Break-through bleeding in relation to predicted factor VIII levels in patients receiving prophylactic treatment for severe hemophilia A

Authors


Peter W. Collins, Arthur Bloom Haemophilia Centre, Department of Haematology, Medical School of Cardiff University, University Hospital of Wales, Heath Park, Cardiff CF14 4XN, UK.
Tel.: +44 2920742155; fax: +44 2920744221.
E-mail: peter.collins@cardiffandvale.wales.nhs.uk

Abstract

Summary. Background: The role of prophylactic factor VIII (FVIII) to decrease hemophilic bleeding and arthropathy is well established. The rationale for this strategy is to convert patients with severe hemophilia A to a moderate clinical phenotype by reducing time spent with a FVIII level <1 IU dL−1. Studies to date, however, have not demonstrated a strong link between FVIII level and the bleeding rate. Objectives: To assess the effect of FVIII level on break-through bleeding in patients with severe hemophilia A on prophylaxis. Patients/methods: This study analysed data from 44 patients aged 1–6 and 99 patients aged 10–65 years with severe hemophilia A (FVIII <1 IU dL−1) who were treated with prophylactic FVIII as part of clinical studies assessing pharmacokinetics, safety and efficacy of a recombinant FVIII (Advate®). Each patient had pharmacokinetic measurements and FVIII infusions recorded, and these were used to calculate time spent with a FVIII below 1, 2 and 5 IU dL−1. Results: The data demonstrate that increasing time with a FVIII below 1 IU dL−1 is associated with increased total bleeds and hemarthroses. Lack of adherence to the intended frequency of FVIII infusion was the most important determinant of low FVIII and increased bleeding. In children aged 1–6 years, the rate of bleeding was also influenced by FVIII half-life and clearance. Conclusions: These data have important implications for the management of patients with severe hemophilia.

Introduction

Hemophilia A is an inherited bleeding disorder caused by a deficiency of factor VIII (FVIII). Patients with severe hemophilia A have a FVIII plasma concentration less than 1 IU dL−1 and experience spontaneous and trauma-induced bleeds. Joint bleeds lead to hemophilic arthropathy resulting in progressive disability. Patients with moderate hemophilia (FVIII level between 1–5 IU dL−1) are characterized by fewer hemarthoses, usually trauma-induced, and a decreased likelihood of developing arthropathy [1]. This clinical observation led to the use of prophylactic FVIII infusions to convert patients’ bleeding phenotype from severe to moderate with the result of decreasing or preventing arthropathy [2].

Retrospective and prospective studies have established that prophylaxis decreases the number of bleeds, severity of arthropathy and improves quality of life [3–9]. Secondary prophylaxis in adults who have already developed hemophilic arthropathy is being used more commonly with the aim of slowing progression of joint deterioration and improving mobility [10].

It is usually assumed that these break-through bleeds while on prophylaxis are dependent on time spent with a low FVIII level. This physiologically plausible hypothesis is supported by the findings of early dosing studies [11–13]. A relationship of this kind is, however, extremely difficult to observe in clinical practice, as it is confounded when dosing is individualized to an optimal outcome in each patient. Accordingly, a large retrospective study on prophylactic treatment [14] could only demonstrate a weak association between time spent with a FVIII level less than 1 IU dL−1 and hemarthroses in normal joints, emphasizing the need for prospective studies.

In patients on prophylaxis, the time spent with a FVIII below a defined level is dependent on the dose and frequency of FVIII infusions and the rate of elimination of FVIII. A better understanding of how FVIII levels and pharmacokinetics affect break-through bleeding in patients on prophylaxis may improve the prescription of cost effective regimens to decrease joint bleeds and improve long-term joint outcomes [15,16].

The aim of this study was to investigate the effect of FVIII levels, as determined by dosage, adherence and individual pharmacokinetics, on the risk of bleeding in patients treated with prophylaxis as part of clinical trials to establish the pharmacokinetics, safety and efficacy of a recombinant FVIII (Advate®; Baxter Healthcare Corporation, Westlake, CA, USA).

Materials and methods

Patient population and treatment regimen

Data for the current study were collated from 143 patients treated in three Advate clinical studies in previously treated patients (PTPs). Data were analysed from the pivotal and continuation studies, which enrolled PTPs aged 10–65 years who had had at least 150 exposure days (ED) [17], and the paediatric study, which enrolled PTPs 1–6 years of age who had experienced at least 50 ED [18]. Patients with a baseline factor VIII <1 IU dL−1 were included in the analyses.

