Different bleeding risk in type 2A and 2M von Willebrand disease: a 2-year prospective study in 107 patients


Giancarlo Castaman, Department of Cell Therapy and Hematology, San Bortolo Hospital, I-36100 Vicenza, Italy.
Tel.: +39 444 753679; fax: +39 444 753922.
E-mail: castaman@hemato.ven.it


Summary.  Background:  Type 2A and 2M von Willebrand disease (VWD2A and VWD2M) are characterized by the presence of a dysfunctional von Willebrand factor (VWF) and a variable bleeding tendency. So far, a head-to-head comparison of the clinical history and bleeding risk between VWD2A and VWD2M has never been provided in a prospective manner.

Aim of the study:  We assessed the bleeding incidence rate and clinical characteristics in two cohorts of 17 families (46 patients) with VWD2A and 15 families (61 patients) with VWD2M prospectively followed-up for 24 months. VWF gene mutations were characterized in all of them.

Results:  Mean bleeding score (BS) and VWF antigen at enrollment were significantly higher in VWD2A patients (P = 0.007). No correlation between VWF activity or factor VIII levels and the severity of BS was observed. The incidence rate of spontaneous bleeding requiring treatment was 107/100 patient-years (95% CI, 88.3–131) in VWD2A compared with 40/100 patient-years (95% CI, 30–53) in VWD2M (P < 0.001). The risk of bleeding was significantly higher in patients with BS ≥ 10 at enrollment compared with those with BS 0–2. Furthermore, 54 episodes of gastrointestinal bleeding occurred in 17/46 (36.9%) VWD2A patients and seven in 2/61 (3.3%) VWD2M patients (P < 0.0001).

Conclusion:  Bleeding tendency in VWD2A is greater than that of VWD2M, is not explained by factor VIII or VWF levels and is mainly due to an increased incidence of gastrointestinal bleeding.


von Willebrand disease (VWD) is an inherited bleeding disorder characterized by a variable bleeding tendency and heterogeneous laboratory phenotypes [1]. According to the ISTH classification, three different types have been recognized in VWD depending on quantitative (VWD1 and VWD3) or qualitative (VWD2, including four subtypes: VWD2A, VWD2B, VWD2M and VWD2N) VWF defects [2]. Among qualitative defects, VWD2A and VWD2M are the most frequent and are characterized by defective VWF-dependent platelet adhesion and/or decreased VWF–platelet interactions with or without the loss of high-molecular-weight (HMW) multimers, although type 2M cases resulting from impaired binding to subendothelium have been recently described [3].

At least two different pathophysiological mechanisms are evident within VWD2A: mutations that impair the production of the larger multimers (Group I) and those with normal synthesis and release but with increased proteolytic susceptibility to ADAMTS13 (Group II) [3,4]. The mutations responsible are predominantly located in the A2 domain of VWF, with some (e.g. S1506L and R1597W) being more largely represented [3]. In the VWD2A category mutations formerly classified as type IIC are also included, characterized by reduced proteolytic susceptibility to ADAMTS13 [5], and the previous type IID, characterized by mutations in the dimerization area in the CK domain [3,5]. Recently, also patients with mutations in the D3 domain associated with the relative reduction of HMW multimers (type IIE) have been included in VWD2A [3,5].

VWD2M is often difficult to diagnose accurately because of its phenotype, which can be difficult to differentiate from VWD1 or VWD2A unless a full range of phenotypic evaluations is carried out [3,6–9]. A few typical mutations (R1315C/L or R1374H/C), mainly located in the A1 domain, represent the large majority of cases [3,9,10], but some authors suggest including these mutations among VWD2A (IIE) cases, because of the relative decrease of larger multimers [6,7,11–13]. Thus, it has been also suggested that VWD2M should be incorporated into the large group of VWD2A [14]. However, typically in VWD2M there should be no loss of HMW VWF multimers, as required for typical VWD2A in the original classification.

The risk of bleeding in VWD has only recently been addressed in a prospective manner in well-characterized patients. By means of an indirect comparison, the risk of bleeding appears to be greater in patients with VWD2B compared with those with VWD1 and increased VWF clearance [15,16]. So far, a head-to-head comparison of the clinical history and bleeding risk between VWD2A and VWD2M has never been provided in a prospective manner. The aim of our study was to evaluate the type and frequency of spontaneous bleeding requiring treatment during a 2-year follow-up period in two large cohorts of patients with VWD2A and VWD2M.

