Guidelines on the laboratory aspects of assays used in haemostasis and thrombosis

Abstract

This guideline amalgamates and updates two previous guidelines1, 2 published on behalf of the British Society of Haematology (BSH). The guideline was compiled according to the BSH process at https://b-s-h.org.uk/guidelines. The writing group produced the first draft of the manuscript which was revised by wider members of the Haemostasis and Thrombosis Task Force before being reviewed by an extended group of UK haematology medical and scientific members of the BSH; amendments were made accordingly. The 'GRADE' system (http://www.gradeworkinggroup.org) does not apply (due the lack of clinical trials to support the best practice recommendations) and was therefore not used. Appendix 1 details the literature search process undertaken. This guideline is intended to help clinical laboratories perform high quality valid assays for the basic procoagulants and anticoagulants as part of a routine diagnostic service. Areas that overlap with or have been included in other BSH (https://b-s-h.org.uk/guidelines/) or United Kingdom Haemophilia Centre Doctors Organisation (UKHCDO) (http://www.ukhcdo.org/guidelines/) guidelines have been omitted, including guidance on: heparin-induced thrombocytopaenia (HIT); lupus anticoagulant (LA) testing; D-dimer assays; platelet function testing; diagnosis of von Willebrand disease (VWD); measurement of factor replacement in haemophilia A and B; monitoring of anticoagulants [vitamin K antagonists (VKA) and direct oral anticoagulants (DOAC)]; and global assays of haemostasis (e.g. TEG, ROTEM, thrombin generation). Preanalytical errors account for the majority of errors in the haemostasis laboratory and it is essential that they are well understood and minimised.3 To produce accurate and meaningful results and increase quality and standardisation within haemostasis laboratories, correct procedures must be followed (Table I). Phlebotomy staff must be well trained and laboratory staff must understand the tests used and potential sources of error. Result interpretation requires understanding of the potential effect of patient factors on the assays. Blood collection Sample handling Storage and preparation Once thawed, do not refreeze plasma unless data are available to demonstrate that results are unaffected.157 The correct filling of tubes (especially in patients with a high haematocrit), effects of haemolysis, and presence of anticoagulant drugs all need to be considered. Before processing, samples should be inspected visually to ensure they are labelled correctly, correctly filled (using manufacturers guides) and not clotted or haemolysed. Samples should be rejected if <90% filled, unless shorter collection volumes have been validated for a particular test.4 Gross icterus and lipaemia may affect results by interfering with optical absorbance or impeding light transmittance, but analyses using mechanical end points are not affected. The use of analysers with automated spectrophotometric detection of haemolysis, icterus and lipaemia (HIL) and with the ability to check fill volume may improve standardisation and quality control as well as prevent over or under-rejection of haemolysed samples.5-7 Guidelines from the Clinical & Laboratory Standards Institute (CLSI) suggest samples with gross haemolysis should not be used because of possible clotting, factor activation and interference with end-point measurement.8 They recommend rejecting all haemolysed samples but the degree to which the haemolysis has a significant effect on results varies according to the assay being performed.9-12 Modern coagulometers now provide optical quantitation of free haemoglobin which removes any subjectivity in sample assessment and rejection. Therefore it may be necessary to locally validate the impact of haemolysis and define rejection cut-offs.13 Causes of in vivo haemolysis still need to be excluded in patients when repeated flagging occurs. We suggest that a coagulation screen, prothrombin time (PT) and activated partial thromboplastin time (APTT) be performed on all samples referred for haemostasis assays. This serves as confirmation of the quality of the sample and may also detect the presence of anticoagulants if this is not known from the clinical details provided. However the sensitivity of PT and APTT reagents to anticoagulants varies and these tests may not detect concentrations that can interfere with other assays; so if there is doubt about their presence an assay using specific calibrators (if known) should be performed.14 If indicated, the presence of a normal thrombin time (TT) effectively excludes the presence of a thrombin inhibitor or significant contamination with unfractionated (but not low molecular weight) heparin. Commercial products such as DOAC-Stop (Haematex, Hornsby, NSW, Australia),15-17 DOAC-Remove (5-Diagnostics, Basel, Switzerland),18 20 mg/ml of activated