Laboratory Variability in the Diagnosis of Type 2 VWD Variants

Abstract

Introduction:

Type 2 von Willebrand disease (VWD) refers to patients with a qualitative defect in von Willebrand factor (VWF). Accurate diagnosis of type 2 VWD subtypes can be challenging.

Aim of the study:

to compare the historical diagnosis of type 2 VWD with current laboratory testing.

Methods:

Subjects were enrolled in the Zimmerman Program either due to a pre-existing diagnosis of VWD (retrospective cohort) or due to evaluation for bleeding symptoms or suspected VWD (prospective cohort). Original diagnosis was assigned by the local center and central diagnosis was based on central laboratory testing.

Results:

217 index cases in the retrospective cohort and 35 subjects in the prospective cohort carried a local diagnosis of type 2 VWD (29% and 6% of enrolled index cases respectively). In the retrospective cohort, the diagnosis was confirmed in 66% of cases with a pre-existing diagnosis of 2A, 77% 2B, 54% 2M, and 72% 2N. In the prospective cohort, 31% were confirmed 2A, 60% 2B, 23% 2M, and 100% 2N. Several genetic variants were repeatedly implicated in subjects with changed diagnosis: p.M1304R, p.R1315C, p.R1374C, and p.R1374H.

Conclusions:

Both the prospective and retrospective cohorts demonstrated consistent variation in subjects whose diagnosis changed between 2A, 2B, and 2M. The importance of accurately diagnosing type 2 VWD may be most significant in the 2B subtype given potential concerns with the use of desmopressin in type 2B VWD. Some genetic variants appear in multiple types of VWD, making specific diagnoses challenging.

Introduction

Von Willebrand disease is a common inherited bleeding disorder, with the prevalence of symptomatic VWD estimated at 1 in 1000 persons [1]. Von Willebrand factor (VWF) mediates adhesion of platelets to sites of endothelial injury and binds Factor VIII (FVIII). VWD is caused by a quantitative or qualitative defect in this protein [2]. Type 2 VWD describes qualitative VWF defects and can be further classified as type 2A VWD for decreased VWF activity with loss of high molecular weight (HMW) multimers, type 2B VWD for increased platelet binding and subsequent loss of HMW multimers, type 2M for decreased VWF activity with maintenance of HMW multimers, and type 2N VWD for decreased binding to FVIII [3]. Correct identification of VWD subtypes is central to appropriate clinical management of these patients.

VWF antigen, VWF activity (either by VWF:RCo or VWF:GPIbM assay), and FVIII activity are included in a typical diagnostic panel, with additional testing available for further classification of type 2 VWD variants [2]. Accurate diagnosis of VWD in the clinical setting can be difficult, however, with multiple studies demonstrating variability in diagnosis. For instance, in a large European study, 15% of subjects with previously diagnosed type 1 VWD were reclassified to type 2 following identification of subtle abnormal multimers, many with reduced satellite bands [4]. In a study by Doruelo et al, gene sequencing of type 2M VWD cases revealed some variants also seen in type 2A VWD [5]. Another element of variability comes from potential issues with the VWF ristocetin cofactor activity assay, which has a high amount of variation [6]. Recently this has been supplemented by the VWF:GPIbM assay which uses a gain-of-function GPIb to bind VWF in the absence of ristocetin [7].

Our aim was to assess variability in diagnosis of type 2 VWD by comparing pre-existing and/or local diagnoses of type 2 VWD with results based on central laboratory testing.

Methods

Subjects for this research project were enrolled in the Zimmerman Program for the Molecular and Clinical Biology of VWD from various sites in the United States and Canada and were designated to either the retrospective or prospective cohort. Subjects with a pre-existing diagnosis of VWD were enrolled in the retrospective cohort (supplemental figure 1). The prospective cohort consisted of cases with a suspected diagnosis of VWD (supplemental figure 2). Informed consent was obtained following review and approval by each institution’s IRB. All eligible subjects were approached with the goal of recruitment of 170 type 2 VWD subjects for the retrospective study and 600 VWD subjects of any type for the prospective study. Subjects were recruited from participating hematology clinics (appendix 1) from 2006–2018 for the retrospective study and from 2013–2017 for the prospective study.

