International Council for Standardization in Haematology (ICSH) recommendations for laboratory measurement of ADAMTS13

This guidance document was prepared on behalf of the International Council for Standardization in Haematology (ICSH), by the ADAMTS13 Assay Working Group, which comprises an international group of both clinical and laboratory experts. The document provides recommendations on best practice for the performance of ADAMTS13 assays in clinical laboratories. ADAMTS13 assays support the differential diagnosis of thrombotic microangiopathies and have utility in the management of thrombotic thrombocytopenic purpura (TTP). There are three types of assay: activity, antigen and autoantibody/inhibitor assays. Methods for activity assays differ in terms of sensitivity, specificity, precision and turnaround time. The most widely used assays involve VWF peptide substrates and either chromogenic ELISA or FRET techniques, although chemiluminescence assays and rapid screening tests have recently become available. Tests for autoantibodies and inhibitors allow confirmation of acquired, immune‐mediated TTP, while antigen assays may be useful in congenital TTP and as prognostic markers. In this document, we have attempted to describe ADAMTS13 assays and the conditions that affect them, as well as: blood collection, sample processing, quality control, standardization and clinical utility; recognizing that laboratories in different parts of the world have varying levels of sophistication. The recommendations are based on expert opinion, published literature and good clinical laboratory practice.

multimeric form (ULVWF), which is rapidly cleaved by ADAMTS13, resulting in progressively smaller multimers. 1 The high shear stress found in small arterioles unravels the globular form of ULVWF so that the metalloprotease domain of ADAMTS13 can come into close proximity with the cleavage site in the VWF A2 domain, allowing cleavage. The other domains of ADAMTS13 play a role in the attachment and unravelling of VWF, while VWF itself induces conformational changes in ADAMTS13, making it fully active. [3][4][5][6][7] Thrombotic microangiopathies (TMA) are a group of disorders characterized by thrombocytopenia and microangiopathic haemolytic anaemia (MAHA) resulting in varying degrees of organ damage; they are rare, but potentially life-threatening unless they are recognized and treated rapidly. They include thrombotic thrombocytopenic purpura (TTP) and haemolytic uraemic syndrome (HUS), 8,9 but may also be secondary to cancer, viral infection (eg HIV), organ or bone marrow transplantation, pregnancy (preeclampsia/eclampsia or HELLP syndrome), severe hypertension, drugs, autoimmune disorders, sepsis and disseminated intravascular coagulation. [10][11][12] In thrombotic thrombocytopenic purpura (TTP), there is a loss of ADAMTS13 function resulting in the prolonged circulation of ULVWF, which has an increased platelet binding capacity and may participate in platelet aggregate formation in the microcirculation with sequestration and consequent thrombocytopenia. Blood flow is limited, causing ischaemic organ damage, and haemolysis occurs due to red cell mechanical destruction. Elevated levels of indirect bilirubin and lactate dehydrogenase are common due to the gross haemolysis. Diagnosis is based on the presence of thrombocytopenia and MAHA, as well as the clinical history and examination of the blood film 13 ; additional laboratory studies are required to distinguish between different causes of TMA.
TTP exists in acquired, immune-mediated (iTTP) and congenital (cTTP) forms. 14 iTTP is caused by the presence of autoantibodies to ADAMTS13, which decrease its function or increase ADAMTS13 clearance from the circulation. These antibodies are heterogeneous, varying in epitope specificity and avidity. iTTP can occur at any age, mostly in the 3rd-5th decade of life, and is frequently associated with infection and pregnancy, but may also occur secondary to a variety of other conditions. Patients with iTTP who enter remission may relapse at varying intervals, most recurrences occurring during the first year due to the reappearance of the autoantibody, loss of ADAMTS13 and development of ULVWF. In cTTP, which usually presents in childhood, the loss of ADAMTS13 activity is caused by pathogenic genetic ADAMTS13 variants, resulting in decreased synthesis, secretion, or function.
Later onset cTTP may also occur, often during pregnancy.
ADAMTS13 assays are useful to distinguish TTP from other TMA and, in the untreated patient at presentation, a severe deficiency (<10 IU/ dL) of ADAMTS13 activity is diagnostic for TTP. 14 ADAMTS13 activity has been listed as a critical test. 15 Plasma levels and function of ADAMTS13 are modulated by hepatic stellate cell damage, cytokine levels and oxidation (which inhibits ADAMTS13 proteolytic activity and makes VWF resistant to cleavage) 2 ; a mild to moderate reduction in activity may occur in cardiovascular disease, liver disease, sepsis and consumptive coagulopathies, with. Guidance on the diagnosis of TTP, management, therapy and prophylaxis are available 13,16,17 and ICSH has published recommendations on schistocyte enumeration in TMA. 18 HUS 9 may be subdivided into infection-associated HUS (IA-HUS) and complement-mediated HUS (CM-HUS, formerly termed atypical HUS). IA-HUS is typically associated with Shiga toxin-induced bloody diarrhoea in children, which is treated with supportive care, and in some cases renal dialysis. CM-HUS, more common in adults, usually results from defective regulation of the complement system and may sometimes show multisystem symptoms similar to TTP, making the differential diagnosis difficult. Unlike TTP, ADAMTS13 levels are usually normal in CM-HUS, although they sometimes show a moderate reduction (30-50 IU/dL). ADAMTS13 assays are thus pivotal in differentiating TTP and CM-HUS, which require different treatments. Plasma exchange (PEX) and immunosuppressive agents are typically used for TTP 8,13 while PEX and complement inhibition (eg with eculizumab) are used in CM-HUS, 19 although PEX may be used until a firm diagnosis is made.
The aim of this ICSH document is to provide laboratory guidance and an up-to-date summary of best practice for the validation, performance, and reporting of ADAMTS13 assays. The consensus recommendations provided are based on information from peer-review publications, the authors' personal experience and expert opinion, as well as good clinical laboratory practice.

