Coagulation and Transfusion Medicine / von Willebrand Factor Proficiency Testing
von Willebrand Factor Assay Proficiency Testing
The North American Specialized Coagulation Laboratory
Association Experience
Wayne L. Chandler, MD,1 Ellinor I.B. Peerschke, PhD,2 Donna D. Castellone, MS, MT(ASCP)SH,2
and Piet Meijer, PhD,3 on behalf of the NASCOLA Proficiency Testing Committee
CME/SAM
Key Words: von Willebrand disease; von Willebrand factor antigen assay; Proficiency testing
DOI: 10.1309/AJCPH5JK4ONENPAE
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.
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%.
862
862
Am J Clin Pathol 2011;135:862-869
DOI: 10.1309/AJCPH5JK4ONENPAE
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.
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
al5-8 reported in-depth on vWF quality assurance results from
approximately 50 clinical laboratories in the Southern hemisphere, concentrating on vWF functional assays. Major findings
© American Society for Clinical Pathology
Coagulation and Transfusion Medicine / Original Article
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
60
No. of Participants
50
40
30
20
10
0
2004
2005
2006
2007
2008
2009
❚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.
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.
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
© American Society for Clinical Pathology
Am J Clin Pathol 2011;135:862-869
863
DOI: 10.1309/AJCPH5JK4ONENPAE
863
863
Chandler et al / von Willebrand Factor Proficiency Testing
❚Table 1❚
Source of Reagents
Survey Responses per Year*
Assay Type/Method
vWF antigen
Diagnostica Stago (Parsippany, NJ) LIA
Homemade/other
Instrumentation Laboratory (Bedford, MA) LIA
Diagnostica Stago EIA
Corgenix (Broomfield, CO) EIA
Total
vWF ristocetin cofactor
Siemens/Dade Behring (Deerfield, IL)
Helena (Collierville, TN)
Homemade/other
Bio/Data (Horsham, PA)
Chrono-log (Havertown, PA)
Total
vWF immunofunctional
Instrumentation Lab LIA
American Diagnostica (Stamford, CT) EIA
Corgenix EIA
American Biochemical and Pharmaceutical (Epsom, England) EIA
Other
Total
vWF collagen binding
Life Diagnostics (West Chester, PA)
Homemade/other
Technoclone (Vienna, Austria)
Total
2004
2005
2006
2007
2008
2009
Total
105
19
4
8
6
142
124
12
5
14
10
165
144
13
6
16
12
191
161
15
14
4
7
201
151
19
25
4
4
203
139
20
22
4
2
187
824
98
76
50
41
1,089
49
32
28
11
5
125
69
27
26
9
3
134
83
24
22
14
4
147
90
22
17
19
1
149
89
8
15
22
11
145
92
4
7
20
11
134
472
117
115
95
35
834
0
3
0
1
2
6
0
5
4
4
0
13
4
7
6
4
0
21
14
8
6
4
0
32
18
7
6
3
0
34
23
5
6
12
0
46
59
35
28
28
2
152
9
4
0
13
18
3
0
21
15
5
0
20
8
5
3
16
18
6
4
28
15
4
6
25
83
27
13
123
EIA, enzyme immunoassay; LIA, latex immunoassay; vWF, von Willebrand factor.
* Each year, 4 surveys were distributed. The number represents the total number of responses per year for all 4 surveys. The average number of participants per survey is
approximately one fourth of the total number of survey responses.
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).
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
864
864
Am J Clin Pathol 2011;135:862-869
DOI: 10.1309/AJCPH5JK4ONENPAE
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
© American Society for Clinical Pathology
Coagulation and Transfusion Medicine / Original Article
❚Table 2❚
Assay and Ratio Imprecision*
Assay type
vWF Ag
Mean level (U/dL)
CV (%)
vWF ristocetin cofactor
Mean level (U/dL)
CV (%)
vWF immunofunctional
Mean level (U/dL)
CV (%)
vWF collagen binding
Mean level (U/dL)
CV (%)
Ratio
vWF RCo/Ag
Mean
CV (%)
vWF Imm/Ag
Mean
CV (%)
vWF CB/Ag
Mean
CV (%)
Normal Plasma (7 Surveys)
Type 1 vWD (7 Surveys)
Type 2 vWD (3 Surveys)
111 ± 11
11 ± 2
38 ± 10
15 ± 6
27 ± 18
19 ± 7
83 ± 6
20 ± 3
28 ± 7
30 ± 7
17 ± 7
42 ± 5
95 ± 9
15 ± 3
34 ± 5
14 ± 5
20 ± 16
49 ± 33
102 ± 16
21 ± 4
31 ± 9
14 ± 6
15 ± 17
30 ± 19
0.76 ± 0.04
22 ± 4
0.74 ± 0.07
32 ± 8
0.73 ± 0.20
49 ± 10
0.90 ± 0.08
19 ± 8
0.90 ± 0.11
20 ± 8
0.76 ± 0.19
53 ± 40
0.94 ± 0.17
23 ± 2
0.85 ± 0.14
12 ± 5
0.49 ± 0.17
43 ± 26
Ag, antigen; CB, collagen binding; CV, coefficient of variation; Imm, vWF immunofunctional activity; RCo, vWF ristocetin cofactor activity; vWD, von Willebrand disease;
vWF, von Willebrand factor.
