Body Mass Index Best Predicts Recovery of Recombinant Factor VIII in Underweight to Obese Patients with Severe Haemophilia A

Abstract

Background Factor VIII (FVIII) products are usually dosed according to body weight (BW). This may lead to under- or over-dosing in underweight or obese patients, respectively.

Objective This article evaluates the pharmacokinetics (PK) of recombinant FVIII concentrate, particularly recovery, in relation to body mass index (BMI) and other body composition descriptors.

Materials and Methods Thirty-five previously treated adults with severe haemophilia A from five BMI categories (underweight, normal, overweight, obese class I and II/III) were included. PK was evaluated after 50 IU per kilogram of BW single-dose recombinant FVIII (turoctocog alfa). The body composition variable was based on measurements of weight, height, bioimpedance analysis, and dual-energy X-ray absorptiometry. A dosing model was derived to achieve similar peak FVIII activity levels across BMI categories.

Results A statistically significant positive association between BMI and C_30min, IR_30min, and AUC_0–inf was observed; CL and V_ss showed a significant negative association with BMI; t_½ was independent of BMI and other parameters. The dosing model introduced a correction factor ‘M’ for each BMI category, based on linear regression analysis of C_30min against BMI, which ranged from 0.55 for underweight to 0.39 for obese class II/III. This model achieved similar peak FVIII activity levels across BMI categories, estimating an average dose adjustment of +243.3 IU (underweight) to –1,489.6 IU (obese class II/III) to achieve similar C_30min.

Conclusion BMI appears to be the best predictor of recombinant FVIII recovery; however, PK endpoints were also dependent on other body composition variables. The model demonstrated that dosing can be adjusted for individual BMI to achieve better FVIII predictability across BMI categories.

Keywords

body mass index - dosing model - recombinant factor VIII - haemophilia - pharmacokinetics

Introduction

Factor VIII (FVIII) products have a long-standing and well-established role in the treatment of haemophilia A as long-term prophylaxis to protect against bleeding episodes, for on-demand treatment to control a bleed or for providing haemostatic cover for surgery.[1] Correct dosing of FVIII is crucial to avoid under-dosing, which may lead to inadequate bleed control, or overdosing leading to a waste of FVIII product.[2]

FVIII products are typically dosed according to total body weight (BW),[2] [3] meaning that heavier patients receive a proportionally higher drug dose than patients with normal weight.[4] The growing population of overweight and obese patients with haemophilia is a clinical concern,[3] [5] and arthropathy occurs more commonly in overweight and obese patients with haemophilia;[6] though it is unclear why joint disease is increased in this patient population.[4]

FVIII is primarily distributed in plasma, with only a small fraction (∼14%) circulating outside the vascular system.[1] The ratio of plasma volume to BW usually decreases with increasing severity of obesity,[7] as increased BW is accounted for by more fatty tissue, which contains less blood volume than muscle of the same weight.[6] Overweight/obese individuals, therefore, typically have a lower plasma volume per kilogram of BW, while underweight individuals have a higher plasma volume per kilogram of BW than normal-weight individuals. Hence, weight-based dosing of FVIII would introduce a systematic error of unknown magnitude and may not result in the desired FVIII plasma levels in people with varying BW and/or body compositions. It has been suggested that dosing should consider fat mass as well as BW.[2]

Few studies have investigated the impact of different morphometric parameters on FVIII pharmacokinetics (PK),[8] [9] [10] [11] [12] and clinical guidance for the dosing of FVIII in patients with haemophilia according to body composition is lacking. Data from a PK trial with the recombinant FVIII (rFVIII) turoctocog alfa (NovoEight, Novo Nordisk, Bagsværd, Denmark) in a small subset of patients with a body mass index (BMI) ≥ 30 kg/m² indicated that typical weight-based dosing leads to higher post-dose FVIII levels in overweight/obese (vs. normal-weight) patients.[13]

Here, we report the findings of a larger PK trial that assessed whether a relationship between PK (particularly recovery) and body composition could be identified and utilised to improve the predictability of plasma FVIII activity after treatment with turoctocog alfa in a dosing model.

Materials and Methods

Trial Objectives and Endpoints

The primary objective of the trial was to evaluate the single-dose PK (particularly recovery) of turoctocog alfa in relation to BMI in patients with severe haemophilia A. Secondary objectives were to evaluate the single-dose PK of turoctocog alfa in relation to other BW and composition variables as potential PK predictors, and to add to the established safety record of turoctocog alfa.

The primary PK endpoint was FVIII activity at 30 minutes (C_30min). Key secondary PK endpoints comprised incremental recovery at 30 minutes (IR_30min), area under the FVIII activity time curve from 0 to infinity (AUC_0–inf), terminal half-life (t_½), clearance per kilogram of BW (CL), and apparent volume of distribution per kilogram of BW at steady state (V_ss) ([Supplementary Material], available in the online version). The safety endpoint was the number of adverse events (AEs) up to day 28.

Patients

Male patients aged ≥ 18 years with severe haemophilia A (FVIII activity < 1%) and > 150 exposure days to FVIII products were included. Patients with a current/past history of FVIII inhibitors (≥ 0.6 Bethesda Units) or any known coagulation disorder (apart from haemophilia A) were excluded. At visit 1, BMI was calculated based on BW and height. Patients were then grouped into five BMI categories: underweight (< 18.5 kg/m²), normal (18.5–24.9 kg/m²), overweight (25–29.9 kg/m²), obese class I (30–34.9 kg/m²), or obese class II/III (≥ 35 kg/m²). Recruitment was stratified by BMI group to ensure a balance between the five categories.

Trial Design

The PK trial was a multi-national, multi-centre, open-label, single-arm trial investigating the single-dose PK of turoctocog alfa in relation to BMI in previously treated adults with severe haemophilia A without inhibitors. The trial was conducted between October 2016 and June 2017 at 13 sites in seven countries (Austria, Bulgaria, Germany, Serbia, Spain, Taiwan and the United States).