Laboratory testing was obtained from subjects at the time of enrollment and included factor VIII activity (FVIII), VWF antigen (VWF:Ag), VWF GPIbM activity (VWF:GPIbM), ristocetin cofactor activity (VWF:RCo), VWF factor VIII binding capacity, VWF propeptide (VWFpp), VWF collagen binding (VWF:CB), VWF-platelet binding, and DNA for genetic sequencing. FVIII was measured by 1 stage clotting assay. VWF:Ag and VWF:GPIbM were assessed with an enzyme-linked immunosorbent assay as described previously [8]. To measure VWF:RCo, ristocetin was added to formalin-fixed platelets using the Dade Behring BCS system [9]. Monoclonal antibodies bound to plasma VWF were used to assess FVIII binding activity [10]. VWFpp assay was evaluated as previously described [11]. Collagen binding was assessed with type 3 collagen as previously described [12]. Spontaneous binding of platelets to VWF was assessed with a platelet binding assay [13]. Genetic sequencing of the VWF coding region was performed with polymerase chain reaction amplification including intron-exon boundaries [9]. The original diagnosis was assigned by the local center, but central diagnosis required meeting specific laboratory criteria at the time of study entry.

For the retrospective cohort, subjects were enrolled based on a historical diagnosis with past laboratory testing, performed anywhere from months to years prior to study enrollment. For the prospective cohort, local diagnosis was made based on laboratory testing done at the time of enrollment, with a sample sent at the same time to the central laboratory (table 1). Local laboratory testing for the retrospective cohort used the VWF:RCo assay. For the prospective cohort, one laboratory used the VWF:GPIbM, and the remainder used the VWF:RCo. Each laboratory utilized their own platform and standards for the assay.

Table 1:

Laboratory diagnoses based on central VWF levels at time of study entry.

Diagnosis	Criteria
Low VWF	VWF levels between 30 IU/dL and lower limit of normal
Type 1	VWF:Ag and/or VWF:RCo and/or VWF:GPIbM <30 IU/dL with normal VWF:RCo/VWF:Ag ratio, normal multimers, and VWFpp/VWF:Ag ratio <3
Type 1C	VWFpp/VWF:Ag ratio >3 with normal VWF:RCo/VWF:Ag ratio, normal multimers and no evidence of type 2 VWD
Type 2A	VWF activity/antigen ratio <0.6 and loss of high molecular weight multimers
Type 2B	VWF activity/antigen ratio <0.6 and loss of high molecular weight multimers with increased VWF-platelet binding
Type 2M	VWF activity/antigen ratio <0.6 and normal multimers
Type 2N	Low FVIII and abnormal VWF-FVIII binding with decreased FVIII/VWF:Ag ratio
Type 3	Undetectable VWF levels
Normal	No abnormal VWF levels on either reported historical or current testing
Other	Includes subjects classified as hemophilia, potential hemophilia, hemophilia carrier, or for whom a definitive diagnosis could not be made

Results

Retrospective Cohort

A total of 3457 subjects in the retrospective cohort were enrolled in the Zimmerman Program at the initiation of this study. This population included 771 index cases with a pre-existing diagnosis of either low VWF, type 1 VWD, type 2 VWD, type 3 VWD, or unclassified VWD, as well as family members of index cases. The index cases were evaluated in this study, and of these subjects, 761 had fully evaluable data. Central lab results were compared to pre-existing diagnoses, and changes in diagnosis among subjects initially classified as type 2 VWD were assessed. Figure 1 depicts the number of type 2 VWD variant diagnoses that were confirmed with central testing in both retrospective and prospective cohorts. With regard to changes in diagnosis, all subjects in this cohort had a pre-existing diagnosis of VWD, therefore no subjects were identified as having VWD based solely on central laboratory testing. All switches in diagnosis were based on the central laboratory data not matching the assigned historical diagnosis. Data for individual subjects are given in supplemental table 1.

Figure 1. Confirmed type 2 VWD diagnoses in the retrospective and prospective VWD cohorts.

Type 2 subjects enrolled in the retrospective cohort (RC) and prospective cohort (PC) are represented on the X axis for each type 2 diagnosis. The Y axis denotes number of local diagnoses confirmed by central testing in black and number without a confirmed diagnosis in grey. Both study groups demonstrated consistent variation in subjects with discrepancies between original and central laboratory diagnosis.

In total, of the 761 index cases evaluated in this study, 217 subjects (29%) had a pre-existing diagnosis of type 2 VWD. Evaluation of central laboratory data at the time of study enrollment showed 178 subjects (23%) with a diagnosis of type 2 VWD. Table 2 shows the breakdown by subtype for both the pre-existing diagnosis and the laboratory diagnosis at time of enrollment for these subjects.

