Use of the collagen-binding assay for von Willebrand factor
in the analysis of type 2M von Willebrand disease:
a comparison with the ristocetin cofactor assay


  • Anne F. Riddell,

    1. Katharine Dormandy Haemophilia Centre and Haemostasis Unit, Department of Haematology, Royal Free Hospital, Pond Street, London NW3 2QG, UK
    Search for more papers by this author
  • P. Vincent Jenkins,

    Search for more papers by this author
    • *

      Present address: Department of Biophysics and Biochemistry, University of Rochester School of Medicine and Dentistry, Rochester, NY 14642, USA.

  • Ioana C. Nitu-Whalley,

    1. Katharine Dormandy Haemophilia Centre and Haemostasis Unit, Department of Haematology, Royal Free Hospital, Pond Street, London NW3 2QG, UK
    Search for more papers by this author
  • Angus H. McCraw,

    1. Katharine Dormandy Haemophilia Centre and Haemostasis Unit, Department of Haematology, Royal Free Hospital, Pond Street, London NW3 2QG, UK
    Search for more papers by this author
  • Christine A. Lee,

    1. Katharine Dormandy Haemophilia Centre and Haemostasis Unit, Department of Haematology, Royal Free Hospital, Pond Street, London NW3 2QG, UK
    Search for more papers by this author
  • Simon A. Brown

    1. Katharine Dormandy Haemophilia Centre and Haemostasis Unit, Department of Haematology, Royal Free Hospital, Pond Street, London NW3 2QG, UK
    Search for more papers by this author

: Professor Christine A. Lee, Haemophilia Centre and Haemostasis Unit, Royal Free and University College Medical School, Pond Street, London NW3 2QH, UK. E-mail: Christine.Lee@


Summary. This study compares the utility of two functional assays for von Willebrand factor (VWF), the ristocetin cofactor assay (VWF:RCo) and the collagen-binding assay (VWF:CBA). We analysed a group of 32 patients with type 2 von Willebrand disease (VWD) (25 patients with type 2M, six with type 2A and one with type 2B) and 22 normal control subjects. VWF:RCo/VWF antigen (VWF:Ag) ratios and VWF:CBA/VWF:Ag ratios were compared between the patient and control groups. In the six patients with type 2A VWD, both VWF:RCo/VWF:Ag ratios and VWF:CBA/VWF:Ag ratios were discordant (≤ 0·7). In the 25 type 2M VWD patients, the VWF:CBA/VWF:Ag ratios were concordant (> 0·7), but the VWF:RCo/VWF:CBA ratios were discordant (≤ 0·7) (P = 0·001) compared with control subjects. Thus, VWF:RCo/VWF:Ag ratios were discordant in both type 2M and 2A VWD patient groups indicating a functional abnormality. However, VWF:CBA/VWF:Ag ratios were discordant in the type 2A VWD group but not in the type 2M VWD group. Our study showed that VWF:CBA is sensitive to functional variants associated with the loss of high-molecular-weight multimers, i.e. type 2A and 2B in VWD, but the assay was unable to discriminate defective platelet-binding VWD variants with normal multimeric patterns such as type 2M VWD. It was concluded that the VWF:CBA assay should be used in association with rather than as a replacement for the VWF:RCo assay.

Von Willebrand factor (VWF) is a large multimeric glycoprotein found in plasma, platelets and the endothelial cells, and it has two main roles in haemostasis. Firstly, VWF is a constitutive carrier of factor VIII, prolonging its circulating half-life. Secondly, VWF plays a critical role in platelet:platelet and platelet:subendothelium binding via its interactions with a number of ligands including platelet glycoprotein (GP)Ib-V-IX, GPIIb-IIIa and collagen types I, III and VI (Sadler, 1998).

Von Willebrand disease (VWD), the most common inherited bleeding disorder, is a very heterogeneous disorder resulting from either a quantitative or a qualitative abnormality of VWF. According to the revised classification of VWD, type 1 is defined as a partial quantitative deficiency of VWF, type 2 as qualitative abnormalities of VWF, and type 3 refers to the virtually complete absence of VWF (Sadler, 1994). Type 2 VWD is divided into four subtypes: type 2A is characterized by reduced platelet-dependent function resulting from the absence of high- (HMW) and intermediate-molecular-weight multimers; type 2B is characterized by an increased ability of VWF to bind GPIb complex on platelets; type 2M is defined as a decreased platelet-dependent function that is not caused by the absence of HMW multimers; and type 2N is defined by decreased binding to factor VIII.

