Mossimo Morfini, Medical Department Wyeth Lederle, Aprilia, Italy. Tel.: +39 055 4277587; fax: +39 055 4277794; e-mail: firstname.lastname@example.org
Summary. When the one-stage clotting assay is used in comparison with the chromogenic and immunological assays, plasma levels of factor (F)VIII are underestimated by 40–50% after infusion of B-domain deleted recombinant FVIII (BDD-rFVIII) in patients with hemophilia. A possible way to counteract the underestimation of FVIII levels by the one-stage assay is the adoption of a recombinant FVIII reference standard instead of a plasma standard. To evaluate the usefulness of such a standard [ReFacto® Laboratory Standard (RLS)], the pharmacokinetic parameters of a single dose of BDD-rFVIII (25 U kg−1) were evaluated in a multicenter study carried out in 18 patients with severe hemophilia A. The very low in vivo recovery, obtained with the combination of the one-stage assay and plasma reference standard, was increased up to the values obtained by the chromogenic assay when the results were expressed in terms of RLS. When the plasma standard was used, the one-stage/chromogenic ratio was 0.82 ± 0.12 for FVIII levels above 25 U dL−1 and 1.42 ± 0.99 for FVIII levels below 25 U dL−1. Using the RLS, the one-stage/chromogenic ratio increased to 1.01 ± 0.19 at FVIII levels above 25 U dL−1, as a consequence of a complete overlap of the two decays; however, at FVIII levels below 25 U dL−1, the one-stage/chromogenic ratio was still 1.6 ± 0.85. After the twelfth hour, FVIII concentrations obtained by chromogenic assay were always lower than those resulting from the one-stage clotting assay, independently of the standard used. Results obtained by chromogenic assay were not affected by the type of standard used. Compared with those obtained by the one-stage assay, higher values of clearance, lower volume of distribution area and shorter plasma half-life or mean residence time were obtained by chromogenic assay because of a shape change of the decay curve due to a shift to higher values in the first part (time interval 0–12 h) and to lower values in the second part of the decay curve (time interval 12–48 h). As a consequence, the slope of the decay curve obtained by means of chromogenic assay was steeper. In conclusion, the more homogeneous results of in vivo recovery and pharmacokinetic analysis, due to the decrease of discrepancy between the two methods when RLS was used, make the cheaper and more widely used one-stage assay preferable to the more expensive chromogenic assay, on condition that the ReFacto specific standard has used.
A B-domain-deleted recombinant factor (F)VIII (BDD-rFVIII, ReFacto®)  was developed on the basis of the knowledge that the B-domain is not required for expression of FVIII coagulant activity and is removed during the naturally occurring activation of the molecule [2,3]. BDD-rFVIII (1432 amino acids) is structurally indistinguishable from full-length FVIII (2332 amino acids). While the interaction between von Willebrand factor (VWF) and BDD-rFVIII is approximately 25% greater than that of plasma-derived FVIII, other functional properties of the two molecules are indistinguishable, such as the association and dissociation rate constants for VWF binding, activation–inactivation by thrombin, inactivation by activated protein C, and FXa generation [4–6]. It was soon apparent that plasma levels of BDD-rFVIII were underestimated by 40–50% when the one-stage clotting assay was used , whereas the levels measured by chromogenic assay were in good agreement with FVIII antigen levels in postinfusion patient samples . Low FVIII activity measured by one-stage clotting assay is thought to be due to the high phosphatidylserine content of the platelet substitute used in the assay [7,9] as suggested by the observation that, when a platelet-rich hemophilia A plasma or a phospholipid mixture with a low content of phosphatidylserine are used, no discrepancy is observed . FVIII assay standardization is a quite old and hot issue. Discrepancies between one-stage and two-stage clotting methods have been reported for many years [10–12].
