Hemorrhagic symptoms and bleeding risk in obligatory carriers of type 3 von Willebrand disease: an international, multicenter study


Francesco Rodeghiero, Department of Hematology and Hemophilia and Thrombosis Center, San Bortolo Hospital, I-36100 Vicenza, Italy.
Tel.: +39 0444 753626; fax: +39 0444 753922; e-mail: rodeghiero@hemato.ven.it


Summary. Objectives: We undertook an international, multicenter study to describe the clinical picture and to estimate the bleeding risk in a group of obligatory carriers of type 3 von Willebrand disease (VWD). Patients and methods: Obligatory carriers (OC) of type 3 VWD were identified by the presence of offspring with type 3 VWD or by being an offspring of a type 3 patient. Normal controls were age- and sex-matched with the obligatory carriers. A physician-administered standardized questionnaire was used to evaluate hemorrhagic symptoms at presentation. A score system ranging from 0 (no symptom) to 3 (hospitalization, replacement therapy, blood transfusion) was used to quantitate bleeding manifestations. Odds ratios were computed for each symptom. Results: Ten centers participated to the study, enrolling a total of 35 type 3 VWD families, with 70 OC. A total of 215 normal controls and 42 OC for type 1 VWD were also included. About 40% of type 3 OC had at least one bleeding symptom compared to 23% of normal controls and 81.8% of type 1 OC (P < 0.0001 by chi-squared test), showing that type 3 OC clearly represent a distinct population from type 1 OC. The clinical situations associated with an increase of bleeding risk in type 3 OC were epistaxis [odds ratio 3.6; 90% confidence intervals (CI) 1.84–21.5], cutaneous bleeding (odds ratio 5.5; 90% CI 2.5–14.1) and postsurgical bleeding (odds ratio 16.3; 90% CI 4.5–59). The severity of bleeding score correlated with the degree of factor (F) VIII reduction in plasma. Conclusions: OC for type 3 VWD represent a distinctive population from type 1 OC. These patients, however, present with more frequent bleeding symptoms in comparison to normal controls, especially in case of significantly low FVIII. Desmopressin and/or tranexamic acid might be useful to prevent or treat bleeding in these cases.


von Willebrand disease (VWD), the most frequent inherited hemorrhagic disorder [1], is caused by deficiency or abnormal function of von Willebrand factor (VWF) [2]. The clinical spectrum of the disease is characterized by an extreme heterogeneity of bleeding symptoms, ranging from a few, doubtful hemorrhagic symptoms to the occurrence of severe life-threatening episodes [3]. The current revised classification of VWD identifies two major categories, characterized by quantitative (type 1 and 3) or qualitative (type 2) VWF defects [4]. Partial quantitative deficiency of VWF in plasma with or without reduced content in platelets identifies type 1 VWD, whereas type 3 VWD is characterized by total absence or trace amounts of VWF in plasma and platelets. Type 1 is easily distinguished from type 3 by its milder VWF deficiency (usually in the range of 20–40 U dL−1), the autosomal dominant inheritance and the presence of milder bleeding symptoms [5].

Type 3 VWD is an extremely rare disease, with no more than three to five cases per million inhabitants [2,3]. These patients present with moderate to severe bleeding symptoms, often requiring replacement therapy with FVIII/VWF concentrates [6]. Unlike type 1 patients in whom usually a single allele is mutated, these patients are usually homozygotes or compound heterozygotes for VWF gene mutations in both alleles, leading to severely reduced or absent plasma and platelet VWF [7]. A wide heterogeneity of plasma and/or platelet VWF levels are usually observed among obligatory carriers (OC) of type 3, pointing at the heterogeneity of the molecular basis responsible for the deficiency [8]. Indeed, these subjects may be asymptomatic and phenotypically normal or may have bleeding symptoms and meet the criteria of type 1 VWD [8–10]. Thus, a clear distinction between type 1 and type 3 OC cannot be invariably made, and a study comparing bleeding history and laboratory phenotype of type 3 OC with type 1 OC and normal controls would be helpful in clarifying these issues. Furthermore, the severity of bleeding and bleeding risk in specific circumstances (e.g. surgery) in type 3 OC has never been formally estimated and compared with type 1 OC or normal population by using standardized measurements.

