Genotype/phenotype association in von Willebrand disease: is the glass half full or empty?


David Lillicrap, Department of Pathology and Molecular Medicine, Richardson Laboratory, Queen’s University, Kingston, K7L 3N6, Canada.
Tel.: +1 613-548-1304; fax: +1 613-548-3156.


Summary.  Since the original description of this disease in 1926, major advances have been made in our understanding of the pathogenetic mechanisms responsible for von Willebrand disease (VWD). We now recognize that this disease comprises a collection of diverse quantitative and qualitative abnormalities of the adhesive protein von Willebrand factor (VWF), the key protein involved in platelet adhesion, and the carrier protein for factor VIII (FVIII) in plasma. Since the mid-1970s there has been a growing appreciation of ‘variant’ forms of the disease and in the 20 years since the cloning of the VWF gene, much progress has been made in characterizing the molecular genetic pathology of type 3 VWD and the various type 2 variants, types 2A, 2B, 2M and 2N VWD. In all of these cases, mechanistic insights into VWF structure/function have been forthcoming. Most recently, preliminary results relating to the mutational landscape of type 1 disease have been published that highlight the complex pathogenic background of this mild/moderate quantitative trait.


Since the 1926 description of the clinical features of a new form of inherited bleeding disease by Erik von Willebrand, the global biomedical community has come to recognize that VWD represents the most common inherited cause of bleeding in humans. Up until the 1970s, there was little attempt to differentiate the various forms of the condition apart from the severe variant representing type 3 disease. However, from the mid-1970s to present day, increasing knowledge of VWF structure/function has revealed a highly heterogeneous array of genetic pathologies that result in either reduced amounts of protein or the synthesis of dysfunctional proteins that show reduced efficiency in either the platelet-dependent function or FVIII binding role of VWF (Fig. 1).

Figure 1.

 Structure of the von Willebrand factor pre-pro-polypeptide with the binding sites for its multiple ligands (SP = signal peptide, CK = cysteine knot).

von Willebrand disease classification

The International Society on Thrombosis and Hemostasis Scientific and Standardization Committee on von Willebrand factor have published two ‘consensus’ documents detailing the characteristics of the various VWD subtypes [1]. The most recent 2006 classification maintains the fundamental features of the 1994 document but incorporates new knowledge relating in particular to type 1 VWD. There are two quantitative forms of VWD: type 1 disease in which VWF levels are reduced to between approximately 0.05 and 0.45 IU mL−1 and type 3 VWD in which VWF is virtually undetectable. The qualitative variants comprise three subtypes in which the platelet-dependent function of VWF is affected: type 2A, 2B and 2M disease, and one subtype, type 2N VWD, in which the FVIII binding capacity of VWF is defective.

In this review, the genotype/phenotype correlations for VWD will be discussed, beginning with the least prevalent but well characterized type 3 subtype and ending with the most common and most contentious type 1 form of the disease.

Type 3 von Willebrand disease

This VWD subtype varies in incidence from frequencies of 1 per 500 000 to 1 per million in many Western countries, to figures as high as 6 per million in countries such as Iran where consanguineous marriages are more frequent [2]. This was the form of disease present in the original young girl described by Erik von Willebrand in 1926. In type 3 VWD, a severe mucocutaneous bleeding tendency is often accompanied by spontaneous musculoskeletal bleeding as a result of the associated very low level of FVIII.

Type 3 VWD is inherited as a recessive trait. Thus, the vast majority of parents of type 3 VWD patients will neither show clinical nor laboratory evidence of VWF deficiency. Patients with VWF levels between 0.01 and 0.05 IU mL−1 will invariably have type 3 VWD, as dominant VWF mutations will rarely, if ever, produce this degree of deficiency. These patients will not show a clinically useful response to desmopressin (DDAVP) infusion.

Following the cloning of the VWF gene in the mid-1980s, type 3 VWD mutations were the first to be described in a series of patients with severe VWF deficiency who had sometimes developed anti-VWF antibodies following protein replacement therapy. These initial reports were of total and partial VWF gene deletions from which VWF protein synthesis was impossible. In the most current VWF mutation database there are >100 reports of type 3 mutations [3]. In addition to VWF gene deletions, this molecular pathology includes a miscellany of nonsense, frameshift, splicing and missense substitutions. In some instances these mutations have been inherited in the homozygous state but in many cases the phenotype is caused by compound heterozygosity for two different null alleles.

The gene deletions associated with type 3 VWD range from single exon deletions to complete deletions extending into the adjacent 3′ TMEM16B locus [4]. No additional phenotypic abnormalities have been reported with these large defects.