A possible approach to counteract the underestimation of FVIII by conventional one-stage assays is the adoption of a recombinant reference instead of a plasma reference standard. In general, a separate standard for recombinant DNA-derived concentrates, according to the like vs. like principle, reduces potency variability [13–15]. With this as background, Wyeth (Cambridge, MA, USA) Genetics Institute, Pharmacia (Stockholm, Sweden), and the National Institute for Biological Standards and Controls (NIBSC, South Mimms, Herts, UK) have produced a ReFacto® Laboratory Standard (RLS) to be used in the one-stage clotting assay to measure FVIII activity in patients receiving BDD-rFVIII [16–18]. This study was designed to evaluate the usefulness of this standard by comparing the pharmacokinetics of a single dose of BDD-rFVIII by one-stage and chromogenic assays using a plasma reference standard and RLS.
Materials and methods
Twenty-one patients were enrolled in 18 Italian hemophilia centers participating in an open-label, multicenter study carried out according to the International Society on Thrombosis and Haemostasis (ISTH) recommendations : age ≥12 years, severe hemophilia A (FVIII <1 U dL−1) without inhibitor [anti-FVIII titer <0.6 Bethesda Units (BU)] more than 50 previous exposures to plasma-derived FVIII or rFVIII, at least 20 exposures during the last 12 months, absence of severe liver disease or full-blown AIDS, presence of HBsAb (otherwise vaccination against hepatitis B virus was strongly recommended) and written informed consent. The protocol of this study was submitted to and approved by the Institutional Review Boards of each participating center.
The pharmacokinetic study was carried out after a wash-out period of at least 3 days in a non-bleeding state. The recommended dose of BDD-rFVIII (25 IU/kg of body weight) was injected intravenously in no more than 10 min. From the opposite arm, venous blood samples were collected in 4.5-mL Vacutainer tubes containing 0.5 mL of sodium citrate 3.8% (1/10 v/v) at 0.25, 0.5, 1, 3, 6, 9, 24, 33, and 48 h after the end of infusion. After centrifugation at 2000 × g at 4 °C, plasma samples were immediately snap frozen, stored at −40 °C and sent within 2–3 months to the central coagulation laboratory of the Florence Hemophilia Center. For three of 21 patients, it was impossible to analyze the samples because they were not properly handled or stored. Hence, only samples from the remaining 18 patients were used for the pharmacokinetic analysis.
Factor VIII assays
The one-stage clotting assay was performed using the following reagents from Dade-Behring (Marburg, Germany): (i) FVIII immune-depleted plasma as substrate; (ii) Pathromtin SL vegetable phospholipids activated by silica dioxide; (iii) imidazole buffer pH 7.35; and (iv) calcium chloride 0.025 m. An automatic BCT Dade-Behring coagulometer, reading at 405 nm, was used according to the manufacturer's instructions. The reference calibration curves were constructed using as a FVIII reference standard an International Plasma Standard (IPS) from Dade-Behring calibrated against the 4th International Plasma (NIBSC code 97/586) and ReFacto Laboratory Standard (RLS) (George King Biomedical, Inc., Overland Park, KS, USA; 9.4 IU/vial) calibrated by chromogenic assay against the 6th International Standard for FVIII (NIBSC code 97/616). Both standards were reconstituted with 1 mL of distilled H2O and diluted to 1 U mL−1 FVIII concentration with plasma from a patient with severe hemophilia. Six dilutions (1 : 5–1 : 10–1 : 20–1 : 40–1 : 80–1 : 160) in imidazole buffer were used to construct the curve for each standard. A nearly linear relationship was obtained in a semilog plot, with FVIII activity (U dL−1) on the logarithmic scale (abscissa) and the activated partial thromboplastin time (APTT) (seconds) on the linear scale (ordinate). The postinfusion plasma samples were diluted with imidazole buffer 1 : 5–1 : 10–1 : 20 (v/v). Equal aliquots (50 µL) of standard or sample dilution and of reagents were automatically dispensed in a reaction minivial by the Behring Clotting Timer in the following order: (i) standard or sample dilution, (ii) FVIII immune-depleted plasma, and (iii) Pathrontin SL. After incubation at 37 °C for 120 s, 50 µL of CaCl2 0.25 m were added and the time for clotting formation was recorded. Each sample dilution was tested in duplicate. Percent FVIII activity was calculated in patient samples by the parallel curves method . The index of parallelism was calculated as the ratio between slopes (normal value 0.8–1.2). Only the curves with a normal index of parallelism were examined by the specific software.
