Thrombosis Research 134 (2014) 1074–1080

Contents lists available at ScienceDirect

Thrombosis Research journal homepage: www.elsevier.com/locate/thromres

Regular Article

Evaluation of a chromogenic commercial assay using VWF-73 peptide for ADAMTS13 activity measurement Bérangère Joly a, Alain Stepanian b, David Hajage c, Sandrine Thouzeau a,d, Sophie Capdenat a, Paul Coppo d,e, Agnès Veyradier a,d,⁎ a

Service d’Hématologie biologique, Hôpital Antoine Béclère, Assistance Publique-Hôpitaux de Paris, Clamart, Université Paris 11, France Service d’Hématologie biologique, Hôpital Louis Mourier, Assistance Publique-Hôpitaux de Paris, Colombes, Université Paris 7, France Département d’épidémiologie et de recherche clinique, Hôpital Louis Mourier, Assistance Publique-Hôpitaux de Paris, Colombes, Université Paris 7, France d French Reference Center for Thrombotic Microangiopathies (CNR-MAT), Hôpital Saint Antoine, Assistance Publique-Hôpitaux de Paris, Paris, Université Paris 6, France e Service d’Hématologie, Hôpital Saint Antoine, Assistance Publique-Hôpitaux de Paris, Paris, France b c

a r t i c l e

i n f o

Article history: Received 7 June 2014 Received in revised form 14 August 2014 Accepted 4 September 2014 Available online 16 September 2014 Keywords: ADAMTS13 ADAMTS13 activity FRETS-VWF73 Chr-VWF73 Thrombotic thrombocytopenic purpura

a b s t r a c t Introduction: Thrombotic thrombocytopenic purpura (TTP) is a thrombotic microangiopathy (TMA), related to a severe functional deficiency of ADAMTS13 activity (b 10% of normal). ADAMTS13 activity is thus crucial to confirm the clinical suspicion of TTP, to distinguish it from other TMAs, and to perform the follow-up of TTP patients. Material and methods: We compared the performance of the commercial chromogenic assay Technozym® ADAMTS13 Activity ELISA (chromogenic VWF73 substrate, Chr-VWF73, Technoclone, Vienna, Austria), to that of our in-house FRETS-VWF73 used as reference method. A large group of 247 subjects (30 healthy volunteers and 217 patients with miscellaneaous TMAs) was studied. Results: The lower limit of detection of the Chr-VWF73 was 3%, which is well adapted to the clinically relevant threshold for TTP diagnosis (10%). Our results showed a reasonable agreement between FRETS-VWF73 and Chr-VWF73 assays to distinguish samples with an ADAMTS13 activity b10% from those with an ADAMTS13 activity N 10%. However, Chr-VWF73 assay provided false negative results in ~12% of acute TTP patients. Inversely, the Chr-VWF73 assay globally underestimated ADAMTS13 activity in detectable values ranging from 11 to 100% (with a great variability compared to FRETS-VWF73), which may be a concern for the follow-up of TTP patients in remission. Conclusion: In-house assays developed and performed by expert laboratories remain the reference methods that should be used without limitation to control values provided by commercial assays when needed. Also, the development of an international reference preparation will be crucial to improve standardization. © 2014 Elsevier Ltd. All rights reserved.

Introduction Thrombotic microangiopathies (TMAs) are rare diseases characterized by a profound thrombocytopenia, erythrocyte fragmentation and organ failure of variable severity. Thrombotic thrombocytopenic purpura Abbreviations: ADAMTS13, A disintegrin and metalloprotease with thrombospondin type 1 repeats, member 13; CV, Coefficient of variation; FRETS-VWF73, Fluorescence resonance energy transfer-VWF73; HELLP, Hemolysis elevated liver enzymes low platelet count; HUS, Hemolytic uremic syndrome; IH, In-house; LLD, lower limit of detection; NPP, Normal pool plasma; NPV, Negative predictive value; PPV, Positive predictive value; QC, Quality control; SD, Standard deviation; TMA, Thrombotic microangiopathy; TTP, Thrombotic thrombocytopenic purpura; USS, Upshaw-Schulman syndrome; VWF, von Willebrand factor. ⁎ Corresponding author at: Service d’Hématologie biologique, Hôpital Antoine Béclère, 157 rue de la Porte-de-Trivaux, 92140 Clamart, France. Tel.: + 33 1 45 37 43 05; fax: +33 1 45 37 40 95. E-mail address: [email protected] (A. Veyradier).

http://dx.doi.org/10.1016/j.thromres.2014.09.006 0049-3848/© 2014 Elsevier Ltd. All rights reserved.

