Dr Ioana Nitu-Whalley Haemophilia Centre and Haemostasis Unit, Royal Free Hospital, Pond Street, London NW3 2QG, UK.
This clinical retrospective study investigated the difficulties in diagnosing type 1 von Willebrand disease (VWD). A total of 246 patients previously diagnosed with type 1 VWD were reclassified into ‘possible’ type 1 VWD (patients with low levels of VWF adjusted for the blood group and either a significant bleeding history or family history) and ‘definite’ type 1 VWD, requiring low levels of von Willebrand factor (VWF), a bleeding history and inheritance. On reclassification, only 144/246 (59%) patients had low VWF levels adjusted for blood group, 88/246 (36%) patients met all the criteria for ‘definite’ type 1 VWD and 51/246 (21%) patients were ‘possible’ type 1 VWD. A significant proportion of patients, 102/246 (42%), remained an indeterminate group with blood type O, VWF levels between 35 and 50 U/dl and personal and/or family bleeding history. This subgroup might require reclassification as ‘not VWD’. However, a similar bleeding tendency was found in two matched groups of patients of blood groups O and non-O and VWF levels between 35 and 50 U/dl. These results suggest that the use of ABO adjusted ranges for VWF levels might not be essential for diagnosis, because bleeding symptoms may depend on the VWF level regardless of the ABO type. Of the diagnostic criteria, the bleeding history was of prime importance in the clinical decision to diagnose and treat type 1 VWD. These observations could help in the reconsideration of how the criteria for diagnosing type 1 VWD could be adjusted in order to maximize their clinical relevance.
Von Willebrand disease (VWD) is the most common inherited bleeding disorder caused by quantitative or qualitative defects of von Willebrand factor (VWF). VWF plays an essential role in the primary haemostasis by promoting platelet adhesion and aggregation under high shear conditions ( Sadler, 1993; Meyer & Girma, 1993). Congenital VWD has been divided into three categories: type 1 and type 3 refer to a partial or virtually complete deficiency of plasma VWF; and type 2 reflects a qualitative defect of VWF. Type 2 VWD is further subdivided into four variants (2A, 2B, 2M and 2N) according to specific phenotypic features ( Sadler, 1994; Sadler & Gralnick, 1994; Nishino & Yoshioka, 1997). This classification is based on the principle that all VWD is the result of mutations at the VWF locus, which is located on chromosome 12. Type 1 VWD is transmitted in an autosomal dominant fashion and is characterized by a partial deficiency of structurally and functionally normal VWF. The incidence of VWD has been reported to be between 1% and 3% in the general population, with type 1 VWD representing over 70% of all VWD ( Holmberg & Nilsson, 1972; Rodeghiero et al, 1987 ; Werner et al, 1993 ). However, type 1 VWD has an incomplete penetrance (≈ 60%) and variable expressivity, largely resulting from a number of both genetic and environmental modifier factors ( Miller et al, 1979 ; Nichols & Ginsburg, 1997).
One of the best characterized genetic modifiers of plasma VWF levels is the ABO blood group ( Gill et al, 1987 ). Previous studies have shown that individuals with blood group O have lower levels of plasma VWF than individuals with blood groups A, B or AB. The mean von Willebrand factor antigen (VWF:Ag) level is 25% lower for persons of blood group O compared with other blood types, and one likely explanation may be related to the presence of glycosyltransferases ( Gill et al, 1987 ; Rodeghiero et al, 1987 ; Yamamoto et al, 1990 ).
These complex interactions account for a wide range of severity observed in type 1 VWD, making a correct diagnosis, especially in the mild forms, a real challenge. One of the main difficulties is to differentiate between mild forms of type 1 VWD and normal subjects with blood group O.
The diagnostic criteria of type 1 VWD are based on the presence of low levels of structurally normal VWF, which imply decreased levels of VWF ristocetin cofactor activity (VWF:RCo) and VWF:Ag of less than 2 SD below ABO type-adjusted population mean, on two determinations, the evidence of a significant mucocutaneous bleeding history and a family history of type 1 VWD or demonstration of an appropriate causative mutation. A significant mucocutaneous bleeding history is defined by at least two bleeding symptoms in the absence of a blood transfusion history, or one symptom requiring treatment with blood transfusion, or one symptom recurring on at least three distinct occasions; and a positive family history that includes either at least one first-degree relative or at least two second-degree relatives with both a personal history of bleeding and laboratory values compatible with type 1 VWD (reviewed by Batlle et al, 1997 ).