Materials and methods


This is a prospective cohort observational study on patients diagnosed and followed at two institutions (Hemophilia and Thrombosis Centers in Milano and Vicenza). Eligible patients were drawn from the two Centres’ databases by reviewing the data of all patients with a diagnosis of VWD2A and VWD2M. Diagnosis of VWD2A and VWD2M was based on the ISTH criteria [2]. Patients unable to participate in the planned visits or to give informed consent were considered ineligible. We excluded from this study patients with concomitant hepatic or renal insufficiency, who used antiplatelet or anticoagulant drugs, were pregnant or had active malignancy at the time of first study visit, or on home-treatment with desmopressin or VWF/FVIII concentrates prior to the start of the study (two with type 2A and one with type 2M).

Baseline visit

Enrolled patients were evaluated at a baseline visit and consecutively followed for 24 months thereafter. A standardized questionnaire, originally devised to determine a bleeding severity score in VWD1 [17], was administered to all patients during the baseline visit, and a blood sample was obtained from all patients. The bleeding score was computed from the symptoms that had occurred in the patients before entering the study.

Follow-up procedures

All the enrolled patients were advised to contact the Center for any significant bleeding, which in the subject’s opinion required at least medical consultation. Bleeding symptoms were also assessed by direct interview every 6 months and all the patients were contacted at the end of the follow-up (24 months) to assess the occurrence of bleeding symptoms by more detailed questions. Patients were informed about the anonymous use of their data and the purpose of the study and gave informed consent in accordance with the Declaration of Helsinki. The study was approved by the Institutional Review Board of the hospitals and was registered at the EMEA, EUDRA (Europe) (ID number EudraCT-2005-004496-38).

Laboratory methods

Factor (F) VIII and VWF were measured at each Centre at baseline in all patients according to previously reported methods [18]. The evaluation of multimer pattern of plasma VWF was carried out by using low-resolution gel electrophoresis using 0.9% low gelling temperature agarose or medium resolution using 1.6% agarose and blotting. The VWF gene was screened for mutations by PCR and direct sequencing of all 52 exons, including intron-exon boundaries in all index cases [19]. Primers used (synthesized by Sigma-Aldrich, St Louis, MO, USA) are listed at http://www.euvwd.group.shef.ac.uk/data.htm. The specific mutation was subsequently identified in all the affected relatives included in the study.

Statistical analysis

Given the low number of investigated type 2A and 2M patients, we compared differences between these two subgroups using one-way analysis of variance and log-transformed data instead of using non-parametric methods. The yearly incidence of bleeding during follow-up (and its 95% confidence interval [CI]) in the two groups was calculated by dividing the total number of bleeding episodes by the total number of patient-years. For each patient, the observation period started on 1 April 2007 and ended on 31 March 2009. The cumulative hazard of bleeding was calculated using the Nelson-Aalen estimator, assuming a recurrent-event model [20]. In this model, patients with a bleeding event are not censored at time of first bleeding but are retained for all the 24-month study time. Differences in incidence rates were computed according to Rothman et al. [21] for univariate analyses, and using a Cox regression model for multivariate analysis. A probability level of 0.05 was taken as cut-off for statistical significance. Statistical analysis was performed using the Stata software package (Stata 11.0; StataCorp, College Station, TX, USA).


General characteristics of the patient population

After review of 621 patients diagnosed as having VWD, a group of 46 patients with VWD2A (24 men and 22 women; median age at enrollment, 47.5 years; range, 13–87) belonging to 17 unrelated families and a group of 61 patients with VWD2M (25 men and 36 women; median age at enrollment, 53 years; range, 19–96) belonging to 15 unrelated families entered the study. Table 1 summarizes the main features of the two groups of patients at recruitment. Male to female ratio and age were similar between the groups. Significantly higher VWF:Ag plasma levels were evident in VWD2A, while FVIII:C and VWF:RCo values were similar in the two groups of patients. In all, baseline VWF:RCo/VWF:Ag was ≤0.6. Bleeding score at enrollment was significantly greater in the population of patients with VWD2A.

Table 1.   Main demographic and phenotypic data in the investigated patients at baseline
 VWD2A (n = 46)VWD2M (n = 61)P
Median age (range)47.5 (13–87)53 (19–96)0.26
Bleeding score (mean ± SD)10.5 ± 5.57.3 ± 5.50.007
FVIII:C (IU dL−1, mean ± SD)54.2 ± 24.546.1 ± 18.40.07
VWF:Ag (IU dL−1, mean ± SD)44.5 ± 22.427.2 ± 13.7<0.001
VWF:RCo (IU dL−1, mean ± SD)9 ± 7.28 ± 3.50.19
VWF:RCo/VWF:Ag (mean ± SD)0.21 ± 0.160.36 ± 0.130.002

Tables 2 and 3 show the gene mutations of the two patient populations, together with the median FVIII and VWF according to the identified mutation. S1506L was the most frequent VWD2A mutation and accounted for 28% of patients, followed by G1631D and V1665E (19.5% each), while the majority of VWD2M mutations were R1374H (38% of cases) and R1315C (20%).