(TTP) is a severe form of TMA [1], which specificity is to be associated with a severe functional deficiency of the von Willebrand factor (VWF)-cleaving protease, ADAMTS13 (A Disintegrin and Metalloprotease with Thrombospondin type 1 repeats, member 13) [1]. A severe functional deficiency of ADAMTS13 activity (defined by an undetectable activity, lower than 10% of normal and corresponding to the clinically relevant threshold) is responsible for the accumulation of unfolded and ultralarge hyper-adhesive multimers of VWF in plasma, inducing the spontaneous formation of VWF-rich platelet thrombi in the microcirculation [2]. These thrombi are responsible for a consumption thrombocytopenia and a mechanical hemolytic anemia, that may be associated with a multivisceral ischemia; these latter symptoms define the acute phase of TTP (contrasting with remission phases occurring after successful treatment, during which the hematologic bicytopenia and the ischemia symptoms disappear). Severe ADAMTS13 deficiency may be either very rarely inherited, resulting from biallelic mutations

B. Joly et al. / Thrombosis Research 134 (2014) 1074–1080

of the encoding gene [1,3], or most frequently acquired, then related to the presence of polyclonal autoantibodies against ADAMTS13 [4–6]. Miscellaneous TMAs, like the hemolytic uremic syndrome (HUS) [2,7] or the “Hemolysis Elevated Liver enzymes Low Platelet count” (HELLP) syndrome [8], are usually associated with detectable ADAMTS13 levels. Consequently, the diagnosis of acute TTP is defined by the combination of the clinical symptoms previously described and an undetectable ADAMTS13 activity. Thereby, the measurement of ADAMTS13 activity in plasma is crucial to confirm the clinical suspicion of acute TTP, to distinguish it from other TMAs in order to support the prescription of specific therapies (i.e. rituximab - Mabthera®, Roche in acquired TTP, or eculizumab - Soliris®, Alexion - in atypical HUS). When patients reach clinical remission (defined by the disappearance of clinical symptoms after appropriate treatment), ADAMTS13 activity measurement is also useful during their follow-up: firstly (short-term follow-up), to check the occurrence of biological remission defined by a progressive normalization of ADAMTS13 activity expected after a few weeks/months (except in hereditary forms); secondly (longterm follow-up), once ADAMTS13 activity has normalized, to detect a potential decrease of ADAMTS13 activity that may be an early sign of TTP relapse (as biological relapse usually preceeds clinical relapse) [6]. This last point is crucial to identify TTP patients at high risk of relapse in order to make them beneficiate from a preemptive Rituximab infusion. [9]. In order to measure ADAMTS13 activity in plasma, an exogenous substrate (VWF full-length or short peptides of VWF including the cleavage site by ADAMTS13) is subjected to degradation by ADAMTS13 contained in the tested plasma. In a second step, the degradation products of VWF are measured by different available methods [10]. These assays vary in terms of substrate, reagent composition, reaction times and calibrators [11–14]; therefore, a standardization remains to be performed. However, today, historical methods using full-length VWF [12] and mainly the fluorescence resonance energy transfer (FRETS)-VWF73 assays developed by Kokame and coll. in 2005 [14] are considered as reference methods by the international medical community devoted to ADAMTS13 and TTP [15–17]. Indeed, the FRETS-VWF73 has been validated on large series of patients worldwide and many related studies (including those of our group) have been published in the international literature [16,17] since the princeps publications of the New England Journal of Medicine in 1998 [2,4]. Since 2006, our laboratory has been qualified as the national reference laboratory for ADAMTS13 investigation of the French Reference Center for TMA certified by the French Ministry of Health (National Plan for Rare diseases, Arrêté du 12 juillet 2006 portant labellisation des centres de reference pour une maladie rare, Journal Officiel de la République Française NOR: SANH0622910A). Our in house (IH) FRETS-VWF73 adapted from Kokame and coll. [14] has been the French reference method for ADAMTS13 activity measurement since 2006; we previously showed that our IH FRETS-VWF73 assay was correlated to our IH immunoradiometric/ELISA assay based on full-length VWF developed in 1999 [7,13]. Since the early 2010’s, some commercial kits for ADAMTS13 activity using small peptides of VWF have been developed and distributed on the market by manufacturers, before any evaluation on large series of TMA patients and normal subjects have been made by expert laboratories. These evaluations led by expert laboratories are crucial because, in most countries, non expert laboratories are in demand to use these kits mostly to be able to provide a first rapid answer to their collaborating physicians to documente a suspicion of acute TTP. Recently, homemade assays i.e. the FRETS-VWF73 assay [14] and the collagen binding assay [18,19] have been compared to commercial kits: Mackie et al. [20] compared both the home-made collagen binding assay and the FRETS-VWF73 assay to the Actifluor® assay (FRETS-VWF86, American Diagnostica Inc., Stamford, CT, USA) and to the Technozym® ADAMTS13 activity ELISA (Chr-VWF73, Technoclone, Vienna, Austria) in 159 plasma samples including 38 healthy volunteers and 121 TMA patients.

1075

Moreover, our group led by Thouzeau et al. [21] compared our IH FRETS-VWF73 assay to the ATS-13® Activity Assay (Gen-Probe Inc., Waukesha, WI, USA) in 209 plasma samples including 30 healthy volunteers and 179 TMA patients. All these three commercial assays for ADAMTS13 activity measurement showed a global correct agreement with the reference methods but unexplained discrepancies were present for some samples and the biological threshold did not perfectly match the clinically relevant threshold [20,21]. Among these three commercial kits, the Technozym® ADAMTS13 activity ELISA (Chr-VWF73, Technoclone, Vienna, Austria) is the only one to use a chromogenic GST-VWF73 substrate which makes it more accessible to hospital hematology laboratories when compared to fluorimetry. In that regard, the aim of the current study was to investigate the Technozym® ADAMTS13 activity ELISA (Chr-VWF73, evaluated method) compared to our IH FRETS-VWF73 used as reference method [13,22], in a large cohort of TMA patients.