Turoctocog alfa was supplied as a sterile, freeze-dried powder in single-use vials and reconstituted in 4.3 mL of 0.9% sodium chloride prior to intravenous injection. Total trial duration for each patient was 28 days, excluding a screening period of ≤ 28 days. The trial included one PK session per patient. There were six planned visits: one screening visit (visit 1); one 72-hour PK session with four visits (visits 2–5); one follow-up visit (visit 6) that took place 28 ± 7 days after visit 2. At visit 2, each patient received a single intravenous bolus injection of turoctocog alfa at 50 IU per kilogram of BW in a non-bleeding state after a ≥ 96-hour washout period for FVIII-containing products. PK sampling was performed 11 times between –1 hour (i.e., pre-dose) and 72 hours.

The trial was approved by independent ethics committees and institutional review boards, and was conducted in accordance with the Declaration of Helsinki[14] and Good Clinical Practice guidelines.[15] Written informed consent was obtained from all patients prior to any trial-related activity. The trial (NCT02941354) was registered at www.clinicaltrials.gov.

Pharmacokinetic Assessments

Blood samples for PK analysis were collected pre-dose, at 15 and 30 minutes and 1, 3, 6, 9, 24, 28, 48 and 72 hours post-dose. PK endpoints were calculated using FVIII activity and measured by one-stage clotting (activated partial thromboplastin reagent, SynthASil [Instrumentation Laboratory, Milan, Italy]) and chromogenic (Coamatic FVIII kit [Instrumentation Laboratory]) assays performed on a Coasys Plus C analyser (Roche Diagnostics, Mannheim, Germany).[16] Human plasma standard, calibrated against World Health Organization international FVIII standards, was used to calibrate both assays. All PK-derived parameters were determined using non-compartmental methods, as such methods are consistent with previous studies,[16] [17] simple to use and require no assumptions to be made.

Measures of Body Composition and PK Predictors

Measurements of body composition and other potential predictors (ideal BW [IBW], body surface area, lean BW, adjusted BW based on fat-free mass measurements from bioimpedance analysis [BIA], predicted blood volume) were investigated for their association with PK parameters based on the chromogenic assay.

Measurements of body composition: Body composition was measured at either visit 3, 4 or 5 using dual-energy X-ray absorptiometry (DXA) and BIA. Where possible, both measurements were taken at the same visit. The whole-body DXA scan, conducted at a local imaging site, measured total fat mass, lean body mass and body fat percentage. BIA, conducted at the trial site using a Tanita DC 430 SMA Body Composition Analyzer (Tanita Corporation, Tokyo, Japan) supplied by Novo Nordisk, measured total fat mass, fat-free mass and body fat percentage. The average of two measurements was used for each parameter to reduce variability.

In addition to the above measurements, body composition variables were calculated using formulae based on BW and/or height measurements ([Table 1]).

Table 1 Formulae for key calculated predictors

Parameter	Formula
BMI (kg/m2)	BMI = weight (kg)/[height (m)][2]
IBW (kg)[2]	IBW = height (cm) − 100–[height (cm) – 150]/4
BSA (m2)[18]	BSA = BW0.425 × height in cm0.725 × 0.007184
LBW (kg)[19]	LBW = (9,270 × BW)/(6,680 + [216 × BMI])
ABW (kg), based on FFM measurements from BIA	ABW = FFM + 40% (BW-FFM)
PBV (L), based on indexed blood volume[20]	PBV = InBV × BW/1,000

Abbreviations: ABW, adjusted body weight; BIA, bioimpedance analysis; BMI, body mass index; BSA, body surface area; BW, body weight; BV, blood volume; FFM, fat-free mass; IBW, ideal body weight; LBW, lean body weight; PBV, predicted blood volume.

Relationship between coagulation parameters and BMI: The influence of BMI on different coagulation parameters, namely pro-thrombin time, pro-thrombin fragment 1 + 2, von Willebrand factor (vWF), plasminogen activator inhibitor-1 (PAI-1) and plasmin-a2-antiplasmin (PAP) complex, was investigated.

Relationship between t_½, blood group and vWF: The relationship between t_½, blood group and vWF was also determined.

Statistical Methods

There were no formal statistical power calculations with respect to sample size. Thirty-five patients were planned to be included in the trial so that 6 ± 1 patients could be included in each of the five BMI categories at baseline. All dosed patients with data after dosing were included in the full and safety analysis sets.

Primary analysis comprised a linear regression of the primary endpoint (C_30min) against BMI (BMI was treated as a continuous variable). Alternative parametric functional models for C_30min were also explored as part of the primary analysis, covering quadratic and linear regression on logarithm-transformed predictors. No formal hypotheses were tested; however, linearity was determined by fitting linear, quadratic and log-linear models, and comparing the R ² for these models to assess goodness-of-fit; p-values assessed statistical significance. Other conditions of validity (homoscedasticity and outlier assessment) were also evaluated by visual inspection. C_30min was also analysed using measured and calculated predictors in the regression analyses. Measured predictors were body fat percentage (measured by DXA and BIA), lean body mass (measured by DXA), fat-free mass (measured by BIA) and total body fat (measured by DXA and BIA). The mean of two measurements was used for all BIA assessments, unless only one was available. Calculated predictors comprised IBW, body surface area, lean BW, adjusted BW and predicted blood volume. A selection of the analyses conducted for the primary endpoint (C_30min) was also conducted for the secondary PK endpoints (IR_30min, AUC_0–inf, t_½, CL and V_ss); all analyses with t_½ were adjusted for age and blood group (O and Non-O). Non-compartmental analysis of PK data was performed using SAS (version 9.4; SAS Institute Inc., Cary, North Carolina, United States), using code made to mimic algorithms used in Phoenix WinNonlin (Certara, Princeton, New Jersey, United States).

