SEARCH

SEARCH BY CITATION

Keywords:

  • ADAMTS-13;
  • hemolytic uremic syndrome;
  • thrombotic microangiopathy;
  • thrombotic thrombocytopenic purpura (TTP);
  • von Willebrand factor;
  • von Willebrand factor cleaving protease

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Participants
  6. Preparation of test plasmas
  7. Common standard
  8. Design of the study
  9. Analyses of data
  10. Results
  11. Methods characteristics
  12. Methods expressing results as continuous variable
  13. Method expressing results as non-continuous variable
  14. Discussion
  15. Conclusion
  16. Addendum: the role of each author
  17. Addendum
  18. Acknowledgements
  19. References

Summary. Background: ADAMTS-13 is a von Willebrand factor (VFW)-cleaving protease. Its congenital or acquired deficiency is associated with thrombotic thrombocytopenic purpura (TTP) and more rarely with the hemolytic uremic syndrome. We report on a survey evaluating 11 methods for ADAMTS-13 measurement performed in different labs. Design: Two plasmas, one normal and one from a patient with familial TTP, were mixed at the co-ordinating center to prepare 6 plasmas with 0%, 10%, 20%, 40%, 80% and 100% ADAMTS-13 levels. Each plasma was aliquoted and assembled into sets of 60 (coded from 1 to 60), each containing 10 copies of the original 6 plasmas. Plasmas were frozen and shipped in dry ice to 10 labs with a common frozen reference plasma. Laboratories were asked to measure ADAMTS-13 with their methods. Results were sent to the coordinating center for statistical analysis. Results: Of the 10 methods performed under static conditions 9 were quantitative and one was semiquantitative. One method performed under flow conditions evaluated the extent of cleavage of endothelial cell-derived ultralarge VWF string-like structures and expressed results as deficient, normal, or borderline. Linearity (expected-vs-observed levels), assessed as the squared correlation coefficient, ranged from 0.98 to 0.39. Reproducibility, expressed as the coefficient of variation for repeated measurements, ranged from < 10% to 83%. The majority of methods were able to discriminate between different ADAMTS-13 levels. The majority were able to detect the plasma with 0% level and some of them to discriminate between 0% and 10%. Overall the best performance was observed for three methods measuring cleaved VWF by ristocetin cofactor, collagen binding, and immunoblotting of degraded multimers of VWF substrate, respectively. The poor interlaboratory agreement of results was hardly affected by the use of the common standard. The method performed under flow conditions identified the plasmas with 0%, 10%, 20% and 40% activity as deficient in 7, 5, 1 and 3 of the 10 replicate measurements. The plasmas with 80% and 100% were identified as normal in all of the 10 replicate measurements. Conclusions: The survey shows varied performance, but supports an optimistic view about the reliability of current methods for ADAMTS-13.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Participants
  6. Preparation of test plasmas
  7. Common standard
  8. Design of the study
  9. Analyses of data
  10. Results
  11. Methods characteristics
  12. Methods expressing results as continuous variable
  13. Method expressing results as non-continuous variable
  14. Discussion
  15. Conclusion
  16. Addendum: the role of each author
  17. Addendum
  18. Acknowledgements
  19. References

ADAMTS-13 (a disintegrin and metalloprotease with thrombospondin type 1 domains) is a plasma metalloprotease [1–4] that cleaves von Willebrand factor (VWF) under physiological conditions at the peptide bond Tyr1605-Met1606 releasing fragments of 176 and 140 kDa [5,>6]. The pathophysiologic significance of this cleavage rests on the ability of uncleaved VWF to promote platelet aggregation under conditions of high shear stress in the circulation [7] and on the observation that congenital or acquired deficiency of ADAMTS-13 is associated with intravascular platelet aggregation manifested clinically as thrombotic thrombocytopenic purpura (TTP) [8–10] and, much more rarely, as the hemolytic uremic syndrome (HUS) [11, 12]. In view of the importance that laboratory investigation may have for the management of patients with these conditions, several assay methods for ADAMTS-13 measurement have been developed [9, 13–20]. Although a recent study has evaluated five of these methods for their diagnostic efficacy in testing plasmas from patients with hereditary and acquired TTP or other conditions [21], little information is presently available about such performance characteristics as reproducibility, discrimination between different ADAMTS-13 levels and results comparability. To cover this gap an international collaborative survey evaluated 11 methods in different laboratories that were asked to measure ADAMTS-13 for the same set of plasmas with different levels of the protease.

