 or U dL−1 FIX activity; dashed lines, mean predicted values; shaded regions, 95% prediction intervals.

Pharmacokinetic simulations to explore the effect of reduced dosing frequency

Pharmacokinetic simulations were performed to predict the minimum and maximum FIX activity levels if a 40 U kg−1 dose of N9-GP was given every second week (Fig. 4). For this dose regimen the mean steady-state peak FIX activity level was 59 U dL−1 (95% PI, 39–85 U dL−1) while the mean steady-state predicted trough level was 4.0 U dL−1 (95% PI, 2.1–6.5 U dL−1) FIX activity (Fig. 4).

Figure 4.

 Predicted FIX activity profiles following 40 U kg−1 N9-GP dosed once every second week. Blue shading, 95% prediction interval; blue dashed line, mean predicted value; blue solid line, 1 U dL−1 FIX activity; red line, 3 U dL−1 FIX activity.

Pharmacokinetic simulations to explore N9-GP for the treatment of knee bleeds

For simulations performed for the on-demand treatment of a mild/moderate knee bleed, a single 40 U kg−1 dose of N9-GP was predicted to provide FIX activity levels above 40 U dL−1 for an average of 23 h (95% PI, 0.5–52 h). In comparison, two doses of rFIX (an initial dose of 70 IU kg−1 followed by a 40 IU kg−1 dose 12 h later) and pdFIX (an initial dose of 55 IU kg−1 followed by a 40 IU kg−1 dose 12 h later) were required to maintain the same target FIX activity level (Fig. 5A). Using these dosing regimens, at 24 h the predicted mean FIX activity with N9-GP was 40 U dL−1(95% PI, 29–54 U dL−1), with pdFIX 31 IU dL−1(95% PI, 21–43 IU dL−1) and with rFIX 29 IU dL−1(95% PI, 22–37 IU dL−1).

Figure 5.

 Predicted FIX plasma activity levels following the on-demand treatment of (A) a mild/moderate knee bleed, (B) a severe knee bleed, and (C) an intracranial hemorrhage with N9-GP (dark pink) vs. rFIX (blue) and pdFIX (green). Blue solid lines, 40 IU dL−1 or U dL−1 for (A) and 50 IU dL−1 or U dL−1 FIX activity for (B); dashed lines, mean predicted values; shaded regions, 95% prediction intervals. The doses predicted to be required for N9-GP, rFIX and pdFIX were as follows. (A) N9-GP, one 40 U kg−1 dose; rFIX, one 70 IU kg−1 dose followed by one 40 IU kg−1 dose 12 h later; pdFIX, one 55 IU kg−1 dose followed by one 40 IU kg−1 dose 12 h later. (B) N9-GP, one 80 U kg−1 dose; rFIX, one 150 IU kg−1 dose followed by five 12-hourly doses with 40 U kg−1; pdFIX, one 110 IU kg−1 dose followed by five 12-hourly doses with 40 U kg−1. (C) N9-GP, one 80 U kg−1 dose followed by 40 U kg−1 doses on days 2, 5, 12 and 17; rFIX, one 140 IU kg−1 dose followed by 50 IU kg−1 doses every 12 h for the first 7 days, then 50 IU kg−1 doses every 24 h for the next 13 days; pdFIX, one 100 IU kg−1 dose followed by 50 IU kg−1 doses every 12 h for first 7 days, then 50 IU kg−1 doses every 24 h for the next 13 days.

For simulations performed for the on-demand treatment of a severe knee bleed, a single 80 U kg−1 dose of N9-GP was predicted to provide FIX activity levels above 50 U dL−1 for a mean of 3 days (95% PI: 1.6–4.6 days). In comparison, six doses of rFIX (one 150 IU kg−1 dose followed by five 12-hourly doses with 40 IU kg−1) and pdFIX (one 110 IU kg−1 dose followed by five 12-hourly doses with 40 IU kg−1) were required to maintain FIX activity above a similar level (Fig. 5B).