All AEs were summarised and listed.

Dosing Model

The required FVIII dose for on-demand treatment is currently determined using the following formula: dose = BW (kg) × 100 (as desired level is 1 IU/mL) × 0.5.[21] [22] This formula indicates that, for a patient to have a C_30min of 1 IU/mL at complete washout (1 IU/mL relates to 100% FVIII activity [i.e., normal levels]), the dose should be 50 IU per kilogram of BW. The proposed new dosing model derived from the results of the current trial used C_30min (as the goal was to achieve similar peak FVIII activity levels across BMI categories) and BMI, because the regression for BMI gave the largest R ² values among the tested predictors; thus, a model using BMI has the largest potential to reduce variability in C_30min. Therefore, the new dosing model was based on linear regression of FVIII activity at 30 minutes, measured using the one-stage clotting assay, with BMI as a continuous predictor. The one-stage clotting assay was used because it is more commonly used for routine FVIII monitoring. The dosing model used the following formula:

where ‘M’ is the correction factor for dosing that is expected to provide similar C_30min across patients with different BMI. ‘M’ was in the following form:

where X is the average BMI in this trial for a given BMI category and were the intercept and slope parameter estimates obtained from a linear regression analysis of C_30min against BMI and scaled so that a desired FVIII rise of 100% corresponded to the average FVIII activity in the normal BMI category. Further details of how linear regression was used to calculate the correction factor (‘M’) and thus establish the new dosing model are provided in the [Supplementary Material], (available in the online version).

The reference value for C_30min (based on the one-stage clotting assay) was the mean value across all normal-weight patients in the trial. With this, it was intended that the proposed dosing model would give similar C_30min values after dosing with turoctocog alfa across all BMI categories and irrespective of other descriptors of body composition in patients with haemophilia A. The dosing model was assessed by scaling the original data according to the new dosing model ([Supplementary Material], available in the online version) and re-analysing the relationship between PK and body measurements.

Results

Patients

Thirty-five patients received a single dose of turoctocog alfa. The overall mean age was 37.4 years (range: 23.0–57.0 years); mean ages were similar across all BMI categories. The majority of patients were white (N = 32 [91.4%]). Patient demographics and baseline characteristics are shown in [Table 2].

Table 2 Patient demographics and baseline characteristics

	BMI categories
	Underweight < 18.5 kg/m2	Normal weight 18.5–24.9 kg/m2	Overweight 25.0–29.9 kg/m2	Obese class I 30.0–34.9 kg/m2	Obese class II/III ≥ 35.0 kg/m2	Total
N	5	7	9	7	7	35
Age (y)
Mean (SD)	34.0 (8.9)	42.4 (11.0)	34.4 (11.4)	35.9 (10.6)	40.3 (8.1)	37.4 (10.2)
Race, N (%)
White	5 (100.0)	5 (71.4)	8 (88.9)	7 (100.0)	7 (100.0)	32 (91.4)
Asian	0 (0.0)	1 (14.3)	1 (11.1)	0 (0.0)	0 (0.0)	2 (5.7)
Black or African American	0 (0.0)	1 (14.3)	0 (0.0)	0 (0.0)	0 (0.0)	1 (2.9)
Height (m)
Mean (SD)	1.81 (0.07)	1.70 (0.09)	1.74 (0.10)	1.76 (0.10)	1.78 (0.06)	1.76 (0.09)
BW (kg)
Mean (SD)	55.26 (5.86)	63.14 (5.95)	82.98 (12.5)	102.3 (12.7)	131.9 (17.4)	88.69 (29.3)
BMI (kg/m2)
Mean (SD)	16.78 (0.77)	21.77 (1.61)	27.12 (1.78)	33.06 (2.04)	41.41 (4.86)	28.62 (8.65)
BFP (%, DXA)
Mean (SD)	17.2 (6.4)	27.1 (8.7)	28.9 (8.3)	35.1 (6.9)	44.7 (8.7)	30.9 (11.3)
BFP (%, BIA)
Mean (SD)	14.1 (6.4)	23.2 (5.2)	25.2 (3.6)	32.2 (3.0)	42.1 (3.8)	27.7 (9.8)
IBW[a] (kg)
Mean (SD)	73.4 (5.01)	65.3 (7.11)	68.3 (7.40)	69.3 (7.50)	71.3 (4.72)	69.2 (6.74)
BSA[a] (m2)
Mean (SD)	1.7 (0.12)	1.7 (0.13)	2.0 (0.20)	2.2 (0.20)	2.5 (0.17)	2.0 (0.32)
LBW[a] (kg)
Mean (SD)	49.6 (4.50)	51.7 (4.96)	61.3 (7.95)	68.5 (7.82)	78.2 (6.09)	62.5 (12.10)
ABW[a] (kg)
Mean (SD)	50.6 (4.56)	54.4 (5.29)	68.8 (9.36)	82.1 (8.53)	101.0 (12.44)	71.7 (19.68)
PBV[a] (L)
Mean (SD)	4.4 (0.38)	4.5 (0.43)	5.2 (0.67)	5.8 (0.67)	6.7 (0.56)	5.4 (1.01)

Abbreviations: ABW, adjusted body weight; BFP, body fat percentage; BIA, bioimpedance analysis; BMI, body mass index; BSA, body surface area; BW, body weight; DXA, dual-energy X-ray absorptiometry; IBW, ideal body weight; LBW, lean body weight; PBV, predicted blood volume; SD, standard deviation.

^a See [Table 1] for formulae for calculated predictors.

Pharmacokinetics

FVIII activity: FVIII activity profiles showed an expected exponential decline over time with a clear trend of increasing FVIII activity levels from underweight to obese class II/III patients. The summary of key PK endpoints (based on the chromogenic assay) by BMI category is presented in [Table 3].