charcoal19 or Adexanet alfa20 have been shown to remove the effect of DOAC. However, if used, results should be interpreted with caution as there may be differing limits to the concentration of the DOAC they can remove.15, 17 Most routine tests should be performed within 4 h of collection with the exception of APTT for monitoring unfractionated heparin (UFH) which preferably should be tested within 1 h and the PT which has a stability of 24 h.9 There is now more evidence that the 4-h window can be extended for many tests; however, if local processes make this necessary validation is required (Table II).3, 21-23 Other patient factors have been shown to interact with haemostasis testing. The list is too extensive to report here but examples include the presence of rheumatoid factor or paraproteins that may interfere with clot-based assays (i.e. routine PT/APTT) and acquired anti-mouse antibodies that have been reported to cross react with immunoassays (for example D-dimer) with the potential to produce erroneous results.6, 24-26 Haemostasis assays using a clotting end-point can be performed by photo-optical and electro-mechanical methodologies, forming the basis of the majority of commercial analysers. Parameters generated from clot waveform analysis in photo-optical systems may give additional information over electro-mechanical systems [examples include its ability to detect changes in patients with underlying disseminated intravascular coagulation (DIC) or monitor factor (F)VIII replacement therapy] but optical systems are also prone to interfering substances (i.e. lipaemia) so in practice, as these parameters have not been universally adopted for clinical use, both systems are acceptable.27, 28 The procurement and validation of commercial coagulation analysers or reagents is outside the scope of this guideline. Regardless of method and target analyte, quantitative assays rely on a general principle of establishing a calibration curve using a sample with known concentration, to provide a clinically useful dose–response curve. For many assays, a parallel-line assay using multiple test dilutions should be employed. The accuracy and precision of the assay should be determined for each assay, and these should inform the assay design for clinical samples (for example whether duplicate testing is required).29, 30 Commercial plasma calibrators, lyophilised or frozen, traceable to the relevant World Health Organization (WHO) International Standards (IS) when available, should be used as reference standards for quantitative haemostasis assays. Where IS are available, activities or antigenic values of coagulation factors and inhibitors should be expressed in International Units (iu), except fibrinogen which is routinely reported in g/l. Normal and pathological quality control (QC) material obtained from commercial sources or prepared in-house is essential to evidence the quality of performance of the assay and should be included in each assay run or sample batch to ensure adequate assay precision and safeguard against reporting invalid results. Commercial QC material is available for some tests with stated values towards critical cut-off points (i.e. D-dimers) when not covered by normal and pathological preparations. Clinical laboratories should participate in accredited external quality assurance (EQA) schemes for each analyte to monitor performance and assess comparability with peer group laboratories. When establishing an assay, laboratories should determine the linear portion of the dose–response curve, which will require multiple dilutions (at least three dilutions are required to establish linearity) e.g. serial doubling dilutions. A wide-ranging dilution curve is desirable as this maximises the quantitation capability of the assay. However, for some assays, linearity of responses can only be achieved over a limited range (Fig 1). In these cases, separate reference curves should be established for high and low concentrations of the analyte. Re-dilution of test samples may also be useful. Ideally, a fresh calibration curve should be carried out for each batch of assays, but a stored calibration curve is acceptable with prior validation and proof of stability [defined by acceptable internal quality control (IQC) and EQA performance]. Calibration curves should be renewed on a regular basis, the frequency determined by the local laboratory workload, but always when changing reagent lots. Multiple dilutions of test samples should be assayed where one-stage factor assays are used, to demonstrate parallelism. The cut-off may be predefined locally by the instrument/reagent combination; however agreement of <20% deviation from each other is often considered linear or parallel.31 As this value is sometimes perceived as too high as a single parameter, other values such as calibrator and test r values, and calibrator to slope test ratios may be of use to identify sample activation or