Pharmacokinetic simulations to explore N9-GP for the treatment of intracranial hemorrhage

The predicted initial mean dose of N9-GP required to achieve the target trough FIX activity level of 60–80 U dL−1 was 80 U kg−1, which was lower than that predicted for rFIX (140 IU kg−1) and pdFIX (100 IU kg−1). rFIX and pdFIX both required a total of 28 doses (total mean FIX consumption of 1490 and 1450 IU kg−1, respectively) to maintain FIX above the target levels for a 3-week period (Fig. 5C). In contrast, N9-GP required a total of five doses with a mean total FIX consumption of 240 U kg−1 to maintain the same target levels (Fig. 5C).

Pharmacokinetic simulations to explore dosage and dosing frequency of N9-GP for surgery

The predicted initial mean dose of N9-GP required to reach the target FIX activity level of 100–120 U dL−1 was 80 U kg−1. This was a lower dose than that for rFIX (150 IU kg−1) and pdFIX (110 IU kg−1). rFIX and pdFIX required 16 subsequent doses of 40 IU kg−1 following the initial infusion to maintain FIX target levels above the desired trough for a 2-week postoperative period. In contrast, two infusions of N9-GP at 40 U kg−1 (given every 5 days) were required to maintain the same target levels (Fig. 6). The total mean FIX usage was predicted to be 790 and 750 IU kg−1 for rFIX and pdFIX, respectively. In contrast, total mean FIX consumption for N9-GP was predicted to be 160 U kg−1.

Figure 6.

 Predicted FIX plasma activity levels following replacement therapy for surgery with N9-GP (dark pink) vs. rFIX (blue) and pdFIX (green). Dashed lines, mean predicted values; shaded regions, 95% prediction intervals. The doses predicted to be required for N9-GP, rFIX and pdFIX were as follows. N9-GP: one 80 U kg−1 dose followed by two 40 U kg−1 doses (on days 5 and 10). rFIX: one 150 U kg−1 dose, followed by six 12-hourly doses with 40 U kg−1 (up to day 3), and 10 24-hourly doses with 40 U kg−1. pdFIX: one 110 U kg−1 dose, followed by six 12-hourly doses with 40 U kg−1 (up to day 3), and 10 24-hourly doses with 40 U kg−1.

Discussion

In this study, population-based pharmacokinetic modeling predicted that once-weekly prophylaxis with N9-GP dosed at 10 or 40 U kg−1 would maintain trough FIX plasma concentrations at a level previously associated with bleed prevention. Further simulations suggested that a single dose of N9-GP may provide improved on-demand therapy for the treatment of acute bleeding episodes. N9-GP may also allow treatment of intracranial bleeds and the provision of hemostatic cover for surgery, with fewer injections and reduced factor consumption. The simulations presented here need to be confirmed in clinical studies and are currently being explored in the ongoing clinical trial program for N9-GP.

The aim of prophylaxis is to convert severe hemophilia B (< 1 IU dL−1 FIX activity) to a milder phenotype. The success of prophylaxis in hemophilia A and B has been shown to be influenced by time per week of factor level below a certain trough level [18,19]. However, the amount of coagulation factor a person is exposed to, as measured by the area under the curve (AUC), or recurrent high peak levels, may also be important. It may be that the most important parameter depends on the aim of prophylaxis. If the aim is to prevent spontaneous bleeds, trough levels might be important. Alternatively, to prevent trauma or sport-induced bleeds, having peak levels at the time of sporting activity may be critical, whereas to treat a subclinical bleed, recurrent peaks or AUC may play important roles. The importance of troughs, peaks and AUC may thus differ depending on personal circumstances and may change through the patient’s life [20,21]. For example, the FIX trough and peak levels required to provide effective prophylaxis for an older, more sedentary person may differ from those of a more physically active adolescent. Given the different pharmacokinetic profile of N9-GP, aiming for the same trough or peak value as that for standard FIX products may not result in a similar hemostatic effect. It may be that a prolonged low FIX activity level, resulting from weekly or less frequent infusions of a product with a prolonged half-life that provides a trough of 1–2 IU dL−1, may provide adequate prophylaxis for some patients to prevent spontaneous bleeds. However, to prevent sport-related bleeds, peak levels or higher troughs may be needed.