Materials and Methods Blood Samples We studied 247 citrated plasma samples, consisting in 30 healthy volunteers and 217 patients with miscellaneaous TMAs from the French Reference Center for Thrombotic MicroAngiopathies biobank (study approved by the Ethics Commitee of Hospital Pitié-Salpêtrière and Hospital Saint-Antoine, Paris, France). Among these 217 patients, diagnoses were previously made, using the measurement of ADAMTS13 activity by our IH FRETS-VWF73 assay (reference method), the titration of anti-ADAMTS13 IgG and ADAMTS13 gene sequencing [22] (Table 1). The samples included: (i) 19 congenital TTP (Upshaw-Schulman syndrome, USS) defined as undetectable ADAMTS13 activity by IH FRETSVWF73, no detectable anti-ADAMTS13 IgG and identified ADAMTS13 gene mutations; (ii) 75 acquired TTP in acute phase (defined by an ADAMTS13 activity b 10% with the reference method, in the presence of autoantibodies against ADAMTS13), including 34 idiopathic and 41 secondary forms; (iii) 52 acquired TTP in remission (defined by an ADAMTS13 activity ranging from 11% to 100% with the reference method and no clinical symptom); (iv) 71 other TMAs like HUS, HELLP syndrome, TMA associated with either sepsis, cancer or auto-immune diseases; (v) 30 healthy volunteers. In total, we studied 94 samples with ADAMTS13 activity b 10%, and 153 samples with a detectable ADAMTS13 activity, either ranging from 11% to 19% (n = 15), from 20% to 49% (n = 53), or from 50% to 100% (n =85) (Table 1).

Table 1 Distribution of tested plasmas, according to diagnosis (1a) and ADAMTS13 activity using in-house FRETS-VWF73 (1b). 1a. Diagnosis

n patients

Normal subjects Congenital TTP (USS) Acquired TTP – idiopathic form Acquired TTP – secondary form Remission Other (HUS, neoplasia, sepsis, dysimmunity) Total

30 19 34 41 52 71 247

1b. ADAMTS13 activity using in-house FRETS-VWF73 assay

n patients

b10% 10 - 19% 20 - 49% 50 - 100% Total

94 15 53 85 247

HUS: Haemolytic and uremic syndrome; TTP: thrombotic thrombocytopenic purpura; USS: Upshaw-Schulman Syndrome.

1076

B. Joly et al. / Thrombosis Research 134 (2014) 1074–1080

ADAMTS13 Activity Assays In-house FRETS-VWF73 Assay Our IH FRETS-VWF73 assay [13,22] is adapted from the original test from Kokame et al. [14]: a normal pool plasma (NPP), composed of 50 healthy volunteers plasmas, was used for the calibration curve with an arbitrary ADAMTS13 activity of 100%. Tested samples were diluted in home-made diluent (about 1/8) and mixed with a recombinant FRETS-VWF73 peptide based on the VWF-A2 domain (Peptide Institute Inc., Osaka Japan) in the well of a microplate (overall dilution 1/16), according to the manufacturer’s instructions. We used as home-made internal quality control (QC): one USS plasma (ADAMTS13 activity b 10%), undiluted and half-diluted NPP (100% and 50% ADAMTS13 activity, respectively). Fluorescence was measured at 30 minutes (350 nm excitation, 450 nm emission) by an endpoint method using a Fluoroskan Ascent fluorimeter (Thermo Scientific, Saint Herblain, France). IH FRETS-VWF73 assay is defined as our reference method. Our normal range (previously defined from 100 healthy volunteers) is 50-100%. The lower limit of detection (LLD) of the test, previously defined as the lowest ADAMTS13 activity detectable when compared to buffer blank, is 10%; the upper limit of detection is 100% [13,14,22]. Technozym® Activity ELISA (Chr-VWF73) (Technoclone, Vienna, Austria) Measurement of ADAMTS13 activity in human plasma was also performed using a chromogenic assay, Technozym® ADAMTS13 activity ELISA (Chr-VWF73, Technoclone, Vienna, Austria), according to the manufacturer’s instructions. Samples were diluted 1/31 in assay buffer. Six lyophilized calibrators, positive and negative control plasmas were provided. We also used as home-made internal QC: one TTP patient plasma (ADAMTS13 activity b 10%), undiluted and half-diluted NPP (100% and 50% ADAMTS13 activity, respectively). This assay used a chromogenic GST-VWF73 substrate, a monoclonal anti-GST antibody coated microplate, a HRP conjugated monoclonal antibody (450/ 620 nm optical density) directed against the cleavage site of VWF73 substrate [22]. The manufacturer indicated that the normal range determined from 40 healthy volunteers was 40 to 130% and the LLD 0.2%. However, as recommended by the manufacturer, we determined our own normal range, found to be 50-100% (from 100 healthy volunteers) and also our own LLD (using serial dilutions of NPP between 0 and 25%), found to be 3%. Any value actually measured between 3% and 10% with the Chr-VWF73 assay was affected b10%, in agreement with the 10% clinically relevant threshold. Also, in our hands, the upper limit of detection was 100%. Both assays were performed in parallel on the same stored sample (IH FRETS-VWF73 was freshly performed for the current study although it had been historically performed in all samples) and all samples were tested in duplicate with both assays. Repeats were systematically performed on outlayers samples. Technozym® ADAMTS13 Inhibitor ELISA Anti-ADAMTS13 IgG were measured by ELISA using the Technozym® ADAMTS13-INH commercial kit (Technoclone, Vienna, Austria), according to the manufacturer’s instructions. The positivity threshold was 15 IU/ml, as indicated by the manufacturer. Statistical Analysis Analyses were done using the SAS 9.1 software package (SAS Institute, Cary, NC). IH FRETS-VWF73 assay (reference method) and the Chr-VWF73 assay (evaluated method) were compared with a different methodology as a function of ADAMTS13 activity range. For undetectable ADAMTS13 activity values (b10% with the FRETSVWF73 assay), the agreement between both assays was assessed by the use of the Cohen Kappa coefficient. Kappa coefficient is usually