The chromogenic assay (Dade-Behring FVIII Chromogenic Assay) consisted of two steps. In the first step, FVIII was activated by thrombin to FVIIIa, which promotes the conversion of bovine FX to activated FXa in the presence of activated bovine FIX (FIXa), phospholipids (PL) and calcium chloride in Tris buffer pH 8.0. In the second step, the activity of FXa was measured from the hydrolysis of a p-nitroaniline substrate specific for FXa. The reagents were added in the following sequence: standard or sample dilution, FX reagent, and FIXa reagent containing 0.6 nmol of bovine thrombin. After incubation for 150 s, the chromogenic substrate specific for FXa were added together a thrombin inhibitor. The initial release rate of p-nitroaniline measured at 405 nm during the first 30 s was proportional to the activity of FXa and thus to the FVIII activity of the sample (kinetic rate method). Six doubling dilutions of plasma standard and RLS in imidazole buffer were used to construct the calibration curves (from 1 : 10 to 1 : 320, corresponding to 100 U dL−1 and 3.12 U dL−1, respectively). Plasma samples were tested in three doubling dilutions (1 : 10–1 : 20–1 : 40). All dilutions were tested in duplicate. Activity values were calculated by the parallel curve method  using a log/log plot. This modification was necessary to obtain a linear relationship between absorbance and FVIII activity. The volume of all reagents dispensed was reduced by one-quarter in comparison with the procedure recommended by the manufacturer.
For pharmacokinetic analysis, area under the curve (AUC), clearance (Cl), mean residence time (MRT), volume of distribution area (VdArea), and terminal half-life (evaluated on the last 4 points) were calculated according to the model independent method using the software PKRD . Percentage and incremental in vivo recovery (IVR) were evaluated according to Nilsson's formula  and Prowse's formula , respectively, on the basis of the peak value (Cmax) identified during the first postinfusion hour. Each kinetic study, obtained using two different methods of FVIII assay, each with two different standards, provided four decay curves for analysis (72 curves in total). Furthermore, all FVIII postinfusion concentrations of the 72 curves (n = 648) were stratified according to the FVIII concentration and the reference used. Four arrays of 162 values each were obtained. From each paired value, the one-stage/chromogenic assay ratio was calculated.
Statistical analysis was performed by means of paired, one-tailed Student's t-test.
Mean age of the 18 patients was 32 ± 12 years. The actual infused dose was 27.4 ± 2.2 IU kg−1 instead of 25.0 IU kg−1 (recommended dose) due to adjustments made by the investigators for the vial dosage.
Reproducibility of FVIII assays
The reproducibility of one-stage and chromogenic assays is reported in Tables 1 and 2, respectively. A linear–log and a log–log plot of one-stage and chromogenic reference curves are displayed in Figs 1 and 2, respectively. A good reproducibility of the one-stage assay was observed at each dilution, the coefficient of variation (CV) ranging from 1.00% to 1.62% for the plasma standard and from 0.84% to 1.64% for RLS. The reproducibility of the chromogenic assay was poorer: for FVIII concentrations of ≥10.6 U dL−1, the CV was <10% (6.3–9.7%), but for FVIII concentrations <10.6 U dL−1 CVs ranged from 10.5% to 17.5%, independently of the standard used.
Table 1. Variability of standard reference curves of one-stage clotting assay
Activated partial thromboplastin times, s
FVIII concentration in IPS (U dL−1)
FVIII concentration in RLS (U dL−1)
Table 2. Variability of standard reference curves of chromogenic assay
The outputs (means and SD) of the software for each FVIII assay are shown in Table 3. The mean values of 97 plasma-derived FVIII kinetics previously performed in our hemophilia population  are given for comparison. In terms of IPS, both the incremental and percentage IVR by one-stage assay were lower than those evaluated by chromogenic assay (P < 0.00001). The IVR by one-stage assay increased up to the range of the values of our hemophilia population  when RLS was used as reference. This notwithstanding, a significant difference between one-stage and chromogenic assays (P < 0.05) still existed. The usage of IPS or RLS did not affect the results of the chromogenic assay. No differences among the AUCs calculated from FVIII concentrations obtained by means of three different methods were observed, except for the one-stage assay in terms of RLS, which gave the highest value of AUC (P < 0.0001). In agreement with this finding, the smaller value of Cl was achieved by the same assay and RLS (P < 0.0001).
