OUP user menu

von Willebrand Factor Assay Proficiency Testing
The North American Specialized Coagulation Laboratory Association Experience

Wayne L. Chandler MD, Ellinor I.B. Peerschke PhD, Donna D. Castellone MS, MT(ASCP)SH, Piet Meijer PhD on behalf of the NASCOLA Proficiency Testing Committee
DOI: http://dx.doi.org/10.1309/AJCPH5JK4ONENPAE 862-869 First published online: 1 June 2011

Abstract

We evaluated the accuracy and precision of von Willebrand disease (vWD) testing performed by up to 50 North American Specialty Coagulation Laboratories from 2004 through 2009, using proficiency samples from healthy subjects (n = 7) and patients with type 1 vWD (n = 7) or type 2 vWD (n = 3). We analyzed 2,212 submitted results. Precision was highest for von Willebrand factor (vWF) antigen assays (coefficient of variation, 14%), which were performed predominantly by latex immunoassays, and lowest for ristocetin cofactor assays (coefficient of variation, 28%), which were increasingly replaced by collagen binding and immunofunctional methods during the 6-year evaluation period. Overall interpretation error rates ranged from 3% for normal samples, 28% for type 1 vWD, and 60% for type 2 vWD. Type 2 vWD samples were correctly identified by all laboratories using collagen binding/antigen ratios but by only one third of laboratories using ristocetin cofactor/antigen or immunofunctional/antigen ratios. In 2009, only 27% (12/45) of laboratories performed vWF multimer analysis, with error rates ranging from 7% to 22%.

Key Words:
  • von Willebrand disease
  • von Willebrand factor antigen assay
  • Proficiency testing

von Willebrand factor (vWF) levels and function may be altered in plasma due to hereditary and/or acquired influences. Low vWF function in blood may reflect a genetic disorder, von Willebrand disease (vWD), or represent a population risk factor for bleeding.1 Most clinical laboratories use a panel of tests to evaluate vWF. The most common assays reported include vWF antigen levels, vWF function assays (ristocetin cofactor, collagen binding, and immunofunctional), factor VIII activity, and vWF multimer analysis. Less common assays include ristocetin-induced platelet aggregation for evaluation of type 2B vWD, vWF gene sequencing, and factor VIII binding to vWF.

Assessment of the probability of vWD and the risk of bleeding due to reduced vWF functional levels in blood requires accurate and precise measurements of vWF levels, function, and structure. A number of studies have reported on the ability of clinical laboratories to evaluate vWF based on the results from external quality assurance proficiency testing programs. In the United States, the College of American Pathologists reported on vWF assay precision and accuracy for clinical laboratories participating in its vWF quality assurance program from 2003 to 2005.2 The External Quality Control for Assays and Tests (ECAT) Foundation reported on proficiency testing precision, accuracy, multimer results, and interpretation for clinical laboratories in Europe and North America from 2003 to 2005.3 The United Kingdom National External Quality Assessment Scheme (UK NEQAS) reported on its vWF proficiency testing results,4 while Favaloro et al58 reported in-depth on vWF quality assurance results from approximately 50 clinical laboratories in the Southern hemisphere, concentrating on vWF functional assays. Major findings from prior studies include problems with the ristocetin cofactor assay, including low precision and diagnostic errors, better results with collagen binding assays, and overall high error rates diagnosing type 1 and type 2 vWD.

The present study provides insight into the current status of vWF testing in North America from the North American Specialized Coagulation Laboratory Association (NASCOLA). It identifies changing trends in vWD testing during 6 years (2004–2009), using samples from patients with type 1 or type 2 vWD rather than artificially depleted samples, and includes results from immunofunctional assays, multimer studies, and final interpretations.

