Genetic linkage and association analysis in type 1 von Willebrand disease: results from the Canadian Type 1 VWD Study

Authors


David Lillicrap, Department of Pathology and Molecular Medicine, Richardson Laboratory, Queen's University, Kingston, Ontario, Canada K7L 3N6.
Tel.: 613-548-1304; fax: 613-548-1356; e-mail: lillicrap@cliff.path.queensu.ca

Abstract

Summary. Background: von Willebrand disease (VWD) is the most common bleeding disorder known in humans, with type 1 VWD representing the majority of cases. Unlike the other variant forms of VWD, type 1 disease represents a complex genetic trait, influenced by both genetic and environmental factors. Aim: To evaluate the contribution of the von Willebrand factor (VWF) and ABO blood group loci to the type 1 VWD phenotype, and to assess the potential for locus heterogeneity in this condition, we have performed genetic linkage and association studies on a large, unselected type 1 VWD population. Method: We initially collected samples from 194 Canadian type 1 VWD families for analysis. After the exclusion of families found to have either type 2 or type 3 VWD, and pedigrees with samples from single generations, linkage and association analysis was performed on 155 type 1 VWD families. Results and conclusion: The linkage study has shown a low heterogeneity LOD score of 2.13 with the proportion of families linked to the VWF gene estimated to be 0.41. Linkage was not detected to the ABO locus in this type 1 VWD population. In the family-based association test, significant association was found between the type 1 VWD phenotype, the quantitative traits, VWF:Ag, VWF:RCo, and FVIII:C and the ABO ‘O’ and ‘A’ alleles and the VWF codon 1584 variant. There was also weak association with the −1185 promoter polymorphism and VWF:Ag, VWF:RCo, and FVIII:C plasma levels. These studies provide further evidence to support the role for genetic loci other than VWF and ABO in the pathogenesis of type 1 VWD.

Introduction

von Willebrand disease (VWD), the most common inherited bleeding disorder in humans, is the result of qualitative or quantitative abnormalities of von Willebrand factor, a large adhesive glycoprotein that plays two key roles in primary hemostasis. The protein is responsible for platelet adhesion to sites of exposed subendothelium and it acts as a protective carrier in the plasma for factor VIII [1]. The gene that encodes von Willebrand factor (VWF) is located on chromosome 12p13.3, and was cloned and characterized by a number of groups simultaneously in 1984 [2–5]. Although the 1994 Revised ISTH Classification of VWD [6] states that ‘All VWD is caused by mutations at the VWF locus’ there has been a growing appreciation that additional genetic loci may play important roles in regulating VWF biosynthesis and ultimately determining the plasma level of the protein [7–9]. Twin, and other family-based studies, in normal and ‘hypercoagulable’ pedigrees, have shown that the heritability of VWF plasma levels (the proportion of variation that can be attributed to additive genetic factors) ranges between 32% and 75% [10–14]. There are additional studies that show the influence of single nucleotide polymorphisms (SNP) in the VWF gene promoter region on plasma VWF levels [15,16].

The diagnosis of VWD requires three components: a personal history of excessive mucocutaneous bleeding, laboratory hemostasis results that are consistent with the diagnosis and a family history of excessive bleeding. While each of these three components may prove difficult to establish [17], the demonstration of a positive family history is often particularly difficult to determine. Factors that contribute to this dilemma include the lack of challenges to the hemostatic system in young children and males, and the genetic phenomena of incomplete penetrance and variable expressivity of mutant VWF alleles [18,19]. In combination, these factors have significantly complicated attempts to characterize the genetic basis of the most common form of VWD, type 1 disease, in which mild to moderate reductions in qualitatively normal VWF are found. The contribution of loci other than the VWF gene has clearly been shown to be important, with the ABO locus contributing about 30–40% of the genetic variability of the plasma VWF level [20]. The recently published GAIT project [21] employed genome-wide linkage analysis to show the influence of regions of chromosomes 1, 2, 5, 6 and 22, in addition to the ABO locus (chromosome 9) on VWF levels in several large Spanish families recruited with idiopathic thrombophilia. Ethnic influences on VWF levels have also been identified [22] and a variety of environmental factors such as age, stress, thyroid hormones, pregnancy or inflammatory states play additional roles in determining these levels [23].

This report describes genetic linkage and association studies of polymorphic markers within the VWF gene and at the ABO blood group locus in a large population of families diagnosed with type 1 VWD. The study was undertaken as a component of a larger research program to determine the molecular genetic basis of this condition, with the objective of evaluating the contribution of locus heterogeneity in type 1 VWD.