Table 2:

Comparison of prior and central laboratory diagnoses for type 2 VWD subjects in the retrospective cohort (n=761 total subjects).

VWD Type	Pre-existing Diagnosis	Central Laboratory Diagnosis
Total type 2 VWD	29% (217)	23% (178)
Type 2A	13% (98)	10% (80)
Type 2B	8% (62)	7% (55)
Type 2M	5% (39)	4% (28)
Type 2N	2% (18)	2% (15)

98 subjects with a pre-existing diagnosis of type 2A VWD had complete test results available from both local and central laboratories. 65 subjects (66%) were confirmed as type 2A VWD. The remainder had the laboratory diagnosis change: 6% to normal, 3% to low VWF, 4% to type 1 VWD, 11% to type 1C VWD, 5% to type 2B VWD, 2% to type 2M VWD, 1% to type 2N, and 1% to unclassified (figure 2A).

Figure 2. Comparison of local vs central laboratory testing results for type 2 VWD subjects in the retrospective cohort.

Figure 2A. 98 index cases in the retrospective cohort carried a pre-existing diagnosis of type 2A VWD. Following central laboratory testing, the majority of cases (66%) were confirmed as type 2A VWD. Figure 2B. 62 index cases in the retrospective cohort carried a pre-existing diagnosis of type 2B VWD. Following central laboratory testing, the majority of cases (77%) were confirmed as type 2B VWD. Figure 2C. 39 index cases in the retrospective cohort carried a pre-existing diagnosis of type 2M VWD. Following central laboratory testing, around half of the cases (54%) were confirmed as type 2M VWD. Figure 2D. 18 index cases in the retrospective cohort carried a local diagnosis of type 2N VWD. Following central laboratory testing, the majority of cases (72%) were confirmed as type 2N VWD.

62 subjects with a pre-existing diagnosis of type 2B VWD had complete test results available from both local and central laboratories. 48 subjects (77%) were confirmed as type 2B VWD. The remainder had the laboratory diagnosis change: 6% to normal, 2% to low VWF, 2% to type 1 VWD, 2% to type 1C VWD, 8% to type 2A VWD, 2% to type 2M VWD, and 2% to unclassified (figure 2B).

39 subjects with a pre-existing diagnosis of type 2M VWD had complete test results available from both local and central laboratories. 21 subjects (54%) were confirmed as type 2M VWD. The remainder had the laboratory diagnosis change: 15% to normal, 10% to low VWF, 8% to type 1 VWD, 10% to type 1C VWD, and 3% to type 2A VWD (figure 2C).

18 subjects with a pre-existing diagnosis of type 2N VWD had complete test results available from both local and central laboratories. 13 subjects (72%) were confirmed as type 2N VWD. The remainder had the laboratory diagnosis change: 11% to normal, 11% to hemophilia A, and 6% to unclassified (figure 2D).

Type 1 VWD

Some subjects had a pre-existing diagnosis of type 1 VWD and received a diagnosis of type 2 VWD following central laboratory testing. Of 459 subjects originally diagnosed with type 1 VWD, 10 subjects (2%) received a new diagnosis of type 2 VWD. Four were type 2A, 1 type 2B, 4 type 2M and 1 type 2N.

Prospective Cohort

The prospective cohort of the Zimmerman Program included 664 index cases with a local laboratory diagnosis of VWD, of whom 610 had complete laboratory data available for analysis. Local diagnoses were compared to those obtained by central laboratory testing with a focus on revised diagnoses in patients with initial designation of type 2 VWD. Percent of confirmed diagnoses is graphically represented in figure 1. With regard to changes in diagnosis, almost all subjects considered as type 2 VWD by the central laboratory results had a diagnosis of some type of VWD based on local laboratory results. The sole exception was a subject coded as type 2M VWD by the central laboratory based on a low VWF collagen binding with type IV collagen. Since this testing is not typically performed commercially, it makes sense that the local laboratory did not have this data to use in classification. Some changes in diagnosis were due to the original center not assigning a specific type 2 variant to those subjects with type 2 VWD. Data for individual subjects are given in supplemental table 1.