Participants

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Participants
  6. Preparation of test plasmas
  7. Common standard
  8. Design of the study
  9. Analyses of data
  10. Results
  11. Methods characteristics
  12. Methods expressing results as continuous variable
  13. Method expressing results as non-continuous variable
  14. Discussion
  15. Conclusion
  16. Addendum: the role of each author
  17. Addendum
  18. Acknowledgements
  19. References

As a preliminary step the literature was searched to identify all the laboratories that had already published methods for ADAMTS-13 measurement. These, and other laboratories that to our knowledge used unpublished methods, were contacted. Eleven laboratories accepted to participate in the survey. One laboratory used two different methods. One laboratory, using a method based on platelets adhesion mediated by test plasmas under flow conditions, chose to withdraw from the survey after the exercise and the statistical analysis were completed.

Preparation of test plasmas

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Participants
  6. Preparation of test plasmas
  7. Common standard
  8. Design of the study
  9. Analyses of data
  10. Results
  11. Methods characteristics
  12. Methods expressing results as continuous variable
  13. Method expressing results as non-continuous variable
  14. Discussion
  15. Conclusion
  16. Addendum: the role of each author
  17. Addendum
  18. Acknowledgements
  19. References

After informed consent, plasma was obtained by plasmapheresis from a 60-year-old woman who had a familial form of chronic recurrent TTP, but was in remission at the time of plasma donation. She and her asymptomatic brother (a sister had died from TTP) had unmeasurable levels of ADAMTS-13 as measured with two different methods [9, 13] and did not have inactivating antibodies. The patient had her first episode in 1967 and until the present time has presented with 10 different episodes of TTP. Genotype analysis was performed by PCR followed by sequence analysis and showed two missense mutations in the ADAMTS-13 gene: a G to A substitution at nucleotide 262 in exon 3 that causes the amino acid substitution Val to Met at codon 88 in the metalloprotease domain [22] and a G to T substitution at nucleotide 3707 in exon 27, not previously reported, that causes the amino acid substitution Gly to Val at codon 1239 in the second CUB (C1r/C1s, Urinary EGF and Bone morphogenetic protein) domain. The plasma was collected into a solution of citrate phosphate dextrose (CPD) (pH 5.6) (ratio of blood: anticoagulant, 8 : 1). Suitable aliquots of citrated (0.32%) plasma samples from healthy donors were concurrently pooled to prepare a normal plasma. The ADAMTS-13-deficient and normal plasmas were stored frozen at −70 °C until preparation of the test plasmas which took place at the Milan co-ordinating laboratory within 2 months after collection. On the day of preparation the deficient and normal plasmas were thawed at 37 °C and mixed to prepare six plasmas coded A through F. Assuming that the deficient and the normal plasmas had 0% and 100% ADAMTS-13 levels, respectively, the relative proportions of the two were arranged to obtain 100% (plasma A), 80% (plasma B), 40% (plasma C), 20% (plasma D), 10% (plasma E) and 0% (plasma F) ADAMTS-13 levels. Each of these plasmas was aliquoted in plastic tubes which were then assembled into sets of 60, each containing 10 copies of the six original plasmas. Plasmas in each set were eventually coded from 1 to 60 and stored frozen at −70 °C until they were shipped in dry ice to the participants.