While the number of unique type 3 VWD mutations continues to rise, there are some recurrent mutations attributable either to founder effects such as deletion mutations reported in Germany, Italy and Hungary, or to substitutions at hypermutable CpG dinucleotides within arginine codons. In addition, the original, Åland Island exon 18, 2435 del C, frameshift mutation had also been reported in many Western countries [5].

The characterization of missense mutations in association with type 3 disease is now well-documented [6]. Several of these mutants remove or introduce new cysteine residues that presumably disrupt disulfide-pairing within the protein. Interestingly, missense substitutions have been reported in sites as diverse as the N-terminus of the propeptide to the extreme C-terminus of the mature VWF subunit. The assumption (not proven in most cases) is that these substitutions disrupt VWF configuration to such an extent that very little, if any, of the protein is secreted into the circulation.

The literature suggests that approximately 3% of type 3 VWD patients treated with protein replacement will develop anti-VWF antibodies [7]. While initial studies suggested that this association was confined to patients with VWF gene deletions, adequate supportive information is not available to substantiate this claim. Furthermore, a lack of standardization of testing for these antibodies further complicates the assessment of this genotype/phenotype correlation.

Type 2 von Willebrand disease (Figs 2 and 3)

Figure 2.

 Location of mutations producing the various type 2 forms of VWD and of the currently-reported accelerated clearance type 1C (clearance) forms of type 1 VWD.

Figure 3.

 Type 2 von Willebrand disease variants localized to domains A1-3.

Three of these VWD subtypes, types 2A, 2B and 2M disease are inherited as dominant traits while type 2N VWD is a recessive disease. There is some evidence to suggest that the phenotypic penetrance of these traits is high and that the variability of expression with specific mutations is minimal.

Type 2A von Willebrand disease

In type 2A VWD, the loss of platelet-dependent function of VWF is attributable, in large part, to a loss of the higher molecular weight forms of the protein. Studies performed over a decade ago indicated that this deficit can result from either a failure to synthesize these multimers (type 2A, group 1) or a tendency for enhanced ADAMTS-13-mediated proteolysis of the secreted high molecular weight forms (type 2A, group 2).

There are now in excess of 80 reports on type 2A VWD mutations in the VWF mutation database [3]. All but four of these mutations involve missense substitutions that adversely affect one of the following processes:

  • 1Multimer formation caused by mutations in the VWF propeptide.
  • 2Multimer formation caused by substitutions in the N-terminus of the VWF monomer.
  • 3A2 domain substitutions resulting in enhanced ADAMTS-13-mediated proteolysis or defective biosynthesis and secretion.
  • 4Dimer formation caused by substitutions at the C-terminus of the VWF monomer.

Approximately 85% of type 2A mutations are located in exon 28, and while many of these substitutions produce their effect through enhancing ADAMTS-13-mediated proteolysis, some of the more disruptive mutations likely result in a group 1 pathogenetic mechanism [8]. As with other genotypes, some of the type 2A mutations are recurrent, attributable to either founder effects or to their occurrence within hypermutable arginine codons (i.e. R1597).

Type 2B von Willebrand disease

This interesting gain-of-function phenotype is the result of missense substitutions in the glycoprotein Ib binding region of VWF, the A1 domain. The VWF A1 domain – GPIbα interaction has been characterized at the level of crystallography and shows that a large concave surface of GpIbα makes contact with the A1 domain at both the apex and base of the A1 structure [9]. The mutations resulting in type 2B VWD, cluster at the base of the A1 domain in the region interacting with the β finger of GpIbα. These mutations probably enhance the affinity of GpIbα binding through the destabilization of interactions within the base of the A1 domain. In some fashion, this destabilization must mimic the structural changes caused by shear stress on immobilized VWF under conditions where platelet adhesion is being promoted.

The VWF mutation database currently lists in excess of 50 reports on mutations resulting in type 2B VWD [3]. All of these mutations except one (insertion of a methionine at codon 1304) involve missense substitutions in exon 28 of the VWF gene, the region encoding the A1 protein domain. Several of these mutations have been reported multiple times, with three of the recurrent mutation sites involving hypermutable arginine codons (R1306W/Q/L, R1308C/P and R1341Q/P/L).

A well-recognized phenomenon in some cases of type 2B VWD is the presence of thrombocytopenia that is exacerbated under conditions of stress or following DDAVP administration. A recent study of 67 type 2B VWD patients possessing 11 different A1 domain missense substitutions showed that the incidence of thrombocytopenia at rest was 30% and that this figure doubled under conditions of stress [10]. Clinical evaluation of these patients over a  24-month period, showed that the risk of bleeding and higher bleeding scores were inversely correlated with the platelet counts in these patients. The missense substitutions H1268D, R1306W, R1308C, I1309V, V1316M and R1341Q/W were especially prone to the development of thrombocytopenia. Within families with these mutations, the phenotypic expression of these abnormalities appeared to be consistent. In contrast to the phenotype associated with most of the type 2B substitutions, the missense mutations P1266Q/L and R1308L were never associated with thrombocytopenia and always showed normal plasma multimer patterns. The former of these substitutions has previously been designated VWD type 1 Malmo/New York and these patients do not develop thrombocytopenia, even after the administration of DDAVP [11,12].