As the association between low levels of VWF and bleeding history can be coincidental, and not causative, a definitive diagnosis of VWD requires that their co-inheritance is demonstrated. Thus, it appears that a ‘definite’ type 1 VWD diagnosis needs documentation of low levels of VWF, the presence of a bleeding history and inheritance (family history or appropriate VWF mutation). The diagnosis of type 1 VWD is only ‘possible’ in patients with laboratory parameters compatible with type 1 VWD (low levels of normal VWF), who have either a significant mucocutaneous bleeding history or a positive family history.
We studied retrospectively a sizeable VWD population currently registered at the Royal Free Hospital Haemophilia Centre and analysed the blood group, the bleeding history, the family history and the laboratory parameters supporting the diagnosis. In an attempt to standardize the criteria used to register a patient as affected with type 1 VWD, we analysed how the specific diagnostic parameters had influenced the decision on diagnosis. Consequently, we reclassified our population into two categories as ‘definite’ and ‘possible’ type 1 VWD in order to evaluate which particular diagnostic criteria were most relevant to the clinical practice. For each subgroup of patients, we analysed the role of the bleeding symptoms and whether the bleeding history differed in these subgroups. We also addressed the issue of whether stratification of patients by ABO blood group and adjustment of the VWF levels for the blood group was useful in the clinical diagnosis of VWD and in the management of patients with bleeding symptoms, irrespective of the presence of a definitive diagnosis.
MATERIALS AND METHODS
This was a retrospective study and included a total of 270 patients representing multiple members of different families previously diagnosed with type 1 VWD and registered at our centre up to December 1997. Each patient had a review of their medical records, and data including VWF levels, blood group type, personal bleeding history and family history were collected. Twenty-four patients were excluded from the study as there were no records of their blood group.
Data on VWF levels were collected from the patients records and included the VWF activity (VWF:Ac) and VWF:Ag levels. The VWF:Ac was measured as a functional assay using a monoclonal antibody-based enzyme-linked immunosorbent assay (ELISA) technique, which represents an alternative to the ristocetin cofactor activity ( Murdock et al, 1997 ). The VWF:Ag assay was carried out by standard ELISA techniques ( Cejka, 1982). The normal ranges for both tests based on a normal local population were established in the past, with a lower limit for both assays of 50 U/dl. As the criteria for diagnosing type 1 VWD were not available in a standard protocol when the original diagnosis of type 1 VWD was made, the influence of age, blood group or other variables was not taken into account. Thus, historically, a diagnosis of type 1 VWD has been accepted at our centre if the VWF activity levels were less than 50 U/dl, irrespective of blood group or age.
For the purpose of this retrospective analysis, we assumed the lower limit of normal of VWF:Ac 35 U/dl for blood group O and 50 U/dl for blood group non-O. The lower level of 35 U/dl for blood group O is the traditionally accepted limit in the literature ( Gill et al, 1987 ), although ideally, the lower limit of reference ranges should be calculated directly in the local population for both O and non-O subjects and age adjusted.
The age of our population ranged between 1 year and 93 years and, although it is known that the VWF levels increase with age ( Gill et al, 1987 ), the age variable was not included in the retrospective analysis.
A bleeding history was considered positive if the subject presented at least two bleeding symptoms or one symptom recurring on at least two occasions, as documented in the medical records.
The data on bleeding symptoms were incomplete with respect to the site, severity and frequency of bleeding. Therefore, we had to resort to some sentinel bleeding symptoms that characterized the different subgroups of patients. Hence, the bleeding events were classified broadly into mucocutaneous and non-mucosal bleeding.
Mucocutaneous bleeding included spontaneous mucosal bleeding (nose bleeds, easy spontaneous bruising and haematoma, bleeding from the oral cavity and gingival bleeding, menorrhagia), traumatic bleeding (prolonged bleeding after cuts and wounds and excessive bruising and haematoma after trauma) and excessive bleeding after mucosal surgery (tooth extraction or oral surgery and ENT surgery). Non-mucosal bleeding included spontaneous deep bleeding (gastrointestinal haemorrhage, bleeding post partum and bleeding into muscle or joints) and bleeding after abdominal surgery.
Another classification of the bleeding symptoms was into ‘bleeding requiring any form of specialized treatment’ (desmopressin and/or clotting factor concentrates) and ‘bleeding episodes not requiring treatment. In the category of bleeding requiring treatment, we included any form of treatment administered either preventively (e.g. for surgical procedures) or for treatment of a spontaneous/traumatic bleed. In the category of bleeding not requiring treatment, the patients presented with a positive bleeding history, with or without a haemostatic challenge, but they received no form of treatment as documented in the notes.