Design of the study

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Participants
  6. Preparation of test plasmas
  7. Common standard
  8. Design of the study
  9. Analyses of data
  10. Results
  11. Methods characteristics
  12. Methods expressing results as continuous variable
  13. Method expressing results as non-continuous variable
  14. Discussion
  15. Conclusion
  16. Addendum: the role of each author
  17. Addendum
  18. Acknowledgements
  19. References

Participants were provided with sets of test plasmas, suitable numbers of aliquots of the common standard and a detailed testing protocol. They were asked to measure ADAMTS-13 in the test plasmas with their own methods performed according to the local procedures (i.e. numbers of replicates and/or dilutions for each sample) and standards. The whole set of measurements were split into 10 different working sessions, each including suitable dilutions of the local and common standards and the six test plasmas. On each session plasmas were tested in ascending consecutive order, i.e. plasmas 1–6 on the first session, through plasmas 55–60 on the last session. On each session the method had to be calibrated by a new set of local and common standards. Results for each sample were entered in the data collection forms as ADAMTS-13 levels calculated as percentage of the local and the common standards. Participants were unaware of the procedure adopted to prepare and assemble test plasmas, so that at the end of the 10 working sessions they performed in blind 10 replicate measurements for each of the six test plasmas. Methods 8 and 11 deviated from the protocol for sample cohort testing because of local practical reasons, and the samples were analyzed consecutively in more than 10 sessions.

Analyses of data

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Participants
  6. Preparation of test plasmas
  7. Common standard
  8. Design of the study
  9. Analyses of data
  10. Results
  11. Methods characteristics
  12. Methods expressing results as continuous variable
  13. Method expressing results as non-continuous variable
  14. Discussion
  15. Conclusion
  16. Addendum: the role of each author
  17. Addendum
  18. Acknowledgements
  19. References

Results were decoded and analyzed at the Milan co-ordinating laboratory. Results for all methods that expressed ADAMTS-13 levels as a continuous variable were used to evaluate (i) the linearity of expected vs. observed ADAMTS-13 levels; (ii) the interassay reproducibility for each method; (iii) the ability of each method to discriminate between different ADAMTS-13 levels; (iv) the between-method comparability of results and (v) the influence of local and common standards on the between-method comparability of results. The linearity of observed vs. expected ADAMTS-13 levels was evaluated by calculating the coefficient of correlation expressed as r2. The interassay reproducibility was evaluated by calculating the mean and standard deviation for the replicate measurements obtained on each test plasma for each method. Reproducibility was expressed as coefficient of variation (CV), calculated as the ratio between the standard deviation and the mean, multiplied by 100. The ability to discriminate between different levels of ADAMTS-13 was evaluated by testing for significant differences between the mean ADAMTS-13 levels of the replicate measurements obtained with each method for the test plasmas (one-way analysis of variance). The F-value generated by this statistical analysis for each method was taken as an index of discrimination to rank the methods: the higher the F-value, the better the discrimination. The influence of different standards on the between-method comparability of results was evaluated by comparing the CV values for each of the test plasmas as determined with ADAMTS-13 levels calculated with each method according to the local and the common standard supplied by the coordinating laboratory. Results which were below the detection limits were omitted from the above analyses.

Methods characteristics

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Participants
  6. Preparation of test plasmas
  7. Common standard
  8. Design of the study
  9. Analyses of data
  10. Results
  11. Methods characteristics
  12. Methods expressing results as continuous variable
  13. Method expressing results as non-continuous variable
  14. Discussion
  15. Conclusion
  16. Addendum: the role of each author
  17. Addendum
  18. Acknowledgements
  19. References