Table 3 Summary of key PK parameters by BMI category (chromogenic assay)

	BMI categories
	Underweight < 18.5 kg/m2	Normal weight 18.5–24.9 kg/m2	Overweight 25.0–29.9 kg/m2	Obese class I 30.0–34.9 kg/m2	Obese class II/III ≥ 35.0 kg/m2
N	5	7	9	7	7
C30min (IU/mL)
Geometric mean (CV%)	1.24 (17.3)	1.65 (9.4)	1.66 (15.9)	1.79 (15.1)	1.96 (13.2)
Range	1.02–1.58	1.47–1.90	1.32–2.02	1.46–2.31	1.59–2.28
IR30min (IU/mL)/(IU/kg)
Geometric mean (CV%)	0.022 (17.7)	0.029 (9.5)	0.029 (16.0)	0.032 (15.5)	0.035 (13.3)
Range	0.018–0.028	0.026–0.034	0.023–0.036	0.026–0.041	0.028–0.041
AUC0–inf (IU × h/mL)
Geometric mean (CV%)	17.8 (32.5)	26.1 (33.6)	21.3 (44.1)	24.9 (58.0)	29.7 (25.6)
Range	14.0–30.9	17.1–40.4	13.3–39.0	11.0–41.1	19.0–42.4
t½ (h)
Geometric mean (CV%)	11.5 (21.2)	10.9 (38.5)	9.2 (35.2)	9.9 (38.1)	10.3 (21.7)
Range	8.3–15.0	5.7–15.3	5.8–15.5	5.2–13.6	8.3–15.4
CL (mL/h/kg)
Geometric mean (CV%)	3.16 (33.1)	2.15 (33.3)	2.64 (44.3)	2.25 (58.6)	1.88 (25.7)
Range	1.80–4.03	1.37–3.30	1.44–4.28	1.36–5.17	1.31–2.94
Vss (mL/kg)
Geometric mean (CV%)	48.1 (22.3)	31.0 (15.0)	31.1 (18.6)	28.5 (17.9)	25.4 (16.0)
Range	35.3–61.1	24.0–36.9	23.0–41.1	22.3–37.3	19.1–32.5

Abbreviations: AUC_0–inf, area under the FVIII activity-time curve from 0 to infinity; BMI, body mass index; C_30min, FVIII activity at 30 minutes; CL, clearance; CV, coefficient of variation; IR_30min, incremental recovery at 30 minutes; PK, pharmacokinetics; SD, standard deviation; t_½, terminal half-life; V_ss, apparent volume of distribution at steady state.

Association of C_30min and AUC_0-inf with Measurements of Body Composition

C_30min, the primary PK endpoint, was found to have a significant positive linear association with measurements of body composition, namely body fat percentage, lean body mass, fat-free mass and total fat mass (all measured by DXA and BIA) ([Supplementary Table S1], available in the online version). Chromogenic and one-stage clotting assays yielded similar results in the statistical analyses of dependencies. While all measurements were significant predictors of C_30min, none had a higher R ² value (indicating how predictive each association is) than BMI, suggesting that BMI is the best predictor of PK endpoints ([Fig. 1A]). The R ² values associated with BIA measurements tended to be higher than those associated with DXA measurements, suggesting that BIA measures give better prediction of PK endpoints than DXA. A statistically significant positive association between AUC_0-inf and BMI was also observed ([Fig. 1B]), as well as with several other measurements ([Supplementary Table S1], available in the online version). Quadratic- and logarithm-transformed models did not add any additional information beyond that provided by the linear model. The association of other PK endpoints with measured predictors is provided in the [Supplementary Material] and [Supplementary Table S1] (available in the online version).

Fig. 1 C_30min and AUC_0–inf versus BMI and IBW (chromogenic assay). (A) C_30min showed significant positive associations with BMI (linear regression line, p < 0.0001; R ²: 0.5086). (B) AUC_0–inf showed significant positive associations with BMI (linear regression line, p < 0.0263; R ²: 0.1409). (C) IBW was not associated with PK endpoints; linear regression line, p = 0.8801 for C_30min; R ², 0.0007 is from the linear regression model; and (D) linear regression line, p = 0.7706 for AUC_0–inf; R ², 0.0026. AUC_0–inf, area under the activity–time curve from 0 to infinity; BMI, body mass index; C_30min, FVIII activity at 30 minutes; IBW, ideal body weight; PK, pharmacokinetic.

Association of PK Endpoints with BMI

All results in this section are based on FVIII measurements derived from the chromogenic assay. Similar results for the primary and secondary endpoints were obtained when using the one-stage clotting assay.

C_30min : A statistically significant positive association between C_30min and BMI, in which C_30min values increased with increasing BMI, was observed. The geometric mean C_30min values ranged from 1.24 IU/mL (coefficient of variation [CV]: 17.3%) in the underweight group to 1.96 IU/mL (CV: 13.2%) in obese class II/III patients ([Table 3]; [Fig. 2A]).

Fig. 2 C_30min, IR_30min, AUC_0–inf, t_½, CL and V_ss by BMI category (chromogenic assay). Statistically significant positive association between BMI and (A) C_30min, (B) IR_30min and (C) AUC_0–inf; lack of association between BMI and (D) t_½; statistically significant negative association between BMI and (E) CL and (F) V_ss. The box and whisker plots are based on standard non-parametric measures, and the circles represent arithmetic means. Arithmetic mean C_30min ranged from 1.25 to 1.98 IU/mL in underweight to obese class II/III patients, respectively. Individual C_30min ranged from 1.02 to 2.31 IU/mL. BMI, body mass index; BW, body weight; C_30min, FVIII activity at 30 minutes; IR_30min, incremental recovery at 30 minutes; V_ss, apparent volume of distribution per kilogram of BW at steady state. BMI categories are derived from BMI calculated at screening: underweight (BMI < 18.5 kg/m²), normal range (BMI: 18.5–24.9 kg/m²), overweight (BMI: 25–29.9 kg/m²), obese class I (BMI 30–34.9 kg/m²) and obese class II/III (BMI ≥ 35 kg/m²).