the presence of inhibitors; this is not required for chromogenic assays insensitive to this type of interference, or for fibrinogen activity assays. The limits of detection must be established for each assay. In clotting assays, a lower indication is obtained from the buffer blank time. However, to establish an accurate quantitation limit in clotting, chromogenic and immunoassays, the lower limit of quantification (LLOQ) i.e. the minimum dilution of the analyte that reliably gives a value statistically significant from zero should be determined.30, 32 When clinical samples are tested, results below the quantitation limit should be reported as being less than this value. Normal reference ranges should be established locally with an appropriate number of healthy normal subjects; apparently normal patients should not be used. A minimum number of 120 subjects is recommended for establishing reference ranges.33 Transference of the reference range may be accepted from ranges quoted by manufacturers or other sources (where resources are limited) using 20–40 normal subjects, depending on the required accuracy by defining acceptable concordance in advance.29, 30, 33 Reference ranges may vary for different reagent and analyser combinations. For some assays age, sex and/or blood group-related ranges may be employed, and should be considered in the interpretation of results; examples including D-dimer, free protein S antigen and von Willebrand factor (VWF) respectively. Pregnancy is also known to affect levels, particularly of free protein S antigen, FVIII and VWF. Specific and separate ranges are required for paediatric populations as differing factors reach adult levels at differing times.34, 35 Generating local reference ranges for each assay for each age range is likely to be unfeasible for most laboratories. Published ranges should be considered if based upon similar analyser/reagent combinations, otherwise referral to a specialist centre should be considered.37 For data fitting a normal or Gaussian distribution, mean ± 1·96 standard deviation (SD) is a usual definition employed for a reference range and commonly rounded to ±2 SD. If data are known to be skewed then normalisation, for example by log transformation or the use of non-parametric evaluation, may be used to calculate the interval.33 Clinical significance of results is an important part of data interpretation, and clinical definitions (e.g. for haemophilia and VWD) do not always match the statistically calculated reference range.38 For some analytes, cut-off values derived from normal and patient populations are more clinically useful. Receiver operating characteristic (ROC) analysis is usually required, for example when studying D-dimer for deep vein thrombosis (DVT) exclusion.39 If it cannot be generated locally due to the numbers or resource then the manufacturer's recommended cut-off should be locally validated on a smaller sample (20 is often suggested). Laboratories should report reference ranges with every assay result. Clinical laboratories predominantly use one-stage clotting assays (OSCA) to measure factor levels based on the ability of test samples to correct the clotting time of factor-deficient plasma. PT-based OSCA are predominantly used for FII, FV, FVII and FX, and APTT-based OSCA for FVIII, FIX, FXI, FXII, high molecular weight kininogen (HMWK) and prekallikrein (PRK). An APTT-based OSCA can also be used to measure FII, FV40 and FX,41 and snake venom-based OSCA assays can be used to measure FII42 and FX.41 Some rare coagulation factor variants may only be detected by these alternative assays.40 Prothrombin time reagents can contain heparin neutralisers and vary in the source of tissue factor (TF) and phospholipids used. The source and composition of TF can influence measured levels in some cases43, 44 and whether the reagent is sensitive to direct factor Xa inhibitors (DFXaI)45 or LA.46 Reagents for the APTT vary in the type of contact activator (e.g. ellagic acid, silica, kaolin), which can influence the sensitivity of the reagent to contact factors (including FXI and FXII,47 HMWK and PRK48, 49). Modifying the incubation time of the APTT has been reported as increasing the sensitivity to contact factor deficiencies and may be of use as a screen in centres where specific deficient plasmas are not available (subject to local validation).50 The source and concentration of phospholipid can also influence how sensitive the reagent is to LA.51 Reagents for APTT do not contain heparin neutralisers but vary in sensitivity to heparin and DFXaI.45 Commercially available factor-deficient plasmas can be from inhibitor-free congenitally deficient patients, or immunodepleted or immuno-absorbed normal plasmas, and can be lyophilised or frozen. Although it may be seen as a cost saving in underresourced circumstances it is not recommended to use in-house patient sourced plasma. The deficient plasma should have <1 