Patients, materials and methods

Ascertainment of study population

The original study population comprised 194 families submitted to The Canadian Type 1 VWD Study aimed at determining the molecular genetic basis of type 1 VWD. These families had been ascertained both retrospectively and prospectively at 13 tertiary care academic health care centers across Canada by hematologists directing inherited bleeding disorder clinics.

The definition of type 1 VWD that has been used in this study involves identifying an individual (an index case) with a personal history of excessive mucocutaneous bleeding and plasma levels of VWF:Ag and VWF:RCo between 0.05 and 0.50 U mL−1 obtained on at least two occasions. The absence of a family history for VWD was not an exclusion criterion for the study in an attempt not to bias our study results toward showing linkage. Type 2 VWD patients were identified and excluded based on a VWF:RCo/VWF:Ag ratio of <0.6 and/or abnormal multimer results. Type 3 VWD patients were identified and excluded based on VWF:Ag <0.05 IU mL−1 in the index case. Bleeding histories were obtained by recording the presence or absence of six cardinal bleeding symptoms for VWD: epistaxis, easy bruising, menorrhagia (women >12 years), postoperative bleeding, postdental procedure bleeding, excessive wound bleeding.

All participants in the study were informed of the experimental nature of the study and gave informed consent. The study was approved by the Research Ethics Board of Queen's University and at each of the source institutions. Whole blood samples were collected by phlebotomy in 3.2% sodium citrate (at a ratio of 9:1 v/v) and ethylenediaminetetraacetic acid (EDTA) from the index case and available family members. Platelet poor plasma samples were also obtained from all available family members and repeat VWD phenotypic studies were performed in a central laboratory at the Kingston General Hospital to confirm the diagnosis of type 1 VWD.

Coagulation studies

Laboratory tests for VWF:Ag, VWF:RCo, and FVIII:C were performed at the source clinic attended by the patient according to local methods. All platelet poor plasmas were prepared by the locally utilized protocols. These tests were repeated on frozen platelet poor plasma samples at the Clinical Hemostasis Laboratory at the Kingston General Hospital and all available laboratory results (from the referring clinic and the central laboratory) were averaged to obtain the ‘reference’ hemostatic phenotype used in the study database. VWF:Ag was measured by the IMUBIND VWF enzyme-linked immunosorbent assay (ELISA) kit according to the procedure supplied by the manufacturer (American Diagnostica, Greenwich, CT, USA). The VWF:RCo was measured by platelet aggregometry using freshly prepared, washed normal platelets, and FVIII:C was measured using a one-stage assay. In the Kingston laboratory, all measurements of VWF:Ag, VWF:RCo, and FVIII:C were made against non-ABO-matched commercial reference plasma that had been calibrated against the 91-666 or 97-586 World Health Organization (WHO) Plasma Standard. Normal reference ranges were derived independently by each participating center. VWF multimers were analyzed by electrophoresis using a 1.6% sodium dodecyl sulfate (SDS) agarose gel followed by electrotransfer to a nylon membrane, and the multimers were visualized using the chemiluminescent visualization kit from Amersham Pharmacia Biotech (Baie D'Urfe, Quebec, Canada) [24]. A ratio of <0.6 for VWF:RCo to VWF:Ag values was used to differentiate between type 1 and type 2 VWD variants.

PCR amplification of genomic DNA and genotype analysis

VWF polymorphism genotyping  An EDTA-anticoagulated whole blood sample was obtained from all enrolled subjects and genomic DNA was isolated from leukocytes using a salt extraction method [25]. DNA corresponding to the VWF promoter and the coding regions of interest were amplified by polymerase chain reaction (PCR). Primer sequences for these PCR studies have been previously published [26]. Using a DNA thermal cycler (Perkin-Elmer Life Sciences, Shelton, CT, USA), DNA was amplified for 35 cycles of 45–60 s at 94 °C, 45–60 s at 53–60 °C, and 45–60 s at 72 °C. The following VWF SNPs and the number of simple sequence repeats found at STR2 within intron 40 were genotyped ( Table 1): promoter nucleotide (nt) −1185 (G+/A−, BstUI) and −1051 (A+/G−, NlaIII), exon 8 aa 318 (T+/A−, MslI), exon 12 aa 471 (G+/A−, AatII), exon 13 aa 484 (G+/A−, RsaI), exon 18 aa 789 (A+/G−, RsaI), exon 28 aa 1547 (C+/T−, BstEII), exon 28 aa 1584 (A+/G−, KpnI), intron 30 nt 18 (A−/C+, Bsp1286I), exon 42 aa 2413 (C+/T−, AciI) and the STR2 simple sequence repeat in intron 40 [27,28].