Materials and Methods

NASCOLA is a nonprofit organization that provides proficiency testing, in partnership with the ECAT Foundation, for North American laboratories performing diagnostic testing for bleeding and prothrombotic disorders and a forum for critical evaluation of coagulation testing procedures, reagents, and instrumentation.9,10 In 2003, NASCOLA began distributing 1 vWF proficiency test 4 times per year. Proficiency testing specimens consisted of lyophilized plasma samples prepared by the ECAT Foundation (Leiden, the Netherlands).3 Several types of samples were distributed, including normal plasma, plasma from patients diagnosed with type 1 or type 2 vWD, and samples with artificially prepared multiple factor deficiency (typically commercial abnormal coagulation control plasma samples). For the purposes of this study, we analyzed proficiency testing results for vWF antigen, vWF ristocetin cofactor activity, vWF collagen binding activity, vWF immunofunctional activity, and vWF multimer analysis, and we evaluated final interpretations from 6 years of surveys (2004–2009) for plasma samples prepared from healthy subjects (n = 7 separate proficiency testing samples) and from patients diagnosed with vWD type 1 (n = 7 separate proficiency testing samples) or type 2 (n = 3 separate proficiency testing samples). Plasma samples artificially depleted of factors were not evaluated.

Figure 1

Number of clinical laboratories participating in the North American Specialized Coagulation Laboratory Association von Willebrand factor antigen testing survey from 2004 through 2009.

To better understand laboratory practices contributing to test performance, laboratories participating in NASCOLA vWF proficiency testing were surveyed by electronic questionnaire in 2004. The questionnaire was designed to elicit information about preanalytic, analytic, and postanalytic variables affecting factor VIII and vWF testing. A total of 39 questionnaires were distributed, of which 33 completed questionnaires were returned for evaluation.

Results

Figure 1 shows the number of laboratories submitting results for vWF proficiency testing surveys during each year of the study period. The data are based on laboratories participating in vWF antigen testing, the assay with the highest number of participants. A total of 2,212 results were analyzed. As in prior studies, results greater than 3 SD from the mean were not evaluated.3,8 Three results that were 6 SD, 7 SD, and 12 SD from the mean were removed from analysis.

Assay Method

Four proficiency testing surveys were distributed by NASCOLA/ECAT each year. Table 1 shows the methods used and number of total survey responses for each assay type for each year.

For vWF antigen, the most common type of assay used was the latex immunoassay. Responses for vWF antigen proficiency testing challenges submitted with this assay code increased from 76.8% to 86.1% of the total responses from 2004 to 2009. The most common vWF antigen reagent used was the Diagnostica Stago latex immunoassay (Parsippany, NJ), which accounted for 75.7% ± 2% of participant responses during the 6-year study period.

For vWF ristocetin cofactor activity, the most common source of reagents was Siemens/Dade Behring (Deerfield, IL), which accounted for 39.2% of participant responses in 2004 and 63.4% in 2009, representing an increase in automated ristocetin cofactor testing.

The use of the vWF immunofunctional assay increased substantially during the 6-year evaluation period. In 2004, only about 5% of results for vWF functional assays used vWF immunofunctional methods. By 2009, 21.3% of vWF functional assay results used immunofunctional methods, with the Instrumentation Laboratory (Bedford, MA) LIA accounting for 50% of immunofunctional assay responses (23/46).

View this table:
Table 1

Use of the vWF collagen binding assay remained constant during the 6 years evaluated, constituting about 11% to 12% of the total reported vWF functional results. Life Diagnostics (West Chester, PA) was the most common source of vWF collagen binding reagents, representing almost 70% of participant responses.

The use of immunofunctional and collagen binding vWF functional assays increased from 13.2% of the total functional responses in 2004 to 32.9% in 2009, whereas ristocetin cofactor responses decreased from 86.8% of total functional assays in 2004 to 67.1% in 2009.

Assay Imprecision

Table 2 shows the all-method mean and coefficient of variation (CV) for each assay separated by assay type and type of proficiency sample—normal, type 1 vWD, or type 2 vWD. The mean reported vWF antigen ranged from 94 to 121 U/dL for the 7 samples from healthy subjects, 21 to 52 U/dL for the 7 samples from patients with type 1 vWD, and 15 to 47 U/dL for the 3 samples from patients with type 2 vWD. The average CV for all vWF antigen testing was 14%.

The mean reported vWF ristocetin cofactor activity ranged from 74 to 91 U/dL for the 7 samples from healthy subjects, 16 to 36 U/dL for the 7 samples from patients with type 1 vWD, and 12 to 25 U/dL for the 3 samples from patients with type 2 vWD. The average CV for all vWF ristocetin cofactor activity testing was 28%.