Table 1 shows that 10 methods were carried out under static conditions and used recombinant, plasma-purified VWF, endogenous plasma VWF, or ultralarge VWF as substrate, mild denaturizing conditions (urea), Ba2+ to promote ADAMTS−13 interaction with VWF, and a variety of systems to measure cleaved VWF: residual collagen binding activity (five methods), residual ristocetin cofactor activity (one method), quantitative immunoblotting of degraded multimers of VWF substrate (one method), immunoblotting of unreduced VWF multimers (one method), immunoradiometric assay (IRMA) with monoclonal antibodies to the C- and N-terminal VWF subunit (one method), SDS-PAGE and quantitative immunoblotting of a recombinant VWF polypeptide including A1-A2-A3 domains (one method). One of the methods (method 7) using residual collagen binding activity is a screening assay considered by the authors to be semiquantitative; it discriminates between normal and abnormal protease activity using a cut-off of =78% for normal protease activity. One method (method 8) carried out under flow conditions evaluated the extent of cleavage of endothelial cell-derived ultralarge VWF string-like structures with attached platelets under fluid shear stress. This method did not express ADAMTS-13 results as a continuous variable. Because of this, results using method 8 were not included in the statistical analysis and are described separately.

Table 1.  Main characteristics of the investigated methods
MethodSubstrateAssay conditionsMeasurement of cleaved VWFReference
 1Purified (plasma-derived) VWFBa2+ +ureaSDS agarose electrophoresis and immunoblotting13
Static conditions
 2Recombinant VWFBa2+ +ureaCollagen binding14
Static conditions
 3Recombinant VWFBa2+ +ureaCollagen binding14
Static conditions
 4Purified VWFBa2+ +ureaCollagen binding14
Static conditions
 5Purified VWFBa2+ +ureaRistocetin cofactor activity15
Static conditions
 6Recombinant VWFBa2+ +ureaIRMA16
Static conditions
 7Patient citrated plasmaBa2+ +ureaCollagen binding (semiquantitative)17
Static conditions
 8Ultralarge VWF stringsParallel-plate flow chamberQuantitation of cleaved ultralarge strings18
systems
 9Plasma-derived VWFBa2+ +ureaCollagen binding14
Static conditions
10Recombinant VWFBa2+ +ureaSDS-PAGE and Western Blot using MoAb to VWF-A312
A1-A2-A3 domainsStatic conditions
11Ultralarge VWF (soluble)Ba2+ +ureaSDS-agarose and immunobloting18
Static conditions

Linearity of expected vs. observed ADAMTS-13 levels. Plots of expected vs. observed ADAMTS-13 levels for test plasmas obtained with methods expressing results as continuous variable are shown in Fig. 1. The linearity may be rated as excellent for methods 5, 4 and 1 (r2 value from 0.98 to 0.97), good for method 2 (r2 value, 0.94). For the other methods r2 values ranged from 0.81 to 0.39.

image

Figure 1. Expected vs. observed ADAMTS-13 levels as measured with 10 methods for test plasmas with expected levels ranging from 100% to 0%. Points for each test plasma represent 10 replicate measurements taken on different working sessions. Bars represent mean values.

Download figure to PowerPoint

Table 2 shows results obtained with each method for 10 replicate measurements performed on test plasma with expected undetectable levels of ADAMTS-13, i.e. the plasma from the patient with familial chronic recurrent TTP. Methods 1, 4, 5 and 6 found ADAMTS-13 levels that were below their detection limit and these values were consistent throughout the whole set of 10 replicate measurements. Method 9, found undetectable levels in all but one occasion. Methods 2, 3, 10 and 11 found low or undetectable levels, but results were less consistent throughout the set of 10 replicate measurements. Method 7 (semiquantitative) found levels ranging from 27% to 36%. Table 2 also shows results obtained with each method for 10 replicate measurements performed on test plasma with an expected 10% ADAMTS-13 level. The majority of the methods found levels that were on average close to the expected 10%. Method 7 (semiquantitative) and method 11 found levels that were on average 53% and 19%, respectively. Methods 2 and 6 found levels that were on average close to the expected 10%, but results were not consistent throughout the set of 10 replicate measurements.

Table 2.  ADAMTS-13 levels (%) as measured by different methods for test plasmas with expected 0% and 10% level
ReplicatesSamplesMethods
1234567*91011
  • *

    Semi-quantitative method.