All 99 patients aged 10–65 years and 44 patients aged 1–6 years received prophylaxis for at least part of the study time. Patients aged 10–65 years were prescribed a fixed standard prophylactic regimen (25–40 IU kg−1 at least three times a week) for the first 75 ED after which their regimen could be changed. For the present study, the first 75 exposures and subsequent exposures as long as the standard prophylactic regimen was continued were analysed. The prophylactic regimens in the patients aged 1–6 years were at the discretion of the treating clinician and hence a variety of regimens were used.

Measurements

At study entry, subjects underwent a physical examination and the presence of physical abnormalities was recorded. A formal joint score examination was not performed. Patients were categorized as either having no arthropathy or detectable arthropathy in at least one joint based on the study entry physical examination. Throughout the study, patients kept home treatment diaries and recorded all FVIII infusions and bleeds. The data reported here are therefore based on the patients (or parents) assessments. It is assumed that an event requiring an infusion of FVIII is a bleed. Bleeds requiring FVIII were described by date, time and site (either joint or non-joint).

Pharmacokinetics and FVIII levels

The pharmacokinetics (PK) of FVIII were evaluated for each patient, based on 4–5 blood samples for the patients aged 1–6 years and 9–11 samples for patients aged 10–65 years, taken for a total period of 48 h after infusion of a single bolus dose of 50 IU kg−1 Advate [17,19]. Factor VIII was measured by the one-stage assay in a central laboratory. Elimination half-life was estimated using a two-phase linear model with a non-parametric regression method [20,21], based on the baseline-adjusted FVIII values at each time point. Area-under-the-curve (AUC) was computed using the linear trapezoidal rule, with extrapolation to infinite time. Clearance was calculated as dose divided by AUC. Weekly AUC is calculated as the area under the predicted FVIII level curve during one week of treatment. During the study period, intra-individual variation in PK parameters was low [18].

The time per week each individual spent below 1, 2 or 5 IU dL−1 FVIII was calculated as previously described [14] from their observed PK and the actual times and doses of their prophylactic infusions as recorded in their treatment diary. For each recorded infusion (prophylactic, bleed treatment and PK infusions), the FVIII concentrations (PK curve) after the infusion was projected based on the infusion dose and the observed PK parameters. The time between the projected PK curve reaching these levels (1, 2 or 5 IU dL−1) and the next infusion, was added up during the study periods and then divided by the number of weeks. For comparison, each patient’s individual PK data were used to calculate the hypothetical time below 1, 2 and 5 IU dL−1 assuming they had taken standard prophylaxis, defined as 30 IU/kg three times a week, with 100% adherence. For each subject, hypothetical infusion records, with administration of this dose at 10.00 hours on Monday, Wednesday and Friday (and then on the next Monday at 10.00 hours), were generated. Then, the time per week was calculated in the same manner.

Statistical analysis

Annualized bleeding rates (bleeds/year) were calculated as the number of bleeding events divided by length of time of the treatment regimen, in years.

In order to explore significant factors associated with bleed rate, potential variables such as average weekly prophylactic dose and frequency, adherence to the dose and frequency of the prescribed treatment regimen, and time spent with factor VIII below 1 IU dL−1 were defined within a treatment regimen for each patient. An individual was required to remain on a prophylactic regimen for 3 months to be included in the analysis.

Adherence to dose was expressed as the percentage of infusions within the specified range for a prophylactic dose; adherence to frequency was expressed as a percentage of weeks when FVIII was given on at least 3 days, irrespective of the day of the infusion. Adherence was based on data recorded in the patients’ dairies.

To estimate the effect that non-adherence to the prescribed standard prophylaxis had on time spent with a FVIII below 1 IU dL−1, a comparison was made between the time per week spent below this level, calculated from patients’ recorded infusions, and the time per week calculated on the assumption that they had taken a standard regimen of 30 IU kg−1 on Monday, Wednesday and Friday. Analyses were performed separately for patients aged 1–6 years and those aged 10–65 years.