| PRE-ANALY TIC AL VARIAB LE S
Citrated (0.109 mol/L sodium citrate) platelet-poor plasma is normally used for ADAMTS13 assays. Samples should be prepared to ensure platelet depletion (<10 × 10 9 /L; ie centrifugation at 2000 g for 15 minutes). This should be performed at room temperature (18-20°C), particularly if other haemostasis assays might be performed, to avoid potential platelet changes and proteolytic activation. Serum and heparinized plasma have been used and may be advantageous in certain types of assay, where high sensitivity is required, with less sample dilution, and where citrate chelation of divalent cations interferes with the assay. However, laboratories must validate blood collection into alternative anticoagulants before clinical application. Plasma or serum samples should be rapidly removed from the cells using a plastic transfer pipette, avoiding disturbance to the buffy coat and stored in appropriate polypropylene containers.
ADAMTS13 is present in serum, but inaccurate activity results could potentially occur, due to platelet and leucocyte activation as well as ADAMTS13 degradation, by thrombin and other proteases.
However, this can be effectively mitigated with protease inhibitors.
It should be noted that ADAMTS13 activity can also vary significantly due to other factors, for example changes in assay buffer conditions (pH, ionic strength and surfactants) can affect ADAMTS13 activity due to allosteric mechanisms. 3-6 EDTA plasma cannot be used for activity assays, as it irreversibly disrupts the quaternary structure of ADAMTS13, causing false low/absent activity. In cases of unexpected absence of ADAMTS13 activity, where results do not   fit the clinical picture and to exclude pre-analytical variables such   as incorrect sample type, a prothrombin time or activated partial   thromboplastin time on the sample can be helpful as grossly prolonged times suggest EDTA contamination or serum. Alternatively,   as EDTA samples have markedly elevated potassium and decreased calcium, 20 electrolyte analysis showing an abnormal sodium to potassium ratio suggests EDTA anticoagulant.
Blood samples should be collected with minimal stasis and processed rapidly to avoid cellular and plasma activation. Any significant delays in processing should be recorded and depending on the laboratory policy, added as a comment to the report. Clotted samples or those containing clumped platelets are not suitable. Samples that are not being assayed on the same day should be stored frozen at below −40°C.
If the samples are shipped to another site, dry ice should be used to avoid slow thawing during transit and comparable results can then be achieved between different laboratories. 21 Although ADAMTS13 has been shown to have good stability at room temperature, 22 some of these studies have employed samples from healthy normal subjects and it is not clear whether stability is good in all clinical samples, particularly those with critical illness and the potential for reduced plasma protease inhibitor levels. Before assay, frozen samples should be warmed to 37°C in a water bath for 5-10 minutes, until completely thawed and mixed gently to ensure that any cryoprecipitate is dissolved. Thawing for longer than this is undesirable as ADAMTS13 activity is lost after prolonged periods at 37°C. 23 Icteric samples may not be suitable for some assay types (see specificity section below) due to interference in absorbance measurement or quenching of fluorescence signal.
Haemolysis is frequently seen in patients with TMA, due to entrapment and lysis of red cells in the microcirculation. However, where it is due to in vitro haemolysis, for example caused by a slow blood draw, fresh samples should be obtained. Variable ADAMTS13 activity may also be obtained in samples with gross lipaemia, although the exact nature of any influence of plasma lipids on activity assays has not been investigated and might possibly vary depending on the composition of the lipid.
Samples for TTP diagnosis are optimally collected prior to initiation of PEX, although samples collected early in treatment may still be informative. It is important that the laboratory has details of the type and timing of treatment.