iu/dl of the deficient factor (locally confirmed for each batch), and normal levels (above 50 iu/dl) of the non-deficient factors. Factor VIII-deficient plasmas used for the Nijmegen modification of the Bethesda assay for FVIII inhibitors should also have normal levels (above 50 iu/dl) of VWF, as anomalous results have been reported without it.52 Chromogenic assays (CA) are available for various clotting factors53 but are predominantly used for FVIII and FIX assays. These are two-stage assays where test plasma is added to a mixture containing the appropriate co-factors, FX, thrombin or prothrombin, calcium ions and phospholipid. After incubation the amount of activated FX (FXa) generated is proportional to the amount of functional FVIII or FIX in the sample.54 The FXa is then assayed in the second step using a chromogenic substrate. The results obtained by the OSCA and CA are concordant in most instances, but in certain situations can differ significantly.55-58 Up to 10% of non-severe haemophilia A cases have a normal APTT and normal FVIII by OSCA, but a reduced FVIII by CA.59 The opposite discrepancy (FVIII reduced in OSCA but normal in CA) has also been described.60, 61 As CA rely on the formation of FXa, DFXaI may interfere with the assay and cause falsely reduced results.45 These assays tend to be LA-insensitive.62 Laboratory monitoring of FVIII or FIX replacement therapy for treatment of haemophilia A or B is performed to ensure optimal therapy, and OSCA or CA of FVIII and FIX are used for this purpose. It is beyond the scope of this guideline to discuss the new modified molecules that are available as treatment options but is worth noting that in many cases these assays are not suitable in their unmodified form for measuring them accurately.63, 64 Factor VIII can be measured in the presence of emicizumab (Hemlibra®) using a CA but only if the components are bovine in origin. A modified form of FVIII OSCA assay can be used to estimate emicizumab concentration in combination with a product-specific calibrator.65 Plasma samples found to have an abnormal screening PT or APTT may be further investigated to define the abnormality by performing mixing tests: abnormal screening tests are repeated on equal volume mixtures (50:50) of normal and test plasma. Correction of the prolonged result into the reference range suggests the absence of an immediate-acting inhibitor, but it is important to note that some FVIII inhibitors are slow-acting (and therefore mixing may correct if the APTT is performed immediately after mixing) and that the dilution of the antibody in a 50:50 mixing study may also normalise the result into the reference range. A lack of correction in mixing studies suggests the presence of an immediate-acting inhibitor, the most common of which is LA. It should also be noted that FIX inhibitors are typically fast-acting,66 as are some FVIII inhibitors in acquired haemophilia A. The results of immediate or incubated mixing studies are often not clear-cut and formulae such as those of Rosner or Chang may be helpful. However, no approach is both 100% sensitive and specific.67, 68 If the result of more than one OSCA is reduced, this may indicate the presence of a non-specific inhibitor (LA) or a high-titre specific inhibitor. The OSCAs should be examined for non-parallelism and repeated at higher plasma dilutions if necessary to dilute the inhibitor and confirm its specificity (Fig 2). Factor VIII or FIX inhibitors should be quantified with a Bethesda assay,69 with Nijmegen modification for FVIII.70 Incubation at 37°C for 2 h is required. This should be repeated with a porcine substrate replacing the human normal plasma to quantify cross-reactive inhibitors to porcine FVIII.71, 72 Commercial assays are available and may be useful to detect FVIII antibodies if a is or inhibitors that increase than deficiencies of and have all been and when due to antibodies can be from deficiencies by PT and/or APTT mixing For slow-acting inhibitors such as those with FVIII incubation at 37°C may be required before testing. should be as often with the of laboratory for inhibitors in and acquired can be found Plasma fibrinogen is a which in different As an fibrinogen can be in and underlying before being and APTT-based assays are not sensitive for the of fibrinogen and are all with a in some may to a or have no clinical A high concentration of thrombin is added to test plasma and the clotting time is The test result is with a calibration curve prepared by clotting a of dilutions (at least of a reference sample of known fibrinogen Samples with clotting outside the linear part of the standard curve should be and may be in and in disease which may the but does not the assay. levels of products or direct thrombin inhibitors may some assays depending on the thrombin concentration in the in of fibrinogen The assay is not usually by levels of heparin of in fibrinogen have been reported in patients