  • **

    Mean of 9 measurements.

 10%< 35–106< 6< 6.25< 536< 3< 60
 2 < 3=57< 6< 6.25< 534< 390
 3 < 3< 53< 6< 6.25< 531< 3100
 4 < 3< 5< 2.5< 6< 6.25< 535< 31212.5
 5 < 3< 5< 2.5< 6< 6.25< 531< 3106.25
 6 < 3< 58< 6< 6.25< 527< 3126.25
 7 < 3< 5< 2.5< 6< 6.25< 5< 3< 66.25
 8 < 3< 58< 6< 6.25< 5319< 612.5
 9 < 3< 5< 2.5< 6< 6.25< 529< 380
10 < 35–109< 6< 6.25< 5< 380–6.25
 110%12.51151111< 570111112.5
 2 15147149< 55711150
 3 12< 58169145172012.5
 4 12< 510111184314159.4
 5 15< 5498< 550151512.5
 6 15< 5687< 55112189.4
 7 1012< 2.589< 562025
 8 7126109< 557< 3750
 9 10< 5810874771150
10 1510111311< 59129.4
Mean 127**1195310**1419

Reproducibility. The coefficient of variation (CV) for each test plasma and method shown in Fig. 2 indicate that the most precise methods were 7 and 5 (average CV less than 10%). Intermediate precision was achieved by methods 4, 2 and 1 (average CV between 12% and 15%), whereas methods 9, 3, 10, 11 and 6 yielded average CV values in the range of 30% and 83%.

image

Figure 2. Inter-assay reproducibility for 10 methods expressed as coefficient of variation (CV%) calculated on 10 replicate measurements for each of the test plasmas with ADAMTS-13 levels ranging from 100% to 10%. CV values for the plasma with an expected 0% level have not been calculated because most of the values reported by each method for this plasma were below the detection limits (see also Table 2).

Download figure to PowerPoint

Discrimination. The overall ability of each method to discriminate between different ADAMTS-13 levels is shown in Fig. 3. High (above 100) F-values, denoting good discrimination, were obtained with methods 5, 1, 4 and 2, intermediate values (between 10 and 100) with methods 7, 9, 10, and 3, low values (below 10) with methods 11 and 6. The ability of each method to discriminate between the two plasmas with expected 10% and 0% ADAMTS-13 levels could not be established by statistical analysis because the values found by most methods for the latter plasma were below detection limits. However, the ability can be subjectively rated as excellent (methods 1, 4 and 5) or good (method 9) (see Table 2 for details).

image

Figure 3. Discrimination between ADAMTS-13 levels achieved by 10 methods. Bars represent F-values (see numbers on the bars for the actual value) as determined by one-way analysis of variance on 10 replicate measurements for the test plasmas. The larger the F-value, the better the discrimination.

Download figure to PowerPoint

Results comparability and influence of standards. When the results for test plasmas with each method were calculated according to the local standard, the interlaboratory overall mean CV value was 40% (range 16% to 55%) (Table 3). When the results were calculated according to the common standard, the interlaboratory CV values were reduced to some extent in some instances, but the overall mean remained high (overall mean 33%, range from 14% to 45%) (Table 3).

Table 3.  ADAMTS-13 levels (%) obtained by participants who completed the whole series of measurements for test plasmas using local and common standards
Test plasmaLocal StandardCommon Standard
NMeanSDCV%NMeanSDCV%
  • *

    Statistical analysis has not been performed because most of the values reported by each method for this plasma were below the detection limits.

