International consensus guidelines on anticardiolipin and anti–β2‐glycoprotein I testing: Report from the 13th International Congress on Antiphospholipid Antibodies
Introduction and scope of the work
Standardized enzyme‐linked immunosorbent assays (ELISAs) for anticardiolipin (aCL) and anti–β2‐glycoprotein I (anti‐β2GPI), though required for diagnosis of the antiphospholipid syndrome (APS) according to international classification criteria (1), have remained elusive, in spite of the publication of several proposals and consensus documents (2-17). Variation in the results of both aCL and anti‐β2GPI testing remains a concern that extends to many other autoimmune disease–related tests and limits their clinical utility (8-12). The emergence of new platforms and detection technologies utilizing semi‐ or fully automated analyzers poses additional challenges with respect to test standardization (18). However, aCL and anti‐β2GPI testing has been indispensable in the diagnosis and management of APS. As these antibodies are measured with immunoassays, anticoagulant therapy (unlike in the case of lupus anticoagulant [LAC]) does not affect the results, making these assays particularly useful in patients who are receiving treatment.
At the 13th International Congress on Antiphospholipid Antibodies (April 13–16, 2010, Galveston, TX), a task force of scientists and pioneers in the field discussed and analyzed critical questions related to standardization of aPL testing (16). It was recognized that although different groups (European, Australasian, American) (2-7) had published recommended protocols for the determination of aCL (first described in 1983) (19) and anti‐β2GPI (developed for the first time in the early 1990s) (20), broad international consensus is still lacking.
This task force has therefore undertaken to develop and publish international consensus guidelines on the recommended best practices for immunoassays for the measurement of aCL and anti‐β2GPI antibodies, and on the most important requirements for technical and performance characteristics (LAC testing was recently addressed at a different forum [21, 22]). The intent of this document is to provide end users (clinical laboratories) and kit developers (manufacturers and research laboratories) with state‐of‐the art recommendations and expectations regarding aCL and anti‐β2GPI testing. We also provide interpretation of aPL antibody test results, in order to assist practicing rheumatologists in understanding test methodologies and results and making decisions regarding the diagnosis and management of APS.
Note on regulatory requirements
The activity of clinical laboratories and kit manufacturers is controlled by various regulatory authorities. The regulation of manufacturers is mostly risk based, i.e., the most important factors are the safety and effectiveness of the product. Some examples of these regulatory bodies are the Food and Drug Administration (FDA) (US), Health Canada (Canada), EU Directive 93/42/EEC (through appointed national Competent Authorities) (European Union), and the Therapeutic Goods Administration (Australia).
The work of laboratories is usually regulated by professional organizations based on recommended professional practices that are required in order to obtain accreditation. (In Australia, for example, the process is managed by the National Association of Testing Authorities and all laboratories have to also comply with International Organization for Standardization [ISO] 15189 standards.) In the US, however, some elements of the work of diagnostic laboratories are actually regulated by law, through the Clinical Laboratory Improvement Amendment of 1988 (CLIA). When laboratories wish to begin implementing a new test, CLIA requires that they verify the manufacturer's claim regarding the precision, accuracy, measuring range linearity, and reference range of the assay. As the law does not specify how these verifications should be made, any professionally and statistically sound method can be used. However, most laboratories follow guidelines published by the Clinical and Laboratory Standards Institute (CLSI), a global, nonprofit professional organization that promotes the development and use of voluntary consensus standards and guidelines within the health care community.
Preanalytical issues
Type of specimen (serum versus plasma) and storage conditions.
Background.
Previous guidelines on aCL and anti‐β2GPI testing recommended the preferential use of serum specimens (3, 4), because of concerns regarding sample dilution (up to 10%) due to the presence of citrate anticoagulant in coagulation tubes when blood is drawn. Besides sample dilution, it is important to consider the possible matrix effect, originating from the presence of clotting factors and various anticoagulants, and the potential effect of contaminating platelets.
Recommendations.
Manufacturers must state in their package insert what specimen types can be used, including the type of permissible anticoagulants in the case of plasma, and laboratories must follow these instructions. The method of specimen preparation (time and force of centrifuging) should also be specified (23). When plasma is used for aCL/anti‐β2GPI testing, platelet‐poor samples are necessary (platelet count <10,000 μl−1) (23, 24).
