Rapid discrimination of the phenotypic variants of von Willebrand disease

Abstract

Approximately 20% to 25% of patients with von Willebrand disease (VWD) have a qualitative defect of the von Willebrand factor (VWF) protein activities. Variant VWD typically is classified as type 1C, 2A, 2B, 2M, or 2N depending on the VWF activity defect. Traditionally, diagnosis has relied on multiple clinical laboratory assays to assign VWD phenotype. We developed an enzyme-linked immunosorbent assay (ELISA) to measure the various activities of VWF on a single plate and evaluated 160 patient samples enrolled in the Zimmerman Program for the Molecular and Clinical Biology of von Willebrand Disease with type 2 VWD. Using linear discriminate analysis (LDA), this assay was able to identify type 1C, 2A, 2B, 2M, or 2N VWD with an overall accuracy of 92.5% in the patient study cohort. LDA jackknife analysis, a statistical resampling technique, identified variant VWD with an overall accuracy of 88.1%, which predicts the assay's performance in the general population. In addition, this assay demonstrated correlation with traditional clinical laboratory VWF assays. The VWF multiplex activity assay may be useful as a same-day screening assay when considering the diagnosis of variant VWD in an individual patient.

Figure 1

ROC curve discrimination between individual assay components. Each ROC curve depicts individual VWF activity tests that together comprise the assay. ROC curve analysis shows each individual test’s ability to make VWF phenotype discriminations. (A) ROC curve: type 1C vs type 2A, VWF:GPIbM. (B) ROC curve: type 2B vs type 1, VWF:GPIbM. (C) ROC curve: type 2B vs type 2A, VWF:GPIbM. (D) ROC curve: HA (mild or carriers) vs type 2N, VWF:FVIIIB. (E) ROC curve: type 1C vs type 2A, VWF:CB3. (F) ROC curve: type 2M vs type 2A, VWF:CB3. (G) ROC curve: type 1C vs type 2A, VWFpp. (H) ROC curve: type 1C vs type 1, VWFpp. (I) ROC curve: type 2A vs type 1C, VWF:Ag. RT, ratio threshold.

Figure 2

Clinical VWF:Ag vs experimental VWF:Ag ratio in 134 VWD subjects.

Figure 3

Clinical VWF:FVIIIB vs experimental VWF:FVIIIB/Ag ratio. (A) Type 2N VWD and HA (mild or carriers) subjects. (B) Type 2N and 2N carrier VWD subjects. Type 2N VWD and type 2N carrier separation by clinical vs experimental VWF:FVIIIB assay. Type 2N carriers demonstrated both higher clinical and experimental VWF:FVIIIB than phenotypic 2N VWD or known compound 2N heterozygote or 2N homozygote subjects.