A non-parametric statistical method (Wilcoxon test) was used for simple comparisons between the groups of patients who experienced bleeding episodes while on study and those who did not. The effects of time below 1 IU dL−1, weekly AUC and half-life on annual bleeding rate were analysed by multivariate analyses. Multivariate analysis on annual bleeding rate (as a dependent variable) was performed separately for each of the potential factors associated with bleed rate such as adherence to treatment schedule, PK parameters (half-life), time with FVIII below 1 IU dL−1 and total weekly dose. Bleed cause, bleed site, age and weight ratio (= actual weight/ideal weight) were used as common independent variables. In all instances a regression model utilizing the negative binomial distribution was used. All analyses were performed separately for patients aged 1–6 years and those aged 10–65 years.

The effect of time with a FVIII below 1 IU dL−1 on break-through bleeds was investigated in a multivariate analysis using data from all patients but analysing 1–6 year olds and 10–65 year olds separately. As prophylactic regimens used in the 1–6 year olds were at the discretion of the treating clinician, these were more diverse in both dosing and frequency. To allow for a comparison with the results from the older patients, the analysis on FVIII levels below 1 IU dL−1 and bleeds was repeated in the subgroup of 20 patients who were treated with prophylaxis at least three times a week throughout the study.

Results

Patient characteristics

Patient baseline characteristics, bleed numbers, days on prophylactic regimen and adherence to prescribed treatment are shown in Table 1. The median age of the patients in the 10–65 year old group was 18.4 years and the majority of this group had clinical evidence of joint pathology. Prophylaxis was given a median of three times per week and median follow-up was about 1 year. Adherence to the prescribed dose of prophylaxis was high but adherence to frequency was lower as a result of missed infusions. The median adherence to frequency for the 1–6 year olds was 89% of the time and for the 10–65 year olds it was 78% of the time. Median number of bleeds on prophylaxis was three per year whereas 25% of patients aged 1–6 years and 16% of patients aged 10–65 experienced no bleeds during follow-up.

Table 1.   Patient characteristics, treatment and bleeding
 1–6 years of age (n = 44) Median (10–90 percentile)10–65 years of age (n = 99) Median (10–90 percentile)
Age at study entry (years)3.0 (1.7–5.7)18.4 (11.8–46.8)
Body mass index16.3 (14.4–20.2)23.1 (17.2–27.9)
Patients with clinical arthropathy n (%)1 (2)66 (67)
Prescribed prophylaxisInvestigator determined25–40 IU kg−1 3–4 × week
Length of time on regimen (days)369 (194–539)394 (258.0–1024.0)
Weekly dose (IU kg−1)108.2 (65.9–177.5)83.5 (60.6–100.2)
Infusions/week administered2.9 (2.0–3.2)2.7 (2.0–3.0)
Adherence to dose (%)100 (21.9–100)98.9 (68.0–100)
Adherence to frequency (%)89.2 (60–100)78.4 (41.9–92.9)
Bleeds per patient per year3.1 (0.0–11.6)3.3 (0.0–14.6)
Patients without bleeds n (%)11 (25)16 (16)
Total bleeding episodes190924
Total hemarthoses52501

Effect of prophylaxis on FVIII level

Calculated from the infusions given and the individual patients’ pharmacokinetics, the median predicted time per week spent with a FVIII below 1 IU dL−1 in patents aged 1–6 years was 19 h and in the patients aged 10–65 years, 16.5 h. Data are also shown for time with a FVIII less than 2 and 5 IU dL−1 (Table 2). Median adherence to frequency was approximately 90% for the patients aged 1–6 years and about 80% for those aged 10–65 years; therefore, a proportion of the time spent with a FVIII less than 1 IU dL−1 was as a result of reduced adherence. In order to estimate the effect that lack of adherence might have on FVIII levels, the predicted time that patients spent with a FVIII below 1, 2 and 5 IU dL−1, calculated on the basis of the FVIII infusions given, was compared with the predicted time each patient would have spent below these levels had they adhered to a standard prophylactic regimen of 30 IU kg−1 on Monday, Wednesday and Friday (Table 2). This comparison demonstrated that in the patients aged 1–6 years 100% adherence to a standard prophylactic regimen made virtually no difference to predicted time below 1, 2 and 5 IU dL−1. However, in the patients aged 10–65 years a difference was seen for time below 1 IU dL−1 with a decrease from a median (10–90 percentile) of 16.5 (1.5–45.9) hours during actual treatment to a hypothetical 10.0 (0–28.2) hours with full adherence (Table 2).