The laboratory diagnosis and classification of VWD relies on phenotypic characterization. This involves using screening, diagnostic and discriminating tests. The diagnostic tests commonly used are determinations of plasma factor VIII level, VWF antigen (VWF:Ag) and assessment of the functional ability of VWF to bind platelets, as determined by the ristocetin cofactor agglutination activity assay (VWF:RCo). Type 1 and type 2M VWD can be difficult to differentiate; however, defects that interfere disproportionately with VWF platelet-dependent function, such as type 2A and 2M VWD are generally characterized by a discordant VWF:RCo/VWF:Ag ratio of < 0·7 (Mancuso et al, 1996; Federici, 1998; Hillery et al, 1998). This ratio is characteristically concordant in type 1 VWD (≥ 0·7). Further diagnosis of type 2 VWD relies on the analysis of functional parameters of VWF, generally by testing platelet-rich plasma for ristocetin-induced platelet aggregation (RIPA), by performing plasma multimer analysis and, in the case of suspected type 2N VWD, the measurement of factor VIII binding capacity (Mazurier et al, 1990).

Owing to limitations in sensitivity, reproducibility and interlaboratory variability of the agglutination-based VWF:RCo assay (Favaloro et al, 1999), there has been renewed interest in the VWF collagen-binding assay (VWF:CBA) as a measure of VWF function (Brown & Bosak, 1986; Favaloro et al, 1990, 1993; Favaloro & Koutts, 1997; Fischer et al, 1998; Favaloro, 2000). The VWF:CBA is based on the ability of VWF preferentially to bind collagen to the HMW multimers of VWF. The assay has been shown to be sensitive in the discrimination of type 1 and types 2A and 2B VWD, where the VWF:CBA/VWF:Ag ratios were significantly lower in the latter two groups (Favaloro & Koutts, 1997).

In this study, we have used the VWF:CBA in parallel with the VWF:RCo assay to test previously characterized patients with types 2A and 2M VWD. The latter group had recently been classified as type 2M VWD on the basis of discriminant VWF:RCo/VWF:Ag of < 0·7 in the presence of HMW multimers (Nitu-Whalley et al, 2000). The current study has demonstrated that the VWF:CBA was insensitive to the functional defect in patients with type 2M VWD.

Patients and methods

Patient selection and control group.  The patient group consisted of 32 patients previously diagnosed as having VWD: six patients with type 2A, 25 patients from 14 kindred with type 2M VWD and a single patient with type 2B VWD, who was not included in the statistical analysis. The characterization and diagnosis of the patients with 2M VWD have been described previously (Nitu-Whalley et al, 2000). All patients classified as type 2M demonstrated normal distribution of low-, medium- and high-molecular-weight multimers. Patients 8 and 9 demonstrated the presence of suprahigh-molecular-weight multimers. The control group consisted of 22 healthy volunteers. All participants gave informed consent.

Methods.  Venous whole blood was collected into 0·106 mol/l sodium citrate (Sarstedt Monovette 9NC/3 ml tubes). After centrifugation at 2000 g for 10 min at 4°C, samples were aliquoted into Sarstedt cryotubes and stored at −70°C until further testing. A 20 normal pool standard calibrated against the 7th British standard (NIBSC, Potters Bar, UK) was used in VWF:Ag and VWF:RCo assays.