Samples that have been heat‐inactivated at 56°C for 30 minutes should not be used since this may cause false‐positive IgM aCL results (25). As with any other autoantibody tests, best results are achieved when nonhemolyzed, nonlipemic samples are used.
Serum samples should be stored at 2–8°C and tested within 2–3 days of collection. Testing of plasma samples should be completed within 24 hours. If a longer storage period is required, serum and plasma samples should be stored at −20°C or below. Samples should not be stored in a frost‐free freezer because of the risk of repeated freezing and thawing, which may result in loss of activity (26). It is recommended that CLSI guidelines H18‐A4 and H21‐A5, regarding specimen handling and processing, be followed (23, 27).
Analytical issues
The isotype of aCL and anti‐β2GPI tested.
Background.
Several studies have demonstrated a stronger association of IgG aCL/anti‐β2GPI with the clinical features of APS compared to IgM isotypes (28-31); hence, measurement of antibody isotype is important in clinical decision making (3, 4). IgA aCL and anti‐β2GPI measurements are not included in current criteria for diagnosis of APS, although IgA aCL has been known to affect thrombus formation in mice (32) and to be associated with certain clinical manifestations of APS (33-37). There are several reports of clinical manifestations of APS in patients with isolated IgA anti‐β2GPI positivity (38, 39). In most cases, however, the IgA isotype is present together with IgG and/or IgM.
Recommendation.
In view of the emerging evidence and recent publications, the task force recommended testing for the IgA isotype (particularly IgA anti‐β2GPI) in cases in which the IgG and IgM isotypes are negative and APS is still suspected (40).
Antigen.
Recommendations.
Developers of in‐house and commercial aCL assays should use the complex of cardiolipin plus (bovine or human) β2GPI as antigen in the aCL kits. No recommendation is made regarding the origin and format of β2GPI (purified human protein or bovine serum).
For anti‐β2GPI ELISAs, whole‐molecule β2GPI of human origin should be used (since not all human anti‐β2GPI bind to β2GPI from other sources) (41-43). Negatively charged (“high” binding or gamma‐irradiated) microtiter plates are usually utilized for anti‐β2GPI assays, to increase antigen density and to mimic the binding to negatively charged phospholipids (44, 45).
New, emerging platforms have utilized other variations of the solid phase, for example, microparticles or single‐use individual wells (18). The type of the solid phase, the concentration of the antigen, and the method of coating are not addressed herein, since this information is usually proprietary and cannot be regulated. However, methods of coating and types of solid phase should be validated through the evaluation of the performance of the assay. Regulatory authorities require that new methods undergo thorough assessment, not only from an analytical, but also from a clinical point of view. This includes comparison of newer assays against existing ones to determine the agreement both in terms of quantitative results (units) and, for aCL, semiquantitative categories (low‐, medium/moderate‐, and high‐positive). Although this process requires comparison with existing assays (which can potentially be flawed), clinical studies (that may demonstrate a stronger correlation with thrombotic manifestations and/or miscarriages) are preferred to determine if a new assay provides superior performance to existing assays.
Quantitation of results and units of measurement.
Recommendations.
Results of aCL and anti‐β2GPI tests are usually expressed in units on a continuous scale. Because the relationship between the amount of analyte and the measurement signal is not linear in immunoassays, the use of a calibration curve is required for accurate determinations (46). The calibration curve establishes the relationship between the antibody activity and the signal that is produced during the measurement procedure. Assays with 1‐point calibration may result in quantitation bias, especially in higher antibody ranges.
As the results of the aCL and anti‐β2GPI tests are not expressed in International Units (because of the lack of an international reference standard), the assays are called semiquantitative in spite of the use of calibration curves. Requirements for the developers of semiquantitative assays of this type include determination of the analytical measuring range (minimum and maximum values), demonstration of measurement linearity (the ability to provide results that are directly proportional to the concentration/amount of the analyte in the test sample), and demonstration of acceptable precision and accuracy throughout this range.