| Consensus recommendations on sample collection and handling for ADAMTS13 assays
• Citrated plasma (centrifuged to ensure platelet depletion) should normally be used.
• Heparin plasma or serum samples may be used depending on the assay type and if it is validated for these sample types.
• EDTA plasma is not suitable for ADAMTS13 assays.
• Samples should be centrifuged and plasma separated from the cells as rapidly as possible after blood collection to avoid in vitro changes.
• Unless assays are performed immediately, plasma samples should be stored and shipped below-40°C to avoid potential proteolysis.

| ADAMTS1A SSAYS
There are a variety of in-house and commercial methods for ADAMTS13 activity, antigen, inhibitors and ADAMTS13 antibody assay.

| Activity assays
The widespread use of ADAMTS13 assays has been limited by the rarity of TTP in the general population, technical difficulty of the assays and the length of time taken to generate an assay result. 24 First generation activity assays involved incubation of patient plasma with VWF and measurement of residual VWF by multimer electrophoresis, 25 SDS PAGE and Western Blotting with VWF antibodies, 26 Immunoradiometric assay, 27 or ristocetin cofactor assay, 28 or collagen binding ELISA (CBA). 29 Electrophoretic detection of VWF multimers and analysis by ristocetin cofactor assay are difficult, with poor precision. Assays based on residual VWF binding to collagen are also time consuming, but the ELISA principle allows higher throughput, with better sensitivity and precision. However, there are numerous potential sources of error, including the use of denaturing agents that besides unravelling VWF could alter ADAMTS13 structure and dissociate ADAMTS13 autoantibodies.
Second generation activity assays employ peptide substrates based on the ADAMTS13 cleavage site in the VWF A2 domain.
These are rapid, producing results within a few hours, with high throughput and good precision. However, these assays may fail to detect certain inherited ADAMTS13 defects, as they measure metalloprotease activity independent of exosite interactions with other VWF domains. Similar problems can occur where ADAMTS13 autoantibodies against these exosites exist. Elevated endogenous VWF levels influence some assays, and increased bilirubin levels can interfere in fluorescence-based techniques.
Some of the current activity assays show relatively poor sensitivity, which can be an issue in the clinical management of TTP.
The key principles of some of the activity assays have been reviewed 20 ; they may be divided into assays utilizing full-length VWF as substrate and those using peptides or fragments derived from VWF. The details of the more popular assays are given in Table 1.
The most commonly used peptide substrate assays employ a 73 amino acid peptide (VWF73), based on the sequence around the cleavage site in the VWF A2 domain, 30 although some methods use slightly longer peptides, modified sequences or fluorophores. 31 The peptides may contain: tags to allow their attachment to microtitre plates (eg His-tag); attached proteins to allow measurement of in- In chromogenic activity ELISA methods, substrate cleavage may be detected by loss of signal (eg removal of GST), or the increased binding of monoclonal antibodies against a neo-epitope generated after substrate cleavage (eg anti-N10 antibody). 32 In FRET-based assays, an increased fluorescence signal occurs due to the loss of quenching when the peptide is cleaved and the quencher is no longer in close proximity to the fluorophore. The activity ELISA has multiple steps (incubation, washing, reagent addition, etc.) and many laboratories have microplate readers, whereas the FRET assays can be set-up with automatic reagent addition, incubation and measurement, minimizing hands-on time, although a fluorescence microplate reader is required.
An ADAMTS13 activity assay based on the VWF73 peptide and surface-enhanced laser desorption ionization time-of-flight Abbreviations: Ab, antibody; CV, coefficient of variation; DF, dilution factor; em, emission; ex, excitation; GST, glutathione S-transferase; HRP horseradish peroxidase; LLOQ, lower limit of quantitation; NR, not reported; PNP, pooled normal plasma; r, recombinant. a Exact detail varies between laboratories.
(SELDI-TOF) mass-spectrometry has also been reported. 33 Although highly sensitive, instrumentation requirements make this assay impractical for clinical diagnostic use in most hospital settings.
More recently, particle-based automated assays have been developed. An automated chemiluminescence assay utilizes a two-step immunoassay involving magnetic particles coated with GST-VWF73 peptide substrate and chemiluminescent detection based on an isoluminol labelled monoclonal antibody that reacts with the cleaved peptide. It is a rapid assay (33 minutes), with good sensitivity and precision, but only has a three point calibration curve. Although there were some discrepancies compared to FRET-VWF73 and a chromogenic activity ELISA, with some samples showing inter-assay disparity at normal and high activity, there was a good correlation between the assays and high agreement in classifying samples with ADAMTS13 levels below 10 IU/ dL. [34][35][36] Due to the reaction principle, these assays are not affected by icterus, lipaemia or plasma turbidity. However, the sample size in two of these studies was small and further evaluations may be needed before their widespread acceptance and regional regulatory bodies may require additional verification or validation procedures.
The ionic strength, pH, divalent cation and chloride ion concentrations are critical in ADAMTS13 assays, and all buffers should be fresh in order to avoid pH changes and potential bacterial contamination.
Alternatively, some buffers can be stored for extended periods lyophilized, or prepared as stock solutions (eg 10× stock buffers) especially if filtered with a 0.22 µmol/L membrane. Beside the pH and ionic environment, the use of denaturants and differences in sample dilution factor between assays might have an impact on the binding kinetics of ADAMTS13 and its autoantibodies, so that varying degrees of dissociation could occur in different assay methods. 37 Finally, some assays utilize heat inactivated normal plasma (treated at 56°C for 30 minutes to denature ADAMTS13) as a diluent, to help preserve the ADAMTS13 environment and allow a linear standard curve.