VWF:Ag was measured in patients and control subjects by an in-house enzyme-linked immunosorbent assay (ELISA). ELISA (Maxisorp nunc immuno plate, through LifeTechnologies, Paisley, UK) plates were coated overnight at 4°C using rabbit polyclonal anti-human VWF (Dako, Glostrup, Denmark) in pH 9·6 carbonate:bicarbonate buffer (Na2CO3 0·15 mol/l, NaHCO3 0·035 mol/l). Plates were washed five times with a high salt phosphate wash buffer (Na2H2Po4 0·3125 mol/l, Na2HPO4 0·0935 mol/l, NaCl 1·25 mol/l), pH 7. Dilutions of the 20 normal pool standard, patient and control plasmas were made in high salt phosphate wash buffer (Na2H2Po4 0·3125 mol/l, Na2HPO4 0·0935 mol/l, NaCl 1·25 mol/l), pH 7, containing 3% (w/v) polyethylene glycol (PEG) (Woodhams & Kernoff, 1983). Samples were aliquoted and incubated at room temperature on a plate shaker at 400 r.p.m. After a further five washes, plates were tagged with anti-VWF horseradish peroxidase antibody (Dako). After a further incubation step for 1 h and a further washing step, the reaction was visualized by the addition of substrate containing 10 mg of O-phenylenediamine dihydrochloride (Sigma Chemical, Poole UK). This reaction was stopped using a 1·5-mol/l H2SO4 solution, and the absorbance was measured at 490 nm.

VWF:RCo was measured by platelet aggregometry using fresh washed platelets from a normal donor (McFarlane et al, 1975). Dilutions of patient and control samples were analysed against a 20 normal pool standard.

VWF:CBA was measured using type III collagen (human placenta type X acid soluble; Sigma Chemicals) as described previously by Brown & Bosak (1986). Briefly, 25 mg of type III collagen was dissolved in 20 ml of 50 mmol/l acetic acid and then diluted to a final concentration of 4 µg/ml. ELISA plates were coated with 100 µl of 4 µg/ml type III collagen in coating buffer (0·14 mol/l NaCl, 0·006 mol/l Na2HPO4, 0·002 mol/l NaH2PO4, 0·03 mol/l KCl, pH 7·3). The coated plates were sealed and left at room temperature for 24 h, then placed at 4°C for a further 48 h before use. Patient and control plasma was diluted 1:200 and 1:400 in high salt phosphate wash buffer (Na2H2PO4 0·3125 mol/l, Na2HPO4 0·0935 mol/l, NaCl 1·25 mol/l), pH 7, containing 3% (w/v) PEG and aliquoted. The ELISA plates were incubated and washed as described for the VWF:Ag assay. Patient and control values were compared with a 20 normal pool standard curve. A value of 100 µ/dl was designated for the collagen binding of the 20 normal pool plasma.

Statistical methods.  The Mann–Whitney non-parametric t-test was used for comparing the patient groups and control group, and the Wilcoxon signed rank test was used for comparing results within patient groups. All VWF:RCo and VWF:CBA results with values of < 5·0 U/dl were assigned a value of 4·0 U/dl during statistical analysis. 95% confidence intervals (95% CI) were calculated for all parameters in both control and patient groups (Gardner & Altman, 1989).


Results of VWF:Ag, VWF:RCo and VWF:CBA in both control and patient groups are shown in Tables I–III (Figs 1 and 2 allow a comparison of the results obtained between patient groups). The normal control subjects gave comparable results for VWF:Ag, VWF:RCo and VWF:CBA, with similar median values. The 95% CIs were identical for VWF:Ag and VWF:RCo, and slightly broader for WF:CBA. The VWF:RCo/VWF:Ag and VWF:CBA/VWF:Ag ratios were similar and demonstrated similar median values and a slightly wider 95% CI (Table I).

Figure 2.

Scattergram showing range of VWF:RCo/VWF:Ag ratio results for type 2A and 2M patients and the control group. The horizontal line represents the median of results for each group.

Figure 1.

Scattergram showing range of VWF:CBA/VWF:Ag ratio results for type 2A and 2M patients and the control group. The horizontal line represents the median of results for each group.

Table I.  Laboratory data for the normal control group.
 VWF:Ag (IU/dl)VWF:RCo (U/dl)VWF:RCo/Ag ratioVWF CBA (IU/dl)VWF:CBA/Ag ratio
  1. Results are shown as range of results, median, standard deviation (SD) and 95% CI.

95% CI75–10075–1000·9–1·084–1180·9–1·2

In the patient group with type 2A VWD, although all measured VWF parameters were below normal, both VWF:RCo and VWF:CBA were disproportionately reduced compared with VWF:Ag (P = 0·031 and P = 0·031 respect- ively). This disproportionate reduction in function was reflected in the low VWF:RCo/VWF:Ag and low VWF:CBA/VWF:Ag ratios (Table II, Figs 1 and 2). The 95% CI and median values of VWF:RCo/VWF:Ag and VWF:CBA/VWF:Ag ratios within the type 2A VWD patient group were similar (P = 0·48).