Table 1.  von Willebrand factor gene polymorphisms analyzed
Exon/ intron (No.)Codon/intron (nucleotide)NucleotidesGenomic position (bp from 12-pter)DbSNP (rs No.)
Promoter−1185G/A  
Promoter−1051A/G  
Exon 8318T/A6053089rs1800387
Exon 12471G/A6043694rs1800377
Exon 13484G/A6042463rs1800378
Exon 18789A/G6023795rs1063856
Exon 281547C/T5998204rs216310
Exon 281584A/G5998094rs1800386
Intron 30−18A/C5994752rs2363334
Exon 422413C/T5961261rs216867
Intron 40 SSR  

ABO genotyping  ABO blood group nt 261 (rs8176719) and nt 703 (rs8176743) SNPs were genotyped using the 5′-nuclease assay to distinguish classical ABO blood groups [29,30]. Two allele-specific probes were designed for each SNP, using primer express 1.5 ( ABI, Foster City, CA, USA) following the manufacturer's guidelines. One probe matched the wild-type sequence and the other matched the mutant sequence. Each probe was labeled at the 5′-end with a fluorescent reporter dye and at the 3′-end with a non-fluorescent quencher with a minor groove binder also present at the 3′-end. The reporter dyes used in this case were 6-carboxyflourescein for the wild type and VIC (ABI) for the mutant. The primer and probe sequences for the rs81767119 and rs8176743 SNPs are detailed in Table 2. ABO genotypes were then generated from the results of these SNP analyses.

Table 2.  ABO blood group SNPs
ABO SNPsPCR primers and TaqMan probes
  1. SNP, single nucleotide polymorphism.

rs8176719Primer F: 5′-ATG TGA CCG CAC GCC TCT C-3′
Primer R: 5′-TCT CTA CCC TCG GCC ACC TC-3′
Probe 1: TCC TCG TGG TGA CCC
Probe 2: TCC TCG TGG T A CCC CT
rs8176743Primer F: 5′-TGG AGA TCC TGA CTC CGC TGT-3′
Primer R: 5′-GTA GAA ATC GCC CTC GTC CTT G-3′
Probe 1: ACC CCG GCT TCT A
Probe 2: ACC CCA GCT TCT ACG

Genetic linkage analysis

Pairwise and multipoint linkage analysis was performed. Parametric linkage analysis for VWF markers was undertaken assuming an autosomal dominant model of inheritance, with penetrance = 0.6, phenocopy rate = 0.003, and a disease allele frequency = 0.001. For ABO parametric linkage analysis, an autosomal recessive model was assumed, with disease allele frequency 0.45, phenocopy rate = 0.0001, and penetrance = 0.1. Both homogeneity and heterogeneity LOD (HLOD) scores were calculated. In addition, a non-parametric statistic, which is based on allele sharing by affected individuals, was calculated (NPL Z-score and P-value). The information content at each polymorphic site provides a measure of the extraction of meiotic information [31]. All linkage analysis was performed using genehunter v2.1_r5 [31,32]. Marker allele frequencies were estimated from founders. Linkage analysis was performed both including and excluding families in which the variant codon 1584 cysteine allele was documented.

Family-based association analysis

Family-based association analysis was performed using fbat v 1.5.5 [33] to test for association between specific alleles of markers and traits. The type 1 VWD phenotype was evaluated as a discrete trait, and VWF:Ag, VWF:RCo, and FVIII:C as quantitative traits. An additive genetic model was used throughout as it is generally most powerful. Analysis was performed both including and excluding families in which the variant 1584 cysteine allele was documented. To account for multiple affected individuals in the same family, the empirical variance option was used for the type 1 VWD phenotype.

Derivation of power estimates

For association analysis we have used pbat v 2.2 to perform power calculations [34]. These indicate that we have good power (1−β = 0.8) to detect an association with the discrete outcome with α = 0.01 when the locus accounts for 10% of the attributable fraction and the disease allele frequency is <2%. To determine the power to detect association with a quantitative trait, we used similar parameters, and there is good power to detect association with a locus which contributes to 10% of the trait heritability for disease alleles with a frequency of >5%.

Parametric linkage simulations for the disease trait performed using slink [35] show that under the assumption of locus homogeneity, the maximum LOD score for all pedigrees is 10.2. Power analysis for non-parametric linkage (NPL) analysis, performed using merlin, estimates a maximum NPL Z-score of 10.1.

Results

Details of the study population

A total of 194 families were submitted to The Canadian Type 1 VWD Study. These families were referred from 13 tertiary care, academic health centers across Canada. The number of families referred from each center ranged from 1 to 66.