| Point of care (POC)/rapid tests
A rapid, semi-quantitative test for ADAMTS13 activity is now commercially available, 38 which utilizes a flow-through cartridge and an activity ELISA principle, being completed within 30 minutes, without the need for specialized instrumentation or personnel. It has four indicator points (zero, 10, 40 and 80 IU/dL) and in comparison with a chromogenic activity ELISA in 220 patients with suspected TMA, showed 88.7% sensitivity, 90.4% specificity and 96.2% negative predictive value. The method is limited by its subjective visual interpretation and potential for interference by lipids (which might block the device membrane), haemolysis and icterus. The test is suitable for use in a POC environment as a screening tool and for the negative exclusion of TTP. When decreased activity is detected, it should be confirmed by bioassay in an accredited laboratory.

| Agreement/disparity between activity assays
A number of studies 21 TA B L E 2 Potential sources of discrepancy between activity assays and handling differences may contribute to these changes as well as a variety of in vivo and in vitro factors (

| Sensitivity
ADAMTS13 assay sensitivity can be attributed to several factors, including the substrate type, assay method, instrument, buffer conditions and volume of plasma. ADAMTS13 has maximal activity at low pH (pH 6) and low ionic strength (zero NaCl) as opposed to pH 7.4 and salt concentrations similar to those in plasma (150 mmol/L NaCl).
Generally, VWF peptide-based assays are more sensitive than those using multimeric VWF digest analysis. The first generation FRETbased assay sensitivity was limited (~3%-5% of normal ADAMTS13 activity), 30

| Specificity
In assays using full-length VWF, it must be high purity, containing all normal multimer sizes and free of ADAMTS13 contamination.

| Antigen assays
Immunoblotting procedures and ELISA assays have been used for

| Autoantibody assays
Western blotting, immunocapture and ELISA assays for determination of ADAMTS13 autoantibody levels have been described, 59,60 but ELISA methods for immunoglobulin (Ig) G class autoantibodies are most widely used in clinical laboratories. The calibration of these assays, their reporting units and cut-off values vary; the results may not be interchangeable. The effects of different incubation temperatures and different assays for measuring residual activity have recently been investigated, 23 showing a good correlation between FRET-based and CBA-based inhibitor assays, although the pre-incubation at 37°C causes a reduction of plasma ADAMTS13 activity when tested in the FRET-based assay. A flow-based assay for ADAMTS13 inhibitor assessment has also been described. 61 The mixing test inhibitor methods are prone to errors because of plasma manipulation and dilution of patient plasma with PNP limits the sensitivity. Dilution could potentially alter the equilibrium between free and bound antibody. Bethesda type assays of ADAMTS13 inhibition may be problematic due to the variability in antibody epitope specificity, affinity and reaction kinetics be- There is currently no IS or reference material for ADAMTS13 antibody and inhibitor assays, meaning that numerical results are difficult to compare between different assay types.