IR_30min : A statistically significant positive association between IR_30min (derived from C_30min) and BMI was observed. The geometric mean IR_30min increased from 0.022 (IU/mL)/(IU/kg) (CV: 17.7%) in underweight patients to 0.035 (IU/mL)/(IU/kg) (CV: 13.3%) in obese class II/III patients ([Table 3]; [Fig. 2B]).

AUC_0–inf : A statistically significant positive association between AUC_0-inf and BMI was also observed. The geometric mean AUC_0-inf values ranged from 17.8 IU × h/mL (CV: 32.5%) in underweight patients to 29.7 IU × h/mL (CV: 25.6%) in obese class II/III patients ([Table 3]; [Fig. 2C]).

t_½ : The geometric mean ranged from 9.2 hours (CV: 35.2%) to 11.5 hours (CV: 21.2%) across the BMI categories ([Table 3]) and was found to be independent of BMI ([Fig. 2D]) and other predictors ([Supplementary Table S1], available in the online version). In addition, there was no relationship between BMI and the time to 1% of normal FVIII activity ([Supplementary Table S2] and [Supplementary Fig. S1] [available in the online version]).

CL: Derived from AUC_0-inf, CL showed a significant negative association with BMI, as expected. Underweight and obese class II/III groups showed a geometric mean CL of 3.16 (CV: 33.1%) and 1.88 mL/h/kg (CV: 25.7%), respectively ([Table 3]; [Fig. 2E]).

V_ss : A significant negative association between V_ss per kilogram of BW and BMI was observed. The highest geometric mean V_ss was found in the underweight group (48.1 mL/kg [CV: 22.3%]) and the lowest in the obese class II/III patients (25.4 mL/kg [CV: 16.0%]) ([Table 3]; [Fig. 2F]).

Association of C_30min with Body Surface Area, Lean BW, Adjusted BW and Predicted Blood Volume

We observed a significantly positive association of C_30min and derived parameter IR_30min with most potential predictors of PK parameters, namely body surface area, lean BW, adjusted BW and predicted blood volume ([Supplementary Table S1], available in the online version). However, there was no significant association of C_30min with IBW (p = 0.8801; R ²: 0.0007) ([Fig. 1C]) and AUC_0-inf with IBW (p = 0.7706; R ²: 0.0026) ([Fig. 1D]). The R ² values for PK predictors were all lower than that for BMI and most other body composition measurements. As with the assessments, quadratic- and logarithm-transformed models of calculated predictors did not add any additional information beyond that provided by the linear model. The association of other PK endpoints with body surface area, lean BW, adjusted BW and predicted blood volume is provided in the [Supplementary Material] (available in the online version).

Coagulation Parameters and BMI

We found no association between coagulation parameters (i.e., pro-thrombin time, pro-thrombin fragment 1 + 2, vWF and PAP [data not shown]) and BMI. However, increased levels of PAI-1 in obese class II/III patients were observed.

Relationship between t_½, Blood Group and vWF

We observed a trend of decreasing t_½ with decreasing vWF antigen levels, with a large variation in t_½ and vWF levels ([Fig. 3A]). Additionally, patients with blood group O tended to have a shorter t_½ than patients with other blood groups ([Fig. 3B]). Adjusting for BMI did not change the significant influence of vWF on t_½ (data not shown).

Fig. 3 t_½ versus vWF (A) and blood group (B) (chromogenic assay) t_½ versus blood group is a box plot (B), ‘ + ’ represents the mean and the horizontal lines within the box represents the median. t_½, terminal half-life; vWF, von Willebrand factor.

Dosing Model

The above PK results were used to derive a new dosing model showing the potential for dose adjustments that would produce more uniform C_30min values across BMI categories. The dosing model used the following formula:

where ‘M’ is the correction factor for dosing that is expected to provide similar C_30min across patients with different BMI:

• Underweight: 0.55

• Normal weight: 0.51

• Overweight: 0.47

• Obese class I: 0.43

• Obese class II/III: 0.39

Between-patient variation in C_30min was reduced by 40.5%, and between-patient variation for AUC_0-inf was reduced by 26.4%. The observed total dose based on an administered dose of 50 IU per kilogram of BW and the new total dose based on the model by BW and BMI category are shown in [Table 4], along with potential dosing adjustment based on the model; the mean total dose adjustment varied from +243.3 IU in underweight patients to –1489.6 IU in obese class II/III patients. None of the other investigated PK predictors further improved the dosing model.

Table 4 Average total dose adjustment (M) across all BMI groups based on the proposed model when dosing turoctocog alfa at 50 IU/kg

	BMI categories
	Underweight < 18.5 kg/m2	Normal weight 18.5–24.9 kg/m2	Overweight 25.0–29.9 kg/m2	Obese class I 30.0–34.9 kg/m2	Obese class II/III ≥ 35.0 kg/m2
N	5	7	9	7	7
Observed dose (IU)
Mean dose (SD)	2,780.2 (277.4)	3,214.1 (279.8)	4,201.7 (622.5)	5,144.1 (639.2)	6,642.9 (838.1)
New dose using dosing model
Mean dose (SD)	3,023.5 (282.5)	3,216.3 (302.9)	3,890.9 (515.7)	4,424.5 (508.4)	5,153.3 (429.8)
Change from observed
Mean (SD)	243.3 (25.3)	2.2 (84.5)	–310.7 (126.4)	–719.6 (182.5)	–1,489.6 (469.1)
Correction factor (M)	0.55	0.51	0.47	0.43	0.39

Abbreviations: BMI, body mass index; M, the correction factor to dosing that is expected to provide similar C_30min across patients with different BMI (i.e., the factor for multiplication with body weight to obtain the dose that is predicted to give the same C_30min from the one-stage clotting assay as the average in the normal BMI category in this study); SD, standard deviation.