Table 2.   Clinical and laboratory characteristics of the 46 patients with VWD2A
MutationNumber of familiesNumber of patientsBleeding score*FVIII:C* (IU dL−1)VWF:Ag * (IU dL−1)VWF:RCo* (IU dL−1)VWF:RCo/VWF:Ag ratio
  1. *Median and range.

S1506L51210 (5–19)48 (21–88)27 (11–102)6 (3–27)0.26 (0.08–0.45)
R1597W136 (5–7)61 (46–87)41 (28–81)6 (6–10)0.07 (0.03–0.36)
V1607D2811 (4–26)54 (38–83)44 (31–106)13 (6–24)0.22 (0.19–0.40)
G1631D198 (5–16)48 (38–109)58 (26–80)6 (6–7)0.10 (0.07–0.23)
V1665E3913 (4–19)45 (22–63)35 (10–47)6 (6–6)0.18 (0.15–0.50)
Table 3.   Clinical and laboratory characteristics of the 61 patients with VWD2M
MutationNumber of familiesNumber of patientsBleeding score*FVIII:C* (IU dL−1)VWF:Ag* (IU dL−1)VWF:RCo* (IU dL−1)VWF:RCo/VWF:Ag ratio*
  1. *Median and range.

L1278P164 (2–13)38 (20–75)28 (15–50)6 (6–10)0.32 (0.12–0.40)
R1315L1310 (9–15)43 (26–51)20 (15–27)7 (6–11)0.47 (0.22–0.55)
R1315C3124 (1–14)49 (28–70)17 (11–45)8 (6–15)0.45 (0.32–0.61)
Y1321C199 (1–23)48 (29–73)19 (11–50)6 (6–6)0.32 (0.12–0.55)
L1361W144 (2–8)57 (41–102)31 (24–64)7 (6–15)0.22 (0.21–0.29)
R1374H6236 (1–18)38 (18–82)21 (10–59)7 (6–26)0.39 (0.15–0.60)
c.3831del-31312 (9–14)40 (36–61)15 (13–18)6 (6–6)0.40 (0.33–0.46)

In all patients with VWD2A HMW VWF multimers were absent in plasma, while VWD2M patients showed the full complement of multimers, in some instances with a relative decrease of the largest forms (data not shown).

Overall outcome

During follow-up, 14/46 (30%) VWD2A subjects and 34/61 (56%) VWD2M subjects (P < 0.001) never contacted our Centers to require consultation or treatment for bleeding or to plan invasive procedures. No deaths occurred during the study period.

Table 4 summarizes the spontaneous bleeding episodes requiring treatment during the follow-up and their annual incidence rates. Among the spontaneous bleeding episodes, gastrointestinal bleeding and menorrhagia were significantly more frequent in VWD2A. There were 61 gastrointestinal bleeding episodes occurring during follow-up: 54 in 17/46 (36.9%) patients with VWD2A (range, 1–7) and seven in 2/61 (3.3%) patients with VWD2M (range, 2–5) (P < 0.0001). Table 5 summarizes the main characteristics of patients with gastrointestinal bleeding. Overall, 63% of these patients were aged ≥65 years. Seven of the 13 patients with S1506L, belonging to six different families, had gastrointestinal bleeding, followed by three patients with G1631D belonging to a single large family. Apart from two patients for whom extensive instrumental diagnosis was not pursued because of very old age, diagnostic procedures revealed the presence of multiple intestinal angiodysplasia.

Table 4.   Incidence rate/100 patient-years (95% CI) of spontaneous bleeding requiring treatment occurring during the follow-up
 All eventsP
VWD2A (n = 46)VWD2M (n = 61)
  1. *Calculated on the number of fertile women.