Recently, we developed a bleeding score system (BS), which takes into account the number and the severity of bleeding symptoms, and validated it in an international multicenter study in OC of type 1 VWD [11]. As a part of the project, we also studied OC of type 3 VWD identified in kindreds with type 3 affected individuals, and compared bleeding history with that of a normal control population and type 1 OC, through a physician-administered standardized questionnaire to produce a descriptive picture of clinical symptoms and to estimate the bleeding risk of these subjects.

Patients and methods

Selection of participating centers

Participation to the International Study was offered to internationally recognized centers among the members of the International Society on Thrombosis and Haemostasis, within the framework of the Scientific and Standardization Subcommittee on VWF. All the centers that responded were accepted. The study included separate enrollment and analysis of type 1 and type 3 families. The results obtained with type 1 VWD families have already been reported [11] and the present study is focused on type 3 VWD families.

Definition of type 3 VWD patients

Type 3 VWD patients were defined as any symptomatic patient with undetectable (< 3 IU dL−1) VWF antigen (VWF:Ag) level.

Definition of obligatory type 3 VWD carriers

A subject was considered as OC if she/he had an offspring with type 3 VWD or if she/he was the offspring of a type 3 VWD patient. All these subjects should have been living and available for direct history taking at the time of enrollment. Informed consent was obtained from investigated subjects according to local institutional review board (IRB) procedures.

Definition of obligatory type 1 VWD carriers

The criteria have been already described [11]. Briefly, a subject was considered as an OC in presence of at least another affected first degree relative and an affected offspring. Thus, a total of three members were evaluated for each family.

Definition of controls

Each participating center enrolled normal controls, matched by age and gender with type 3 OC. Controls were eligible if in ostensible good health and had never been referred for evaluation because of hemorrhagic symptoms.

Bleeding history and score system

The same physician-administered standardized questionnaire was administered to type 3 and type 1 OC and normal controls by a physician aware of carriership for the disease status (either of type 1 or 3), but unaware of the previous bleeding history. The average and most severe presentation of each symptom was required. Specific symptoms included epistaxis, cutaneous symptoms, bleeding from minor wounds, oral cavity bleeding, gastrointestinal bleeding, postpartum hemorrhage, muscle hematomas/hemarthrosis, bleeding after tooth extraction, bleeding after surgery, menorrhagia. The severity of each symptom was evaluated, using a score system ranging from 0 to 3 as follows: 0, absent or trivial; 1, present; 2, requiring medical attention; 3, requiring blood transfusion, replacement therapy or desmopressin administration. Bleeding symptoms were recorded on a standardized form (available at http://www.med.unc.edu/isth/SSC/collaboration/Bleeding_Type1_VWD.pdf; accessed 27 June 2006).

FVIII:C and VWF measurement

FVIII:C and VWF were measured at each participating center, with results expressed as IU dL−1. The centers used a one-stage activated partial thromboplastin time clotting assay for FVIII:C, ristocetin cofactor using formalin-fixed platelets and aggregometry, ELISA using polyclonal antibodies for VWF:Ag. A two-stage chromogenic peptide substrate method for FVIII:C and latex immunoassay test for VWF:Ag were used at the Karolinska University Hospital. The median value of available data at each center was included for analysis.

Statistical methods

The chi-squared test was adopted to assess differences of bleeding symptoms frequencies. Fisher's exact test was used to assess differences in the severity of bleeding score. Kruskal–Wallis test was used to assess differences in FVIII and VWF levels between type 1 and type 3 OC. Logistic regression was used to compute the age-adjusted odds ratio for each bleeding symptoms and to compare them between type 3 OC and normal controls. Logistic regression was also used to evaluate the relative contribution to bleeding symptoms of VWF:Ag, FVIII:C and presence of a disease allele (either type 1 or type 3). For this purpose, we modeled the probability of having an increased bleeding score (defined as greater than 3 or 5 in males and females, see [11]) as a function of being a normal control, an OC of type 1 or type 3, and of the levels of VWF:Ag or FVIII:C.