Table 2.   Pharmacokinetic parameters and the affect of adherence on trough factor VIII (FVIII) levels
 1–6 year olds Median (10–90%)10–65 year olds Median (10–90%)
During actual treatmentWith full adherence*During actual treatmentWith full adherence*
  1. AUC, area under the curve.

  2. *Estimated based on assumption of 100% adherence to standard prophylaxis, 30 IU kg−1 on Monday, Wednesday and Friday.

Time per week FVIII <1 IU dL−1 (h)19.0 (2.9–53.9)19.2 (0–45.7)16.5 (1.5–45.9)10.0 (0–28.2)
Time per week FVIII <2 IU dL−1 (h)35.9 (14.8–71.1)35.6 (12.2–67.6)29.0 (5.6–58.9)20.8 (0–51.5)
Time per week FVIII <5 IU dL−1 (h)66.1 (37.5–96.7)72.1 (38.5–96.8)57.7 (23.6–91.9)57.4 (17.9–90.0)
Elimination half-life (h)9.3 (7.5–12.7)NA11.1 (8.8–15.4)NA
Clearance (ml h−1 kg−1)4.2 (2.7–6.3)NA3.3 (2.2–5.2)NA
AUC/dose [(IU*h dL−1)/(IU kg−1)]23.6 (16.0–36.7)NA30.5 (19.2–46.0)NA
Weekly AUC (IU*h dL−1)2884 (1579–3756)NA2442 (1478–3586)NA

Break-through bleeding rate according to time spent with low FVIII

The hypothesis that break-through bleeding is associated with time spent with low FVIII levels was initially investigated by comparing the patients who had had no bleeds with those who had had at least one bleed in the follow-up period (Table 3). Despite similar follow-up times, the patients who had at least one bleed during the observation period spent a longer time with a FVIII below 1, 2 and 5 IU dL−1 than those who had no bleeds. In the 10–65 year olds, patients who had no bleeds had a FVIII less than 1 IU dL−1 on average for less than half the time of those that experienced at least one bleed, reaching statistical significance for time below 1 IU dL−1 (P = 0.045). Although a proportionally larger difference was observed in the 1–6 year olds, statistical significance was not reached, probably because of the lower number of patients and relatively few bleeds observed in this group. The independent effect of time with a FVIII below 1 IU dL−1 on bleeding rate was further investigated for both age groups (Table 4).

Table 3.   Comparison of time spent below different factor VIII (FVIII) levels according to bleeding pattern
 Age 1–6 years, n = 44 Median (10–90 percentile)Age 10–65 years, n = 99 Median (10–90 percentile)
Bleeds N = 33No bleeds N = 11Bleeds N = 83No bleeds N = 16
  1. *P < 0.05; **P < 0.01.

Age (years)2.7 (1.7–5.7)3.2 (2.0–5.7)16.8 (11.7–43.9)23.0 (12.4–50.8)
Regimen time (days)396 (210–685)364 (165–504)394 (258–1024)464 (277–988)
Average weekly frequency2.9 (1.9–3.2)3.0 (2.0–3.1)2.7 (2.0–3.0)2.8 (2.6–3.1)
Average weekly dose (IU kg−1)103.9 (58.2–155.5)164.0* (91.5–227.1)83.4 (60.6–99.6)85.8 (67.9–101.1)
Adherence to frequency (%)87.1 (60.0–100)96.1 (82.9–100)77.8 (41.9–91.4)83.1 (65.4–96.4)
Time spent with a FVIII below 1 IU dL−1 (h)21.8 (2.9–56.2)12.3 (4.4–21.6) 17.7 (2.1–45.9)8.1* (1.5–31.1)
Time spent with a FVIII below 2 IU dL−1 (h)40.8 (14.8–71.1)23.2 (15.1–41.3)31.3 (5.8–58.9)20.1 (4.2–56.2)
Time spent with a FVIII below 5 IU dL−1 (h)70.4 (43.5–96.7)53.2 (37.5–70.5)61.2 (27.5–91.9)50.6 (16.8–89.7)
Table 4.   Multivariate analysis of the effect of factor VIII (FVIII) trough levels and pharmacokinetic parameters on bleed rate
  Age 1–6 years old any prophylaxisAge 10–65 years old standard prophylaxis
All bleedsHemarthosesAll bleedsHemarthroses
CoeffCoeffCoeffCoeff
  1. NS indicates not statistically significant. AUC, area under the curve.