A (Expected 100%)611155.951610338.237
B (Expected 80%)68445.65567731.541
C (Expected 40%)63819.95263616.245
D (Expected 20%6194.7256185.229
E (Expected 10%6101.6166101.414
F (Expected 0%)*
Overall mean CV%   40   33

Method expressing results as non-continuous variable

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Participants
  6. Preparation of test plasmas
  7. Common standard
  8. Design of the study
  9. Analyses of data
  10. Results
  11. Methods characteristics
  12. Methods expressing results as continuous variable
  13. Method expressing results as non-continuous variable
  14. Discussion
  15. Conclusion
  16. Addendum: the role of each author
  17. Addendum
  18. Acknowledgements
  19. References

Method 8, which was carried out under flow conditions, evaluated the extent of cleavage of endothelial cell-derived ultralarge VWF string-like structures with attached platelets under fluid shear stress. Results were expressed as percentage of cleaved ultralarge VWF strings after 2 min perfusion. According to the authors, cut off values for the interpretation of results are as follows: results equal or less than 80% should be considered as ADAMTS-13-deficient; results between 80% and 90% as borderline-deficient; and results greater than 90% as normal. As shown in Fig. 4, this method identified the plasmas with expected 0%, 10% and 20% ADAMTS-13 levels as deficient in 7, 5 and one of the 10 replicate measurements. The plasma with an expected 40% ADAMTS-13 level was identified as deficient in three of the 10 replicate measurements. The plasmas with expected 80% and 100% levels were identified as normal in all of the 10 replicate measurements.

image

Figure 4. Method 8 performed under flow conditions. Classification of the six test plasmas over 10 replicate measurements as ADAMTS-13-deficient, -borderline, or –normal, according to the percentage of cleaved ultralarge VWF string-like structures with attached platelets under fluid shear stress.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Participants
  6. Preparation of test plasmas
  7. Common standard
  8. Design of the study
  9. Analyses of data
  10. Results
  11. Methods characteristics
  12. Methods expressing results as continuous variable
  13. Method expressing results as non-continuous variable
  14. Discussion
  15. Conclusion
  16. Addendum: the role of each author
  17. Addendum
  18. Acknowledgements
  19. References

Although the specificity of low ADAMTS-13 levels for the diagnosis of TTP, HUS, or other thrombotic microangiopathies is still debated [23, 24], the management of patients with these conditions may benefit from the laboratory measurement of the protease. Because of the many methods that have been developed [9, 13–20] it is essential to know their relative performance characteristics. We chose to investigate the reproducibility of repeated measurements, the discrimination between different levels and the ability to identify a severe deficiency of ADAMTS-13. The opportunity was also taken to assess the comparability of results obtained by different laboratories and methods, and this was also evaluated in relation to the use of local and common standards. We reasoned that to be meaningful the evaluation had to be based on the use of the same set of well characterized plasmas unknown to the participants. Test plasmas were prepared by mixing varying quantities of plasma from a patient with a severe familial deficiency of ADAMTS-13 without detectable inhibitors into normal pooled plasma. The patient plasma was evaluated blindly before the study in two laboratories using two different methods (described in [9, 13]) that found ADAMTS-13 levels below detection limits (6% and 3%, respectively). Since the plasma donor is a compound heterozygote for two mutations in the exons of ADAMTS-13 gene encoding domains known to be important for the functional activity of the protease (the catalytic and the second CUB domains) [25–27], it is plausible (albeit not yet shown by expression studies) that the plasma used in this study had very low levels of ADAMTS-13 levels. For the purpose of this study we arbitrarily assumed that the plasma was completely ADAMTS-13-deficient, with 0% levels of the protease activity. Accordingly, appropriate mixtures of patient and normal plasmas gave the opportunity to prepare six test plasmas with graded levels of ADAMTS-13 spanning from 0% to 100%. Coding of the plasmas at the co-ordinating laboratory ensured unbiased estimates of the performance characteristics.