A positive family history of VWD was accepted if there was at least one first-degree family member recorded in the notes who was registered with a diagnosis of type 1 VWD based on a personal bleeding history, laboratory diagnosis or both. For a few patients, the relatives were registered elsewhere, so their coagulation parameters were not always available. In these rare cases, a confirmation of the specific diagnosis of VWD based on the bleeding history was sought from the index patient.
Parametric statistics were used to compare the bleeding histories between blood group O and non-O putative VWD patients. Comparisons of non-categorical values were made using the chi-squared test.
The study comprised a total of 246 patients previously registered with type 1 VWD on the basis of at least a VWF activity and antigen levels of less than 50 U/dl and a documented blood group.
The eligible population was classified according to blood group: 173/246 (70%) patients blood group O and 73/246 (30%) patients blood group non-O.
Type 1 VWD patients blood group O (n = 173)
For the 173 individuals with blood group O, the diagnosis was re-evaluated based on the VWF:Ac values adjusted for ABO blood group (normal range 35–150 U/dl). Of the 173 patients, 71 had VWF:Ac < 35 U/dl, and 102 patients had VWF:Ac levels between 35 and 50 U/dl, representing two distinct subsets (Fig 1).
Individuals with blood group O and VWF:Ac between 35 and 50 U/dl (n = 102)
Of the 102 individuals who did not have a low VWF:Ac level when adjusted for the blood group, 45/102 (44%) would otherwise be eligible for a ‘definite’ type 1 VWD diagnosis, and 53/102 (52%) would qualify for a ‘possible’ type 1 VWD diagnosis, 18/102 (18%) on the basis of positive FH and 35/102 (34%) on the basis of positive BH. Four subjects from this group had no family history or a bleeding history ( 1Table I).
Table 1. Table I. Classification of previously diagnosed type 1 VWD into ‘possible’ and ‘definite’ type 1 VWD categories and stratification by blood group type. FH, family history; BH, personal bleeding history.
Individuals with blood group O and VWF:Ac < 35 U/dl (n = 71)
Seventy-one patients with blood group O had a VWF:Ac level of less than 35 U/dl, but only 40/71 (56%) were assigned to a diagnosis of ‘definite’ type 1 VWD. Of the remainder, 27/71 (38%) were ‘possible’ type 1 VWD: 12/71 (17%) patients had a documented family history, and 15/71 (21%) individuals had a recorded bleeding history. Four subjects had no bleeding or family history demonstrated ( 1Table I).
Type 1 VWD patients blood group non-O (n = 73)
Seventy-three patients with blood group non-O had a VWF:Ac less than 50 U/dl. Within this group, 48/73 (66%) patients were classified as ‘definite’ type 1 VWD, with both a personal and a family bleeding history. A total of 14/73 (19%) patients had a positive family history, and 10/73 (14%) patients demonstrated a significant bleeding history. Thus, 24/73 (33%) patients with blood group non-O were ‘possible’ type 1 VWD. One subject had no bleeding or family history demonstrated ( 1Table I).
Diagnosis of ‘possible’ and ‘definite’ VWD
Of the original 246 patients previously diagnosed as type 1 VWD, 133/246 (54%) fulfilled all the criteria for a ‘definite’ type 1 VWD diagnosis, and 104/246 (42%) had only ‘possible’ type 1 VWD: 44/246 patients (18%) had a positive family history alone, and 60/246 (24%) patients had a personal bleeding history.
Of note, we found 14 individuals who had VWF:Ac levels between 5 and 10 U/dl but did not fulfil the criteria for a ‘definite’ type 1 VWD diagnosis. Despite low levels of VWF:Ac, four of them were asymptomatic.
Assessment of the bleeding history
A direct comparison between individuals who experienced any kind of bleeding episodes vs. individuals with no bleeding symptoms showed very similar results among different categories of patients stratified by blood group (P = 0.959, 2 Table II).
Table 2. Table II. Classification of different subgroups of type 1 VWD patients according to the presence of bleeding episodes that did or did not require specialized treatment.
Table 3. Table III. Classification of different subgroups of patients with type 1 VWD according to the type of bleeding pattern (mucosal, non-mucosal or both).
Thus, we found a similar clinical pattern of bleeding symptoms in the blood group O and blood group non-O patients, irrespective of the VWF levels.
Type 1 VWD comprises a heterogeneous group of patients in whom the clinical diagnosis is often difficult because of a considerable intraindividual phenotypic variation. By definition, low levels of VWF resulting from modifying influences are not included in the diagnosis of type 1 VWD. Thus, the demarcation between mild type 1 VWD and normal poses a diagnostic dilemma.
We reviewed a population of 246 individuals who had been previously diagnosed as type 1 VWD and analysed the criteria that were used for diagnosis. In accordance with other studies, we observed that the majority of patients (70%) were blood group type O, in comparison with the expected frequency of 45% blood group O in the normal population ( Gill et al, 1987 ).