interpretated as good if ranging between 0.61 and 0.80 and as very good if ranging between 0.81 and 1. For detectable ADAMTS13 activity values (from 11% to 100% with the FRETS-VWF73 assay), the comparison between both assays was performed using the Spearman’s rank correlation coefficient (using a scattered diagram, Fig. 1) and also using limits of agreement calculated with the Bland Altman method (using a Bland Altman plot, Fig. 2). TTP patients were defined on clinical presentation of TTP associated with an ADAMTS13 activity b 10% with the IH FRETS-VWF73 assay. True positive patients were defined as TTP patients whose ADAMTS13 activity was b 10% with the Chr-VWF73 assay. False negative patients were defined as TTP patients whose ADAMTS13 activity was detectable with the Chr-VWF73 assay. True negative patients were defined as non-TTP patients whose ADAMTS13 activity was detectable with the Chr-VWF73 assay. False positive patients were defined as non-TTP patients whose ADAMTS13 activity was b 10% with the Chr-VWF73 assay. These 4 categories of patients may be visualized on the scattered diagram expressing ADAMTS13 activity by Chr-VWF73 as a function of ADAMTS13 activity by IH FRETS-VWF73 (Fig. 1). Sensitivity was calculated as the percentage of correctly detected TTP patients by Chr-VWF73 assay, using the formula: sensitivity (%) =(true positive X 100)/(true positive + false negative). Specificity was calculated as the percentage of correctly detected non-TTP patients by Chr-VWF73 assay, using the formula: specificity (%) = (true negative X 100)/(true negative + false positive). Positive predictive value (PPV) was estimated as the percentage of TTP patients among the subjects whose ADAMTS13 activity was b 10%, using the formula: PPV = (true positive X 100)/(true positive + false positive). Negative predictive value (NPV) was estimated as the percentage of non-TTP patients among the subjects whose ADAMTS13 activity was N 10%, using the formula: NPV = (true negative X 100)/(true negative + false negative). Results Chr-VWF73 Assay Analytic Features Limits of Detection In our hands, the LLD of the Chr-VWF73 assay was 3% (in contrast to the 0.2% value indicated by the manufacturer), while this limit was previously defined as 10% with our IH FRETS-VWF73 assay. However, for both methods, the clinically relevant threshold (b10%) was the one considered for all samples which ADAMTS13 activity was measured lower than 10% (as values measured between 3% and 10% with the ChrVWF73 assay have no clinical relevance for the biological diagnosis of TTP). The upper limit of detection of both assays was 100%. Inter-assays and Intra-assay Reproducibility Ten assay runs were performed. The manufacturer’s own lyophilized calibrators were used for each assay, with a coefficient of variation (CV) for each of the six calibrators ranging between 3 and 5%. The inter-assay reproducibility was also good for the low QC (range 14-25%) and the high QC (range 60-81%), with a mean CV of 4% for both QC; NPP 100% and NPP 50% showed a mean CV of 8% and 7%, respectively. The TTP patient used as internal negative control (IH FRETS-VWF73 ADAMTS13 activity b 10%), was systematically found lower than the LLD (3%). The duplicate samples showed a correct reproducibility (mean CV: 9%) for all plasma tested. When NPP was assayed as a test sample in Chr-VWF73 assay, slightly lower activities than expected were found. When the commercial calibrators were assayed as test samples in the IH FRETSVWF73 assay, lower values of ADAMTS13 activity than those expected were obtained against the NPP calibrant (N100.0%, 63.5%, 13.6%, 10.5%, b LLD, b LLD versus 110.5%, 64.9%, 22.5%, 15.0%, 11.1%, b LLD respectively).