Table II.  Laboratory data obtained in patients with type 2A and 2B VWD.
PatientVWF:Ag (IU/dl)VWF:RCo (IU/dl)VWF:RCo/VWF:Ag ratioVWF:CBA (U/dl)VWF:CBA/VWF:Ag ratio
Type 2A
 2643100·2< 5< 0·1
 272050·2< 5< 0·3
 28850·6< 5< 0·6
 2929180·6< 5< 0·2
Median22·5< 50·45·00·3
95% CI8–435–180·2–0·65–130·1–0·6
Type 2B

As expected in this previously defined group, all VWF:RCo levels in the patients with 2M VWD were below normal, and the VWF:RCo/VWF:Ag ratio was disproportionately reduced (≤ 0·7). However, no significant difference between VWF:CBA and VWF:Ag was shown in this group (P = 0·19) (Table III). The VWF:CBA/VWF:Ag ratio was ≥ 0·7 in all type 2M VWD cases (Table III, Fig 1). No significant difference between VWF:CBA/VWF:Ag ratios in the 2M patient group was shown compared with the ratio in the normal control group; the 95% CIs were 0·8–1·3 and 0·9–1·2 respectively (P = 0·30). Comparison within the type 2M VWD patient group of VWF:RCo/VWF:Ag with VWF:CBA/VWF:Ag ratios showed that they were significantly different (P = 0·001). Three patients (2, 4 and 10) displayed VWF:CBA/VWF:Ag ratios of 0·7, and another four patients (3, 8, 9 and 22) displayed ratios of 0·8. Both these values were slightly lower than the lowest value in the control group of 0·9, but the reduction was not comparable with those seen in the group of patients with type 2A VWD.

Table III.  Laboratory data for patients with type 2M VWD.
PatientVWF:Ag (IU/dl)VWF:RCo (IU/dl)VWF:RCo/VWF:Ag ratioVWF:CBA(U/dl)VWF:CBA/ VWF:Ag ratio
  • *

    Indicates patients with normal VWF:Ag and VWF:CBA (see text for details).

Type 2M
38< 5< 0·660·8
49< 5< 0·660·7
615< 5< 0·3191·3
712< 5< 0·4131·1
819< 5< 0·3160·8
917< 5< 0·3140·8
1122< 5< 0·2180·8
1233< 5< 0·2551·7
1440< 5< 0·1350·9
1522< 5< 0·2291·3
1623< 5< 0·2190·8
1713< 5< 0·4241·8
1811< 5< 0·5211·9
95% CI18–405–100·2–0·419–530·8–1·3

Patients 12, 17 and 18 demonstrated slightly higher VWF:CBA/VWF:Ag ratios compared with the normal range. Densitometric analysis of the VWF multimers in these patients has demonstrated a normal multimer pattern and distribution (data not shown).


Owing to the poor reproducibility and considerable interassay and interlaboratory variability of the VWF:RCo assay, there has been renewed interest in alternative assays for measuring VWF function, in particular the use of the VWF:CBA assay (Favaloro et al, 1993, 1999, 2000; Fischer et al, 1998). We have retested a group of 32 patients with previously characterized VWD and 22 healthy control subjects using the VWF:Ag, VWF:RCo and VWF:CBA assays. The source and types of collagen have been shown to be important variables in the performance of the VWF:CBA (Favaloro, 2000). In this study, we used type III collagen from human placenta for the VWF:CBA, which has been shown previously to be sensitive to the loss of VWF HMW multimers (Favaloro, 2000). The VWF:Ag, VWF:RCo and VWF:CBA assays were all found to be normal in the control group, reflecting the presence of normal binding of VWF to GpIb-V-IX and human type III collagen, consistent with normal levels of fully functional HMW VWF multimers. The median results and 95% CI for absolute values ofall three VWF parameters tested in the normal subjects showed similar values. The VWF:RCo/VWF:Ag ratio results were similar, but four control subjects had slightly higher VWF:CBA results compared with their VWF:Ag (data not shown), which is reflected in the slightly wider 95% CI. This is in agreement with similar previous studies (Favaloro et al, 1993; Federici et al, 2000).