Only 59% of the population in our cohort had low VWF levels when adjusted for their blood group. By definition, only these patients would have been eligible for a diagnosis of VWD and could be classified further into ‘definite’ and ‘possible’ type 1 VWD (36% exhibited all the criteria for a ‘definite’ type 1 VWD diagnosis, and 21% of the population qualified for a ‘possible’ type 1 VWD diagnosis). In 2% of the studied population, the diagnosis was unacceptable, as it was made solely on the detection of low levels of VWF, without the presence of a bleeding or family history.
In our cohort, 41% of the population had blood group O and ‘normal value’ VWF:Ac (between 35 and 50 U/dl). On reclassification, they represent an indeterminate group, in whom the VWF level reduction comprises a mix of ‘genetic’ and ‘non-genetic’ VWD causes. This group might require reclassification as ‘not VWD’ and possibly a search for an alternative diagnosis in the presence of bleeding symptoms. However, asymptomatic patients with borderline normal values of VWF should be regarded with caution, as many of them might not have had a haemostatic challenge to manifest a bleeding tendency. In these doubtful cases, patients should be reassured and the investigations repeated. However, when the VWD diagnosis can neither be confirmed nor excluded and the risk of bleeding is unknown, empirical treatment is recommended ( Nishino & Yoshioka, 1997).
The ABO locus affects the level of plasma VWF, and the combination of VWF mutation and blood type O may be associated with bleeding symptoms. Blood group O alone could cause low levels of VWF and suggest a VWD diagnosis. It is estimated that about 20% of normal subjects experience excessive bleeding in a lifetime ( Miller et al, 1979 ). Bleeding symptoms and low levels of VWF alone are common and may occur together by chance, so 0.5% of all classified VWD patients are made on a chance combination alone ( Rodeghiero et al, 1987 ).
In our series of patients, a large number of blood group O individuals with VWF ranged between 35 and 50 U/dl did not have definitive VWD, but they were symptomatic. Moreover, this subgroup of patients had a similar bleeding pattern (type of bleeding and requirement for treatment) to the other groups of patients with blood type non-O and VWF levels between 35 and 50 U/dl. Although the sample size differed and the kind and severity of bleeding symptoms among different subgroups were not analysed, it was noticeable that there were similar clinical manifestations in individuals with similar levels of VWF, irrespective of their blood group. These results suggest that the use of ABO-adjusted ranges for VWF levels might not be essential for diagnosis, because bleeding symptoms may depend on VWF level regardless of the ABO type.
Other investigators have also encountered similar problems with the definition and classification of type 1 VWD. Dean et al (1997 ) reported difficulties in classifying a high percentage of paediatric cases, and Ingerslev & Gursel (1999) exemplified family studies that posed a diagnostic dilemma. Fressinaud et al (1998 ) described two patients with blood group O, VWF:Ac > 35 U/dl and positive bleeding history as ‘borderline normal subjects’, because they could not be diagnosed using the highly sensitive PFA-100. In a recent ongoing study looking at the effect of adjusting the VWF levels for the blood group, about 30% of patients with menorrhagia had subnormal VWF levels, but nearly half did not fit the ABO-adjusted laboratory criteria for VWD. However, such patients had similar bleeding features and warranted consideration for similar therapies to the VWD patients ( Kouides et al, 1999 ).
In summary, this retrospective clinical study underlined the practical difficulties in the diagnosis of type 1 VWD. Full criteria for a ‘definite’ type 1 VWD diagnosis (low levels of VWF for the blood group, bleeding history and inheritance) were met in a small number of patients. Of the three parameters, the bleeding history was of prime importance in the clinical decision to diagnose and treat type 1 VWD. We speculate that the distinction between different blood groups and separate normal ranges for VWF should not influence the decision of diagnosis in symptomatic individuals.
On the basis of limited laboratory data, bleeding and family history, the diagnosis of type 1 VWD, especially in the mild forms, is very difficult. New diagnostic tools, such as the PFA-100 test, could be useful in the diagnosis and therapeutic monitoring of patients with VWD ( Cattaneo et al, 1999 ). The study of platelet VWF content can further assist in the diagnosis ( Rodeghiero et al, 1990 ). In addition, characterization of molecular genetics may provide further insights to clarify how the criteria for diagnosing type 1 VWD, especially the mild forms, could be adjusted to maximize their clinical relevance.
Dr I. Nitu-Whalley is supported by the European Union Grant no. BMH4-CT97-2256. We are grateful to Professor Francesco Rodeghiero for helpful comments on the manuscript.