Most of the methods displayed a linear relationship of observed vs. expected ADAMTS 13 levels. Exceptions were one method using collagen binding, one using IRMA and one using SDS-agarose and immunoblotting to measure the cleavage of ultralarge VWF by the protease. The best concordance between observed and expected levels was achieved by two methods (4 and 5) using collagen binding and VWF ristocetin cofactor, respectively. These two methods were also among the most reproducible. Other methods that displayed acceptable or good reproducibility (methods 1 and 2) had no common features in terms of assay design (see Table 1). Method 7 displayed good reproducibility, but is semiquantitative. Method 10 was the only one that directly evaluated the enzymatic activity of the protease on a recombinant polypeptide encompassing the A1 through A3 domains of VWF, including the A2 domain peptide bond cleaved by the protease. The performance of this method in terms of concordance between expected and observed levels and ability to detect the patient with severe deficiency was acceptable, but it was no better than that of the other methods that indirectly evaluated VWF cleavage. Among all quantitative assay systems evaluated, method 11 was the only one performed under static conditions that used ultralarge VWF multimers, the likely physiologic substrate for ADAMTS-13. Whereas its detection rate for the plasma with expected 0% ADAMTS-13 levels was high, further improvement is needed in methodological reproducibility and linearity between expected and observed ADAMTS-13 levels.

Discrimination between different ADAMTS-13 levels can be regarded as an important characteristic because it takes into account both reproducibility and accuracy. In this respect, methods 5, 1, 4 and 2 performed better than the others, but again they have little in common in terms of assay design. Another important characteristic of methods measuring ADAMTS-13 is the ability to detect patients with severe deficiency. In this respect, it is reassuring that the majority of quantitative methods found very low ADAMTS-13 levels in the plasma from the patient with severe deficiency. Some of the methods (1, 4, 5 and to some extent 9) were also able to distinguish 0% from 10% ADAMTS-13 levels.

The poor agreement between results obtained with different methods on plasmas with intermediate or normal ADAMTS-13 levels may be a cause for concern. Even though we found that this drawback may be circumvented, at least to some extent, by using a common standard to construct calibration curves, the influence of different standards does not appear to be a major determinant of the interlaboratory variability. In this respect, the variability was high even for those labs performing the methods that used the same system to measure cleaved VWF, i.e. collagen binding. This supports the conclusion that the interlaboratory variability is probably due to the way individual labs set up the same technique rather than to the between-technique differences.

Method 8 was performed under flow conditions and used ultralarge VWF multimers. Results were expressed as normal, borderline, or deficient according to the percentage of strings cleaved after 2 min perfusion. Because of this, we could not evaluate the method for the reproducibility or for the linearity between expected and observed ADAMTS-13 levels. The ability of this method to classify the test plasma with expected 0% level of ADAMTS-13 as deficient was acceptable. Test plasmas with expected 10% or 20% ADAMTS-13 levels were classified as deficient in few occasions and plasmas with expected levels higher than 20% as normal in most occasions. Overall, these results support the conclusion that this method may be reliable in discriminating ADAMTS-13 levels higher or lower than 20%.

Conclusion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Participants
  6. Preparation of test plasmas
  7. Common standard
  8. Design of the study
  9. Analyses of data
  10. Results
  11. Methods characteristics
  12. Methods expressing results as continuous variable
  13. Method expressing results as non-continuous variable
  14. Discussion
  15. Conclusion
  16. Addendum: the role of each author
  17. Addendum
  18. Acknowledgements
  19. References

In summary, this study involving 11 methods each run in an expert laboratory and testing the same set of well characterized plasmas shows that the majority of the currently used methods for ADAMTS-13 measurement did identify plasmas with severe deficiency. Less consistency was observed for plasmas with normal or moderately reduced levels. Agreement between expected and observed levels as well as methodological reproducibility varied between methods. In this respect, some of the methods based on residual collagen binding, the one based on VWF ristocetin cofactor and the one based on immunoblotting of degraded multimers of VWF substrate, displayed the best performance characteristics. Finally, there was a rather poor between-method agreement of results, and this is only in part explained by the use of different standards to construct calibration curves.

Our conclusions based on results obtained with most of the methods that are presently available, confirm and extend those of Studt et al. [21] drawn from a smaller number and spectrum of methods and support a moderately optimistic view [28] about the reliability of current methods that measure ADAMTS-13 levels. The role of assaying ADAMTS-13 in the management of patients with thrombotic microangiopathies remains to be clarified by additional clinical studies, multicenter and multinational in their design and organization.