Twenty-four of the families from the original study population in which samples were only obtained from individuals within a single generation were excluded from the subsequent analysis.

Upon re-evaluation of the hemostasis studies derived from both the referral clinic and the central laboratory, 12 families were reclassified as type 2 VWD and three families as type 3/type 1 because of the presence of individuals with both type 1 and type 3 phenotypes within the same family. These 15 families were excluded from the analysis presented in this report. Therefore, 155 type 1 VWD families were included in these linkage and association studies.

The 155 families described in this study comprise 651 individuals. Two hundred and ninety-one individuals are affected with type 1 VWD (presence of both mucocutaneous bleeding symptoms and consistent laboratory values), 168 are unaffected (no reported bleeding symptoms and/or normal laboratory values), and the phenotypic status of 192 individuals is unknown. The mean age of the index cases is 22 years (range: 1–60). Epistaxis was reported by 61% of the index cases, easy bruising by 46%, menorrhagia by 36% (women >12 years), postoperative bleeding by 34%, postdental procedure bleeding by 24%, and excessive bleeding from wounds by 21%. In 35% of the index cases, only one of the six bleeding symptoms was reported.

The average family size in this study is 4.2 individuals (range: 3–11). One hundred and thirty-seven of the families have two generations, 16 have three generations, and two have four generations. The numbers of type 1 VWD cases in each study family is summarized in Table 3.

Table 3.  Type 1 VWD family structures
Number of familiesNumber of type 1 VWD patients in family
641
622
213
34
35
16
17
Total 155 

The mean VWF:Ag level of the index cases in these 155 families is 0.40 IU mL−1 (range 0.03–0.88), the mean VWF:RCo level is 0.35 IU mL−1 (range: 0.05–0.74) and the mean FVIII:C level is 0.59 IU mL−1 (range: 0.09–1.34; Fig. 1A–D; for study inclusion, at least two values for VWF:Ag and VWF:RCo must have been <0.50 IU mL−1). Multimer analysis in the 155 index cases showed no obvious loss of the high-molecular weight VWF multimers and no abnormality of the multimer triplet structure. Furthermore, index cases where the mean VWF:RCo to VWF:Ag ratio was <0.6 were reclassified as type 2 VWD and excluded from this analysis.

Figure 1.

Mean von Willebrand factor (VWF) and factor VIII levels for affected and unaffected study subjects. (A) VWF:Ag, (B) VWF:RCo, (C) factor VIII:C, and (D) VWF:RCo to VWF:Ag ratios.

Genetic studies

Genotyping was performed on the study population with a panel of 10 biallelic polymorphisms and one multiallelic VWF simple sequence repeat variant that spanned the gene from nt −1185 in the promoter sequence to codon 2413 in exon 42. The population was also genotyped for ABO blood group alleles.

Linkage analysis of type 1 VWD  Pairwise and multipoint linkage studies were performed to evaluate co-segregation of VWF and ABO alleles and type 1 VWD within pedigrees.

Ninety-two of the initial 155 study families were informative for parametric linkage analysis with the VWF markers and 35 families were informative for the NPL study. With the codon 1584 cysteine families removed ( Table 4B), 76 families were informative for parametric linkage and 31 for NPL. For the ABO linkage studies, 92 families were informative for parametric analysis and 29 for the NPL analysis. After exclusion of codon 1584 cysteine families, there were 76 informative kindred for parametric linkage and 26 for non-parametric analysis.

Table 4.  (A) Genetic linkage analysis of all families; (B) genetic linkage analysis of families without aa 1584 variant
MarkerHomogeneity LOD scoreHeterogeneity LOD scoreαNPL Z-scoreP-valueInformation content
  1. VWF, von Willebrand factor.

(A)
ABO−0.910.000.00−0.470.760.28
−1185−3.740.310.270.220.370.26
−1051−3.810.290.260.190.380.26
aa 3180.510.511.000.130.420.05
aa 4710.210.420.630.110.430.13
aa 4840.991.720.750.890.0890.25
aa 789−4.680.000.000.120.430.27
aa 1547−4.260.230.290.380.280.30
aa 15841.121.131.000.340.300.08
Intron 300.330.341.000.200.380.04
Intron 40−5.981.010.371.450.0150.54
aa 2413−3.530.020.090.250.340.16
VWF multipoint−7.572.130.411.830.00310.87
(B)
ABO−0.810.000.00−0.470.750.29
−1185−2.950.500.360.340.310.25
−1051−3.030.470.350.320.320.25
aa 3180.350.351.000.140.410.05
aa 4710.200.420.630.110.430.13
aa 4840.421.250.700.690.160.25
aa 789−4.360.000.000.080.450.28
aa 1547−4.080.240.310.400.280.31
Intron 300.020.021.000.020.490.00
Intron 40−6.190.830.351.480.020.56
aa 2413−3.310.030.130.280.340.14
VWF multipoint−7.661.660.381.770.00560.87