Ten centers participated in the study, enrolling a total of 35 type 3 VWD families. All included families fulfilled previously established enrollment criteria. The characteristics of the study population are reported in Table 1. A total of 285 subjects were available for analysis: 70 type 3 OC and 215 normal controls. Furthermore, a group of 42 type 1 OC was available for comparison [11]. As expected, FVIII and VWF levels were very low in type 3 index cases while OC had levels around 50%, in keeping with the recessive transmission of type 3 VWD. Obligatory carriers for type 1 VWD showed intermediate levels, all measurements being significantly lower than in type 3 OC. VWF ristocetin cofactor (VWF:RCo)/VWF:Ag ratio, although statistically significantly different between type 1 and type 3 OC, was around the unit in both groups. The possibility of having some type 2M patients mimicking type 1 OC cannot however be excluded.

Table 1.   Main characteristic of the study population
 Control subjects (n = 215)Type 3 affected (n = 35)Type 3 obligatory carriers (n = 70)Type 1 obligatory carriers (n = 42)
  1. The values are given as mean ± SD (range). **P = 0.0001 vs. type 3 obligatory carriers. *P = 0.01 vs. type 3 obligatory carriers. For laboratory measurements, ranges refer to the individual median values.

  2. VWF, von Willebrand factor; Ag, antigen; RCo, ristocetin cofactor.

Gender (M/F)90/12517/1835/3518/24
Age at diagnosis, median 3 (1–22)32 (5–54)27 (3–67)
Age at study, median49 (2–72)15 (3–45)47 (25 –73)47 (12–71)
Factor VIII:C (IU dL−1) 4.24 ± 4.09 (1–15)88.2 ± 33.4 (18–171)44.4 ± 25.4** (5–106)
VWF:Ag (IU dL−1)<358.1 ± 33.2 (12–192)26.8 ± 17.7** (5–72)
VWF:RCo (IU dL−1)<356.3 ± 27.7 (12–160)18.9 ± 18.2** (3–60)
VWF:RCo/VWF:Ag ratio 1.05 ± 0.28 (0.44–1.9) 0.89 ± 0.37* (0.43–2)

Hemorrhagic symptoms were analyzed from a qualitative point of view (presence/absence of a specific hemorrhagic symptom) and a quantitative point of view (hemorrhagic score).

Qualitative analysis

Table 2 summarizes the number of hemorrhagic symptoms in the investigated groups. About 77% of normal subjects had never suffered from bleeding, in comparison to 60% of type 3 OC and 12% of type 1 OC (P < 0.0001). Out of 215 normal controls, one female only (0.46%) reported three or more hemorrhagic symptoms in comparison to 6 (8.5%) type 3 and 21 (50%) type 1 OC. Thus, type 3 OC had more bleeding symptoms than normal controls (P = 0.002) but less than type 1 OC (P < 0.0001).

Table 2.   Distribution of hemorrhagic symptoms compared in the three groups of investigated subjects
Symptoms (n)Control subjects [n = 215 (%)]Type 3 obligatory carriers [n = 70 (%)]**Type 1 obligatory carriers [n = 42 (%)]*
  1. **P = 0.002 vs. control subjects. *P < 0.0001 vs. type 3 obligatory carriers.

0165 (76.7)42 (60)5 (11.9)
133 (15.4)12 (17.2)5 (11.9)
216 (7.4)10 (14.3)11 (26.2)
31 (0.5)4 (5.7)7 (16.7)
41 (1.4)7 (16.7)
51 (1.4)6 (14.2)
61 (2.4)

Quantitative analysis and estimation of bleeding risk in obligatory carriers of type 3 VWD

The score for each hemorrhagic symptom is reported in Table 3 and is limited to the comparison between type 3 OC and normal controls, as the results with type 1 OC have been extensively reported [11]. Overall, there was a strong significant difference in the severity of bleeding symptoms between the two groups for epistaxis (odds ratio 3.6; 90% CI 1.84–21.5), cutaneous bleeding (odds ratio 5.5; 90% CI 2.5–14.1) and postsurgical bleeding (odds ratio 16.3; 90% CI 4.5–59).

Table 3.   Bleeding score distribution in type 3 obligatory carriers and in control subjects and odds ratio for a specific bleeding symptom (type 3 obligatory carrier vs. controls)
SymptomPatientsBleeding scoreχ2P*Odds ratio (95% CI)
  1. Odds ratios are calculated for presence (score 1, 2 and 3 together) vs. absence of symptoms (score 0). *Fisher's exact test.