Table 3. Pharmacokinetic parameters according to the four factor VIII assays
The MRT and half-life were shorter when the chromogenic assay was used. Even with the large variability, the difference between the two methods was statistically significant (P < 0.001). The values of VdArea obtained by chromogenic assay, both in terms of IPS and RLS, were in the range of our general hemophilia population, but those obtained by one-stage assay were very high. As for IVR, the larger VdArea was obtained by one-stage assay (P < 0.001 in comparison with the results from both chromogenic assays). The use of RLS greatly reduced the VdArea with respect to IPS (P < 0.0001), but there was still a small difference between chromogenic and one-stage assays (P < 0.05).
The distribution of FVIII one-stage/chromogenic assay ratios, stratified according to the FVIII concentration, is shown in Fig. 3. The cumulative analysis of one-stage/chromogenic assay ratios showed that only at high FVIII levels, above 25 U dL−1, was the ratio <1 (0.82 ± 0.12), the chromogenic assay giving higher and one-stage assay lower FVIII concentrations. Only if the results of the one-stage assay were expressed in terms of RLS was the ratio very near to 1 at high FVIII concentrations. When the FVIII concentration was below 25 U dL−1, the one-stage/chromogenic assay ratio was always higher than 1 independently of the standard used (1.42 ± 0.99 for IPS and 1.60 ± 0.85 for RLS). A plot of FVIII concentration vs. time course according to the four different assays is reported in Fig. 4. The upper decay curve obtained by one-stage and RLS is parallel to that obtained by one-stage and IPS. The difference in AUC is easily understandable. It is also very evident that both decay curves by chromogenic assay are completely overlapping. Only the first part of these curves, corresponding to the first 10 h after the end of infusion, is higher than that obtained by one-stage and IPS and overlaps exactly that of one-stage and RLS. Within postinfusion hours 10 and 15, both chromogenic curves decline below that of one-stage and IPS.
A discrepancy between one-stage and chromogenic assay in the assessment of the FVIII potency of high-purity plasma-derived concentrates was identified several years ago [10–12,25–28]. The presence of activated FVIII in these products was thought to be the cause of higher potency estimation by one-stage assay [25–27]. A much larger discrepancy, approximately 50%, was reported in the potency estimation of BDD-rFVIII [7,8], similar discrepancies being subsequently observed also for full-length rFVIII concentrates [14,15]. Recently, attention has been focused on the results of FVIII assay in postinfusion samples drawn from patients participating in pharmacokinetic studies , such as one of BDD-rFVIII [7–9,30,31] and another of full-length rFVIII . In both studies the observed discrepancy between one-stage and chromogenic assay was reversed using product-specific standards. Two pharmacokinetic studies meant to investigate the discrepancy between one-stage and chromogenic assay have been carried out but published only in abstract form [16,17].
In this study, the one-stage assay had a good reproducibility. The CV of the chromogenic assay was acceptable (6.0–10.6%) for FVIII concentrations above 10 U dL−1 but poorer (10.6–17.5%) for FVIII concentrations below 10 U dL−1. The procedure for chromogenic assay used in this study, i.e. the kinetic rate method instead of the end-point method, is perhaps the reason for the large variability observed at low FVIII concentrations.