Epistaxis19.5 (12.3–31.1)16.3 (10.5–25.4)0.31
Menorrhagia*40.9 (25.7–64.9)22.0 (12.1–39.7)0.05
Hematuria2.4 (0.8–7.6)0.08
Gastrointestinal58.7 (44.9–76.6)5.7 (2.7–12.0)<0.0001
Oral4.3 (1.6–11.6)3.2 (1.2–8.7)0.32
Hemorrhoidal2.2 (0.5–8.7)3.3 (1.2–8.7)0.37
Table 5.   Main characteristics at baseline of the 19 patients with gastrointestinal bleeding during follow-up
PatientFamilyAgeSexBleeding scoreVWD typeVWF mutationFVIII:C (IU dL−1)VWF:Ag (IU dL−1)VWF:RCo (IU dL−1)

The incidence rate of spontaneous bleeding requiring treatment was 107/100 patient-years (95% CI, 88.3–131) in VWD2A compared with 40/100 patient-years (95% CI, 30–53) in VWD2M (P < 0.001) (Fig. 1). This risk was significantly higher in patients with BS ≥ 10 at enrollment compared with those with BS 0–2 (Fig. 2). Treatment with plasma-derived VWF/FVIII concentrates was more frequent (88% vs. 38%; P < 0.001) and amount of units infused greater in patients with VWD2A (P = 0.009). The more severe clinical phenotype observed in patients with VWD2A is also emphasized by the number of patients requiring treatment with packed red cells (33% vs. 8%; P = 0.021), although the mean amount transfused per patient was not significantly different. During follow-up, one VWD2A and one VWD2M patient were administered VWF/FVIII concentrate prophylactically twice weekly because of the severity and frequency of gastrointestinal bleeding, which had required 24 U and 8 U of packed red cells, respectively.

Figure 1.

 Cumulative hazard of spontaneous bleeding during the study period calculated according to VWD2A (solid line) and VWD2M (dashed line) status. In all curves, each step represents a bleeding event.

Figure 2.

 Cumulative hazard of spontaneous bleeding during the study period calculated according to the bleeding score at recruitment. In all curves, each step represents a bleeding event.


Qualitative defects of VWF are characterized by a wide heterogeneity of laboratory phenotypes because of the different pathophysiology underlying the disorder [2,3,10]. However, the impact on the clinical outcome of the different defects is not known. The extreme heterogeneity of VWD has hampered gathering and comparing large numbers of phenotypically homogeneous patients in a prospective manner. Only recently, two studies have addressed prospectively the determinants of bleeding risk and the frequency of spontaneous bleeding requiring treatment in two cohorts of patients with well-characterized VWD. The contributory role of thrombocytopenia in worsening the bleeding risk in a large population with VWD2B was established [15], while on the whole the risk of spontaneous bleeding in a population of VWD1 with increased VWF clearance appeared small even after a 7-year follow-up study [16].

In this study, we prospectively followed-up two large populations with VWD2A and VWD2M. All the patients showed a discrepant VWF:RCo/VWF:Ag ratio (≤0.6) as typically observed in VWD2A, which shows the lowest ratio, and VWD2M compared with other VWD types [10]. VWD2A patients were all characterized by the absence of high and sometimes intermediate molecular weight multimers and most of the mutations were located in the A2 domain of VWF, suggesting their classification as ‘classical’ VWD2A [2,3,10]. The population of VWD2M patients included mainly mutations located in the A1 domain. There is still controversy regarding whether or not some of these mutations should be included among VWD2A cases [3,10,11,14], but these cases show a clearly different multimer pattern when compared with typical VWD2A. It should be emphasized that the relative reduction of the largest multimers associated with mutations included among VWD2M is clearly appreciable only by using sophisticated multimer analysis [22] with a lot of variations across different studies [6–8,10,22].

Overall, this study shows that the risk of bleeding was significantly higher in patients with VWD2A, and that this difference was mainly attributable to the high incidence of gastrointestinal bleeding and, to a lesser extent, of menorrhagia. For the estimation of the risk of spontaneous bleeding we adopted a conservative approach based on the need for treatment with desmopressin and/or VWF/FVIII concentrates [23]. The significant higher risk of spontaneous bleeding requiring treatment in patients with VWD2A was not attributable to different baseline FVIII or VWF activities between the two cohorts, the only differences being a greater VWF:Ag and loss of HMW multimers in VWD2A. Most spontaneous bleeding episodes were mucosal, consistent with the views that HMW VWF may be more important than FVIII in determining this type of bleeding [24,25]. In fact, although FVIII:C at baseline was as high as 50 IU dL−1 or more in 18/46 (39%) VWD2A patients but in only 17/61 (28%) VWD2M patients, the incidence of mucosal bleeding was greater in VWD2A patients.