The mean reported vWF immunofunctional activity ranged from 87 to 110 U/dL for the 6 samples from healthy subjects (with more than 1 participant responding), 26 to 40 U/dL for the 6 samples from patients with type 1 vWD (with more than 1 participant responding), and 10 to 39 U/dL for the 3 samples from patients with type 2 vWD. The average CV for all vWF immunofunctional activity testing was 19%.

The mean reported vWF collagen binding activity ranged from 89 to 133 U/dL for the 7 samples from healthy subjects, 15 to 39 U/dL for the 7 samples from patients with type 1 vWD, and 5 to 34 U/dL for the 3 samples from patients with type 2 vWD. The average CV for all vWF collagen binding testing was 20%.

The vWF antigen assays had the lowest overall imprecision, whereas the vWF ristocetin cofactor activity assays had the highest overall imprecision of the 4 assay types tested. For the 7 normal samples and 7 type 1 vWD samples, there was no evidence of significant change in the CV of results over time, ie, no trend toward improvement for any of the 4 assays.

View this table:
Table 2

vWF Function/vWF Antigen Ratio Imprecision

Table 2 summarizes the all-method mean and CV for ratios comparing vWF function with vWF antigen, separated by functional assay method and type of proficiency sample—normal, type 1 vWD, or type 2 vWD. This ratio is commonly used in the evaluation of vWF dysfunction (type 2 deficiency), with an empirically established cutoff of 0.5 to 0.7. Ratios falling below this cutoff would be suggestive of type 2 vWD and might trigger further testing for confirmation. As expected, the mean reported vWF ristocetin cofactor/vWF antigen ratio was similar for samples from healthy subjects and patients with type 1 vWD, ranging from 0.69 to 0.81 for the 7 samples from healthy subjects and 0.63 to 0.82 for the 7 samples from patients with type 1 vWD. However, for the 3 type 2 vWD samples, the average vWF ristocetin cofactor/vWF antigen ratio was less than 0.6 for 1 sample (0.51) and greater than 0.6 for the other 2 samples (0.78 and 0.90). The average CV for all vWF ristocetin cofactor/vWF antigen ratios was 31%.

The mean reported vWF immunofunctional activity/vWF antigen ratio was similar in samples from healthy subjects and patients with type 1 vWD, ranging from 0.79 to 1.02 for the 6 samples from healthy subjects (with more than 1 participant responding) and 0.78 to 1.07 for the 6 samples from patients with type 1 vWD (with more than 1 participant responding). For the 3 type 2 vWD samples, the average vWF immunofunctional activity/vWF antigen ratio was less than 0.7 for 1 sample (0.56) but greater than 0.7 for the other 2 samples (0.79 and 0.94). The average CV for all vWF immunofunctional activity/vWF antigen ratios was 23%.

The mean reported vWF collagen binding activity/vWF antigen ratio was similar for samples from healthy subjects and patients with type 1 vWD but lower in patients with type 2 vWD, ranging from 0.78 to 1.18 for the 7 samples from healthy subjects, 0.70 to 1.06 for the 7 samples from patients with type 1 vWD, and 0.39 to 0.69 U/dL for the 3 samples from patients with type 2 vWD. The average CV for all vWF collagen binding activity/vWF antigen ratios was 22%. The vWF ristocetin cofactor/vWF antigen ratios had the highest overall imprecision of the 3 ratios tested.

For vWF immunofunctional and collagen binding assays, the average vWF function/antigen ratio was more than 0.7 for all of the normal and type 1 vWD samples. The average ristocetin cofactor activity/antigen ratio was greater than 0.6 for all normal and type 1 vWD samples. For the type 2 vWD samples, only collagen binding assays showed an average ratio to antigen of less than 0.7 for all 3 samples, but one was close to the cutoff, at 0.69. The vWF ristocetin cofactor/antigen and immunofunctional activity/antigen ratios were less than 0.6 and 0.7, respectively, for 1 of 3 type 2 vWD samples but they were different samples.

For the 7 normal samples and 7 type 1 vWD samples, there was no evidence of significant change in the CV of ratios over time, ie, no trend toward improvement for any of the 3 ratios.