Note: Dosing model was based on linear regression results for one-stage clotting assay versus BMI. Results from the one-stage clotting assay were used for the dosing model as these seemed the most relevant, showing the greatest decrease in standard deviation (vs. results using the chromogenic assay).

Safety

No inhibitor development or AEs leading to death or withdrawal were reported, and no new safety concerns (including no adverse drug reactions) were observed.

Discussion

The current trial was conducted to investigate the single-dose PK of the rFVIII turoctocog alfa in relation to BMI in previously treated adults with severe haemophilia A, with a focus on recovery. Overall, the trial showed increased C_30min, AUC_0–inf and IR_30min with increasing BMI, while CL and V_ss decreased with increasing BMI. Of all the measured and calculated predictors assessed (IBW, body surface area, lean BW, adjusted BW and predicted blood volume), BMI was the most robust and best predictor of PK endpoints. The exception to this was t_½, which seemed to be independent of all predictors.

The Dosing Model

The combined BW and BMI dosing model resulted in predictable PK parameters in patients with different body compositions. The innovative component of the dosing model was incorporation of the correction factor ‘M’ to calculate dosage adjustments based on the difference in C_30min across five BMI categories. The model reduced total variation by 40.5% for C_30min and by 26.4% for AUC_0-inf. No other predictor significantly improved the dosing model beyond the addition of BMI (data not shown). The new dosing model also indicates that, when using current standard dose calculations, underweight patients might receive an insufficient FVIII dose for effective bleed control, overweight/obese patients might receive more FVIII than needed and normal-weight patients would require little or no dose adjustment compared with the current dosing recommendation ([Table 4]).

Why BMI Performs Well as a Predictor of FVIII PK

IBW has been suggested instead of actual BW for FVIII dose calculations.[10] [11] A recent study performed on simulated patients found that dosing based on IBW was cost effective and provided the highest proportion of time spent above 1% FVIII on standard prophylaxis.[3] In contrast, in the current trial, which included real patients and measured PK parameters, IBW was not found to be a useful metric for FVIII dose adjustment when the goal is to achieve uniform PK across weight categories. IBW is directly correlated to height and is not based on any aspect of body composition or weight. In the current trial, IBW was the only predictor that did not show any association with any PK endpoints. A possible explanation for BMI being a good predictor of PK endpoints might be that the trial participants had body compositions (i.e., proportions of highly vascularised, lean tissue vs. poorly vascularised, fatty tissue) that were typical of their BMI category. In this way, some fatty tissues are accounted for by BMI, whereas IBW would completely ignore the fatty tissue and thereby over-adjust the dose. FVIII recovery has been found to increase with increasing BMI and BW.[9] [11] [23] A regression-tree analysis conducted by Henrard et al to examine the impact of various morphometric parameters (different parameters to those chosen in the current trial) on FVIII recovery among 201 adults (> 18 years) with haemophilia A found BMI to be the strongest predictor of FVIII recovery, in line with current findings.[11] The regression-tree analysis also found significantly different (p < 0.001) median FVIII recovery values of 1.60, 2.14 and 2.70 among patients with BMI < 20.3, 20.3 to 29.5 and ≥ 29.6 kg/m², respectively.

Exploring the Measures of Body Composition

Although DXA can measure fat and muscle compartments,[24] [25] BIA body composition measurements gave slightly better predictions (R ²) of PK parameters than DXA measurements in the current trial, which is unexpected. As FVIII activity is primarily distributed in blood, the BIA assessment may have performed better than DXA due to the dependency of BIA on the body's fluid content (including blood). BIA measurements can also be affected by factors such as physical activity, hydration status and consumption of food and beverages.[24] To limit the interference of these factors in the current trial, patients were advised to drink fluid normally during the 24 hours prior to the scan (∼2 L), avoid alcohol consumption (within 12 hours prior to the scan), avoid large meals (within 4–6 hours prior to the scan) and were asked to empty their bladder prior to the scan.

Measures of fat and lean mass directly obtained by DXA and BIA performed weaker as predictors of FVIII PK than the measurements of BMI. This may be explained by the fact that these measures account for lean or fatty tissue but not for both, while blood is unequally distributed in both tissues.

Relationship of BMI to t_½, Blood Group and vWF

Increased adipose tissues in obese patients might contribute to the release of adipokines and cytokines, resulting in changes in coagulation factor levels in the liver.[26] [27] Considering this, the influence of BMI on different coagulation parameters in the current trial was explored. However, no influence of BMI on coagulation parameters was seen, except increased PAI-1 levels in the obese class II/III patients versus non-obese patients, consistent with a study that also found elevated PAI-1 levels in obese individuals, possibly indicating reduced fibrinolysis in this group.[5]

When investigating the relationship between t_½, blood group and vWF, we found a large variation of t_½ and vWF antigen levels. Nevertheless, t_½ appeared to decrease with decreasing vWF levels, which would be expected based on the tight association between FVIII and vWF; vWF maintains FVIII stability and prevents its degradation and CL.[28] Patients with blood group O have lower vWF levels than those with other blood groups.[29] As expected, patients with blood group O in the current trial had lower vWF levels and therefore a shorter t_½ than patients with other blood groups.

Limitations

The study showed an impact of BMI on peak levels after dosing, but not on t_½. Achieving certain peak levels is primarily relevant for managing surgery or acute bleeds in patients with haemophilia. Routine prophylaxis, however, targets certain trough levels that are mainly influenced by individual t_½. Therefore, our data, including the new dosing model, are not relevant for routine prophylaxis dosing.

