Prospective Diagnosis of VWD in a Large Cohort of Patients with Bleeding Symptoms through the Zimmerman Program

Background: Diagnosis of von Willebrand disease (VWD) is challenging in clinical practice due to variability in laboratory testing and clinical bleeding history. We investigated the prospective diagnosis of VWD in academic hematology clinics across the U.S. and Kingston, ON and report on the final cohort here.

Methods: Subjects were enrolled as new consults to their hematologist for evaluation of a bleeding disorder from 11 centers. Laboratory results including VWF antigen (VWF:Ag) and VWF platelet binding activity were determined both locally (VWF ristocetin cofactor activity [VWF:RCo] or VWF:GPIbM) and centrally (VWF:GPIbM) to determine the comparative effectiveness in VWD diagnosis. Some centers defined type 1 VWD with levels <30 and low VWF as VWF:Ag and/or activity ≥30 but less than lower limit of normal while others used only type 1 VWD. Bleeding scores were obtained using the ISTH bleeding assessment tool. DNA sequencing of all exons including intron/exon boundaries was performed and variants classified as presumably pathogenic if present in <1% of the general population using human genetic variation databases.

Results: A total of 1826 subjects undergoing evaluation for VWD or a suspected bleeding disorder were prospectively enrolled. Median age, demographics, and bleeding score did not differ between the VWD and non VWD groups. Since pediatric subjects comprised a large segment of our cohort, bleeding scores were examined for subjects ≥18 years old and showed no difference with a median of 7 for those with and without VWD (p=NS). Subjects with VWD had no significant difference in VWF:Ag between local and central lab testing. Differences in VWF platelet binding was likely due to most local labs using VWF:RCo vs central lab testing with VWF:GPIbM.

The diagnosis of VWD was determined locally. A total of 36% of subjects received a diagnosis of VWD. Of those, 63% received a diagnosis of type 1 VWD, 27% low VWF and 10% other (type 2A, 2B, 2M, 2N, 3 or unclassified). The percent of subjects with VWD was consistent across centers. One subject was diagnosed as type 3 VWD. For type 2 VWD, 4% were 2A, 1% 2B, 2% 2M, and <1% 2N. Pathogenic genetic variants in VWF were most common (71%) in type 2 and 3 VWD subjects as compared to type 1 (45%) or low VWF (25%) cohorts.

We then separated these subjects into 3 groups by diagnostic testing: 1) type 1 VWD (using criteria of VWF:Ag or activity <30 by either local or central lab), 2) low VWF or 3) non VWD. Type 2 VWD required VWF activity/VWF:Ag ratio <0.6, loss of high molecular weight multimers for types 2A and 2B and increased VWF platelet binding for type 2B.

When both local and central lab results were taken into account and study definitions applied, only 29% of the enrolled subjects had abnormal VWF levels. Of these, 78% had low VWF, 15% type 1, and 7% type 2 VWD. Median bleeding score for both the low VWF and type 1 cohorts was 4 (p=NS). Type 2 subjects with confirmed phenotypic diagnoses had a median bleeding score of 5 (p=NS compared to the type 1 or low VWF cohorts).

Discussion: This unique large prospective study of bleeding disorder subjects, designed as an inception cohort, highlights the diagnosis of VWD but presents several clinical challenges: 1) The lack of difference in bleeding scores between VWD and non VWD suggests that bleeding symptoms are common and possibly due to other diagnoses or not necessarily related to VWF levels. 2) The variability in VWD diagnosis with borderline levels is common when local and central lab testing are applied, potentially due to varying use of the VWF:RCo vs VWF:GPIbM to assess platelet binding activity. The comparative effectiveness of methods of diagnosis will be highlighted by followup testing on these subjects. 3) As only 1/3 of subjects merited a VWD or low VWF diagnosis, mild bleeding may be the result of other conditions; platelet defects, vascular disorders, collagen defects, other unidentified mechanisms or potentially non-pathogenic. We hypothesize that the presence of bleeding symptoms increases the likelihood that a diagnosis of VWD will be made, regardless of the level. This cohort will continue to be followed longitudinally which may help elucidate the importance of serial VWF measurement. However, the stability of VWD diagnosis across centers in 1/3 of subjects suggests a broader application.