Haemophilia A and von Willebrand’s disease
Abstract
Summary. Deficient or defective coagulation factor VIII (FVIII) and von Willebrand factor (VWF) can cause bleeding through congenital deficiency or acquired inhibitory antibodies. Recent studies on type 1 von Willebrand’s disease (VWD), the most common form of the disease, have begun to explain its pathogenesis. Missense mutations of varying penetrance throughout VWF are the predominant mutation type. Other mutation types also contribute while about one-third of patients have no mutation identified. Enhanced clearance and intracellular retention contribute to pathogenic mechanisms. Chromogenic substrate (CS) methods to determine FVIII coagulant activity have several advantages over one-stage methods, which include minimal influence by variable levels of plasma components, notably lupus anticoagulant. Direct proportionality between FVIII activity and FXa generation results in high resolution at all FVIII levels, rendering the CS method suitable for measuring both high and low levels of FVIII activity. FVIII inhibitors in patients with inherited or acquired haemophilia A present several challenges in their detection and accurate quantification. The Nijmegen method, a modification of the Bethesda assay is recommended for inhibitor analysis by the International Society on Thrombosis and Haemostasis. Understanding potential confounding factors including heparin and residual FVIII in test plasma, plus optimal standardization can reduce assay coefficient of variation to 10–20%.These areas are all explored within this article.
Introduction
Type 1 VWD is a common autosomally inherited bleeding disorder resulting from a reduced quantity of essentially normal plasma VWF. Individuals registered at a specialist haemostasis centre with the disorder comprise ≤1 in 10 000 of the population. Although type 1 disease is the most common of the three VWD types [1], little was known about its molecular pathogenesis until the last decade. Recent multicentre studies have led to enhanced understanding of the disease phenotype and genotype.
Accurate measurement of FVIII activity is important in several areas including the diagnosis and management of haemophilia A and potency determination for FVIII containing clotting factor concentrates. Two-stage CS methods determine ability of FVIII to potentiate activation of FX by FIXa in the presence of calcium ions and phospholipid. Use of high plasma dilutions enables the CS assay measurements to reflect only tenase activity making utility of the test widely applicable.
FVIII inhibitors are the most frequently occurring blood coagulation inhibitors with an incidence of up to 30% in severe haemophiliacs [2]. Inhibitors against other coagulations factors including FIX, FXI, FV, FII and Fibrinogen have been described; however, their incidence is low and all occur exclusively after substitution therapy of the respective factor. In contrast, FVIII inhibitors not only occur as a result of substitution therapy in haemophiliacs (allo-antibodies), but may also develop as autologous inhibitors, mostly in elderly people in association with an autoimmune disease or a malignancy, but frequently without an underlying associated disease [3]. Each of these areas will be discussed.
Lessons learnt from type 1 VWD studies
Three multicentre studies on type 1 VWD conducted in the European Union (EU), Canada and the UK each recruited index cases (IC), previously diagnosed with type 1 VWD, their affected and unaffected family members (AFM, UFM) plus healthy controls (HC) [4–6]. Recruitment was based on the 1994 VWD classification [7] and included patients considered to fit the criteria (EU), or used upper (≤0.50 IU mL−1, Canada and UK) and lower (0.05 IU mL−1, Canada) assay limits for plasma VWF. 305 IC were recruited.
Bleeding score (BS)
A previously designed BS tool was further developed for the EU study [8]. Participants were scored on 11 different bleeding symptoms with possible summed scores from −3 to 45. IC with a median BS of 9 had significantly more bleeding than HC (median −1). BS was useful for determining extent and significance of bleeding in an individual and correlated inversely with ristocetin cofactor activity (VWF: RCo). IC bled more than their AFM in many cases suggesting that further factors influence disease severity. BS tools are in routine use by several haemostasis specialists and are being further developed to enhance their utility.
Mutation analysis
Candidate mutations were sought in 305 IC and identified in 65%, leaving a significant proportion with no mutation identified. 75% of mutations were missense alterations; other variants included splice, small deletions and insertions, nonsense and promoter region changes. Some splice, small deletion and insertion mutations predicted translation of an in-frame protein; others led to reduced/absent protein expression from the affected allele. Approximately 15% IC had ≥1 candidate mutation with about half having allelic mutations, while the remainder were compound heterozygotes. VWD penetrance was more often complete in families with VWF levels <0.40 IU mL−1, with all individuals who inherited the familial mutation having bleeding symptoms and mutations being commonly identified (>70% IC). In contrast, among those with VWF levels ≥0.40 IU mL−1, candidate mutations were present in fewer cases (<50%) and were more likely to be incompletely penetrant. Several missense mutations were seen in all three studies including p.Y1584C (13% of 305 IC), p.R1205H (6%), p.R924Q (5%) and p.R854Q (3%).