B. Joly et al. / Thrombosis Research 134 (2014) 1074–1080

1077

ADAMTS13 activity (%) (Technozym® ADAMTS13 activity ELISA) 100

90

80

70

60

50

40

30

20

10

0 10

0

[94]

20

[15]

40

30

50

60

70

[53]

80

90

100

ADAMTS13 activity (%) (In-house FRETS-VWF73) [Number] of tested samples Total = 247

[85]

Fig. 1. Scattered diagram illustrating the comparison of the Chr-VWF73 Technozym® (evaluated method) to the in-house FRETS-VWF73 (reference method) for the measurement of ADAMTS13 activity in 217 plasma samples from patients with thrombotic microangiopathy and 30 healthy subjects. Each plasma sample is represented by a dot. The line of identity is shown as a solid diagonal line. The broken lines indicate the clinical relevant threshold (10%). With the in-house FRETS-VWF73 reference method, 94 patients had an undetectable ADAMTS13 activity (b10%), 68 patients exhibited a partial ADAMTS13 deficiency (activity ranging from 11 to 49%) and 85 subjects (TMA patients and healthy subjects) had a normal ADAMTS13 activity (N50%). Among samples which ADAMTS13 activity was b 10% with the reference method (n = 94), 83 samples were also found b10% whereas 11 samples (false negative) were found higher than 10% with the evaluated method. In contrast, 4 samples were found false positive (b10%) with the Chr-VWF73 assay although their ADAMTS13 activity was detectable with the reference assay. The 98 samples found b10% with either the reference assay and/or the evaluated assay were excluded from the calculation of the Spearman’s rank correlation coefficient (R2). R2 (involving 149 samples with values ranging from 11 to 100%) was calculated at 0.59 (p b 0.05). Deviation from line of identity showed that the Chr-VWF73 had a clear tendency to underestimate results in the 50-100% range when compared to the in house FRETS-VWF73 assay.

60

+1.96 SD

FRETS-VWF73 - Chr-VWF73

40

41.7

20

Mean 6.6 0

-20

-1.96 SD -28.5 -40

0

25

50

75

100

(FRETS-VWF73 + Chr-VWF73)/2 Fig. 2. Bland and Altman plot illustrating the comparison of the Chr-VWF73 Technozym® (evaluated method) to the in-house FRETS-VWF73 (reference method) for the measurement of ADAMTS13 activity in 119 plasma samples from patients with thrombotic microangiopathy and 30 healthy subjects. For each patient (plasma sample), the difference between the FRETSVWF73 and the Chr-VWF73 is expressed as a function of the mean value between both assays. Each plasma sample is represented by a dot. The mean difference between both ADAMTS13 activity assays is shown as a solid diagonal line. The broken lines indicate the limits of agreement between both methods (95% confidence interval). As the 98 patients with an undetectable ADAMTS13 activity (b10%) with the reference method and/or the evaluated method were excluded from the current plot, 149 patients were considered for the comparative analysis. The 6.6% mean difference between both assays indicates that the Chr-VWF73 globally underestimated ADAMTS13 activity when compared to the in house FRETS-VWF73. The very large limits of agreement (−28.5% - +41.7%) reflects the great variability between both assays for the measurement of detectable (between 11% and 100%) ADAMTS13 activity levels.

1078

B. Joly et al. / Thrombosis Research 134 (2014) 1074–1080

Table 2 ADAMTS13 activity discrepancies focused on the 10% clinical threshold between in-house (IH) FRETS-VWF73 and Technozym® ADAMTS13 activity ELISA (Chr-VWF73) in 15 patients with thrombotic microangiopathies. Patient

ADAMTS13 activity IH Frets-VWF73

ADAMTS13 activity Chr-VWF73

Anti-ADAMTS13 IgG titers (UI/ml) (normal b15UI/ml)

Clinical context and complementary ADAMTS13 investigation

Congenital TTP (USS)

b10%

11%

3

Exon 3 c.262G N C p.Val88Leu Exon 24 c.3178C N T p.Arg1060Trp

Acquired TTP – idiopathic form Acquired TTP – secondary form Acquired TTP – secondary form Congenital TTP (USS)

b10% b10% b10% b10%

11% 12% 12% 13%

N150 7 84 2

Acquired TTP – idiopathic form Acquired TTP – secondary form Acquired TTP – secondary form Acquired TTP – secondary form Acquired TTP – secondary form Acquired TTP – secondary form Other TMA Other TMA TTP remission Other TMA

b10% b10% b10% b10% b10% b10% 12% 15% 20% 40%

13% 13% 22% 24% 25% 28% b10% b10% b10% b10%

15 4 75 8 45 8 4 6 8 3

Leukemia HIV – Castelman Exon 20 c.2434G N T p.Glu812X Exon 24 c.3178C N T p.Arg1060Trp Wegener disease Renal graft Pulmonary embolism Renal graft HIV HUS Leukemia HELLP and CREST syndromes

CREST: Calcinosis, Raynaud phenomenon, Esophageal dysmotility, Sclerodactyly, Telangiectasia; HELLP: Hemolysis Elevated Liver Enzymes Low Platelet count; HIV: Human Immundeficiency Virus; HUS: Hemolytic Uremic Syndrome.