It is recommended that results of the aCL test be expressed in IgG phospholipid (GPL) units/IgM phospholipid (MPL) units (1) (see also Calibration, below). One GPL/MPL unit is defined as the cardiolipin‐binding activity of 1 μg/ml of affinity‐purified IgG or IgM aCL antibody (13). It is recommended that in addition to GPL/MPL units, results also be reported as low‐, medium/moderate‐ or high‐positive, with the range stated for each (14).
For anti‐β2GPI assays, universal units of measurement are not available. In‐house assay users and commercial kits express the results in arbitrary units (units/ml, standard IgG, IgM, and IgA units, ng/ml, optical density values, etc.) (47). With some anti‐β2GPI assays that utilize monoclonal antibody preparations, results are reported in μg/ml (6, 7, 48, 49). Based on surveys of large numbers of users and experts, this task force unanimously recommended the development and establishment of international units for measurement for anti‐β2GPI antibodies (16). The development of international standard units for measurement of anti‐β2GPI will greatly facilitate the uniformity and comparability of results among different assays.
Calibration of the assays.
Background.
Tests for aCL are traditionally calibrated based on calibrators produced by Louisville APL Diagnostics, which are traceable to the original Harris standards (13-15). Three generations of polyclonal calibrators have been developed and widely distributed based on the values derived from the original standards (15). Typically, these “reference calibrators” are not used for routine day‐to‐day purposes, but rather are utilized by manufacturers and laboratories to assign calibrant units to their own “working (kit) calibrators.”
Alternatively, various monoclonal antibody preparations can be used as standards, including those distributed by the Centers for Disease Control and Prevention (CDC) under the auspices of an International Union of Immunological Societies/World Health Organization/CDC/Arthritis Foundation committee. The use of monoclonal aCL and anti‐β2GPI preparations may have some theoretical advantages compared to the polyclonal reference standards, such as a potentially unlimited supply and better reproducibility over longer periods of time (16). However, these preparations may not behave like patient‐derived material and may not include some of the diverse specificities of aPL antibodies present in APS. Some monoclonal preparations may also be unreliable, as recently demonstrated in a workshop at the 2010 International Congress on Antiphospholipid Antibodies (50). The concentrations of the aCL/anti‐β2GPI monoclonal antibody preparations are expressed in protein concentration units, and as yet, no effort has been made to cross‐validate these against GPL/MPL units, further complicating the issue of quantitation and interpretation of results.
Recommendations.
Standards and calibrators.
Kit developers (manufacturers and in‐house assay users) are strongly encouraged to select a reliable “primary standard” (whether human‐derived polyclonal or monoclonal) to use for standardization of their assays (51). When preparing secondary standards and working calibrators, the task force recommends that the proposed new calibrant first be compared against the primary standards to ensure accuracy, and that selected groups of actual patient sera then be used (given the heterogeneity of aPL) to further demonstrate comparability (correlation and agreement) of results produced with the use of primary standards and working calibrators, as well as of different lots/batches (51, 52). The production and quality control of primary standards and working calibrators should be performed using Quality System Management guidelines (from the FDA or ISO). The task force suggests that unit values of secondary standards and working calibrators should be rigorously defined prior to use, to ensure traceability to the primary standard (51).
Calibration curves.
Appropriate quantitation of aCL and anti‐β2GPI antibodies requires multipoint calibration and the use of statistically correct curve fitting and calculation methods (46). A calibration curve should be included in each run. It is recommended that a calibration curve should be rejected if the correlation coefficient between assay readings and expected values of the calibrators is <0.90 (2, 14).
If personnel at a laboratory choose to develop an in‐house semiquantitative aCL or anti‐β2GPI assay, they should follow the recommendations above regarding calibration curve, traceability, precision, and linearity. If they decide to use commercial kits, they should select a kit that fulfills the criteria above and are advised to verify the manufacturer's claims regarding measurement linearity and precision. The CLSI EP6‐A document Evaluation of the Linearity of Quantitative Measurement Procedures: A Statistical Approach (53) provides a recommended protocol for the verification of linearity.
Precision.
Background.