on concentration in plasma can be measured using an with similar results to a These assays are essential in or where antigen levels be that of the activity assay, and where both antigen and activity are The assays must be performed on from the sample to due to fibrinogen being an If the PT is measured using a photo-optical system then a fibrinogen can be derived from a calibration curve. The results vary according to the PT In one EQA PT-based higher fibrinogen values than assays, and a in and patients and those with in the derived fibrinogen is to the antigenic value and higher than the The derived fibrinogen is not recommended for clinical and can be used to monitor fibrinogen levels the of use is reviewed in the BSH the clot assay for factor has the being and to It the of plasma clotted with thrombin and calcium ions to or This assay has been shown to be and to results may only be seen in cases iu/dl) and International Society on Thrombosis and Haemostasis and guidance is that it is not assays are based upon the activity of in the clot assays based on this have been including activity and and activity assays are and to and may be to falsely results in the presence of the most common CA based upon have been for use on routine coagulation analysers and are reported to be sensitive to However is to to of values at levels so laboratories should confirm a step based on 1 is included in the antigen levels can be assayed using to of and the and them if necessary but should not be used as a screen on its analysis of the and has that over of reported deficiencies result from and may be in of high where the specific is parameters are to as well as and from changes in tissue activator and activator and so collection and handling of samples must be and deficiencies are rare but significant Commercial and are available for measurement of activity and antigenic levels which should be considered in the of a patient with a if other tests In the of activator inhibitor however, the lower end of the normal range is often zero which it to possible cases of from the normal requires The of tests for is in a previous BSH guideline and may include assays of protein protein S and tests for the and factor As when performing into a a routine clotting screen should be performed to the effects of underlying or This may include a fibrinogen and the of a to the presence of or Specific of protein and protein S may value. for if often included as part of a screen on performing testing is Chromogenic heparin assays are and are automated and are or FXa is incubated with plasma and and activity is measured using a chromogenic substrate. human assays and therefore these are not and human or bovine assays should be used. activity may be reduced due to in the or with effects In type activity assays may be with levels measured by bovine assays both being reported as lower or higher than those measured with the incubation time of the sample dilution with the may increase the sensitivity of assays to type however no single assay can be to detect all type may be in assays in the presence of and by assays in the presence of antigen may be measured by with and assays the The assay is only to identify type which has clinical particularly for which have a low in An alternative is molecular analysis of the To measure is to activated by the from snake The generated can be measured using a CA or clotting assay. Chromogenic assays are recommended for functional assays because they are more specific than clotting assays. However, the have specificity so the assays include a number of inhibitors but in some circumstances the substrate can be by various other (e.g. factor to of In assays, this may be to but may be by linearity of the for patient calibrators and a control should be considered where effects of non-specific are for example with paediatric and in patients with or pathological may be by CA in patients because of the presence of which has similar to the Although are more specific than clotting assays, type can only be detected in the clotting assay. The of to detect this rare functional by CA must be against the specificity and precision of clotting In the clotting assay the clotting time of plasma of or and on assay if the levels are reduced the clotting time will be The phospholipid composition of the reagent in APTT-based assays is as it sensitivity to and There is also in the of the of may in samples with high FVIII or in the presence of of patient plasma in plasma may or remove the influence of assays can also be activated by which has the of not being by high levels of FVIII still be by and As for all clotting assays, at least three dilutions of patient plasma should be tested so that can be may be by depending on the reagent phospholipid concentration and and DFXaI can interfere with results. from activated samples due to may cause of and should be repeated if activation is age is

Topics & Concepts