Comparison of Both Methods In the undetectable range of ADAMTS activity (b 10% with the reference method, n = 94 samples), 83 samples were correctly measured b10% with the Chr-VWF73. Consequently, 11/94 patients (~12%) were false negative with the Chr-VWF73 assay (Fig. 1, Table 2). The Cohen Kappa coefficient was calculated at 0.87 which reflects a “very good” statistical agreement between both assays to measure if ADAMTS13 activity is lower or higher than 10%. In the detectable range of ADAMTS activity (from 11% to 100% with the reference method, n = 153 samples), there was a positive correlation (R2 = 0.59, p b 0.05) between the IH FRETS-VWF73 assay and the Chr-VWF73 assay (Fig. 1). However, the Chr-VWF73 assay showed a global tendency to underestimate ADAMTS13 activity, especially within the normal range values between 50% and 100% (Fig. 1). Surprisingly, 4 patients whose ADAMTS13 actvity was detectable with the IH FRETSVWF73 assay (from 12% to 40%) were found undetectable and thus false positive with the Chr-VWF73 assay (Table 2, Fig. 1). Also, in the same detectable range of ADAMTS13 activity, the Bland and Altman plot (Fig. 2) illustrated that ADAMTS13 activity levels were globally lower when measured with the Chr-VWF73. The absolute mean difference between both assays was 6.6% and the limits of agreement (95% confidence interval) were − 28.5% and + 41.7%, reflecting a great variability between both assays for the measurement of detectable ADAMTS13 activity values. Data Analysis According to the Type of Samples Healthy Normal Subjects Samples from 30 healthy volunteers (ADAMTS13 activity N 50% with the reference method) showed that the Chr-VWF73 assay globally underestimated ADAMTS13 activity (mean ± SD: 67 ± 9% versus 76 ± 17% with the IH FRETS-VWF73) (Figs. 1, 2). Congenital TTP Cases Among the 19 USS patients, 17 patients (89%) exhibited an ADAMTS13 activity lower than 10% with the Chr-VWF73 assay, including 12 patients with values lower than 3% LLD. Surprisingly, 2 USS patients out of 19 were false negative, exhibiting an ADAMTS13 activity slightly higher than 10% (11% and 13%, respectively) with the Chr-VWF73 assay (Fig. 1, Table 2).

Acquired TTP in the Acute Phase. Among the 75 patients with an acute acquired TTP, a majority (66/75 = 88%) also exhibited an ADAMTS13 level b 10% with the Chr-VWF73 assay (including 54 patients with an ADAMTS13 activity level lower than 3% LLD). However, in 9 patients (12%), ADAMTS13 activity was found N 10% with values ranging from 11 to 28% (Fig. 1, Table 2) leading to 12% of false negative results. The discrepancies observed were apparently independent of the antiADAMTS13 IgG titer, the clinical context (Table 2), or the technical interferences such as lipemia, hemolysis, icterus or preanalytical process (rapid processing, clean venipuncture) in tested samples (data not shown). Acquired TTP in Remission. Samples from 52 TTP patients in remission were tested (ADAMTS13 activity ranging from 20% to 90% with the reference method). The Chr-VWF73 assay globally underestimated ADAMTS13 activity, compared to our IH FRETS-VWF73 assay (mean ± SD: 37 ± 20% versus 50 ± 30%, respectively) (Figs. 1, 2). The discrepancies observed were independent of the clinical context or the technical interferences (data not shown). However, one patient with an acquired TTP in remission, exhibiting a detectable ADAMTS13 activity with the IH FRETS-VWF73 assay (20%) was surprisingly found lower than 10% and thus false positive for TTP biological diagnosis with the Chr-VWF73 assay (Table 2). Other TMAs. Among the 71 other TMAs (ADAMTS13 activity ranging from 12% to 100% with the reference method), the Chr-VWF73 assay also globally underestimated ADAMTS13 activity when compared to our IH FRETS-VWF73 assay (mean ± SD: 48 ± 22% versus 50 ± 28%, respectively) (Figs. 1, 2). Three patients exhibiting a detectable ADAMTS13 activity with the IH FRETS-VWF73 assay (12%, 15% and 40%, respectively) were found b 10% with the Chr-VWF73 assay (Fig. 1, Table 2). These false positive results appeared independent of the clinical context, or the technical interferences (data not shown). Sensitivity, Specificity, PPV and NPV of the Chr-VWF73 Assay for the Biological Diagnosis of TTP The sensitivity of the Chr-VWF73 assay for the biological diagnosis of TTP was 88% and its specificity was 97%. PPV and NPV were 95% and 93%, respectively.