In this study, there was no evidence for linkage to the ABO locus either in all families or in those after removal of the codon 1584 cysteine-positive families (Table 4A,B). Multipoint HLOD scores in all families (HLOD = 2.13, α = 0.41) are consistent with linkage to the VWF gene, as is the NPL analysis (Z-score = 1.83, P = 0.0031). The majority of linkage information has been extracted from the families, with an information content of 0.87. Even after removal of the ‘1584 cysteine’ families (Table 4B), there was still evidence for linkage (HLOD = 1.66, α = 0.38), NPL Z-score = 1.77, P = 0.0056, with an information content of 0.87. Aside from the fact that these studies confirm linkage of the type 1 VWD phenotype to the VWF locus in both parametric and non-parametric analysis of this population, the other result of note is that linkage to VWF was only present in 41% of families under the genetic model used for these studies.

We have also estimated heritability for the traits (VWF:Ag; VWF:RCo; FVIII:C) in the families, and performed variance components quantitative trait linkage analysis of the traits (using merlin v 0.10.2) [36] with the markers at the VWF and ABO loci. We find that in these families, the traits are moderately heritable (additive h2 = 12–20%), but do not find significant linkage to either VWF or ABO, with all LOD scores <0.7.

Family-based association analysis of discrete and quantitative traits  Family-based association testing was performed to evaluate the association between the discrete trait, type 1 VWD, the quantitative traits VWF:Ag, VWF:RCo, and FVIII:C and marker alleles at the ABO and VWF loci ( Table 5A,B). In all families, there is significant excess transmission of the ABO ‘O’ allele to affected individuals (P = 0.001), and an association of this allele with VWF:Ag (P = 0.0016), VWF:RCo (P = 0.0023), and FVIII:C (P = 0.05). As expected, there was also significantly reduced transmission of ABO ‘A’ alleles in type 1 VWD. At the VWF locus, there was excess transmission of the mutant codon 1584 cysteine allele in type 1 disease (P = 0.003), as well as weak excess transmission of allele ‘9’ of the intron 40 simple sequence repeat (P = 0.042; Table 5A). The codon 1584 cysteine variant revealed weak association with VWF:Ag (P = 0.021) and VWF:RCo (P = 0.025) and borderline association with FVIII:C (P = 0.076).

Table 5.  (A) Family-based association analysis of discrete outcomes for all families (n = 155 families); (B) family-based association analysis of discrete and continuous outcomes without aa 1584 variant (n = 137 families)
MarkersAlleleAllele frequencyType 1 VWDTraitsVWF:AgVWF:RCoFVIII:C
Family (No.)Z-scoreP-valueFamily (No.)Z-scoreP-valueZ-scoreP-valueZ-scoreP-value
  1. Additive genetic model, empirical variance model.

  2. Markers with less then 10 informative meioses were not analyzed (i.e. aa 318, intron 30). Z is a standard score. A positive score indicates excess transmission of that allele to affected individuals, a negative score indicates paucity of transmission of that allele to affected individuals.