Type 2M von Willebrand disease

Type 2M mutations result in the opposite phenotype to that seen with type 2B substitutions. These missense mutations are again located predominantly in the VWF A1 domain (19/23 of reported 2M mutants) but in this instance, they interfere with the interaction of VWF and GpIbα without affecting the multimeric profile. Also in contrast to the A1 domain type 2B mutations, the 2M substitutions are distributed throughout this structure, and it is presumed that they mediate their loss-of-function phenotype through long-range disruption of the GpIbα binding regions.

While ‘classical’ type 2M mutations reflect hypofunctional GpIbα-binding mutants, there have also been reports of three missense mutations in the A3 domain that adversely interfere with VWF’s binding to collagen and result in mild bleeding phenotypes: S1731T, W1745C and S1783A [13,14].

Type 2N von Willebrand disease

Type 2N VWD presents with a phenotype that is very similar to mild hemophilia A. The only clinical laboratory manifestation is likely to be a low FVIII level of between 0.05 and 0.30 IU mL−1.

As indicated above, this VWD variant is a recesssive trait and thus phenotypic expression requires one of several possible combinations of mutant VWF alleles: homozygosity for two identical 2N alleles, compound heterozygosity for two different 2N alleles or compound heterozygosity for a type 2N allele and a VWF null allele. All three of these allelic patterns have been described in the literature.

The FVIII-binding region of VWF is located at the N-terminus of the VWF monomer. This region is encoded by exons 18–27 of the VWF gene and there are now in excess of 50 reports of missense mutations in this region of the gene giving rise to a type 2N phenotype [3]. In addition, two mutations at codons R760 and R763 interfere with cleavage of the VWF propeptide whose persistence sterically hinders factor VIII-binding to the VWF monomer.

Once again, the most common recurring type 2N mutations occur at hypermutable arginine codons (R816W and R854Q). There is also a correlation between the particular type 2N mutation, the plasma level of FVIII and expectation for responsiveness to DDAVP. Thus, levels of FVIII associated with the R854Q substitution are usually approximately 0.20 IU mL−1 and respond adequately to DDAVP administration. In contrast, with the T791M and R816W substitutions the levels of FVIIII can be below 0.10 IU mL−1 and in this situation, the therapeutic effect of DDAVP may be inadequate.

Type 1 von Willebrand disease (Fig. 4)

Figure 4.

 Potential pathogenetic mechanisms in type 1 von Willebrand disease.

The most incidental form of VWD is the mild/moderate quantitative trait, type 1 disease. Approximately 80% of all VWD cases represent type 1 VWD. However, despite the fact that this is a very common condition, with population incidence figures of between 1% and 1 in 1000, attempts to generate consensus definitions for the disease have been challenging. Two factors continue to plague the definition of the type 1 phenotype: the level of plasma VWF that defines the upper limit of the ‘disease range’, and the extent of structural abnormality acceptable for inclusion as a type 1 VWD variant. Indeed, the most recent International Society on Thrombosis and Haemostasis (ISTH) classification guidelines have yielded somewhat on the latter issue by accepting that some cases of type 1 VWD will show subtly abnormal VWF multimer patterns [15]. The other significant revision to the type 1 VWD classification is the realization that some of the mutations responsible for this phenotype will occur at loci other than the VWF gene.

In the past couple of years, reports on the genotypes in excess of 300 type 1 VWD index cases have appeared from three population-based studies in Europe, the UK and Canada [16–18]. Encouragingly, the results from all the three studies have documented similar findings. These can be summarized as follows:

  • 1Approximately 65% of type 1 VWD cases have candidate mutations within the coding region, promoter or splice sites of the VWF gene.
  • 2More than 100 different candidate VWF mutations have now been reported.
  • 3Approximately 60% of the candidate mutations are missense substitutions.
  • 4Between 15% and 20% of type 1 VWD patients have more than one candidate VWF gene mutation.
  • 5Candidate VWF mutations are more likely to be found with VWF plasma levels <0.30 IU mL−1.