Gastrointestinal bleeding due to angiodysplasia is a major cause of digestive tract bleeding and is commonly observed in elderly people, with incidences ranging from 2.6% to 6.2% in the general population [26,27]. This type of bleeding is also a well-recognized complication of VWD, but it does often occur at a younger age and almost exclusively occurrs in patients lacking HMW multimers [28–32]. In our population, 63% of patients were 65 years old or even older, suggesting that older age is not the only reason for the occurrence of angiodysplasia in VWD [32]. There are several reports of bleeding from gastrointestinal angiodysplasia occurring mainly in VWD2A and VWD2B, both characterized by the lack of HMW VWF multimers in plasma [28–32]. Of interest, in a recent prospective study lasting 7 years no gastrointestinal bleeding from angiodysplasia was observed in a large population of patients with VWD Vicenza, characterized by lower FVIII and VWF in plasma than in VWD2A patients due to increased VWF clearance, but with the presence of ultralarge VWF multimers [16]. The importance of the presence of HMW multimers in preventing gastrointestinal bleeding has been suggested also in acquired conditions, as demonstrated by the occurrence of gastrointestinal bleeding in aortic stenosis [33,34]. In some of these patients HMW multimers are lacking as a result of high shear stress exerted by the stenotic valve, with increased stretching of the VWF molecule and increased susceptibility to ADAMTS-13 [35]. Surgical correction of valve stenosis is accompanied by the restoration of a normal multimeric pattern and the abolishment of bleeding complications [36].

It has been recently demonstrated that VWF regulates angiogenesis and that endothelial cells from VWD patients with reduced VWF levels in their cells show an enhanced in vitro angiogenesis [37], thus supporting a pathophysiological role for low VWF present in endothelial cells in developing angiodysplasia. Among our patients, the VWD2A S1506L mutation appeared to be frequently associated with angiodysplasia. However, the possible role of specific mutations or of a qualitatively abnormal VWF in endothelial cells in producing a higher risk of this enhanced in vitro angiogenesis is not known and further studies are needed in this regard.

Treatment of bleeding from angiodysplasia with VWF/FVIII concentrates is not always effective [32]. Because of recurrent bleeding, one patient with VWD2A and one with VWD2M underwent a prophylactic regimen during follow-up, with only partial reduction of the severity of bleeding. This is not surprising because even the full correction of VWF activity and of multimer pattern in plasma of type 3 patients with replacement therapy does not guarantee a normalization of bleeding time, considered a sensitive parameter for mucosal bleeding control [38,39]. Furthermore, it has been shown that transfused VWF in type 3 VWD patients does not enter endothelial cells, thus probably failing to transiently influence the neo-angiogenic process [40].

An additional important finding of the present study is the demonstration that the severity of bleeding score at enrollment correlated with the risk of bleeding during follow-up. While there is evidence of the utility of BS in designing a clinically useful diagnostic work-up of VWD1 [41,42], little is known about its predictive role in the severity of clinical manifestations in VWD. The BS correlates with the severity of VWF deficiency in VWD1 [17], but in the present study no correlation was found between BS and VWF at enrollment. On the contrary, having a BS ≥ 10 at enrollment, regardless of the type of VWD, was strongly correlated with the risk of bleeding during follow-up compared with patients with BS 0–2 at enrollment. As the analysis was adjusted also for age, this effect was not due to a possible older age of the patients in the BS ≥ 10 subgroup at enrollment, at variance with the finding in women with VWD1 [23]. However, a limitation of these findings is represented by the fact that BS was not collected at diagnosis, because several patients had been diagnosed several years before. Thus, it cannot be excluded that at least in some patients a more severe BS at enrollment was associated with the previous occurrence of gastrointestinal bleeding.

In conclusion, bleeding tendency in VWD2A is greater than that in VWD2M, is not explained by different FVIII or VWF levels and is mainly due to increased incidence of gastrointestinal bleeding. The lack of HMW VWF multimers observed in VWD2A could play an important role in this different bleeding risk.


G. Castaman designed the study, enrolled the patients, analyzed the data results and wrote the manuscript; A. B. Federici designed the study, enrolled patients, analyzed the data results and reviewed the manuscript; A. Tosetto performed statistical analysis, analyzed the data results and reviewed the manuscript; S. La Marca carried out the genetic studies in the patients and F. Stufano performed VWF multimeric analysis and both reviewed the manuscript; P. M. Mannucci and F. Rodeghiero analyzed the data results and reviewed the manuscript.

Disclosure of Conflict of Interests

The authors state that they have no conflict of interest.