  3. VWF, von Willebrand factor.

(A) All families
ABO‘O’0.68563.280.0010643.160.00163.050.00231.960.050
‘A’0.2748−3.020.002557−2.750.0060−2.750.0059−1.850.063
‘B’0.0416−1.230.2218−1.430.131.150.250.50.61
−1185G nt0.6356−0.310.7560−1.230.221.220.22−1.590.11
−1051A nt0.6356−0.310.7560−1.040.30−0.840.40−1.230.22
aa 471G nt0.86300.600.55300.910.360.720.471.060.29
aa 484G nt0.63460.700.49530.140.890.060.960.710.48
aa 789A nt0.67510.700.48570.690.490.870.390.440.66
aa 1547C nt0.60541.100.27611.030.300.570.570.980.33
aa 1584G nt0.04182.940.0033142.310.0212.240.0251.780.076
Intron 4090.14372.040.042401.520.130.980.331.170.24
100.0523−1.600.1121−0.450.65−0.380.71−0.860.39
110.3874−1.040.30790.140.89−0.070.940.240.81
120.29660.730.4673−0.380.70−0.360.72−0.260.79
130.0517−1.060.2917−1.710.09−1.650.10−1.290.20
140.08340.730.46310.440.661.080.280.650.52
aa 2413C nt0.8632−1.000.3235−1.180.24−0.990.320.520.60
(B) Exclude aa 1584
ABO‘O’0.68462.620.0087552.510.0122.440.0151.410.16
‘A’0.2740−2.470.01450−2.210.027−2.280.023−1.350.17
‘B’0.0414−0.890.3816−1.070.29−0.780.43−0.300.76
−1185G nt0.6341−1.320.1946−2.020.043−2.000.046−2.200.028
−1051A nt0.6341−1.320.1946−1.800.072−1.580.11−1.800.071
aa 471G nt0.86250.190.84270.420.680.240.810.740.46
aa 484G nt0.63400.770.44470.370.710.280.780.990.32
aa 789A nt0.6739−0.550.5848−0.150.88−0.010.99−0.030.98
aa 1547C nt0.60460.630.53540.750.450.270.790.840.40
Intron 4090.14241.070.29260.790.430.460.640.910.36
100.0520−1.490.1418−0.260.79−0.330.74−0.760.45
110.38590.170.86661.180.240.760.450.840.40
120.29550.340.7464−0.640.52−0.560.57−0.560.57
130.0516−1.260.2116−1.940.053−1.860.063−1.440.15
140.08300.480.63280.230.820.910.360.660.51
aa 2413C nt0.88220.060.95240.430.660.480.63−0.260.80

After removal of families with the codon 1584 cysteine variant, the positive and negative associations between the ABO ‘O’ and ‘A’ alleles, respectively, with type 1 VWD, remained (Table 5B). There is also weak association of these two alleles with VWF:Ag and VWF:RCo, but not with FVIII:C. After removal of the codon 1584 variant analysis, there is no significant association of type 1 VWD with any of the VWF markers. This negative finding is consistent with allele ‘9’ of the intron 40 simple sequence repeat being in linkage disequilibrium with the codon 1584 sequence. Regarding association of the quantitative traits with the VWF markers in the codon 1584 cysteine-negative population (Table 5B), there is also weak association of the –1185 promoter polymorphism with VWF:Ag, VWF:RCo, and FVIII:C (P = 0.043, 0.046, and 0.028, respectively). However, given the number of statistical comparisons performed, these latter associations could either be true- or false-positives and the analysis of additional families would be required to clarify this.

Using a recessive model for the association analysis with ABO as shown in Table 5A revealed significant excess transmission of the ‘O’ allele to affected individuals (Z = 3.52, P = 0.00044), but no significant paucity of transmission of the ‘A’ allele (Z = −1.41, P = 0.16). There were no transmissions of the ‘B’ allele.

Discussion

Knowledge of the genetic basis for the most common inherited bleeding disease in humans, type 1 VWD, has been slow to be derived. While there is a significant body of evidence to indicate that types 2A and 2B VWD are almost always autosomal dominant traits because of mutations within the VWF gene [37–41] and that type 2N [39,42] and type 3 [43,44] VWD variants are recessive VWF gene traits, there is little information concerning the location and types of mutation that result in the quantitative type 1 phenotype. Reports of both dominant [45,46] and recessive [47,48] inheritance patterns have been published, and in at least one mouse model of type 1 disease, the causative genetic locus is not the VWF gene [7,9]. Nevertheless, there is a growing appreciation that this condition represents a complex genetic trait [49,50] and there has also been recent renewed debate about the definition of the diagnosis [51,52]. This large population-based study has employed genetic linkage and association studies to evaluate the role of the VWF and ABO gene loci in type 1 disease and also to estimate the contribution of locus heterogeneity to this phenotype. Our results lead to three main conclusions: first, that the type 1 VWD phenotype is linked to the VWF gene locus, but that linkage is present in a minority of type 1 VWD families (41% in this study). This result implies that there is a significant role for other genetic loci or even non-genetic factors in the pathogenesis of this condition. As an example, variability of platelet receptor expression has been documented to act as a modifier of the bleeding phenotype in type 1 VWD [53,54]. Secondly, the association studies performed in this project have confirmed a significant role for the ABO blood group locus in the pathogenesis of this disorder and finally, this study has confirmed, through both linkage and association analysis, the important role of the Y1584C variant in the development of type 1 VWD.