* Data shown represent the all-survey mean and CV for vWF assay types. These data were obtained by calculating the mean and CV for all assay results submitted for individual
surveys conducted from 2004 through 2009 and subsequently determining the mean of all survey averages and CVs.
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.
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
© American Society for Clinical Pathology
Am J Clin Pathol 2011;135:862-869
865
DOI: 10.1309/AJCPH5JK4ONENPAE
865
865
Chandler et al / von Willebrand Factor Proficiency Testing
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
866
866
Am J Clin Pathol 2011;135:862-869
DOI: 10.1309/AJCPH5JK4ONENPAE
❚Table 3❚
Major Variables Affecting vWF Test Results*
No. (%) of
Responding
Laboratories
Specimen handling
Placed at room temperature before analysis
Placed on ice before analysis
vWF standard/calibrator value assignment
Assigned by manufacturer
Assayed against World Health Organization
reference plasma
Assayed against previous laboratory standard
Standard curve preparation
New curve prepared with each run
New curve prepared when quality control fails
23 (70)
9 (27)
26 (79)
6 (18)
2 (6)
21 (64)
11 (32)
vWF, von Willebrand factor.
* Results are based on a questionnaire distributed to 33 North American Specialized
Coagulation Laboratory Association member laboratories in 2004.
❚Table 4❚
Result Interpretation*
Assay Type/Interpretation
vWF antigen
Normal
Abnormal
vWF ristocetin cofactor
Normal
Abnormal
vWF immunofunctional
Normal
Abnormal
vWF collagen binding
Normal
Abnormal
Normal Plasma Type 1 vWD Type 2 vWD
(7 Surveys)
(7 Surveys) (3 Surveys)
325
0
38
272
11
130
246
5
10
226
1
106
47
0
1
42
0
22
35
0
2
32
0
19
vWD, von Willebrand disease; vWF, von Willebrand factor.
* Values represent the number of individual participant responses submitted for the
total number of surveys shown.
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 highmolecular-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.
© American Society for Clinical Pathology
Coagulation and Transfusion Medicine / Original Article
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
❚Table 5❚
vWF Multimer Analysis*
No. (%) of Responding
Laboratories
Percentage of samples submitted for vWF
multimer analysis
<10
10-20
21-40
>40
Condition triggering multimer analysis
vWF activity/antigen ratio <0.5
vWF activity/antigen ratio <0.7
vWF activity/antigen ratio not specified
Other (new patient, low vWF level,
physician request, contract)
13 (39)
10 (30)
7 (21)
3 (10)
5 (15)
4 (12)
9 (27)
15 (45)
vWF, von Willebrand factor.
* Results are based on a questionnaire distributed to 33 North American Specialized
Coagulation Laboratory Association member laboratories in 2004.
❚Table 6❚
Final (Diagnostic) Interpretation*
Interpretation
Normal
vWD type
1
2A
2B
2M
2N
3
Normal Plasma
(7 Surveys)
Type 1 vWD
(7 Surveys)
Type 2 vWD
(3 Surveys)
239
24
1
1
3
3
0
0
0
125
14
5
3
2
0
25
14
11
2
2
8
vWD, von Willebrand disease.
* Values represent the number of individual participant responses for the total number
of surveys shown.
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.
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
© American Society for Clinical Pathology
Am J Clin Pathol 2011;135:862-869
867
DOI: 10.1309/AJCPH5JK4ONENPAE
867
867
Chandler et al / von Willebrand Factor Proficiency Testing
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.2-4 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,5-8 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.12-14 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.
868
868
Am J Clin Pathol 2011;135:862-869
DOI: 10.1309/AJCPH5JK4ONENPAE
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
© American Society for Clinical Pathology
Coagulation and Transfusion Medicine / Original Article
of vWF multimer structure and is, therefore, subject to the
uncertainty of measurement inherent in each of the contributing assays.