Addendum: the role of each author

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Participants
  6. Preparation of test plasmas
  7. Common standard
  8. Design of the study
  9. Analyses of data
  10. Results
  11. Methods characteristics
  12. Methods expressing results as continuous variable
  13. Method expressing results as non-continuous variable
  14. Discussion
  15. Conclusion
  16. Addendum: the role of each author
  17. Addendum
  18. Acknowledgements
  19. References
  • 1
    AT: Study design and manuscript preparation
  • 2
    VC: Study design, data collection and statistical analysis
  • 3
    MB: Supervised laboratory testing and reviewed manuscript
  • 4
    UB: Supervised laboratory testing and reviewed manuscript
  • 5
    J-FD: Supervised laboratory testing and reviewed manuscript
  • 6
    KDF: Supervised laboratory testing and reviewed manuscript
  • 7
    MG: Supervised laboratory testing and reviewed manuscript
  • 8
    J-PG: Supervised laboratory testing and reviewed manuscript
  • 9
    JM: Supervised laboratory testing and reviewed manuscript
  • 10
    MER: Supervised laboratory testing and reviewed manuscript
  • 11
    J-DS: Supervised laboratory testing and reviewed manuscript
  • 12
    PLT: Supervised laboratory testing and reviewed manuscript
  • 13
    PMM: Study design and manuscript preparation

Addendum

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Participants
  6. Preparation of test plasmas
  7. Common standard
  8. Design of the study
  9. Analyses of data
  10. Results
  11. Methods characteristics
  12. Methods expressing results as continuous variable
  13. Method expressing results as non-continuous variable
  14. Discussion
  15. Conclusion
  16. Addendum: the role of each author
  17. Addendum
  18. Acknowledgements
  19. References

The following investigators (alphabetic order) participated in the collaborative study:

D Angerhaus/U. Budde (Hamburg, Germany); M. Böhm/I. Scharrer (Frankfurt, Germany); M.A. Budish/K.D. Friedman (Milwaukee, USA); M.T. Canciani/P.M. Mannucci (Milano, Italy); J.-f. Dong/.J. Moake (Houston, USA); M. Galbusera/C. Capoferri/G. Remuzzi (Bergamo, Italy); J. A. Kremer Hovinga/B. Lämmle/J.-D. Studt (Bern, Switzerland); B. Obert/A. Houllier/J.-P. Girma (Paris, France); M. Rick/D. Aronson (Bethesda, USA); K. Varadi/J. Schreiner/M. Billwein/P. Turecek (Vienna, Austria).

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Participants
  6. Preparation of test plasmas
  7. Common standard
  8. Design of the study
  9. Analyses of data
  10. Results
  11. Methods characteristics
  12. Methods expressing results as continuous variable
  13. Method expressing results as non-continuous variable
  14. Discussion
  15. Conclusion
  16. Addendum: the role of each author
  17. Addendum
  18. Acknowledgements
  19. References

The recombinant VWF-A1-A2-A3 and the monoclonal antibody anti-A3 domain used with method 10 were kindly provided to the participating laboratory by Dr Z. M. Ruggeri, The Scripps Research Institute, La Jolla, USA.

The authors wish to thank Dr E. Rossi (Ospedale L. Sacco, Milan, Italy) and Dr P. Turecek (Baxter BioScience, Vienna, Austria) who helped collecting plasma from the patient with familial TTP and shipping the material to the participating laboratories.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Participants
  6. Preparation of test plasmas
  7. Common standard
  8. Design of the study
  9. Analyses of data
  10. Results
  11. Methods characteristics
  12. Methods expressing results as continuous variable
  13. Method expressing results as non-continuous variable
  14. Discussion
  15. Conclusion
  16. Addendum: the role of each author
  17. Addendum
  18. Acknowledgements
  19. References