Missing mutations?
An in-frame deletion of exons 4–5 was recently described in two of 32 IC in the UK study [9]. The same or similar mutations may account for some of the 35% type 1 VWD cases with no mutation identified [4–6]. However, in a proportion of patients, factors other than the VWF gene including blood group O and platelet bleeding disorders probably contribute to reduced VWF level and symptoms.
Multimer analysis
VWF multimers were analysed in all three studies and patients with abnormal profiles were excluded from Canadian and UK studies. EU patients with abnormal profiles were retained and characterized [10]. Partly as a result of these three studies, VWD classification was amended in 2006 [1] and type 1 VWD now includes patients where the proportion of high molecular weight multimers is not significantly decreased, but can demonstrate subtle abnormalities. Additionally VWF has a normal ratio of function to protein quantity. 57 of 150 EU IC (38%) were originally classified as having abnormal multimers (AbM). By the 1994 criteria [7], these IC would not have type 1 VWD. However, only 22 (15%) fall outside the 2006 type 1 VWD criteria. These include cases now classified as type 2A(IIE), resulting from missense mutations in exons 26–28 or type 2A resulting from exon 28 missense mutations. Recent analysis of Canadian IC using the same multimer technique as in the EU study identified AbM in 29 of 75 (39%). Identification of a multimer abnormality correlated strongly with detection of a VWF mutation(s); >95% IC with AbM in Canadian and EU studies having mutations identified [5,10,11].
Desmopressin response and VWF clearance
Response to desmopressin was monitored over a 4-h period in 77 EU IC [12]. 83% patients had a complete (VWF:RCo and FVIII:C > 0.50 IU mL−1) and 13% partial (<0.50 IU mL−1, but at least three-fold baseline) response and response correlated with mutation location and multimer abnormality. Some IC with D3 domain mutations, notably p.R1205H and missense substitutions affecting p.C1130 demonstrated a complete initial VWF response, but a rapid return to baseline levels within 2–3 h, now recognized as a ‘clearance’ phenotype. Most partial and non-responders had A1–A3 domain mutations.
ABO blood group
Blood-group O was more common in type 1 VWD IC (65%) than in HC (38–46%) [4–6] and was particularly prevalent in individuals lacking an identified mutation (76% of EU IC) [5]. Blood group O may add to the effect of VWF variants including p.Y1584C and to non-VWF factors to reduce VWF plasma levels.
In summary, ‘type 1 VWD’ includes a heterogeneous patient group with VWF levels from just detectable into the normal range, some with minor multimer abnormalities, a wide range of bleeding severities and variable desmopressin responses. Phenotype-genotype relationships are being identified.
Assay design and use of chromogenic factor VIII assays
Background and method principle
Two-stage chromogenic substrate (CS) methods, which can be considered as variants of the two-stage (TS) clotting method, for the determination of FVIII activity in plasma and concentrates have been commercially available as kits for up to 25 years [13,14]. All kit methods measure the ability of FVIII to potentiate activation of FX by FIXa in the presence of calcium ions and phospholipids. Similar to the TS clotting method, the first step comprises activation of FVIII and FX and the generated FXa is measured in a second step through hydrolysis of a chromogenic FXa substrate. Thrombin required for activation of FVIII is generated during the assay [13] or present in a reagent.
Assays are designed such that the amount of FXa formed should be directly proportional to FVIII activity in the sample. Chromogenic methods typically provide two measuring ranges, indicating levels of 0.2–2.0 IU mL−1 in one range and down to 0.005–0.01 IU mL−1 in a low range, the latter being used for e.g. diagnosis and classification of haemophilia A. All CS methods are easy to automate and therewith offer cost-efficient use, e.g. when applied on microplates.