Precision is one of the most important performance characteristics of a laboratory test that reflects the “closeness of agreement” between independent test results (54), often referred to as repeatability and reproducibility. Precision is usually expressed quantitatively in terms of “imprecision,” by means of the coefficient of variation (CV). In quantitative and semiquantitative assays, acceptable precision should be demonstrated throughout the measuring range, but especially at clinically important decision levels, for example around positive/negative and low‐positive/moderate‐positive cutoffs. For diagnostic immunoassays, CV of ≤10% is usually recommended (55). It should be noted that 20% is usually considered as the maximum CV for quantitative (or semiquantitative) measurements. The concentration at which the CV goes beyond this value (usually toward the lower end of the calibration curve) is considered the functional sensitivity of the test (the “limit of quantitation,” to use new terminology) (56). Below this level the measurement cannot be considered quantitative (or semiquantitative), and should be reported as “<x value.”
Recommendations.
It is required that the precision of commercially developed assays be evaluated at several levels by the manufacturers, and published in the instruction manuals. The between‐run (or total) precision reflects the precision of the assay during routine use (within‐run precision is usually too optimistic), so ideally it should be the basis of judging the performance of a test. Technological advances in the field of automation, manufacturing, and measuring procedures make the above‐mentioned goal of 10% CV realistic even for aPL antibody measurements. Developers of commercial and in‐house assays should aim for this performance goal. For traditional, manually performed assays, a level of 15% imprecision would still be acceptable.
When purchasing and using commercial kits, precision claims should be an important part of the information that influences kit selection by laboratory personnel. Verification of the manufacturer's claim regarding precision is a compulsory part of the method implementation process in the US. Although not currently mandated, this practice is also recommended outside of the US, as it is the only way to determine if the particular laboratory is able to achieve precision similar to that claimed by the manufacturer. The CLSI EP15‐A2 guideline Verification of Performance for Precision and Trueness (54) provides recommended protocol and statistical considerations for the verification process.
Use of external positive and negative controls.
Recommendations.
The incorporation of at least one “external,” non–kit‐supplied (either from commercial sources or prepared and properly stored in‐house) positive control in every run, to monitor interassay variation, is recommended (3, 4, 6). This external non‐kit control should ideally have a value approximately at the low‐positive to medium/moderate‐positive cutoff of the assay/kit. Similarly a “negative” control with a value below the cutoff of the assay should be included in each run. A run should be rejected if the result with either the positive control supplied in the kit or the external control falls outside its acceptable range or if the result with the negative control is above the established cutoff value (2).
Singlet/duplicate measurements.
Recommendations.
The issue of singlet versus duplicate testing remains a critical one, in terms of the opposing desires to reduce cost and to achieve a low CV. Recommendations regarding singlet/duplicate testing must be based on performance characteristics (especially precision). If the assay was developed, validated, and showed acceptable precision based on singlet testing, duplicate measurements are not required. If (during the method implementation and performance verification process) the precision does not meet the above‐mentioned performance goals, precision could be re‐verified by testing in duplicate. Running an assay in duplicate because of unsatisfactory precision in singlet testing is discouraged unless it is determined that duplicate testing actually provides better performance than singlet testing.
Cutoff calculation.
Background.
Reference ranges for aCL and anti‐β2GPI antibodies should be established by the nonparametric percentile method, as autoantibody values usually do not follow normal distribution (2-4, 6). Proper establishment of the reference range requires a minimum of 120 reference subjects (taking into consideration the age and type of population most representative for each laboratory) and a carefully planned protocol regarding inclusion and exclusion criteria, specimen collection and storage requirements, and statistical methods (57).
Recommendations.
Kit developers.
Kit developers (manufacturers and laboratories using in‐house methods) who have previously established the cutoff for their already marketed kits using a different method are encouraged to reanalyze the data using the nonparametric percentile method and publish the 95% and 99% values in the instruction manual, or to make these data, in addition to the previously established reference range, available upon request. Details of the reference range study, including the number of subjects and demographic information and the statistical method used, should also be disclosed in the instruction manual or be made available upon request, since this information is crucial for laboratories in their decisions with regard to accepting the manufacturer's reference range.
End users.
End users (diagnostic laboratories) rarely have the resources to conduct a proper reference range study. Because of the difficulties involved in correct establishment of the reference interval, the task force suggests that laboratories should focus on reference range verification or transference instead, using professional guidelines (57).