Chromogenic and one-stage clotting assays yielded similar results in the PK assessments, although there was a shift in FVIII concentration values between the two assays. As the chromogenic assay provided more stable results and led to exclusion of fewer samples than the one-stage assay, results obtained with the chromogenic assay have been presented for PK analyses. However, results from the one-stage clotting assay (more commonly used for routine FVIII monitoring) were used for the dosing model as they showed the greatest decrease in standard deviation (vs. results using the chromogenic assay); hence, we presented results based on the one-stage assay for the dosing model.

Given the negative association observed between age and dose requirement in the literature,[30] lack of adjustment for age in the current study may also be a potential limitation of our findings; age did not affect C_30min, AUC and CL, but the range of ages in the current trial may have been too small to show differences.

Another limitation of the trial was that the study population did not include extremely underweight patients (for example, elderly individuals or those with low fat mass) or extremely muscular individuals (such as body builders and athletes) or severely anaemic patients (to factor haematocrit levels); hence, the results are only applicable to a population with similar attributes. There is a need to validate the model in all BMI populations. BMI may be a poorer predictor of FVIII PK in patients with differing body compositions (e.g., those with greater muscle mass). However, high BMI due to a high volume of muscle mass is rare in patients with severe haemophilia.

Finally, some studies found two-compartmental analysis superior to non-compartmental analysis (e.g., Björkman et al[31]), while others did not (e.g., Morfini et al[32]). For turoctocog alfa PK data, there was no clear distribution phase or divergent trend regarding observed PK overlaid with predicted PK based on non-compartmental methods. As such, non-compartmental analysis had a good fit for turoctocog alfa PK data.

Conclusion

The trial confirmed that the PK of FVIII (C_30min, AUC, IR_30min, CL and V_SS) depends on BMI and body composition. A statistically significant positive association between C_30min and BMI, and between C_30min and body surface area, lean BW, adjusted BW and predicted blood volume, was observed. Furthermore, of the parameters assessed in the trial, BMI was found to be the best predictor of PK endpoints, except for t_½, which appears to be independent. We also propose a novel and simple dosing model, which allowed improved predictability of plasma FVIII activity after treatment with rFVIII across all BMI categories. The dosing model introduced a correction factor ‘M’ for each BMI category ranging from 0.55 for underweight to 0.39 for obese class II/III. Future studies to explore the relationship between FVIII activity levels and clinical efficacy are warranted.

Conflict of Interest

A.T. has received research support, honoraria or consultation fees from Alnylam, Bayer, Biogen Idec, Biotest, Boehringer Ingelheim, Bristol-Myers Squibb, CSL Behring, Leo Pharma, Novo Nordisk, Octapharma, Pfizer, Roche, Shire and SOBI. A.R.C. has received reimbursement for attending symposia/congresses and/or honoraria for speaking or consulting from Novo Nordisk and other companies. G.G. has received reimbursement for attending symposia/congresses and/or honoraria for speaking or consulting from Novo Nordisk and other companies. V.J.-Y. has received reimbursement for attending symposia/congresses and/or honoraria for speaking and/or honoraria for consulting, and/or funds for research from Bayer, CSL Behring, Grifols, Novo Nordisk, Octapharma, Pfizer, Roche, Shire and SOBI. M.P. is a paid consultant for Novo Nordisk A/S. T.L. has received investigator fees as a participant of the clinical trial from Novo Nordisk. M.M. has no conflicts of interest to declare. I.M. is an employee of Novo Nordisk A/S. P.M. has no conflicts of interest to declare. I.P. has received honoraria or consultation fees from Bayer, Biotest, CSL Behring, Novo Nordisk, Octapharma, Pfizer, Roche, Shire and SOBI; also research support from CSL Behring and Novo Nordisk. P.P. is an employee of Novo Nordisk A/S.

Acknowledgments

The authors would like to thank Lars Korsholm for his scientific advice and critical review of the manuscript. Medical writing support was provided by Jo Fetterman, PhD (Parexel, United Kingdom).

Note

All authors confirm they have had full access to the data and contributed to the drafting of this manuscript.