35 subjects had a local diagnosis of type 2 VWD. 33 of those have complete data available for central laboratory classification, of whom 15 (45%) retained a classification of type 2 VWD (table 3). Local laboratory classification yielded 16 subjects with type 2A, 5 subjects with type 2B, 13 subjects with type 2M, and 1 subject with type 2N VWD. Following central laboratory testing, 5 subjects (31%) retained a diagnosis of type 2A VWD, 1 subject was reclassified as type 2B, 3 as likely type 2 but presently unclassified, and one as normal, while the remainder were classified as type 1 VWD. Two of the type 2B subjects were classified as normal based on central laboratory testing while the remainder retained a diagnosis of type 2B VWD (60%). Of the type 2M subjects, only 3 (23%) retained a diagnosis of type 2M VWD. One was classified as type 1 and the remainder were normal. The type 2N subject was considered to be a type 2N carrier (figure 3).

Table 3:

Comparison of local and central laboratory diagnoses for type 2 VWD subjects in the prospective cohort (n=610 total subjects, 35 total type 2 subjects).

VWD Type	Local Diagnosis as percent of enrolled subjects (# of subjects)	Central Laboratory Diagnosis as percent of enrolled subjects (# of subjects)
Total type 2 VWD	6% (35)	4% (24)
Type 2A	3% (16)	1% (8)
Type 2B	1% (5)	1% (4)
Type 2M	2% (13)	2% (11)
Type 2N	0.2% (1)	0.2% (1)

Figure 3. Percent of confirmed type 2 VWD diagnoses in the prospective VWD cohort.

35 index cases in the prospective cohort carried a local diagnosis of type 2 VWD. Following central laboratory testing, 45% retained a diagnosis of type 2 VWD. The pie chart shows diagnoses of this group following central laboratory testing.

When the central laboratory results for the entire prospective cohort were considered, 24 subjects (4%) met criteria for type 2 VWD. 8 received a diagnosis of type 2A VWD, 4 type 2B VWD, 11 type 2M VWD, and 1 a type 2N carrier. 6 subjects had a local diagnosis of type 1 VWD and a central diagnosis of type 2 VWD (all type 2M).

Common Sequence Variants

Among the subjects in the retrospective cohort whose diagnosis of a type 2 VWD variant changed following central testing, common sequence variants were identified. In the type 2A cohort, p.M1304R was classified as type 1 (n=1), type 1C (n=1) and type 2A (n=1) on central laboratory testing. The p.R1315C variant was classified as type 1 (n=2), type 1C (n=1), and type 2A (n=2). The p.R1374C variant was classified as type 1C and type 2A (n=1 each). The p.R1374H variant was classified as only type 2A (n=2).

In the prospective group, p.R1374H was seen in type 2A VWD (n=1), while p.R1315C and p.R1374C were seen in subjects classified as type 2A on local labs and type 1C on central labs (n=1 each). 4 subjects with p.D1472H were classified as type 2 based on local labs but as normal based on central laboratory testing using the VWF:GPIbM assay.

There were some genetic variants where consistency in diagnosis was maintained. For type 2A VWD, p.R1597W remained consistently associated with a type 2A diagnosis (11 subjects). 4 subjects with type 2A VWD had p.G1180R. For type 2B, those with known genetic type 2B variants were almost universally classified as type 2B both locally and centrally. All type 2N VWD subjects had p.R854Q.

Discussion

The data from this study has demonstrated that there is variable reproducibility in type 2 VWD diagnosis, as reflected in both the prospective and retrospective cohorts. This was most strikingly seen in subjects from the retrospective cohort that carried a pre-existing diagnosis of type 2A VWD, type 2B VWD, or type 2M VWD.

There are a number of factors that can influence the reliability and reproducibility of type 2 VWD diagnoses. Stress, exercise, race, ABO blood type, pregnancy, menstruation, and oral contraceptive medications have all been found to affect VWF levels [2,14]. The clinical variability of the VWF:RCo can be as high as 30% [15], and while the VWF:Ag assay has less variability, it has been reported in the past to be as great as 20% [6]. The quality of individual laboratory testing, including collection, storage, processing, and transportation, can also contribute to high clinical variability [2]. Subjects in the retrospective cohort who had a number of years between laboratory tests may have experienced increase in VWF:Ag with age, but generally type 2 VWD patients retain their laboratory abnormality throughout their life.