End users of commercial assays should start with a careful review of the instruction manual for the selected assay. If the details of the reference range study are included (or can be obtained from the manufacturer upon request) and the method of establishing the cutoff appears to be appropriate, the laboratory can accept the manufacturer's reference range after a verification process. This can be a subjective assessment that includes a search and review of the published literature and comparison of the demographic characteristics of the reference subjects with those of the population served by the laboratory. Alternatively, the reference range can be verified on a small number of appropriately selected reference individuals (minimum 20). In this situation, at least 18 of the 20 results should fall into the reference range for successful verification. (57). Another option is reference range transference. In this case the laboratory would have already established a reference range for the same analyte with a specific method. If the laboratory personnel want to switch to a new (but similar) method, they can use data obtained by comparison of the old and the new methods, and can calculate the new reference limits by using the regression equation describing the correlation between the old and the new assays.
However, if the verification is unsuccessful, if information on the manufacturer's reference range study is lacking or incomplete, if it was not performed appropriately, or if the demographics of the reference population are significantly different from those served by the laboratory (for example, a pediatric population), a formal reestablishment of the reference range by the laboratory may be necessary. The CLSI C28‐A3 document contains professionally and statistically sound protocols for reference range establishment and verification (57), and it is recommended that both manufacturers and laboratories follow these protocols.
Rheumatoid factor (RF) interference.
Background.
IgM‐RF has been mentioned as one of the factors capable of causing interference in IgM aCL assays. Published aPL antibody testing guidelines are, however, inconsistent, with some not addressing RF interference at all (1, 6, 7). Even when interference is mentioned, the nature, degree, or mechanism of the phenomenon is not described. Laboratory personnel and clinicians generally have not had evidence‐based information on how IgM aCL results should be interpreted when RF is present.
RF is one of the main sources of interference in immunometric (sandwich‐type) immunoassays (58, 59). It has also been recognized as an interfering factor in tests for IgM antimicrobial antibodies (60), which has eventually led to the development of capture immunoassays for the detection of IgM antibodies (61). Its influence on autoantibody testing, however, has not been well studied. For a long time, there were only 2 publications in the scientific literature on the influence of RF on aPL testing, containing limited data (62, 63). A study aiming to clarify the issue of RF interference on aPL testing has recently been published. It suggests that the coexistence of medium or high levels of IgG aCL/anti‐β2GPI antibodies and high levels of IgM‐RF results in a significant, dose‐dependent positive bias in the IgM aCL/anti‐β2GPI measurement, and can consequently cause false‐positive IgM aCL/anti‐β2GPI results (64). The mechanism of the interference is probably IgM‐RF binding to IgG antibodies bound to the solid‐phase antigen, as has previously been suggested by Agopian et al (63). The degree of interference depends on the level of both the IgG aCL/anti‐β2GPI antibodies and the IgM‐RF (64). Given these findings, it is possible that previously unrecognized RF interference has contributed to findings of lower clinical specificity of IgM aCL/anti anti‐β2GPI antibodies.
Recommendation for manufacturers.
Based on this newly published information, IgM‐RF interference should be considered an important issue. Manufacturers and developers of in‐house assays should therefore properly evaluate their assays for RF interference and provide data on the level of IgM‐RF that may result in interference in their IgM aPL kits. As every test system is different regarding the amount of solid phase–bound antigen, serum dilution, incubation conditions, etc., interference data determined in various aPL tests may vary according to the specific assay conditions; i.e., data obtained with one aPL assay are not interchangeable with those obtained with an alternative assay (64).
Recommendation for laboratories.
Although IgM‐RF can cause falsely elevated IgM aCL/anti‐β2GPI results, this is unlikely to result in misclassification or misdiagnosis, and the interference appears to be most significant when the IgG aCL antibody level is at least moderately positive, which by itself would already be sufficient for a diagnosis of APS. However, the combination of IgM‐RF and IgG aCL in lower concentrations may still be sufficient to produce minor shifts in the IgM aCL values and thus affect the accuracy of the findings. The issue of RF interference should therefore be addressed in interpretative comments when reporting IgM aPL results (see below). Although IgG can be removed from patient samples to prevent interference, those procedures are labor intensive and not standardized, and their efficiency is difficult to measure.