  2. +P < 0.05; ++P < 0.02; *P < 0.005; **P < 0.0001.

Time per week factor VIII <1 IU dL−10.04**0.05**0.02+0.03++
Time per week factor VIII <2 IU dL−10.04**0.05**NS0.02+
Time per week factor VIII <5 IU dL−10.04**0.06**NSNS
Weekly AUCNS−0.0008*NSNS
ClearanceNSNSNSNS
Half-life−0.28+NSNSNS
Average weekly doseNS−0.02*NSNS
Average weekly frequency−0.76**−1.25**−0.73*−0.6+
Adherence to frequency−0.04**−0.04**−0.02**−0.02**

Children aged 1–6 years  In the 44 children aged 1–6 years on prophylaxis, the median annual bleed rate was 3.1 and the median time below 1 IU dL−1 was 19 h per week. Increasing time with a FVIII <1 IU dL−1 was associated with an increased rate of all bleeds and hemarthroses. Similar results were found if the analysis was performed at <2 and <5 IU dL−1 although there was a strong correlation between time <1, <2 and <5 IU dL−1 in these patients (Table 4). The strength of the association implied that for each additional hour spent with a FVIII <1 IU dL−1 the annual bleed rate increased by 2.2% (CI 1.58–2.78%). The predicted effect of time per week with a FVIII less than 1 IU dL−1 on bleeds in the 1–6 year olds is shown in Fig. 1. The probability of having no bleeds per year dependent on the time spent with a FVIII less than 1 IU dL−1 is shown in Fig. 2.

Figure 1.

 Predicted bleed count per year vs. time spent with a factor VIII (FVIII) less than 1 IU dL−1 for patients aged 1–6 years. The predicted hemarthoses per year (represented with the open circle, ‘°’) dependent on time per week spent with a FVIII less than 1 IU dL−1 are shown for the patients aged 1–6 years.

Figure 2.

 Predicted probability of having no bleeds per year dependent on time per week spent with a factor VIII (FVIII) less than 1 IU dL−1. The calculated probability of having no bleeds per year is shown compared with increasing time per week spent with a FVIII less than 1 IU dL−1. Open circles (°) and asterisks (*) represent hemarthroses in patients aged 1–6 years and 10–65 years, respectively.

Because treatment three times a week was the most frequently used regimen, and the association of time with a FVIII below 1 IU dL−1 may have been strongly affected by some patients aged 1–6 years being on lower frequency regimens, the analysis was repeated in the subgroup of 20 1–6 year olds who were treated with prophylaxis at least three times a week throughout the study. In this subgroup, there were only 65 bleeding events, but despite the small numbers of bleeds and patients, the association between time spent with a FVIII less than 1 IU dL−1 and the rate of all bleeds (P < 0.002) and hemarthroses (P < 0.0001) was even stronger.

Patients aged 10–65 years  In the 99 patients aged between 10 and 65 years on standard prophylaxis, increasing time with a FVIII <1 IU dL−1 was associated with an increased rate of all bleeds and hemarthroses (Table 4). The median time spent with a FVIII less than 1 IU dL−1 was 16.5 h and the median bleed rate was 3.3 per year. For each additional hour below 1 IU dL−1 there was a 1.4% increase in the annual bleed rate (CI 0.21–2.62%). The effect of time per week with a FVIII less than 1 IU dL−1 on bleeds in the 10–65 year olds is shown in Fig. 3. The probability of having a zero annual bleed incidence as a function of time spent with a FVIII less than 1 IU dL−1 is shown in Fig. 2. In both age groups, the probability of experiencing no hemarthrosis or any bleed, declined rapidly with increasing time of FVIII less than 1 IU dL−1.

Figure 3.

 Predicted bleed count per year dependent on time spent with factor VIII (FVIII) less than 1 IU dL−1 for patients aged 10–65 years. The predicted hemarthoses per year (represented with the asterisk, ‘*’) dependent on time per week spent with a FVIII less than 1 IU dL−1 are shown for the patients aged 10–65 years.