B. Joly et al. / Thrombosis Research 134 (2014) 1074–1080

Discussion In the current study involving 247 miscellaneous subjects, the Chr-VWF73 assay exhibits a variable performance as a function of ADAMTS13 activity range. The agreement with the FRETS-VWF73 reference method was globally good (0.87 Cohen kappa coefficient) to identify values either lower or higher than the 10% clinically relevant threshold. However, the Chr-VWF73 assay overestimated 12% (11/94) of samples with an ADAMTS13 activity b 10% with the reference method, in both inherited and acquired TTP patients therefore diagnosed as false negative in the acute phase of TTP. In the detectable range of ADAMTS13 activity (values above 10%, ranging between 11% and 100% with the reference assay), the ChrVWF73 assay showed a clear tendency to systematically underestimate ADAMTS13 activity. In our study, this underestimation was not dependent on the type of sample (healthy volunteers, acute TTP, remission TTP, other TMAs). In this detectable range of ADAMTS13, extreme discrepancies leading to false positive samples (ADAMTS13 b 10% with the Chr-VWF73 assay) were very rare (4/153). In contrast, random discrepancies within detectable values were very frequent and concerned almost all samples (Figs. 1 and 2). Moreover, the statistical analysis emphasized a great variability between both assays as, within the range of detectable ADAMTS13 values, the correlation coefficient was not more than 0.59 and the limits of agreement (Bland and Altman test) ranged from about −30% to +40%. This point may be an important problem for the biological follow-up of TTP patients, potentially leading to overestimate both the risk of relapse and consequently the indication of preemptive treatment. Previous evaluations of commercial assays for ADAMTS13 activity measurement comprise 2 studies. Mackie et al. [20] published the evaluation of 2 commercial assays for ADAMTS13 activity, and showed a reasonable agreement between the FRETS-VWF73 assay and the Chr-VWF73 assay in 159 samples from healthy subjects and acquired TMA patients. However, in agreement with our data, with the ChrVWF73 assay, they found firstly, a slight underestimation of some detectable ADAMTS13 activity values and secondly, detectable values for some severe deficiencies leading to false negative results. We also agree with Mackie et al. about the good LLD of the Chr-VWF73 assay (3%) [20], which is well adapted to the clinically relevant threshold for TTP diagnosis (10%). Moreover, our group also investigated another commercial kit for ADAMTS13 activity, the ATS-13® ADAMTS13 activity assay (Gen-Probe Inc., Wankesha, WI, USA) [21]. The main limitations of the ATS-13® ADAMTS13 activity assay were both a strong discrepancy between the LLD (15%), and the clinically relevant threshold for TTP diagnosis (10%) and the overestimation of detectable values. Also, more recently, Mancini et al. [23] also described discrepancies between FRETS-VWF73 assay (ADAMTS13 activity b 10%) and an in-house collagen binding assay (ADAMTS13 activity N 20%) in a series of 20 TTP patients with ADAMTS13 activity b 10%. In that study, they showed that urea could represent one cause of the mismatch; samples showed the accumulation of unfolded and ultralarge hyper-adhesive multimers of VWF in plasma, confirming the diagnosis of TTP. This study further underlines the position of the FRETS-VWF73 (over Collagen Binding Assay) as the reference method for ADAMTS13 activity measurement. There is no doubt that the industrial development of both an international reference preparation for ADAMTS13 activity and external quality controls will be crucial to improve the standardization of ADAMTS13 assays. To date, these reagents are not available on the international market yet; however, in 2013, the National Institute for Biological Standards and Controls (NIBSC, United Kingdom) on behalf of the International Society for Haemostasis and Thrombosis (ISTH) has initiated an international multicenter study (our laboratory has participated in this study as the French referent investigator) to provide these reagents, hopefully soon expected. Also, at a national level, since 2012, our laboratory has been organizing a quality control for ADAMTS13 activity to help volunteer non expert French laboratories (Congress of the

1079

Groupe d’Etude sur l’Hémostase et la Thrombose GEHT, may 2012, data not shown). In the meantime, considering the results of the current study, we suggest the following recommendations: In terms of technical practice, to improve the performance of the Chr-VWF73 assay, we suggest to add one calibrator (target value of about 80%) and above all, to systematically include additional internal QC i.e. at least one severely ADAMTS13 deficient sample (ADAMTS13 activity b 10%), and one known NPP. In terms of medical recommendations, the diagnosis of TTP in the acute phase may be supported by the measurement of ADAMTS13 activity using the Chr-VWF73 assay by non-expert laboratories, considering the LLD of the assay, its reasonable agreement with the FRETS-VWF73 reference assay to distinguish values b 10% from values N10%, its correct sensitivity together with its good specificity, PPV and NPV for the diagnosis of acute TTP. However, considering that the Chr-VWF73 assay may provide false negative results in ~ 12% of acute TTP patients, if the result does not match perfectly with the clinics, we recommend to systematically control ADAMTS13 activity with a referent method by referring to an expert laboratory certified for rare diseases. Furthermore, the main limit of the Chr-VWF73 assay is to globally underestimate normal and subnormal ADAMTS13 activity levels with a random and great variability when compared to the FRETS-VWF73 assay. This is also a concern. Firstly, in our acquired TTP in remission series, one patient out of 52 was not considered in remission with the Chr-VWF73 assay. Secondly, the underestimation of ADAMTS13 activity during the follow-up of TTP patients may potentially lead to overestimate the risk of relapse and consequently, to induce physicians to prescribe prophylactic immunomodulatory treatment in excess. In that regard, we recommend to perform the biologic follow-up of one given patient using the same assay, and to control all ADAMTS13 activity levels below 30% with reference methods performed by expert laboratories. Conflict of Interest Statement No conflict of interest relevant to this article was reported. Acknowledgements The authors are grateful to Sylvaine Savigny for technical assistance and helpful discussion, and to all physicians, members of the reference center for thrombotic microangiopathies for providing samples and clinical data. References [1] Sadler JE. Von Willebrand factor, ADAMTS13, and thrombotic thrombocytopenic purpura. Blood 2008;112:11–8. [2] Furlan M, Robles R, Galbusera M, Remuzzi G, Kyrle PA, Brenner B, et al. von Willebrand factor-cleaving protease in thrombotic thrombocytopenic purpura and the hemolytic-uremic syndrome. N Engl J Med 1998;339:1578–84. [3] Levy GG, Nichols WC, Lian EC, Foroud T, McClintick JN, McGee BM, et al. Mutations in a member of the ADAMTS gene family cause thrombotic thrombocytopenic purpura. Nature 2001;413:488–94. [4] Tsai HM, Lian EC. Antibodies to von Willebrand factor-cleaving protease in acute thrombotic thrombocytopenic purpura. N Engl J Med 1998;339:1585–94. [5] Rieger M, Mannucci PM, Kremer Hovinga JA, Herzog A, Gerstenbauer G, Konetschny C, et al. ADAMTS13 autoantibodies in patients with thrombotic microangiopathies and other immunomediated diseases. Blood 2005;106:1262–7. [6] Peyvandi F, Lavoretano S, Palla R, Feys HB, Vanhoorelbeke K, Battaglioli T, et al. ADAMTS13 and anti-ADAMTS13 antibodies as markers for recurrence of acquired thrombotic thrombocytopenic purpura during remission. Haematologica 2008;93: 232–9. [7] Veyradier A, Obert B, Houllier A, Meyer D, Girma JP. Specific von Willebrand factorcleaving protease in thrombotic microangiopathies: a study of 111 cases. Blood 2001;98:1765–72. [8] Lattuada A, Rossi E, Calzarossa C, Candolfi R, Mannucci PM. Mild to moderate reduction of a von Willebrand factor cleaving protease (ADAMTS-13) in pregnant women with HELLP microangiopathic syndrome. Haematologica 2003;88:1029–34. [9] Hie M, Galicier L, Provot F, Presne C, Poullin P, Bonmarchand G, et al. French Thrombotic Micronagiopathies Reference center. Preemptive rituximab infusions after