From the 1Department of Laboratory Medicine, University of
Washington, Seattle; 2Department of Pathology, the Mount Sinai
School of Medicine, New York, NY; and 3ECAT Foundation,
Leiden, the Netherlands.
Address reprint requests to Dr Chandler: Dept of Laboratory
Medicine, Box 357110, University of Washington, Seattle, WA
98195.
References
1. Sadler JE. Low von Willebrand factor: sometimes a risk factor
and sometimes a disease. Hematology Am Soc Hematol Educ
Program. 2009:106-112.
2. Hayes TE, Brandt JT, Chandler WL, et al. External peer
review quality assurance testing in von Willebrand disease:
the recent experience of the United States College of
American Pathologists proficiency testing program. Semin
Thromb Hemost. 2006;32:499-504.
3. Meijer P, Haverkate F. An external quality assessment
program for von Willebrand factor laboratory analysis:
an overview from the European Concerted Action on
Thrombosis and Disabilities Foundation. Semin Thromb
Hemost. 2006;32:485-491.
4. Kitchen S, Jennings I, Woods TA, et al. Laboratory tests for
measurement of von Willebrand factor show poor agreement
among different centers: results from the United Kingdom
National External Quality Assessment Scheme for Blood
Coagulation. Semin Thromb Hemost. 2006;32:492-498.
5. Favaloro EJ, Bonar R, Meiring M, et al. 2B or not 2B?
disparate discrimination of functional vWF discordance using
different assay panels or methodologies may lead to success
or failure in the early identification of type 2B vWD. Thromb
Haemost. 2007;98:346-358.
6. Favaloro EJ, Bonar R, Kershaw G, et al. Reducing errors in
identification of von Willebrand disease: the experience
of the Royal College of Pathologists of Australasia Quality
Assurance Program. Semin Thromb Hemost. 2006;32:505-513.
7. Favaloro EJ, Bonar R, Kershaw G, et al. Laboratory diagnosis
of von Willebrand disorder: use of multiple functional assays
reduces diagnostic error rates. Lab Hematol. 2005;11:91-97.
8. Favaloro EJ, Bonar R, Kershaw G, et al. Laboratory diagnosis
of von Willebrand’s disorder: quality and diagnostic
improvements driven by peer review in a multilaboratory test
process. Haemophilia. 2004;10:232-242.
9. Peerschke EI, Castellone DD, Ledford-Kraemer M, et al.
Laboratory assessment of factor VIII inhibitor titer: the North
American Specialized Coagulation Laboratory Association
experience. Am J Clin Pathol. 2009;131:552-558.
10. Van Cott EM, Ledford-Kraemer M, Meijer P, et al. Protein
S assays: an analysis of North American Specialized
Coagulation Laboratory Association proficiency testing. Am J
Clin Pathol. 2005;123:778-785.
11. Favaloro EJ, Bonar R, Sioufi J, et al. Laboratory diagnosis of
von Willebrand disorder: current practice in the southern
hemisphere. Am J Clin Pathol. 2003;119:882-893.
12. Preston FE. Assays for von Willebrand factor functional
activity: a UK NEQAS survey. National External Quality
Assessment Scheme [published correction appears in Thromb
Haemost. 1999;81:478]. Thromb Haemost. 1998;80:863.
13. Federici AB, Canciani MT, Forza I, et al. Ristocetin cofactor
and collagen binding activities normalized to antigen levels
for a rapid diagnosis of type 2 von Willebrand disease: single
center comparison of four different assays [letter]. Thromb
Haemost. 2000;84:1127-1128.
14. Nitu-Whalley IC, Riddell A, Lee CA, et al. Identification of
type 2 von Willebrand disease in previously diagnosed type
1 patients: a reappraisal using phenotypes, genotypes and
molecular modelling. Thromb Haemost. 2000;84:998-1004.
15. Piñol M, Sales M, Costa M, et al. Evaluation of a new
turbidimetric assay for von Willebrand factor activity useful
in the general screening of von Willebrand disease [letter].
Haematologica. 2007;92:712-713.
16. Sucker C, Senft B, Scharf RE, et al. Determination of von
Willebrand factor activity: evaluation of the HaemosIL assay
in comparison with established procedures. Clin Appl Thromb
Hemost. 2006;12:305-310.
© American Society for Clinical Pathology
Am J Clin Pathol 2011;135:862-869
869
DOI: 10.1309/AJCPH5JK4ONENPAE
869
869