While it is not known exactly which pharmacokinetic parameters may be important to prevent bleeding, in line with suggestions that trough levels of 1 or 3 IU dL−1 FIX activity may be required for prophylaxis and the prevention of hemarthroses [3,4,19,22–28], a previous post hoc analysis of dose-normalized data (to 50 U kg−1) from the FHD N9-GP trial was performed. This post hoc analysis predicted an extended time to reach 1 and 3 IU dL−1 FIX activity (22 and 16 days, respectively) for N9-GP [14]. Here, population-based pharmacokinetic modeling was used to further explore whether N9-GP could enable prophylaxis with fewer injections. For both 10 and 40 U kg−1 infused weekly, trough levels above 3 U dL−1 FIX activity were predicted for the entire period. Together, these findings suggest that once-weekly dosing with N9-GP may be appropriate prophylaxis in patients with hemophilia B. If the maintenance of trough activities above a certain level is important, then both N9-GP dose regimens (10 and 40 U kg−1 once weekly) are predicted to produce a hemostatic effect at least as good as that of standard FIX prophylaxis regimens. However, if FIX peak levels are important, differences between the two regimens of N9-GP and standard FIX prophylaxis may be observed. This is because, in the 40 IU kg−1 group, the day 3 FIX level is similar to the peak level provided by the standard regimen, whereas in the 10 IU kg−1 group, the trough level is similar to the standard regimen. Data from the ongoing clinical trial of N9-GP prophylaxis exploring once weekly dosing with two dose levels of N9-GP (10 and 40 U kg−1) may enable a better understanding of the relative importance of peaks and troughs for prophylaxis and bleed prevention.

Enhanced on-demand treatment has been suggested as a way to decrease inflammation and prevent joint damage by preventing rebleeding in hemophilia A [29]. If replicated for hemophilia B, multiple infusions of currently available FIX products would be required. Further, treatment of acute bleeds has been reported to require more than one injection to achieve and maintain adequate hemostasis in some cases [8]. To simulate possible on-demand scenarios we considered hypothetical bleeds of different severities. Clinicians treat bleeds on a case-by-case basis and the simulations described in this paper are for illustrative purposes only. To give some indication of the potential role of N9-GP we modeled the treatment of a mild/moderate knee bleed as requiring the maintenance of FIX above 40 IU dL−1 for about 1 day. A severe knee bleed might require FIX to be maintained above 50 IU dL−1 for about 3 days, whereas an intracranial bleed would often be treated for at least 3 weeks.

Simulations performed for rFIX and pdFIX suggest that two doses of the currently available FIX products would be required for on-demand treatment of a mild/moderate knee bleed, while six doses would be needed to treat a severe knee bleed. In contrast, a single 40 or 80 U kg−1 dose of N9-GP was predicted to provide effective hemostasis for the treatment of mild/moderate and severe knee bleeds, respectively. N9-GP may also allow for a significant reduction in the dose level and frequency for the on-demand treatment of intracranial hemorrhage. On the basis of the modeling findings, the initial dose levels of N9-GP chosen for clinical trials on the treatment of mild/moderate and severe bleeding episodes were 40 and 80 U kg−1 N9-GP, respectively. However, decisions on re-dosing are based on the clinical presentation and the judgment of the investigator in addition to measured N9-GP levels, if required.