1080

[10] [11]

[12]

[13]

[14] [15] [16]

B. Joly et al. / Thrombosis Research 134 (2014) 1074–1080 remission efficiently prevent relapses in acquired thrombotic thrombocytopenic purpura. Blood 2014;124:204-10. Peyvandi F, Palla R, Lotta LA, Mackie I, Scully MA, Machin SJ. ADAMTS-13 assays in thrombotic thrombocytopenic purpura. J Thromb Haemost 2010;8:631–40. Studt J-D, Böhm M, Budde U, Girma J-P, Varadi K, Lämmle B. Measurement of von Willebrand factor-cleaving protease (ADAMTS-13) activity in plasma: a multicenter comparison of different assay methods. J Thromb Haemost 2003;1:1882–7. Tripodi A, Chantarangkul V, Böhm M, Budde U, Dong J-F, Friedman KD, et al. Measurement of von Willebrand factor cleaving protease (ADAMTS-13): results of an international collaborative study involving 11 methods testing the same set of coded plasmas. J Thromb Haemost 2004;2:1601–9. Mahdian R, Rayes J, Girma J-P, Houllier A, Obert B, Meyer D, et al. Comparison of FRETS-VWF73 to full-length VWF as a substrate for ADAMTS13 activity measurement in human plasma samples. Thromb Haemost 2006;95:1049–51. Kokame K, Nobe Y, Kokubo Y, Okayama A, Miyata T. FRETS-VWF73, a first fluorogenic substrate for ADAMTS13 assay. Br J Haematol 2005;129:93–100. Zheng XL. Structure-function and regulation of ADAMTS-13 protease. J Thromb Haemost 2013;11:11–23. Muia J, Gao W, Haberichter SL, Dolatshahi L, Westfield LA, Covill SC, et al. An optimized fluorogenic ADAMTS13 assay with increased sensitivity for the investigation of patients with thrombotic thrombocytopenic purpura. J Thromb Haemost 2013;11:1511–8.

[17] Crawley JT, Scully MA. Thrombotic thrombocytopenic purpura: basic pathophysiology and therapeutic strategies. Hematology Am Soc Hematol Educ Program 2013; 2013:292–9. [18] Siekmann J, Turecek PL, Schwarz HP. The determination of von Willebrand factor activity by collagen binding assay. Haemophilia 1998;4(Suppl. 3):15–24. [19] Gerritsen HE, Turecek PL, Schwarz HP, Lämmle B, Furlan M. Assay of von Willebrand factor (vWF)-cleaving protease based on decreased collagen binding affinity of degraded vWF: a tool for the diagnosis of thrombotic thrombocytopenic purpura (TTP). Thromb Haemost 1999;82:1386–9. [20] Mackie I, Langley K, Chitolie A, Liesner R, Scully M, Machin S, et al. Discrepancies between ADAMTS13 activity assays in patients with thrombotic microangiopathies. Thromb Haemost 2013;109:488–96. [21] Thouzeau S, Capdenat S, Stépanian A, Coppo P, Veyradier A. Evaluation of a commercial assay for ADAMTS13 activity measurement. Thromb Haemost 2013;110: 852–3. [22] Moatti-Cohen M, Garrec C, Wolf M, Boisseau P, Galicier L, Azoulay E, et al. Unexpected frequency of Upshaw-Schulman syndrome in pregnancy-onset thrombotic thrombocytopenic purpura. Blood 2012;119:5888–97. [23] Mancini I, Valsecchi C, Lotta LA, Deforche L, Pontiggia S, Bajetta M, et al. FRETSVWF73 rather than CBA assay reflects ADAMTS13 proteolytic activity in acquired thrombotic thrombocytopenic purpura patients. Thromb Haemost 2014:112–5.