Type 1 disease accounts for approximately 80% of the cases of VWD encountered in clinical practice. The precise population incidence of the condition is unknown. Two prospective, pediatric studies have documented a VWD incidence of approximately 1% but these children had not presented to physicians with clinically significant problems [55,56]. The most challenging issue with regard to many of the biologic and genetic questions that remain unanswered in type 1 VWD, concerns the difficulty in determining the diagnosis [51]. The excessive mucocutaneous bleeding manifestations are often overlooked or underestimated, and the standard laboratory measurements of VWF and FVIII that form the central component of the diagnostic criteria vary over time, and are significantly influenced by environmental factors such as stress and estrogenic hormones [17]. There are also well-documented effects of the ABO blood group and ethnicity [20,22]. In addition, there is long-standing evidence that the phenotype exhibits incomplete penetrance and variable expressivity [18,19], thus further complicating individual and family ascertainment of the disease. How much of this variability can be explained by inadequate definition of the disease phenotype is yet another confounding factor.

In this large population-based study, the initial diagnosis of type 1 VWD has been made at tertiary care academic centers by hematologists directing inherited bleeding disorder clinics. These clinicians have evaluated the mucocutaneous bleeding manifestations of the study subjects, and their laboratory hemostatic phenotypes have been derived from a combination of studies from the referring center and a central study laboratory. The family structures and range of hemostatic values present in this study population suggest that this is a collection of families that is representative of the spectrum of type 1 VWD cases that is likely to be seen in regular, tertiary care clinical practice. We have not sought out large multigenerational families with many affected subjects, at least in part, because these kindred may well possess highly penetrant disease variants that only represent a minor proportion of type 1 VWD alleles [45,46]. Nevertheless, there are also disadvantages to the type of study population that we have assembled, one of the more important being the reduced statistical power to demonstrate genetic linkage. These differences in population ascertainment are likely to be of critical significance in comparing results obtained in previous type 1 VWD linkage studies [18] and in the recently completed European Union Type 1 VWD Study [57].

The appropriate genetic analysis of complex traits, such as type 1 VWD, continues to be a subject of debate [58–60]. In this study, we have used both linkage and association analysis to optimize our ability to evaluate the contribution of the VWF and ABO loci on the type 1 phenotype. The results of our multipoint linkage analysis has documented that the disease co-segregates with markers at the VWF locus in 41% of families. Further to this result, in eight of the pedigrees, negative LOD scores between −1.3 and −1.59 were documented. These results extend previous evidence to suggest that the type 1 VWD phenotype does not co-segregate consistently with markers at the VWF locus [61] and also supports prior documentation of relatively minor effects on plasma VWF levels being attributable to the VWF gene [21,62]. In the pair-wise linkage analysis, weakly positive results were obtained with three loci, the codon 1584 variant (LOD: 1.13), codon 484 (LOD: 1.72), and the intron 40 repeat (LOD: 1.01). As these three markers are tightly linked, then depending upon their information content, and specifically which families they are informative in, we would expect them to show some evidence for linkage.

Interestingly, in this study, we did not find linkage of the type 1 phenotype to the ABO locus. Given the recognized influence of the ABO blood group on VWF plasma levels [20], previous documentation of linkage of this trait to the ABO locus in the GAIT study (LOD = 3.46, P = 0.00003) [21], and recent evidence for a significant contribution of blood group ‘O’ to the type 1 VWD phenotype [63], this result was unexpected. Nevertheless, both our parametric linkage analysis (using an autosomal recessive inheritance model) and the model-independent non-parametric analysis were both consistent in their lack of detection of linkage. We propose that at least two factors have contributed to the lack of linkage observed in this population: first, that the relatively small family structures assembled in this study may have compromised the power of the linkage analysis to detect this locus. Secondly, the influence of the ABO locus in this population may be relatively minor in nature. This latter hypothesis seems unlikely given the significant association that we have demonstrated between the ABO locus alleles and both the discrete type 1 VWD trait and the quantitative traits VWF:Ag, VWD:RCo, and FVIII:C.

Interestingly, in the association studies excluding the codon 1584 ‘cysteine’ pedigrees, the promoter polymorphism at nt −1185 showed a weak association with the quantitative traits, VWF:Ag (P = 0.043), VWF:RCo (P = 0.046), and FVIII:C (P = 0.028). These findings provide further support for previous observations that we have made concerning the role of variability in the VWF 5′-upstream sequence in regulating plasma levels of VWF [15,16].

When this study was initiated, one of the polymorphic markers that we choose to examine was the Y1584C coding sequence variant in exon 28. Our subsequent detailed examination of this sequence variation now indicates that this change is pathogenic in nature [26]. Fourteen percentage of our original study population carries this mutation, and the same variant has subsequently been found as a recurrent mutation in other type 1 populations [64]. In this study, the critical nature of this variant is re-emphasized by the results of our linkage and association studies, both of which show a significant role for the 1584 cysteine residue in the development of type 1 disease.