As far as the pharmacokinetic analysis is concerned, it appears that IVR and peak values are affected by the type of FVIII assay, with the chromogenic assay giving higher values. These findings are in good agreement with the previously published BDD-rFVIII kinetic study , taking into account that in the latter study the FVIII plasma concentrations were increased by 18% to correct for dilution of blood samples by sodium citrate. The discrepancy can be reduced but not abolished by expressing results in terms of RLS, according to the like vs. like principle. The implementation of a product-specific standard, RLS, made the results of the ReFacto Italian PK study more consistent with the data coming from our hemophiliac population . According to the like vs. like principle, each patient sample should be considered as a dilution of injected product in FVIII/IX-deficient plasma. When this principle has been applied, the large variability of FVIII estimation of concentrates' potency and concentration in patients' plasma observed in multicenter collaborative exercises decreased significantly. The rationale of like vs. like principle seems to be definitively understandable as far as the concentrate potency and its in vivo recovery is concerned. The clinical efficacy of ReFacto and the relationship between FVIII postinfusion plasma levels and hemostatic effect are beyond the aims of this study.
Our findings confirm, to a lesser extent, the data observed in previously reported pharmacokinetic studies of the potency estimation of ultrapure plasma-derived or rDNA-derived FVIII concentrates and in assays of FVIII activity in postinfusion samples [13–15,31,33–36].
Even though the AUC is the principal and stronger output of the model-independent pharmacokinetic method, the difference between the results of chromogenic and one-stage assay, expressed as IPS, were not as evident as expected. The decay curve obtained by chromogenic assay was above the curve obtained by the one-stage assay only at the early time points. After the first 10 h, when the FVIII concentration had fallen below 25 U dL−1, there was a crossing of the two curves (Fig. 4). This is well demonstrated also by the analysis of the one-stage/chromogenic ratio, which is <1.0 at high FVIII concentrations (60–25 U dL−1) but becomes >1 when the FVIII postinfusion concentrations decline to lower values (25–1 U dL−1). The use of RLS or IPS did not change the results of the chromogenic assay.
The lack of good reproducibility of the chromogenic assay in the low range of FVIII levels may be the cause of the faster decay of FVIII levels in the terminal part of the curve after the first 10 h. Thrombin activation of FVIII, occurring in the chromogenic assay, may have split a portion of the molecule causing a decrease that was particularly evident at low levels. So, in comparison with the decay curve plotted using the results of the one-stage assay and IPS, the chromogenic assay is higher in the initial (left) part and lower in the terminal (right) part, whereas no changes in AUC were observed. Furthermore, the higher FVIII concentrations by chromogenic assay observed in the terminal part of the curve caused some differences in the decay slope. The MRT obtained by chromogenic assay is shorter when RLS is used as standard instead of IPS because of the increase of FVIII concentration in the initial part of the decay curve. In the same way, the longer MRT obtained by one-stage assay both in terms of IPS or RLS is due to the increase of FVIII concentration in the terminal part of the curve.
In conclusion, this is the first fully reported pharmacokinetic study of BDD-rFVIII conducted according to the ISTH recommendations. The outcomes are comparable to those of other FVIII concentrates, both plasma- and rDNA-derived, in that no correction was done for sodium citrate dilution. This confirms that the one-stage clotting assay does underestimate BDD-rFVIII activity in plasma, when expressed in terms of plasma reference, at least in the early part of the decay curve. When PK results are expressed in terms of RLS, they are in the range of our reference hemophilia population, like those of the chromogenic assay, which is not affected by the use of plasma or recombinant standard. The chromogenic assay/one-stage clotting assay ratio is not persistently >1 during the decay of infused FVIII: this is true in the initial part of the decay curve, when the FVIII concentration is high (>25 U dL−1) but the ratio is <1 in the terminal part when the FVIII concentration is low (<25 U dL−1). When the FVIII concentrations are measured by chromogenic assay, the slope is steeper and this shape change of the decay curve results in a shorter half-life. In the postinfusion samples, FVIII concentration falls from 70 U dL−1 to 5 U dL−1 and the chromogenic assay accuracy and reproducibility fall accordingly. On the basis of our findings, the one-stage clotting assay seems to be the best approach to the measurement of FVIII concentration in postinfusion samples when a specific product standard, such as RLS, is used to construct the reference curve.