Specificity of chromogenic methods
In contrast to one-stage clotting (OS) methods, CS methods are not sensitive to preactivation of FVIII due to fast and complete FVIII activation during the assay. The sensitivity of the OS method for preactivated FVIII results in overassignment of FVIII potency, noticed for intermediate purity plasma-derived FVIII concentrates in the 1980s and again observed in the calibration of the plasma-derived standard Mega 2/BRP 3, where partial activation/structural modifications during manufacturing resulted in ∼30% over-assignment of FVIII potency [15].
For reasons of use of a relatively high plasma dilution and involving only the tenase complex, CS methods are minimally influenced by variable levels of plasma components. This also holds for lupus anticoagulants, which may result in a pronounced underestimation of FVIII activity in OS methods [16].
Assay of FVIII activity in FVIII concentrates
Robustness combined with high assay precision and accuracy led to adoption of the chromogenic method as the reference method in the European Pharmacopoeia in 1994 [17]. Importantly, this method requires predilution of FVIII concentrates in FVIII deficient plasma to 1 IU mL−1 followed by further dilution in buffer containing 1% albumin, the quality of which should always be carefully checked.
The increased complexity through the advent of rFVIII preparations combined with lack of adherence to the reference method has often resulted in highly variable results and hence the CS reference method has been subject of much debate and review [18,19]. The method has therefore been made more stringent by the further requirement of selecting activation times, which warrant maximal rate of FX activation [20].
A relatively greater inter-laboratory variation is obtained on analyses of rFVIII preparations with both OS and CS methods as compared with plasma derived concentrates. Reason(s) for this remain(s) to be identified.
Assay of FVIII activity in plasma
Direct proportionality between FVIII activity and FXa generation enables high resolution for the CS method at both high and low levels of FVIII activity. CS methods show strong correlation with OS methods for analysis of samples from haemophilia A patients both before and after FVIII concentrate infusion [13] and also in samples from VWD patients [21].
Combined use of CS and OS methods is important in diagnosis of new haemophilia A patients, as discrepant results are obtained in certain subgroups. This has been shown in comparisons of OS and TS clotting methods and also for OS and CS methods [22]. TS and CS clotting methods showed lower FVIII activities and were in better agreement with clinical phenotype. Mutation analyses revealed point mutations on the A1, A2 and A3 domain interfaces causing the discrepancy. Interestingly, reversed findings have also been reported and mutations close to thrombin cleavage sites have been identified [23]. In the latter study thrombin was present in one CS reagent and it might be informative to perform analyses with the original CS method [13] which may be suitably modified to explore the initial FVIII activation phase.
Altogether, the CS method has demonstrated wide applicability for determining FVIII activity in plasma samples and FVIII concentrates. Recognizing the diversity of FVIII both regarding its source and its formulation, a humble attitude is recommended on assay of FVIII activity, including careful optimization of preanalytical variables.
Laboratory detection of inhibitors to FVIII and other clotting factors
Both classes of FVIII inhibitors, allo- and auto-antibodies, may present as fully neutralizing inhibitors (type 1) or as inhibitors that only partially inhibit FVIII activity (type 2). The difference in kinetics between the two inhibitor types is probably related to their epitope specificity. Type 1 inhibitors, which occur predominantly in haemophiliacs are directed against the FVIII A2 domain in >70% of patients, whereas type 2 inhibitors, which occur predominantly in acquired haemophilia, are directed against the VWF and phospholipid-binding FVIII C2 domain, making the epitope less accessible and resulting in incomplete FVIII inactivation.
FVIII inhibitors manifest themselves by unexpected moderate-severe bleeding in individuals with previously normal haemostasis (autologous inhibitors) or by excessive bleeds, bleeding in unusual sites or low recovery and/or half-life of infused haemostatic products in haemophiliacs substituted with FVIII. However, these findings are not pathognomonic for presence of an inhibitor since antibodies without inhibitor activity may also decrease clotting factor concentration and increased clearance of infused FVIII may result from low VWF concentration. Therefore, clinical suspicion of an inhibitor must be confirmed by objective laboratory tests.
Inhibitor investigation always starts with screening tests, followed, if needed, by specific tests to quantify and identify the exact nature of the inhibitor. A prolonged activated partial thromboplastin time (APTT) clotting time that is not corrected in a mixing study can indicate presence of an inhibitor, provided that the presence of heparin has been excluded. Special care with APTT mixing tests has to be taken when assessing acquired haemophilia with type 2 inhibitors that do not completely inactivate FVIII:C. Residual FVIII may cause normal or borderline abnormal mixing tests, leading to false-negative screening results. An abnormal mixing test is not specific for individual factor inhibitors as lupus anticoagulants show the same phenomenon.