Effect of adherence on bleed rate

The time spent with a FVIII below 1 IU dL−1 is affected by the patients’ adherence to their prescribed prophylactic regimens. A multivariate regression model utilizing the negative binomial distribution was used to evaluate the influence of factors on annual bleeding rate. The analyses were adjusted for cause, site, age and weight in comparison to ideal weight. Analysis of the bleed rates in both 1–6 and 10–65 year olds showed that decreased adherence to frequency of the prescribed prophylaxis was associated with an increase in all bleeds and hemarthroses (Table 4). The data imply that regardless of the dose and frequency of the prophylactic regimen, in the 1–6 year olds (coefficient –0.0348, median 89% adherence to frequency and 3.1 annual bleed rate), if the patients had fully adhered to their prescribed regimen there would have been 0.97 (95% CI 0.63–1.27) fewer bleeds per year. Similarly, in the 10–65 year olds the average patient would have had 1.19 (CI 0.66–1.61) fewer bleeds per year had they adhered fully to their prescribed regimen. The effect of adherence to dose could not be analysed because almost all patients adhered to the prescribed dose when an infusion was given.

Association of pharmacokinetic parameters with bleed rates

Multivariate analysis on pharmacokinetic parameters, independent of site and cause of bleed and age and body habitus (measured by weight in comparison to ideal weight), on bleeding rate demonstrated that in the 1–6 year olds on any prophylaxis, all bleeds (P = 0.01) increased as elimination half-life decreased. There was no association between bleed rates and pharmacokinetic parameters in the 10–65 year olds.

The time spent with low levels of FVIII appeared to be more important than the amount of FVIII the individual was exposed to during the week, because the area under the FVIII curve per week was somewhat associated with hemarthroses in the 1–6 year olds. In the 10–65 year olds there was no association between area under the FVIII curve per week and bleed rate (Table 4).

Discussion

This is the first study to demonstrate an association between the time a patient with severe hemophilia A on prophylaxis has a FVIII level below 1 IU dL and the rate of bleeding. This association was observed for all bleeds combined and for hemarthroses alone in patients aged 10–65 years on standard prophylaxis and children aged 1–6 years on any prophylaxis.

The rationale for prophylaxis is to convert a patient from a severe to moderate phenotype and reduce the risk of bleeds. This can be considered conceptually as either sustaining a trough FVIII above a desired level at all times or reducing the time that that FVIII is low. The association between time <1 IU dL and bleed rate has so far only been investigated retrospectively [14]. A group of 51 hemophilia A patients on high-dose prophylactic treatment were followed-up for 6 years. Dosing was based more on the aim to minimize bleeding frequency than on maintaining a 1 IU dL1 target FVIII level. The relationship between time below a FVIII level of 1 IU dL−1 and incidence of joint bleeding was significant but very weak, even after stratification of the patients according to joint score. There was no relationship between time below 1 IU dL−1 and incidence of other bleeds. In accordance with previous findings [16], some patients did not bleed in spite of a trough level of <1 IU dL−1 and others did in spite of trough levels >3 IU dL−1.

Establishing an unbiased relationship between time below a certain FVIII level and risk of bleeding requires a prospective study in which dosing is fixed rather than adjusted according to clinical outcome, and which also records any non-adherence to the prescribed treatment. Suitable data for such a ‘retrospective-prospective’ study were available from the safety, efficacy and pharmacokinetic trials performed as part of the Advate clinical trial program, although it must be recognized that the studies were not originally designed with these analyses in mind. Factor VIII levels over time were predicted by previously described methods [14,22,23]. To actually pinpoint, by blood sampling, the time when the FVIII level falls below a certain value would not be possible in a large number of patients on home treatment. A statistical investigation of this type thus has to rely on predicted and not on measured coagulation factor levels. Earlier experience [22,23] however indicates good agreement between predicted and actual FVIII levels during prophylactic treatment.

Within the context of the prescribed regimens used in the studies, which were predominantly 3 days a week, lack of adherence to the frequency of infusions appeared to be the most significant variable that affected time spent with a FVIII <1 IU dL−1 and bleed rate. Almost all patients adhered to their prescribed dose of FVIII when taken but missed infusions meant that adherence to frequency was lower. The coefficients derived in the analyses can be interpreted in a clinical context as predicting an increase in annual rate of bleeding as a result of non-adherence equal to an extra 0.97 (CI 0.63–1.27) bleeds per year in the average 1–6 year old and an extra 1.19 (CI 0.66–1.61) bleeds per year in the average 10–65 year old.