Variables Affecting vWF Testing Results

Table 3 summarizes the major variables in laboratory practices that could affect test results. Whereas most laboratories (23/33 [70%]) placed plasma samples for vWF testing at room temperature before analysis, a significant number (9/33 [27%]) placed samples on ice. The latter preanalytic variable has been shown to reduce higher molecular weight multimers.11 There was also significant variation with regard to vWF standard/calibrator value assignment. The majority of laboratories relied on values assigned by the manufacturer of the commercial standard/calibrator in use, while others (6/33 [18%]) assayed their reference plasma against World Health Organization reference plasma. A minority (2/33 [6%]) of laboratories assayed their reference plasma against a previous laboratory standard/reference plasma. Further affecting analytic performance was variability in standard curve preparation. About two thirds of laboratories responding to the questionnaire prepared a new standard curve for each assay or run, and one third did so only when quality control failed.

vWF Assay Interpretation

Table 4 shows the interpretation (normal vs abnormal) separated by assay type and type of proficiency sample—normal, type 1 vWD, or type 2 vWD. All participants correctly identified normal plasma using vWF antigen assays, vWF immunofunctional assays, and vWF collagen binding assays. For vWF ristocetin cofactor assays, 2.0% of responses identified normal plasma as abnormal. For plasma from patients with type 1 vWD, 12.3% of responses reported normal levels using vWF antigen testing, 4.2% using vWF ristocetin cofactor assays, 2% (1/43) using vWF immunofunctional assays, and 6% (2/34) using vWF collagen binding assays. One of the samples from a patient with type 1 vWD had an average vWF antigen level in the survey of 52 U/dL (2006–3), which led 51% (24/47) of participants to interpret results on this sample as normal. For plasma from patients with type 2 vWD, 7.8% of responses reported normal levels using vWF antigen testing and 0.9% with vWF ristocetin cofactor testing, but all participants correctly identified the samples as abnormal using vWF immunofunctional and vWF collagen binding assays.

vWF Multimer Interpretation

Only about one quarter of the laboratories participating in the NASCOLA vWF proficiency testing perform the vWF multimer assay in house. This number did not change during the 6-year study period (average, 11 laboratories reporting multimers; range, 8–15). For the 7 normal samples, 93% (75/81) of responses correctly identified a normal multimer pattern, 5% (4/81) reported loss of high-molecular-weight multimers, and 2% (2/81) reported loss of medium- and high-molecular-weight multimers. For the type 1 vWD samples, 81% (65/80) of responses correctly identified a normal multimer pattern, 18% (14/80) reported loss of only high- or medium- and high-molecular-weight multimers, and 1 participant reported inability to detect any multimers despite reporting near-normal vWF antigen levels. For the type 2 vWD samples, 78% (28/36) indicated loss of high- or medium- and high-molecular-weight vWF, and 22% (8/36) reported a normal multimer pattern. Overall, 14.7% of vWF multimer survey responses were in error.

View this table:
Table 3
View this table:
Table 4

Conditions triggering vWF multimer analysis by NASCOLA laboratories, locally or as a send-out test, were elicited by questionnaire in 2004. Findings are summarized in Table 5. Laboratories indicated requesting vWF multimer analysis on 10% to 40% of patient samples submitted for vWF testing. Approximately half of the laboratories relied on the vWF antigen/activity ratio (cutoff, 0.5–0.7) to determine whether vWF multimer analysis was indicated. The remainder requested vWF multimer analysis based on physician orders or local testing panels.