MenorrhagiaType 3 obligatory carriers281320.961.4 (0.5–3.9)
PostpartumType 3 obligatory carriers290120.073.8 (0.7–20.7)
Tooth extractionType 3 obligatory carriers333100.0912.1 (0.6–7.2)
EpistaxisType 3 obligatory carriers501370<0.00013.6 (1.84–21.5)
Cutaneous bleedingType 3 obligatory carriers58462<0.00015.5 (2.5–14.1)
WoundsType 3 obligatory carriers643110.251.6 (0.5–4.9)
SurgeryType 3 obligatory carriers21602<0.000116.3 (4.5–59)

Laboratory parameters predictive of bleeding in obligatory heterozygotes for type 3 VWD

Figure 1 shows the relationship between the severity of bleeding score and FVIII/VWF level. It appears that the lower the level for each measurement, the higher the bleeding score observed. One-way analysis of variance showed an overall significant difference among the groups (FVIII:C P < 0.0001; VWF:Ag P < 0.01; VWF:RCo P = 0.02). In a multivariate model, including carriership of a disease allele, VWF:Ag and FVIII:C, carriership of either a type 1 or type 3 allele was the most predictive variable associated with an increased BS (P < 0.0001 for both type 1 or type 3 vs. normals). Type 1 OC were, however, more symptomatic than type 3 OC (OR for an increased BS of type 1 vs. type 3 OC 6.2; P = 0.008). FVIII:C was independently associated with an increased BS (P = 0.008), while VWF:Ag level was not after adjustment for FVIII:C. This suggests that FVIII:C has a greater influence than VWF:Ag in modulating the bleeding risk in both type 1 and type 3 OC.

Figure 1.

 Relationship between bleeding score and the level of factor VIII:C and von Willebrand factor (VWF) measurements in obligatory carriers of type 3 von Willebrand disease (VWD). For each quintile of bleeding score, the boxes span from 25th to the 75th percentile. The center line of each box represents the median values.


The bleeding tendency of VWD is highly heterogeneous, ranging from a minor post-traumatic bleeding to severe life-threatening episodes requiring substitutive treatment and hospitalization [2,3]. Patients with type 3 VWD, the recessive severe form of the disease, are considered to be at high bleeding risk due to markedly low FVIII and VWF levels. The actual risk of bleeding in obligatory carriers for type 3 VWD is still a matter of debate. These subjects usually present with FVIII and VWF levels around 50%, as expected in subjects supposedly having a single ‘non-functional’ VWF allele [8,12]. As a consequence, only normal VWF molecules, although reduced in concentration, are predicted in plasma. In the past, discordant results have been reported on the risk of bleeding of these subjects, not always related to the actual level of circulating VWF [10,11]. Some authors have considered type 3 patients as offspring of two type 1 patients, but this occurrence seems to be an exception rather than a rule. In fact, in Swedish and Finnish type 3 VWD population only [13] it has been demonstrated that more than two mutations can occur in a few propositi. Furthermore, in some cases type 1 and type 3 VWD may share the same mutations [14]. A formal estimation of the bleeding risk of these subjects has never been obtained and the comparison with normal population only assessed in descriptive terms [15]. Indeed, it has been repeatedly reported that normal controls may suffer minor bleeding symptoms [15–17], but a thoroughly carried out examination through a physician-administered standardized questionnaire and numeric score has been reported only recently [11].

In this study, we analyzed the bleeding history of 70 OC for type 3 tracked through the identification of 35 type 3 index cases and compared them with 215 normal controls. Type 3 OC were identified on the basis of a pedigree showing that they were parents or offspring of a type 3 patient. In this regard, true paternity was confirmed by VWF gene mutation tracking in the 14 families for which mutation was available. The severely reduced levels of FVIII and VWF in type 3 VWD and mildly reduced levels observed in OC are in accordance with the diagnosis of recessive severe type 3 VWD (Table 1). Some type 3 OC had very low FVIII and VWF levels mimicking some type 1 VWD patients. Yet, the average levels of FVIII and VWF were significantly greater in type 3 OC in comparison to type 1 OC (Table 1), suggesting that the genetic bases for the two groups should be different. However, although most families with type 3 VWD have null mutations, a great heterogeneity of VWF levels and reported bleeding history could be observed in heterozygotes carrying the same mutation, without a clear relationship between VWF levels and bleeding history [9]. This heterogeneity adds to the difficulties of clinical and laboratory diagnosis of VWD [18].