Using the VWF:RCo assay, all six patients with type 2A VWD, the single patient with type 2B VWD and the 25 patients with type 2M VWD displayed disproportionately reduced VWF:RCo/VWF:Ag ratios (≤ 0·7), and these patients would have been classified as having a qualitative defect on this basis. We expected this finding, as the VWF:RCo and VWF:Ag assays were used to identify and, together with multimer and RIPA analysis, to classify these patients. The six patients with type 2A VWD and the single patient with type 2B VWD displayed an abnormal VWF:CBA and disproportionately reduced VWF:CBA/VWF:Ag ratios, consistent with the qualitative defect in the VWF of these individuals. However, none of the 25 type 2M VWD patients displayed a significantly reduced VWF:CBA/VWF:Ag ratio. Six were borderline for a qualitative defect using the VWF:CBA as a replacement for the VWF:RCo assay and, in the remaining 19, the VWF:CBA results were concordant with the VWF:Ag results. Use of the VWF:CBA as a replacement for the VWF:RCo assay would have led to the misclassification of these patients with type 2M VWD as having either type 1 VWD or, in the case of patients 10, 19, 20, 21 and 22, as being normal (asterisked in Table III).

These data suggest that the VWF:CBA detects qualitative defects associated with loss of HMW multimers in type 2A VWD, where there is decreased synthesis or increased proteolysis of VWF leading to the loss of HMW multimers and, in type 2B VWD, where increased clearance of HMW multimers occurs as a result of increased binding of VWF with GpIb. However, in type 2M VWD, where HMW multimers are present despite defective platelet–VWF interaction, the VWF:CBA reflects the HMW VWF present. This observation is important, as few cases of type 2M VWD are described, and this subtype is likely to be underdiagnosed and commonly mistaken for type 1 VWD (Sadler, 1994; Hilbert et al, 2000; Nitu-Whalley et al, 2000). Despite the recognized limitations of the VWF:RCo assay, based on these findings and similar findings by Federici et al (2000), the VWF:CBA should not be considered as a replacement for the VWF:RCo assay in the diagnosis and classification of VWD.

The VWF:RCo assay, as performed by platelet aggregometry, measures the ability of VWF to interact via GPIb-V-IX with platelets in the presence of ristocetin, whereas the VWF:CBA is a direct reflection of the ability of VWF to bind to collagen. Thus, each of these assays measures a different function of VWF. On account of the complexity of VWF, no single assay will assess the complete functional integrity of the VWF protein. Thus, the use of more than one functional test can ensure a fuller diagnosis of VWD. Although insensitive to type 2M VWD variants, the use of the VWF:CBA in combination with the VWF:RCo assay may provide a more complete approach to the diagnosis of qualitative variants. The use of both VWF:RCo and VWF:CBA assays allows the discrimination of 2A, 2B and2M subtypes. In type 2A VWD, both VWF:CBA and VWF:RCo are disproportionately reduced in association with the loss of HMW multimers. In the more specific defect of type 2M VWD, only the VWF:RCo is disproportionately reduced, and the VWF:CBA remains concordant with the VWF:Ag, owing to the presence of HMW multimers. Furthermore, a recently described ELISA method based on a recombinant GPIb fragment is expected to be of value as a replacement for the VWF:RCo assay by measuring the same functional ability but in a simpler and more reproducible manner (Vanhoorelbeke et al, 2000).

In conclusion, we have shown that the VWF:CBA is a sensitive test in the diagnosis of type 2A VWD but of limited use in discriminating cases of type 2M VWD. The VWF:CBA should not be considered as a replacement for the VWF:RCo assay but, rather, as another useful diagnostic test in the profile of VWD diagnosis and classification, enhancing the ability to determine functional variants in VWD.


Dr I. Nitu-Whalley is supported by EU grant number BMH4-CT97-2256. This study was funded by the Katharine Dormandy Trust for Haemophilia and Allied Disorders, reg. no. 262434. The authors wish to thank Paul Murdock for technical help.