Final (Diagnostic) Interpretation

As the complexity of the sample increased, the number of laboratories reporting final (diagnostic) interpretations decreased. For the normal samples, 75.7% of participants provided a final interpretation; for type 1 vWD samples, 56.0% provided a final interpretation; and for type 2 vWD samples, only 44.4% provided a final interpretation. Table 6 shows the final interpretation based on all vWF testing separated by type of proficiency sample—normal, type 1 vWD, or type 2 vWD. For the 7 normal samples, 97.2% of responses correctly indicated a normal sample, 1 participant indicated type 1 vWD, and 2.4% indicated type 2 vWD typically based on low vWF function/antigen ratios or multimer analysis. For the type 1 vWD samples, 72.3% of responses correctly identified type 1 vWD, 13.9% reported a normal sample, and 13.9% reported type 2 vWD. For the type 2 vWD samples, 40% reported type 2A or 2B vWD, the best answers; 40% (25/63) reported type 1 vWD; 13% (8/63) reported type 3 vWD; and the remainder reported normal, type 2M vWD, or type 2N vWD. Error rates on final (diagnostic) interpretations showed an increase as the sample complexity increased: 2.8% for normal samples, 27.7% for type 1 vWD samples, and 60% (38/63) for type 2 vWD samples.

Discussion

Based on review of 6 years of recent vWF proficiency testing data, the state-of-the-art for vWF testing in North America can be summarized as follows: vWF antigen testing is performed mainly by latex immunoassays and demonstrates good analytic performance. The imprecision of vWF ristocetin cofactor analysis remains high and leads to poor discrimination between type 1 and type 2 deficiencies. Alternative vWF functional assays are performed with increasing frequency as a substitute for or supplement to ristocetin cofactor activity evaluation. These newer assays seem to perform better than the ristocetin cofactor assay when used in vWF activity/antigen ratios to discriminate between type 1 and type 2 vWD. Finally, in-house vWF multimer analysis is performed by a minority of laboratories. Diagnostic error rates based on vWF multimer analysis remain high. A more detailed discussion of each of these observations in the context of information obtained from other proficiency testing surveys performed in the United States and abroad follows.

View this table:
Table 5
View this table:
Table 6

Among NASCOLA laboratories, vWF antigen testing by latex immunoassays increased from about 77% to 86% between 2004 and 2009. Similar observations have been made in the United Kingdom.4 vWF antigen latex immunoassays provided the best precision of any assay in the current and prior studies.2,3 Hayes et al2 reported vWF testing results from 160 to 219 laboratories participating in the College of American Pathologists program between 2003 and 2005. Their analysis of a survey sample consisting of the International Society on Thrombosis and Haemostasis Secondary Coagulation Standard, lot 2, demonstrated that vWF antigen latex immunoassays were also accurate, with average reported results agreeing well with the established value for the standard.

The use of ristocetin cofactor assays by NASCOLA laboratories decreased from about 86% of total vWF functional assay responses in 2004 to about 67% in 2009. In contrast, vWF immunofunctional and collagen binding assay use more than doubled from about 14% to 33% of the total. Favaloro et al8 reported a similar drop in vWF ristocetin cofactor testing performed by laboratories in Australasia. In our study, vWF ristocetin cofactor assays continued to show poor overall precision, with no evidence of improvement compared with prior studies. Indeed, vWF ristocetin cofactor was the only assay that produced individual assay interpretation errors for the normal samples.24 Meijer and Haverkate3 reported vWF testing results between 2003 and 2005 for the ECAT Foundation vWF quality assurance program. They reported that vWF ristocetin cofactor assays had the worst precision and that errors in the diagnosis of samples from patients with type 1 vWD were most often due to discordance in the vWF ristocetin cofactor/antigen ratio.3 Kitchen et al4 reported similar problems for vWF ristocetin cofactor/antigen ratios from UK NEQAS data. Favaloro et al,58 in a series of articles summarizing the Australasian experience, reported that the vWF ristocetin cofactor assay was often the source of error in the vWF proficiency testing and that vWF collagen binding activity/vWF antigen ratio was 4- to 10-fold less likely to produce an error in diagnosis than the vWF ristocetin cofactor/vWF antigen ratio. In our data, the vWF collagen binding/vWF antigen ratio was better than the vWF ristocetin cofactor/vWF antigen and vWF immunofunctional assay/vWF antigen ratios at identifying type 2 vWD.