Thus, in these initial surveys of the mutational background of type 1 disease, the clinical heterogeneity that is well recognized appears to be matched by both mutational and locus heterogeneity. While some of the approximately 35% of patients with no apparent VWF gene mutation may have changes in regions of the gene not evaluated in these studies (i.e. within introns or distant regulatory sequences) there is growing expectation that mutations in other genes involved in VWF biosynthesis and processing will also play a role in the pathogenesis of this condition. One such example already exists in a murine model of type 1 disease in which the ectopic expression of a glycosyltransferase enzyme is associated with accelerated clearance of VWF from the plasma [19]. In humans, the contribution of ABO glycan modifications on VWF levels has been appreciated for  well over 20 years [20] and it seems likely that effects of other post-translational modifications may also play a role in regulating the circulating levels of this protein.

Within the mutations, in excess of 100, described to date in type 1 VWD, some recurrent changes have already been characterized. In the Canadian type 1 study, 24% of the characterized mutations were found in multiple index cases [17]. All but one of these recurrent mutations involved missense substitutions.

The problem that remains with type 1 VWD pathogenesis is that in many instances the mechanism through which candidate mutations produce low levels of VWF is unresolved. A good example is the most frequently occurring candidate missense substitution at Y1584C in the VWF A2 domain [21]. This substitution has been documented in 8–25% of type 1 cases in Canada and Europe [16–18]. The VWF plasma levels associated with heterozygosity for Y1584C are usually around 0.40 IU mL−1 and in rare cases homozygous for the 1584C allele, levels around 0.20 IU mL−1 have been documented (i.e. a co-dominant pattern of expression). The 1584C allele is almost always associated with coinheritance of blood group O and its phenotypic penetrance is approximately 70%. Finally, in vitro studies performed by several groups have suggested that the low plasma levels of VWF may result from a combination of impaired secretion and enhanced ADAMTS-13 proteolysis [21,22]. For other common recurring missense mutations, such as R924Q, we have even less knowledge of the mechanisms underlying VWF deficiency.

One pathogenetic theme that has emerged recently is the contribution from accelerated clearance of VWF. The R1205H ‘Vicenza’ variant represents the prototypic example of this phenotype and this mutant has now been shown to exhibit a markedly abbreviated plasma half-life of <2 h [23]. With the information currently available, the accelerated clearance phenotype in type 1 VWD appears to be associated with the following features [24,25]:

  • 1Levels of baseline VWF:Ag <0.30 IU mL−1 and often <0.15 IU mL  mL−1.
  • 2An elevated (between 2-fold and tenfold) ratio of the plasma VWF propeptide to VWF:Ag.
  • 3An exaggerated initial response to DDAVP (>4-fold and often >in excess of tenfold) but subsequent rapid decline in VWF levels (by 4 h).
  • 4Plasma VWF half-lives of <3 h.
  • 5Subtly abnormal VWF multimer patterns (either presence of supra-high molecular weight forms or reduced satellite triplet bands).
  • 6Missense mutations in the D3 or D4-CK domains.

To date, the following VWF missense substitutions have been associated with this phenotype: R1205H, C1130F/G/R, W1144G, C1149R, all in the D3 domain, and S2179F, C2671Y in the D4-CK domains. Although systematic assessments of this phenotype are awaited, it seems that accelerated clearance may be responsible for 5–10% of type 1 cases. It has been proposed that this variant be referred to as type 1C (clearance) disease.

Future priorities in von Willebrand disease genotype/phenotype investigation

While substantial progress has been made in our understanding of the pathogenesis of VWD in the past 30 years, there is still much to learn. For each of the three VWD subtypes important clinical and biologic questions currently remain unanswered.

In type 3 disease, the mechanism of action of the various reported missense substitutions requires investigation and we need more information relating to the genotype-inhibitor correlation.

For type 2 VWD, the recent clinicopathologic studies of type 2B variants indicate that even within a supposed discrete group of mutants, significant phenotypic heterogeneity exists. Thus, it seems likely that additional evaluation of these variants in terms of their clinical features (bleeding scores and penetrance) and functional characteristics such as susceptibility to ADAMTS-13-mediated proteolysis, GpIbα and FVIII binding and clearance rates will reveal additional levels of complexity of these phenotypes.

Finally, knowledge of the pathogenesis of type 1 disease is still in its infancy. Ongoing studies in North America and Europe will continue to add knowledge in the next few years but there is every expectation that this complex genetic trait will be associated with an array of genetic and environmental factors resulting in reduced VWF levels. Plans are already underway to organize genome-wide association studies to detect additional genetic loci contributing to this trait, but if the results of recent similar studies are applicable, these investigations will need to include very large study populations and even then, may only detect loci that contribute small (1–5%) variances in the VWF level.

In closing, advances in our understanding of the pathophysiology of this key hemostatic protein have taught us much in the past three decades. We can look forward with anticipation to further insights in the years ahead.

Disclosure of Conflict of Interests

The author states that he has no conflict of Interest.