Simulations of N9-GP to cover major surgery showed that, if target levels recommended by the WFH were accepted as an illustrative example, then three infusions only could be used to cover a procedure and the 2-week postoperative period, as opposed to 17 infusions using either rFIX or pdFIX. This potentially offers significant benefit to patients, especially those with difficult venous access. It may also allow earlier discharge from hospital. Many joint replacement procedures require prolonged periods of postoperative physiotherapy and fewer infusions would also be needed for this period. It is important to recognize, however, that these simulations assume that clearance during surgery is similar to that seen in the non-bleeding state and this is unlikely to be the case. More rapid clearance on the day of surgery would be expected and so treatment would need to be guided by regular FIX monitoring. The more rapid clearance expected at the time of surgery would affect rFIX and pdFIX as well as N9-GP and so the potential benefit of a longer-acting agent would remain. The clinical efficacy of N9-GP in major surgery is currently being investigated in a clinical trial.

As tailoring of the dose of FIX according to the individual patient pharmacokinetics has the potential to markedly improve the efficacy and cost-effectiveness of prophylactic treatment, it has been recommended that to optimize therapy and maintain adequate trough levels, individual pharmacokinetic data are needed for each patient [30]. However, such tailored dose regimens must be balanced against the practical difficulties, including frequent sampling, associated with generally accepted pharmacokinetic sampling strategies. A population-based model, such as the one reported here, may allow tailored prophylaxis for patients with hemophilia B with less frequent sampling, as reported for patients with hemophilia A [30,31]. Such sparser sampling may be especially useful for pediatric patients.

A potential limitation of our model is that the data used were single-dose pharmacokinetic data. The data utilized covered a 4-week period (corresponding to approximately seven N9-GP half-lives), so N9-GP CL is well characterized, and potential N9-GP accumulation, which may be anticipated with repeated dosing, was incorporated in the modeling. However, any time-dependent changes in the parameters would not have been captured. As the patients in the clinical trial were in a non-bleeding state, it is possible that the model may not apply to actively bleeding patients or in a surgical setting. However, our method of population-based pharmacokinetic modeling has been utilized in other clinical trials in patients with hemophilia A and B to predict the implications for factor levels of various doses and frequencies during prophylaxis [15,30,32].

The simulations presented here suggest that N9-GP may allow prophylaxis with less frequent dose regimens, enhanced on-demand treatment of acute bleeds with a single dose, and surgery without the need for continuous infusion or multiple injections for patients with hemophilia B. While other FIX molecules with prolonged half-lives are also under investigation [33], these new agents are likely to show different pharmacokinetic properties to N9-GP and, as such, the simulations presented here cannot be translated across to these other molecules. Instead, each new FIX molecule will require its own pharmacokinetic model to be developed. It would be interesting to compare the simulations performed here for N9-GP with those for other products under the same clinical scenarios.

With the possibility of prophylaxis with N9-GP requiring less frequent dose regimens (which may obviate the need for indwelling central venous catheters), patients (and/or their parents) may consider switching from on-demand treatment regimens to a prophylactic regimen, which could result in less frequent joint bleeds and fewer life-threatening bleeds, and would be likely to have a positive impact on patient quality of life.

Acknowledgements

The authors wish to acknowledge the hemophilia B patients, investigators and study nurses who participated in the phase 1 trial. This modeling was supported by Novo Nordisk A/S. Medical writing assistance was provided by Sharon Eastwood, of PAREXEL, and Lene Klixbüll Amby, Novo Nordisk A/S, in compliance with international guidelines for good publication practice, and was financially supported by Novo Nordisk A/S.

Disclosure of Conflict of Interests

P.W. Collins has received research support and honoraria from, and acted as a consultant and scientific advisory board member for, Baxter, Bayer Healthcare, CSL Behring, Inspiration Pharmaceuticals and Novo Nordisk. P.W. Collins has also acted as a speaker for Baxter, Bayer Healthcare, CSL Behring and Novo Nordisk. K. Knobe and T. Colberg are Novo Nordisk employees. A. Groth, J. Møss and E. Watson are Novo Nordisk employees and shareholders.

Ancillary