A calculation was made to compare the effect on time with a FVIII below 1 IU dL−1 assuming that the patients had either received their actual treatment or a standard prophylaxis of 30 IU dL−1 on Monday, Wednesday and Friday with 100% adherence. Compared with the actual treatment, the 10–65 year old patients, who missed roughly two out of 10 infusions, would have spent a markedly shorter time with low FVIII had they received the standard regimen with full adherence. In contrast, the hypothetical use of a standard prophylaxis regimen with 100% adherence in patients aged 1–6 years made virtually no difference. This may be as a result of the higher observed adherence (roughly 1 in 10 infusions were missed), the higher median FVIII dose of 108 IU week−1 or both.

In addition to the dose and frequency of infusions, the time spent with a FVIII <1 IU dL−1 is determined by the pharmacokinetics of FVIII in the individual patient. These variables where analysed with respect to bleeding rate on prophylaxis. In the 10–65 year olds there was no association between any pharmacokinetic parameter and bleed rates. In the 1–6 year olds, however, an association was found between rate of elimination of FVIII expressed as either half-life or clearance and the annual incidence of all and joint bleeds.

The strength of the associations between bleed rate and time spent with a FVIII <1 IU/dL−1 imply that if, for example, in the 16 year olds, the time spent below 1 IU dL−1 increased by 10% (17 hours per week) there would be 44% (CI 30.258.6) more bleeds per year in the average patient, whereas if no time was spent with a FVIII below 1 IU dL−1 there would be 1.03 (CI 0.791.24) fewer bleeds. In the 1065 year olds if the time spent <1 IU dL−1 increased by 10% there would be 26% (CI 3.654.3) more bleeds per year in the average patient, whereas if no time was spent with a FVIII below 1 IU dL−1 there would be 0.67 (0.111.13) fewer bleeds per year. The results imply that optimizing prophylactic regimens using pharmacokinetic measures or trough FVIII levels and reducing the length of time a patient spends with a FVIII below 1 IU dL−1 will reduce bleeding episodes and hence potentially improve clinical outcomes. This was not addressed in the current study and the clinical utility of this approach requires further study.

The design of prophylactic regimens that limit the time spent with low FVIII levels also requires consideration of the prescribed regimen. Monday – Wednesday – Friday treatment results in lower levels on a Sunday and the use of alternate day rather than three times a week dosing [22,23] will reduce this time in a cost-effective manner. Adherence to the frequency of prophylactic infusions was strongly associated with bleed rate and has a significant impact on the length of time with a FVIII below 1 IU dL−1, strategies to improve adherence are likely to reduce bleed rates.

Because the number of hemarthroses that a patient experiences has been shown in previous studies to result in progression of arthropathy assessed both clinically and radiologically [24], the implication is that maintenance of measurable FVIII levels will be of clinical value. However, the long-term implications on arthropathy can not be assessed from this study because long-term follow-up of joint pathology was not undertaken. Data from in vitro and animal models suggest that the joints of young children may be more vulnerable to the effect of break-through bleeds than those of adults [25–27], and maintaining trough FVIII levels above 1 IU dL−1 may be more important in this age group.

In conclusion, this study demonstrates that increasing time with a FVIII less that 1 IU dL−1 is associated with an increased rate of break-through bleeding during prophylaxis. Adherence to the prescribed regimen is an important factor determining both FVIII level and bleed rate. Improved knowledge about the factors that affect break-through bleeds on prophylaxis will help to improve the cost-effective use of prophylaxis while optimizing clinical outcomes. Until further prospective data are available, adjusting prophylactic regimens to an individual’s observed bleeding pattern complemented by FVIII measurements in some patients appears to be best practice.

Addendum

P.W. Collins was involved in concept development, the critical analysis of data and wrote the first draft of the paper. V.S. Blanchette, K. Fischer, S. Björkman, J. Astermark contributed to concept development, data and content analysis and critical comment on manuscript. M. Oh performed statistical analyses and was involved in review of manuscript. S. Fritsch and P. Schroth were involved in concept development, data and content analysis, supervision of statistical analysis and critical comment on manuscript. G. Spotts and B. Ewenstein were involved in concept development, analysis of clinical data and critical review of manuscript.