Previous surveys estimated a prevalence of self-reported bleeding symptoms up to 51% in normal subjects, although almost all clinically irrelevant. Conservatively, it has been suggested that at least 25% of normal subjects may report at least one bleeding symptom [19]. The present study shows that 23% of normal subjects has at least one bleeding symptom, as assessed by a physician-administered questionnaire, in contrast with about 40% of type 3 OC and 89.1% of type 1 OC (Table 2) [11]. Thus, the prevalence of bleeding symptoms in type 3 OC is strikingly different when compared with OC in type 1 VWD families with autosomal dominant inheritance and tested with the same methodology [11]. This again underlines the fact that the genetic basis for the two forms of the disease should be different in most cases.

In a previous report, data was collected from 191 OC for type 3 VWD [19]. The risk of bleeding was not significantly different between 117 subjects with VWF less than 50% and 74 subjects with VWF more than 50% (relative risk 2.0, P = 0.051). However, these data were derived from the collection of clinical histories reported by several heterogeneous studies, without any possibility of standardization. We collected data on frequency and severity of bleeding in 70 OC of type 3 through a detailed standardized questionnaire directly administered to the patients by a physician. It appears that some of these subjects may have significant bleeding, with surgery being the most significant situation at bleeding risk (Table 3). The present study shows that number and severity of bleeding symptoms are significantly different between type 3 VWD OC and normal controls, as previously observed with OC for type 1 VWD (11). Furthermore, in a multivariate model the level of FVIII:C in type 3 OC was independently associated with an increased bleeding score, while VWF:Ag did not. This correlation has been recently observed also in a study with type 1 VWD in which a similar bleeding score system has been used [20]. Unfortunately, the small number of clearly symptomatic subjects (12 type 3 OC with bleeding score > 3) did not allow any systematic analysis of factors predicting bleeding in type 3 OC. In subjects with significant bleeding history, in addition to tranexamic acid, an infusion with desmopressin should be considered for cases with particularly low FVIII:C to cover surgical procedures or to control bleeding episodes [2].

In conclusion, the present study suggests that obligatory carriers of type 3 VWD represent a distinctive population from patients with type 1 VWD, even though, in general, mutations in only one of the two alleles is expected to be present in both type of patients. However, rare type 1 patients can have two mutant alleles, as in some families from Åland [9]. Thus, definite conclusions require further genetic investigations. Furthermore, although most episodes are clinically mild, an increased risk of bleeding may occur in some obligatory carriers of type 3 VWD, showing increased frequency of postoperative bleeding, epistaxis and cutaneous hemorrhages.


F. Rodeghiero: study initiation and coordination; G. Castaman, A. Cappelletti, F. Baudo, J. C. J. Eikenboom, A. B. Federici, S. Lethagen, S. Linari, J. Lusher, M. Nishino, P. Petrini, A. Srivastava, J. S. Ungerstedt: data collection; G. Castaman, F. Rodeghiero, A. Tosetto, A. Cappelletti: analysis and interpretation of results; A. Tosetto: statistical analysis; G. Castaman: lead author of initial manuscript; G. Castaman, F. Rodeghiero, A. Tosetto, J. C. J. Eikenboom, A. B. Federici, P. Petrini: revisions of draft manuscript; G. Castaman, F. Rodeghiero, A. Tosetto, F. Baudo, J. C. J. Eikenboom, A. B. Federici, S. Lethagen, S. Linari, J. Lusher, M. Nishino, P. Petrini, A. Srivastava, J. S. Ungerstedt: review and approval of the final manuscript.


We thank all the members of participating laboratories for their excellent work. Dr A. Cappelletti was the recipient of a research grant by AVEC (Associazione Veneta per l'Emofilia e le Coagulopatie). We thank Dr M. Blombäck for critical reading of the manuscript.

Disclosure of Conflict of Interests

The authors state that they have no conflict of interest.