| Criteria for validation of each assay run
A calibration curve should be performed with each batch of tests to avoid reagent and analyser variability. Some activity assay methods may be particularly temperature sensitive. IQC values should be within their target range. Test results should be within the linear portion of the calibration curve; if they have levels above this range, they should be repeated at higher dilution; if below the range, they must be reported as less than the lower limit of quantification. In kinetic assays, the baseline readings should be checked and very low or high values may indicate an analytical problem.

| Consensus recommendations on assays for ADAMTS13, autoantibodies and inhibitors
• Functional FRET-based assays or chromogenic activity ELISA methods are recommended as front line assays as they are sensitive, show good precision and are simpler to use, being completed in a few hours.
• Rapid point of care assays may have utility as screening methods or "out of hours" emergency tests, but there is currently limited performance data.
• Every calibrator should be traceable to the International Standard Plasma (12/252) for assaying ADAMTS13 activity in citrated plasma samples.
• When reporting results: indicate the type of assay performed and use the correct units (eg IU/dL) for activity and antigen assays. If calibrants traceable to the IS are not available, use percentage of pooled normal plasma. State the reference range for the method.
• High and low activity controls should be included in each assay run. Do not use commercial controls in methods other than those intended for their use.
• Protocols should be validated after any modification.
• If gross icterus interferes in some FRET assay methods, the problem can sometimes be resolved by assaying at a higher dilution, treatment with bilirubin oxidase, or using a chromogenic activity ELISA. A comment regarding potential assay interference should be added to the laboratory report.
• If artefactual, in vitro haemolysis is likely (eg secondary to difficult venipuncture or use of a small gauge needle), fresh blood samples should be obtained. If this is not possible, a comment regarding potential assay interference should be added to the laboratory report.
• Where assay results do not match the clinical picture, or congenital TTP is suspected and ADAMTS13 activity results are normal or only show a mild reduction, collagen-binding assays should be considered.
• If decreased ADAMTS13 activity (<20 IU/dL) is detected in a new patient, an ADAMTS13 antibody assay or inhibitor test should be performed. If the results do not match the clinical picture, potential EDTA contamination should be considered and where possible, fresh blood samples obtained.
• Wherever possible, use the same ADAMTS13 assay when studying a patient longitudinally to manage treatment.

| CLINI C AL UTILIT Y OF A SSAYS
Due to the rarity of TTP, ADAMTS13 assays are not widely performed and remain mainly confined to specialized laboratories. When a patient presents with a clinical scenario and blood film of suspected TMA (Fig. 1) Differentiating between cTTP and iTTP relies upon the detection or absence of an ADAMTS13 autoantibody or inhibitor. 13,70 The determination of a clinically significant positive IgG antibody varies between laboratories depending on the assay and reagents

| Consensus recommendations on clinical utility of assays
• ADAMTS13 activity <10 IU/dL is diagnostic for TTP in patients presenting with a clinical scenario and blood film consistent with TMA.
• An alternative diagnosis and repeat ADAMTS13 assay should be considered in TMA patients presenting with an indeterminate range of 10-20 IU/dL ADAMTS13 activity.
• Alternative diagnosis/therapy should be considered in patients with an ADAMTS13 activity >10 IU/dL, especially when >20 IU/ dL, as they rarely respond to PEX.
• The overall clinical picture/treatment of the patient should always be reviewed with close interaction with clinicians. This is important to avoid unnecessary tests and to ensure that appropriate tests meet clinical needs, especially when non-TTP expert clinicians request assays. Clinical scoring systems such as the PLASMIC score may assist in providing guidance for the necessity of ADAMTS13 testing.
• ADAMTS13 antibody assays should be used to decide whether the patient has cTTP or iTTP; definite cTTP diagnosis requires ADAMTS13 genetic testing.
• Nomenclature, normal range, reporting units and methodology (eg inhibitory and noninhibitory antibodies) need further standardization.
• Follow-up ADAMTS13 activity testing is important as persistently low levels increase the risk of exacerbation and TTP relapse.
• ADAMTS13 antigen levels are not clinically useful in the absence of activity assays, but may prove helpful in predicting mortality risk in iTTP patients with high IgG ADAMTS13 antibody level, although further study is needed.

| CON CLUS ION
Our knowledge concerning ADAMTS13 assays has advanced considerably in recent years, but there are still considerable knowledge gaps concerning: the impact of all pre-analytical variables, calibration of autoantibody assays, the place of epitope-specific assays, and which assays to use in different clinical scenarios.

ACK N OWLED G EM ENTS
All authors were part of the ICSH ADAMTS13 Assays Working Group chaired by IM and all contributed to the design, drafting and editing of this document.