With regard to functional assays, the biggest change in use was for vWF immunofunctional assays, which increased from about 5% of the total in 2004 to 21% in 2009. There was also a shift in the immunofunctional method, moving from 100% manual enzyme immunoassays in 2004 and 2005 to 50% automated latex immunoassay by 2009. Preston12 reported an increase in the use of immunofunctional methods in the UK NEQAS survey but found that different results could be obtained on individual samples using different functional assays. We found a similar pattern in our results on samples from patients with type 2 vWD. In 1 of 3 samples, the ristocetin cofactor showed an average ratio to vWF antigen of less than 0.6, whereas in a different sample, the immunofunctional assay showed an average ratio to antigen of less than 0.7. Only the collagen binding/antigen ratio was less than 0.7 in all 3 cases. Early versions of the vWF immunofunctional assay were reported to show poor agreement with vWF ristocetin cofactor assays.1214 The more recent automated latex immunofunctional assay has been reported to show better identification of vWD subtypes and better agreement with glycoprotein Ib–binding enzyme immunoassays.15,16 The accuracy of vWF function/vWF antigen ratios impacts laboratory interpretation and further testing, such as requests for vWF multimer analysis, performed locally or by a reference laboratory.

vWF multimer assays are performed in only about one quarter of NASCOLA participating laboratories, and this number is not increasing. This is higher than the number of laboratories performing multimers in the larger ECAT group, which was about 17%. Overall, the error rate on multimer assays was relatively high, on average 15%, but this error rate was lower than rates reported by ECAT, up to 23% in normal samples and up to 52% in type 1 vWD samples.3

Final (diagnostic) interpretations were provided by only a subset of laboratories participating. In part, this is because not all laboratories provide a complete panel of vWF tests able to differentiate the different subtypes of vWD. Even those providing a complete panel did not always provide a final interpretation. It is interesting, however, that a greater number of laboratories provided final interpretations of vWF subtypes than performed multimer analysis or other testing. Overall, similar error rates were observed for NASCOLA laboratories identifying type 1 vWD (∼28%) compared with the ECAT study (30%).3

Given the imprecision of vWF assays and the error rate of vWF multimer analysis, it is advisable that diagnostic testing be repeated if results do not correlate with clinical findings. The limitations of the current study and factors that may have influenced proficiency test results include the use of lyophilized plasma samples and the adequacy of local assay calibrators and standards. Loss of high-molecular-weight vWF multimers may occur as a consequence of lyophilization and may, therefore, affect the results of functional vWF assays and multimer analysis. Incorrect vWF value assignment of assay calibrators or standards will impact results of vWF antigen and functional assays, regardless of assay sensitivity, precision, and stability.

Determining whether a patient has a genetic disorder, vWD, or simply a low level of vWF that represents a risk factor for bleeding is a complex problem. Bleeding signs and symptoms associated with vWD are not specific and can be similar to those seen in other coagulation factor and platelet disorders. There are many different vWF abnormalities from simply low levels of normal vWF protein to a host of abnormal functional types. Testing for vWD must accurately measure the level of the protein and its function and determine whether the abnormality is likely due to a genetic defect. No single assay can be used to screen for all abnormalities, and a relatively large panel may be needed to detect and classify all the abnormal types that have been identified. Assays used to diagnose vWD are improving, as evidenced by the results obtained with collagen binding assays vs ristocetin cofactor methods. These assays measure different but related vWF functionality. There is much room for improvement, as indicated by the relatively high number of errors on vWF multimer and final interpretations. The latter requires integration of results for vWF function and antigen tests and evaluation of vWF multimer structure and is, therefore, subject to the uncertainty of measurement inherent in each of the contributing assays.

CME/SAM

Upon completion of this activity you will be able to:

  • discuss which assays used to diagnose von Willebrand disease (vWD) have the best and worst precision.

  • discuss which von Willebrand factor activity assay is best at detecting type 2 vWD.

  • analyze the combination of assays that may be useful for diagnosis of vWD in the laboratory.

The ASCP is accredited by the Accreditation Council for Continuing Medical Education to provide continuing medical education for physicians. The ASCP designates this educational activity for a maximum of 1 AMA PRA Category 1 Credit ™ per article. This activity qualifies as an American Board of Pathology Maintenance of Certification Part II Self-Assessment Module.

The authors of this article and the planning committee members and staff have no relevant financial relationships with commercial interests to disclose.

Questions appear on p 975. Exam is located at www.ascp.org/ajcpcme.

References

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
View Abstract