Postanalytical issues
Reporting of results.
Background.
The current classification criteria define clinically significant titers of aCL as >40 GPL units and >40 MPL units, or higher than the 99th percentile of the reference range obtained with normal subjects (1). However, this statement is somewhat confusing, as a value of 40 GPL or MPL units can be substantially different from the 99th percentile value (65, 66). Moreover, the main purpose of the classification criteria is to establish common ground for conducting studies, and not to make diagnoses. It is important to realize that the risk associated with aCL and anti‐β2GPI antibodies increases on a continuous scale, and the presence of multiple aPL antibodies is also associated with higher risk compared to single positivity (67-69). This is one of the reasons testing for LAC is necessary whenever APS is suspected. Moreover, different levels of an antibody may be associated with different symptoms (65, 70, 71). According to one study, for example, IgG aCL levels <21.4 GPL units were associated with low (7%) probability of thrombosis, while IgG aCL levels of 21.4–65.0 GPL units and >65 GPL units were associated with an increased probability of thrombosis (by 20% and 75%, respectively) (69).
Traditionally, IgG or IgM aCL levels of ≥80 GPL or MPL units have been defined as high‐positive, levels between ≥20 and <80 GPL or MPL units as medium/moderate‐positive, and levels that are higher than the cutoff of the assay for each isotype but <20 GPL or MPL units as low‐positive (15). However, the basis of this classification was not determined scientifically, and given the variability of aCL test results between different assays, it may not be valid for every assay. In an effort to better define the “low‐positive” range, Budd et al conducted a study in which the prevalence of positive IgM aCL antibodies was determined in a large cohort of healthy individuals (total 1,141 samples) using 3 different IgM aCL assays (2 commercial kits and an in‐house assay) (66). The 95th and 99th percentile aCL levels were calculated. It was also demonstrated that the vast majority of the IgM aCL–positive samples in this large population of healthy individuals fell between the 95th and the 99th percentiles, but were not considered to be clinically significant. The authors proposed that the range of values between the 95th and the 99th percentile be considered (or reported) as an “indeterminate” or “grey” zone.
Recommendations.
It is recommended that aCL levels be reported as numerical (unit) values, as well as semiquantitatively in defined ranges. It is also recommended that low‐positive, medium/moderate‐positive, and high‐positive ranges be established according to the percentile values of reference results, as follows: negative = lower than the cutoff; low‐positive or indeterminate range = between 95th and 99th percentiles; medium/moderate‐positive = from the upper value of the indeterminate range to 80 GPL/MPL units; high‐positive = >80 GPL/MPL units.
With respect to anti‐β2GPI, there is little or no evidence to support the use of defined semiquantitative ranges. Hence, anti‐β2GPI results should be reported as negative (below the cutoff) or positive, as well as in numerical units.
Interpretative comments.
Background.
Among the existing aCL/anti‐β2GPI testing guidelines, only those published by the Australian Anticardiolipin Working Party have addressed the issue of interpretative comments (3, 4).
Recommendation for manufacturers.
Manufacturers should disclose information and/or data from clinical studies, if such results are available.
Recommendation for laboratories.
Instead of giving specific instructions, the task force recommends addressing the following issues in the interpretive comments: 1) the risk associated with a particular test result, with emphasis on the continuous scale of risk associated with increasing antibody concentrations, and the increased risk associated with positivity for multiple different aPL; 2) the issue of RF interference (including preferably stating the threshold IgM‐RF concentration at which this may become an issue); and 3) recommended further diagnostic actions (for example, repeating the test, or testing for LAC or anti‐β2GPI).