In conclusion, this study has documented, for the first time, in a large, relatively unselected type 1 VWD population, evidence for locus heterogeneity in a substantial number of families. While it is impossible to place formal confidence limits on this data, it appears very likely, based on this study that a significant proportion of type 1 VWD cases will not be linked to the VWF gene. The comparison of these results with a similar, European population-based type 1 VWD study [54] will be particularly informative. These results provide further impetus for the continued search for additional genetic loci that contribute to the development of the type 1 VWD phenotype.

Addendum

Role of coauthors: D.L. designed and supervised the study and wrote the manuscript; P.D.J. supervised the study, performed experiments, interpreted data, and helped write the manuscript; A.D.P. performed the statistical analysis and helped write the manuscript; C.N., C.C., C.H., S.T., C.B., L.O'B., J.L. performed and analyzed the experiments; members of the AHCDC (see below) contributed cases for analysis.

Members of the Association of Hemophilia Clinic Directors of Canada

Kaiser Ali, Division of Pediatric Hematology/Oncology, University of Saskatchewan; Dorothy Bernard, Division of Pediatric Hematology, Dalhousie University; Victor Blanchette, Division of Pediatric Hematology/Oncology, University of Toronto; Mason Bond, Division of Pediatric Hematology/Oncology, University of British Columbia; Josee Brossard, Department of Pediatrics, University of Sherbrooke; Manuel Carcao, Division of Pediatric Hematology/Oncology, University of Toronto; Robert Card, Division of Hematology, University of Saskatchewan; Anthony Chan, Department of Pediatrics, McMaster University; Jeffrey Davis, Division of Hematology/Oncology, University of British Columbia; Michael Delorme, Cancer Center for the Southern Interior, Kelowna, BC; Christine Demers, Departement d'Hematologie, Hopital du Saint Sacrement, Quebec; Sean Dolan, Department of Hematology, Saint John; Nancy Dower, Department of Pediatrics, University of Alberta; Bernadette Garvey, Department of Medicine, University of Toronto; Kulwant Gill, Laurentian Hospital, Sudbury; Jack Hand, Department of Pediatrics, Memorial University; Rosemary Henderson, Department of Laboratory Medicine, Queen Elizabeth Hospital, Charlottetown; Donald Houston, Department of Oncology/Hematology, University of Manitoba; Sara Israels, Division of Pediatric Hematology, University of Manitoba; Lawrence Jardine, Department of Pediatrics, University of Western Ontario; Mariette Lepine-Martin, Departement d'hemato-oncologie, CHUS, Sherbrooke; Reinhard Lohmann, Department of Medicine, University of Western Ontario; Koon-Hung Luke, Department of Hematology/Oncology, Children's Hospital of Eastern Ontario; Patricia McCusker, Division of Pediatric Hematology/Oncology, University of Manitoba; Mohan Pai, Pediatric Hematology/Oncology, McMaster University; Man-Chiu Poon, Department of Medicine, University of Calgary; Bruce Ritchie, Department of Medicine, University of Alberta; Georges Rivard, Hopital Ste-Justine, Montreal; Sue Robinson, Department of Medicine, Dalhousie University; Sheldon Rubin, Moncton Hospital, Moncton; Morel Rubinger, Department of Hematology/Oncology, University of Manitoba; Mary Frances Scully, Department of Medicine, Memorial University; Mariana Silver, Division of Hematology/Oncology, Queen's University; Jean St-Louis, Hematologie-Oncologie, Hopital du Sacre-Coeur de Montreal; Jerome Teitel, Department of Medicine, University of Toronto; Alan Tinmouth, Clinical Epidemiology Research Unit, University of Ottawa; Dimitrios Vergidis, Regional Cancer Care HSC, Thunder Bay; Linda Vickars, Department of Medicine, University of British Columbia; Irwin Walker, Department of Medicine, McMaster University; Margaret Warner, Division of Hematology, McGill University; Philip Wells, Ottawa Hospital, University of Ottawa: Blair Whittemore, Division of Hematology, McGill University; Rochelle Winikoff, Ste-Justine Hopital, Montreal; John Wu, Division of Pediatric Hematology/Oncology, University of Calgary; John Wu, Division of Pediatric Hematology/Oncology, University of British Columbia. Margaret Yhap, Department of Pediatrics, Dalhousie University.

Acknowledgements

This project was funded by a Canadian Institutes for Health Research (CIHR) Operating Grant (MOP-42467). The authors acknowledge the contributions of Dr Dilys Rapson and Mr Kerry Benford. A.D.P. holds a Canada Research Chair in the Genetics of Complex Diseases. D.L. holds a Canada Research Chair in Molecular Hemostasis and is a Career Investigator of the Heart and Stroke Foundation of Ontario.

Ancillary