Acknowledgements

The data provided for analysis in this study were obtained from the the rAHF-PFM pivotal, continuation and paediatric studies, sponsored by Baxter Healthcare Corporation. The authors thank the patients who participated in these studies and the hemophilia center personnel who performed the studies and collected the data. A list of contributing centers and principal investigators is shown below:

T. Abshire (Emory University, Atlanta, USA), C. Altisent (Hospital Vall D’Hebron, Barcelona, ES), E. Berntorp (University Hospital MAS, Malmö, SE), V. Blanchette (Hospital for Sick Children, Toronto, CA), H. H. Brackmann (Institut für Experimentelle Hämatologie und Transfusionmedizin der Universität Bonn, DE), D. Brown (Children’s Memorial Hospital, Chicago, USA), P. W. Collins (University Hospital of Wales, Cardiff, UK), D. Di Michele (The New York Hospital – Cornell Medical Center, New York, USA), J. Di Paola (University of Iowa Hospitals and Clinics, Iowa City, USA), B. Ewenstein (Brigham and Women’s Hospital, Boston, USA), P. Giangrande (Oxford Haemophilia Centre-The Churchill Hospital, Oxford, UK), J. Gill (Comprehensive Center for Bleeding Disorders, Milwaukee, USA), A. Gringeri (Ospedale Maggiore Di Milano, IT), R. Gruppo (Children’s Hospital Medical Center, Cincinnati, USA), C. R. M. Hay (Central Manchester Healthcare NHS Trust, Manchester, UK), F. Hernández (Hospital Universitario La Paz, Madrid, ES), K. Hoots (University of Texas Health Science Center, Houston, USA), A. M. Hurlet-Jensen (Mount Sinai Medical School, New York, USA), J. Ingerslev (University Hospital SKEJBY, Aarhus, DK), R. Kulkarni (Michigan State University, East Lansing, USA), T. Lambert (Hôpital Bicêtre, Le Kremlin Bicêtre, FR), C. Lee (The Royal Free Hospital, London, UK), R. Liesner (Great Ormond Street Hospital for Children NHS Trust, London, UK), M. Manco-Johnson (Mountain States Regional Hemophilia and Thrombosis Center, Aurora, USA), C. Manno (Children’s Hospital of Philadelphia, USA), P. M. Mannucci (Ospedale Maggiore di Milano, IT), P. W. Marks (Brigham and Women’s Hospital, Boston, USA), C. Negrier (Hôpital Edouard Herriot, Lyon, FR), R. Nuss (Mountain States Regional Hemophilia and Thrombosis Center, Aurora, USA), I. Ortiz (University Pediatric Hospital, San Juan, PR), I. Pabinger (Allgemeines Krankenhaus der Stadt Wien, Vienna, AT), P. Petrini (Karolinska Hospital, Stockholm, SE), C. Philipp (University of Medicine and Dentistry of New Jersey, New Brunswick, USA), S. Pipe (University of Michigan Hemophilia Treatment Center, Ann Arbor, USA), H. Pollmann (Institut für Thrombophilie und Hämostaseologie, Münster, DE), M. V. Ragni (Hemophilia Center of Western Pennsylvania, Pittsburgh, USA), C. Rothschild (Centre d’ Hémophilie François Josso, Paris, FR), S. Seremetis (Mount Sinai Medical School, New York, USA), A. Shapiro (Indiana Hemophilia and Thrombosis Center, Indianapolis, USA), M. Siimes (Helsinki University Central Hospital Hus, Helsinki, FI), M. Tarantino (Comprehensive Bleeding Disorders Center, Peoria, USA), A. Thompson (Children’s Memorial Hospital, Chicago, USA), A. Thompson (Puget Sound Blood Center and Children’s Memorial Hospital, Seattle, USA), M. van den Berg (Universitair Medisch Centrum Utrecht, NL), J. Vermylen (KU Leuven Universitaire Ziekenhuizen, Leuven, BE), I. Warrier (Children’s Hospital of Michigan, Detroit, USA) and W. Y. Wong (Children’s Hospital Los Angeles, USA).

Disclosure of Conflict of Interests

S. Bjorkman (consultancy Baxter Octopharma); M. Oh (employee of Baxter); P. Schroth (employee of Baxter); P. Collins (consultant work and honoraria from Baxter); S. Fritsch (employee of Baxter). Research support – P.W. Collins (Baxter, Novonordisk), V.S. Blanchette (Bayer), K. Fischer (Bayer, Wyeth), J. Astermark (Baxter, Wyeth, Aventis, Bayer). Some authors are employed by Baxter Healthcare Corporation (M. Oh, S. Fritsch, P. Schroth, G. Spotts, B. Ewenstein).

Ancillary