Summary
The task force's evidence‐based recommendations for aCL and anti‐β2GPI testing are summarized in Table 1. Adoption of these guidelines by test users, developers, and manufacturers will help in the standardization and harmonization of assays.
| Assay characteristic | Recommendations |
|---|---|
| Specimen requirements | Serum: Heat inactivation at 56°C for 30 minutes should be avoided. Use of nonhemolyzed, nonlipemic samples is recommended. |
| Plasma: Manufacturers must specify the specimen type, including the anticoagulant used. Platelet‐poor plasma (<10,000/μl−1) is required. Use of plasma should take into consideration the dilution factor that may be produced because of the anticoagulant. | |
| Isotype of aCL and anti‐β2GPI tested | The IgG and IgM isotypes are recommended for both aCL and anti‐β2GPI. |
| The IgA isotype is recommended for both aCL and anti‐β2GPI when results of all other tests are negative and APS is still suspected. | |
| Antigen | aCL: Cardiolipin + (bovine or human) β2GPI should be used. |
| Anti‐β2GPI: β2GPI of human origin should be used, on a negatively charged (“high” binding or gamma‐irradiated) plate. | |
| Quantitation of results | aCL: GPL/MPL units should be measured and ranges (low‐positive, medium/moderate‐positive, high‐positive) established. |
| Anti‐β2GPI: Universal units of measurement are not available. Development/establishment of international/universal units of measurement is recommended. | |
| Standards | Manufacturers and test users are strongly encouraged to select a reliable standard to prepare secondary calibrators (polyclonal or monoclonal). |
| The proposed secondary calibrators should be compared and validated against the primary standard, using published and accepted procedures. | |
| Selected groups of actual patient sera should be used if possible, to further establish the extent of agreement in the assay/test system. | |
| Most importantly, the production and the quality control of the standards should be subjected to FDA Good Manufacturing Practices guidelines or an equivalent quality assurance program. | |
| A record of traceability from the recommended standards to any secondary calibrators is required. | |
| Calibration curves | Multipoint calibration and use of statistically correct fitting and calculation methods are required. |
| A calibration curve should be included in each run. | |
| The calibration curve should be rejected if the correlation coefficient between assay readings and expected values of the calibrators is <0.90. | |
| Precision | CV of <10% is recommended for ELISAs. |
| Precision should be evaluated at multiple levels of positivity, including low‐positive to medium/moderate‐positive cutoff level. | |
| For commercial kits, expected precision should be published in manufacturers' instruction manuals. | |
| Positive/negative controls | Incorporation of at least 1 “external” positive control in every run to monitor interassay variation is recommended. |
| Ideally, use of a sample with a value approximately at the low‐positive to medium/moderate‐positive cutoff of the assay/kit and aiming to achieve the recommended precision (CV) is strongly encouraged. The value of this control should be also within the linear range of the assay/kit. | |
| Similarly a “negative” control with values below the cutoff of the assay should be used in each run. | |
| A run should be rejected if the result with either the positive or the negative control falls out of the established range. | |
| Singlet/duplicate measurements | If precision is satisfactory based on single testing, then duplicate measurements may not be required. |
| Cutoff calculation | Cutoffs should be established using a nonparametric percentile method. |
| Manufacturers are encouraged to establish and to report cutoffs based on the 95th and 99th percentiles. | |
| Clinical/diagnostic laboratories using commercial kits should validate/verify cutoffs provided by the manufacturer. | |
| Rheumatoid factor interference | Rheumatoid factor can affect the results of the tests, and this should be addressed in the interpretative comments. |
| Reporting of results | Results should be reported in units and in semiquantitative ranges. |
| Interpretations should include negative, “indeterminate” or grey zone, medium/moderate‐positive, and high‐positive. | |
| Interpretative comments | Inclusion of comments to assist clinicians in the interpretation of test results is strongly recommended. |
| Manufacturers should disclose information based on evidence derived from clinical studies that may assist with the interpretation of results. |
- * aCL = anticardiolipin; anti‐β2GPI = anti–β2‐glycoprotein I; APS = antiphospholipid syndrome; GPL = IgG phospholipid; MPL = IgM phospholipid; FDA = Food and Drug Administration; CV = coefficient of variation; ELISAs = enzyme‐linked immunosorbent assays.
AUTHOR CONTRIBUTIONS
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Pierangeli had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Favaloro, Harris, Meroni, Tincani, Wong, Pierangeli.
Acquisition of data. Lakos, Favaloro, Harris, Meroni, Wong, Pierangeli.
Analysis and interpretation of data. Lakos, Favaloro, Meroni, Tincani, Wong, Pierangeli.
Acknowledgements
We thank Dr. Elizabeth Hsu for help with the preparation and revision of the manuscript.