• References

• 1 Björkman S, Berntorp E. Pharmacokinetics of coagulation factors: clinical relevance for patients with haemophilia. Clin Pharmacokinet 2001; 40 (11) 815-832
• 2 Henrard S, Speybroeck N, Hermans C. Body weight and fat mass index as strong predictors of factor VIII in vivo recovery in adults with hemophilia A. J Thromb Haemost 2011; 9 (09) 1784-1790
• 3 McEneny-King A, Chelle P, Henrard S, Hermans C, Iorio A, Edginton AN. Modeling of body weight metrics for effective and cost-efficient conventional factor VIII dosing in hemophilia A prophylaxis. Pharmaceutics 2017; 9 (04) E47
• 4 Young G. New challenges in hemophilia: long-term outcomes and complications. Hematology (Am Soc Hematol Educ Program) 2012; 2012: 362-368
• 5 Tuinenburg A, Biere-Rafi S, Peters M. , et al. Obesity in haemophilia patients: effect on bleeding frequency, clotting factor concentrate usage, and haemostatic and fibrinolytic parameters. Haemophilia 2013; 19 (05) 744-752
• 6 Wong TE, Majumdar S, Adams E. , et al; Healthy Weight Working Group. Overweight and obesity in hemophilia: a systematic review of the literature. Am J Prev Med 2011; 41 (06) (Suppl. 04) S369-S375
• 7 Feldschuh J, Enson Y. Prediction of the normal blood volume. Relation of blood volume to body habitus. Circulation 1977; 56 (4 Pt 1): 605-612
• 8 Björkman S, Blanchette VS, Fischer K. , et al; Advate Clinical Program Group. Comparative pharmacokinetics of plasma- and albumin-free recombinant factor VIII in children and adults: the influence of blood sampling schedule on observed age-related differences and implications for dose tailoring. J Thromb Haemost 2010; 8 (04) 730-736
• 9 Blanchette VS, Shapiro AD, Liesner RJ. , et al; rAHF-PFM Clinical Study Group. Plasma and albumin-free recombinant factor VIII: pharmacokinetics, efficacy and safety in previously treated pediatric patients. J Thromb Haemost 2008; 6 (08) 1319-1326
• 10 Collins PW, Fischer K, Morfini M, Blanchette VS, Björkman S. ; International Prophylaxis Study Group Pharmacokinetics Expert Working Group. Implications of coagulation factor VIII and IX pharmacokinetics in the prophylactic treatment of haemophilia. Haemophilia 2011; 17 (01) 2-10
• 11 Henrard S, Speybroeck N, Hermans C. Impact of being underweight or overweight on factor VIII dosing in hemophilia A patients. Haematologica 2013; 98 (09) 1481-1486
• 12 Kepa S, Horvath B, Reitter-Pfoertner S. , et al. Parameters influencing FVIII pharmacokinetics in patients with severe and moderate haemophilia A. Haemophilia 2015; 21 (03) 343-350
• 13 Zupančić-Šalek S, Ozelo MC, Korsholm L. , et al. An evaluation of the pharmacokinetics of turoctocog alfa in relation to body mass index (BMI) in patients with haemophilia A (HA). Res Pract Thromb Haemost 2017; 1 (01) 671
• 14 World Medical Association. World Medical Association Declaration of Helsinki: ethical principles for medical research involving human subjects. JAMA 2013; 310 (20) 2191-2194
• 15 International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH), Harmonised Tripartite Guideline: Guideline for Good Clinical Practice E6(R1); 1996. Available at: http://apps.who.int/medicinedocs/documents/s22154en/s22154en.pdf . Accessed November 17, 2019
• 16 Kulkarni R, Karim FA, Glamocanin S. , et al. Results from a large multinational clinical trial (guardian™3) using prophylactic treatment with turoctocog alfa in paediatric patients with severe haemophilia A: safety, efficacy and pharmacokinetics. Haemophilia 2013; 19 (05) 698-705
• 17 Lentz SR, Misgav M, Ozelo M. , et al. Results from a large multinational clinical trial (guardian™1) using prophylactic treatment with turoctocog alfa in adolescent and adult patients with severe haemophilia A: safety and efficacy. Haemophilia 2013; 19 (05) 691-697
• 18 Du Bois D, Du Bois EF. A formula to estimate the approximate surface area if height and weight be known. 1916. Nutrition 1989; 5 (05) 303-311
• 19 Janmahasatian S, Duffull SB, Ash S, Ward LC, Byrne NM, Green B. Quantification of lean bodyweight. Clin Pharmacokinet 2005; 44 (10) 1051-1065
• 20 Lemmens HJ, Bernstein DP, Brodsky JB. Estimating blood volume in obese and morbidly obese patients. Obes Surg 2006; 16 (06) 773-776
• 21 Srivastava A, Brewer AK, Mauser-Bunschoten EP. , et al; Treatment Guidelines Working Group on Behalf of The World Federation Of Hemophilia. Guidelines for the management of hemophilia. Haemophilia 2013; 19 (01) e1-e47
• 22 NovoEight^® Summary of Product Characteristics. Novo Nordisk A/S, Bagsværd, Denmark. Available at: https://www.ema.europa.eu/en/documents/product-information/novoeight-epar-product-information_en.pdf . Accessed July 9, 2019
• 23 Henrard S, Hermans C. Impact of being overweight on factor VIII dosing in children with haemophilia A. Haemophilia 2016; 22 (03) 361-367
• 24 Lee SY, Ahn S, Kim YJ. , et al. Comparison between dual-energy X-ray absorptiometry and bioelectrical impedance analyses for accuracy in measuring whole body muscle mass and appendicular skeletal muscle mass. Nutrients 2018; 10 (06) pii: E738
• 25 Singer JP, Peterson ER, Snyder ME. , et al. Body composition and mortality after adult lung transplantation in the United States. Am J Respir Crit Care Med 2014; 190 (09) 1012-1021
• 26 Balistreri CR, Caruso C, Candore G. The role of adipose tissue and adipokines in obesity-related inflammatory diseases. Mediators Inflamm 2010; 2010: 802078
• 27 Fantuzzi G, Mazzone T. Adipose tissue and atherosclerosis: exploring the connection. Arterioscler Thromb Vasc Biol 2007; 27 (05) 996-1003
• 28 Lalezari S, Martinowitz U, Windyga J. , et al. Correlation between endogenous VWF:Ag and PK parameters and bleeding frequency in severe haemophilia A subjects during three-times-weekly prophylaxis with rFVIII-FS. Haemophilia 2014; 20 (01) e15-e22
• 29 Pipe SW. The hope and reality of long-acting hemophilia products. Am J Hematol 2012; 87 (Suppl. 01) S33-S39
• 30 Björkman S, Folkesson A, Jönsson S. Pharmacokinetics and dose requirements of factor VIII over the age range 3-74 years: a population analysis based on 50 patients with long-term prophylactic treatment for haemophilia A. Eur J Clin Pharmacol 2009; 65 (10) 989-998
• 31 Björkman S, Oh M, Spotts G. , et al. Population pharmacokinetics of recombinant factor VIII: the relationships of pharmacokinetics to age and body weight. Blood 2012; 119 (02) 612-618
• 32 Morfini M, Marchesini E, Paladino E, Santoro C, Zanon E, Iorio A. Pharmacokinetics of plasma-derived vs. recombinant FVIII concentrates: a comparative study. Haemophilia 2015; 21 (02) 204-209