Quantitative FVIII inhibitor assays are based on a universal method of measuring decrease in FVIII activity in a mixture of an exogenous FVIII source (e.g. normal pooled plasma) and the putative inhibitor plasma in a certain time period. A reference measurement is performed with the same method substituting patient plasma with control plasma lacking inhibitor. Residual factor activities in assay mixtures are measured by OS-based clotting assays (mostly APTT) or CS assays.
The Nijmegen method [24], a modification of the Bethesda assay, has been recommended as the standard assay by the International Society on Thrombosis and Haemostasis Factor VIII/IX Scientific Subcommittee. The method has recently been reviewed [25]. Important features of the assay are the use of buffered normal pool plasma as FVIII source and use of FVIII deficient plasma as control sample. In contrast with other coagulation inhibitors, FVIII inhibitors are time- and temperature-dependent because of the binding of FVIII to VWF. Therefore, it is extremely important to standardize both parameters; 2 h incubation at 37°C is optimal. Care must be taken with quantification of type 2 inhibitors as these do not show parallelism with the calibration curve. Therefore, patient plasma dilutions that give residual activity of ∼50% are used to obtain reliable results.
Presence of heparin and lupus anticoagulant may interfere with the inhibitor assay. Heparin may be a problem in patients with catheters as their access seal is mostly heparin-filled to prevent occlusion. Heparin may contaminate the blood sample when puncturing this seal and thus it is advisable to screen these samples for heparin to exclude its presence. Presence of lupus anticoagulant may also give false-positive results. However, these effects can easily be bypassed using a CS to assay residual FVIII.
Residual FVIII activity in test plasma may influence inhibitor titre. As immunoglobulins are heat resistant, plasma samples can be inactivated by heating (90 min at 58°C) to remove residual FVIII activity. Inhibitor that has already complexed with FVIII will not dissociate from the complex during heating [25]. As VWF is also inactivated during this procedure, the use of VWF-containing FVIII-deficient plasma as control sample and as substrate in the FVIII assay is recommended, as VWF concentration in the test system influences inhibitor data.
Besides the methods based on activity measurements, FVIII antibodies can be quantified by immunological techniques like ELISA [26]. These methods have some technical advantages over inhibitor assays and also detect non-neutralizing antibodies. Sahud [27] recently published a study correlating the inhibitor activity assay and a commercially available ELISA (GTI Diagnostics, Waukesha, WI, USA) in Bethesda positive samples (>0.6 BU mL−1), showing positive ELISA results in 235 of 246 samples. Negative ELISA samples probably reflect lack of specificity of the Bethesda method as at least some of these samples were also negative with the Nijmegen assay. In contrast, several samples with low inhibitor titres showed a strong ELISA antibody signal and two of fifty samples from normal donors with negative Bethesda titres were ELISA-positive indicating the low specificity of ELISA methods for inhibitors.
Until now, all inhibitor and antibody tests lack sensitivity to detect low levels of antibodies and inhibitors. Therefore, we have recently developed a more sensitive method, based on the Nijmegen assay, by concentration of the patient sample, using an alternative patient plasma/FVIII source ratio and analysing residual FVIII activity with a CS method (H. Verbruggen unpublished data). This method enables detection of low inhibitor activities (cutoff 0.04 BU mL−1) and may help resolve the problem of the clinical significance of low titre inhibitors.
Unfortunately, the results of inter-laboratory surveys of FVIII inhibitor assays organized by the ECAT Foundation and by RCPA Haematology QAP show an inter-laboratory coefficient of variation of about 30% for the Nijmegen assay and more than 40% for the original Bethesda method, probably caused by a lack of local standardization of the assay.
It may be concluded that the assay conditions for inhibitor testing are now well known, but need implementing in local laboratories. Further investigations are required to develop more sensitive methods.
Acknowledgements
AG acknowledges support from the European Union under the fifth Framework Programme (QLG1-CT-2000-00387) and the NIH Zimmerman Program for the Molecular and Clinical Biology of VWD (HL-081588). SR acknowledges Dr Marianne Mikaelsson for valuable discussions. BV acknowledges Dr Britta Laros-van Gorkom for reviewing the manuscript.
Disclosure
The authors stated that they had no interests which might be perceived as posing a conflict or bias.