VWF activity with the ristocetin cofactor assay can be artificially reduced, particularly with the p.D1472H variant [9,16]. Several groups have developed assays that bypass ristocetin. In 2010 Flood et al published an ELISA assay using recombinant GPIb [8]. In 2014, Patzke et al developed a similar but automated assay [7]. These assays are unaffected by the p.D1472H variant [8] and demonstrate good correlation with the gold standard ristocetin assay in the assessment of subjects with type 1 VWD and type 2 VWD variants, as well as normal controls, with within-device clinical variability of 2–7% [7]. The ISTH has subsequently named this assay the VWF:GPIbM since it involves a gain of function GPIb variant [17]. Favaloro et al also showed that this assay could be an alternative to VWF:RCo assays [18].

Variability in the VWF:RCo could certainly result in subjects having an abnormal VWF activity to antigen ratio and being assigned a either a diagnosis of type 2 VWD or a diagnosis of type 1 VWD, or no VWD on other occasions. Hypothetically, use of the VWF:GPIbM would reduce this variability as seen in the prospective cohort, with a number of subjects changing a diagnosis of type 2M VWD based on VWF:RCo data to normal based on VWF:GPIbM testing. It is important to distinguish type 1 from type 2 VWD as type 1 VWD is typically treated with desmopressin whereas type 2 VWD is generally treated with VWF concentrates [19]. The difference between type 2 VWD and no VWD is also clearly significant.

Genetic testing has emerged as a useful adjunct in the diagnosis of type 2 VWD. We identified four sequence variants which altered the original diagnosis following central laboratory testing. These were p.M1304R, p.R1315C, p.R1374C, and p.R1374H. Chen et all described p.M1304R as a driver of phenotypic variability in VWD amongst family members [20]. The variant p.R1315C has also demonstrated phenotypic variability in the literature and has been associated with type 1C, type 2A, type 2M, as well as “unclassifiable” diagnoses [21]. The variants p.R1374C and p.R1374H mutations have been found in type 2M VWD [22,23]. In our subjects p.R1374C was associated with type 2A and type 1C and p.R1374H with type 1C.

Genetic analysis can present certain disadvantages. Notably, in healthy people, there is significant variability in the VWF gene, with polymorphisms once classified as pathologic now frequently identified in some healthy populations [24]. Additionally, genetic testing is not yet available in all centers.

Interestingly, the percent of subjects diagnosed with type 2 VWD varied dramatically between the retrospective and prospective cohorts. While 217 out of 761 subjects in the retrospective cohort carried a diagnosis of type 2 VWD (29%), only 35 of 610 subjects received a diagnosis of type 2 VWD in the prospective cohort (6%). This could in part be due to the nature of the cohorts. The retrospective cohort was enrolled from participating centers and their population of known VWD patients, which are likely enriched for the more severe forms of VWD, and the study specifically requested additional type 2 subjects for analysis. Since one goal of the Zimmerman project was to evaluate all types of VWD, centers were enrolling patients based on their diagnosis, not based on relative frequencies in their communities. The prospective cohort was enrolled from patients presenting with bleeding symptoms and therefore likely reflects the true prevalence of type 2 VWD in the community. Previous reports of the frequency of type 2 VWD vary between 9 and 30% of total VWD cases [25]. Our data suggest that approximately 4–6% of patients diagnosed with VWD have a type 2 VWD variant.

Accurate diagnosis of VWD is important to assess bleeding risk as well as provide appropriate therapeutic management. For instance, in our retrospective cohort, 3 subjects were reclassified from type 2A VWD to type 2M VWD. Castaman et al demonstrated that the incidence of bleeding is higher in type 2A VWD than in type 2M VWD [26]. Therefore, these subjects may benefit from more accurate risk stratification prior to a procedure. Additionally, use of desmopressin in patients with type 2B VWD could result in a decrease in platelet count [27].

In conclusion, there is variable reproducibility in the diagnosis of type 2 VWD and its variants. While we have shown that there may be conflicting data resulting in differing diagnoses among laboratories, this does not imply that our central laboratory standards are more accurate than those at other institutions. Rather, these results argue for further improvement in diagnostic strategies to minimize the discrepancies that are evident from our study. Re-evaluation of the original VWD diagnoses may be helpful for tailoring of management and risk stratification, especially in patients with higher risk bleeding phenotypes or with type 2B VWD.

Essentials:

1. Patients with von Willebrand disease were enrolled in our study.

2. Type 2 VWD diagnoses were based on original test results.

3. Repeat evaluation resulted in many patients receiving a different type 2 diagnosis.

4. Some genetic variants were particularly likely to move type 2 subcategories.