Article Text
PDF
Abstract
Anti-nuclear antibodies (ANA) are fundamental for the diagnosis of autoimmune diseases, and have been determined by indirect immunofluorescence assay (IIFA) for decades. As the demand for ANA testing increased, alternative techniques were developed challenging the classic IIFA. These alternative platforms differ in their antigen profiles, sensitivity and specificity, raising uncertainties regarding standardisation and interpretation of incongruent results. Therefore, an international group of experts has created recommendations for ANA testing by different methods.
Two groups of experts participated in this initiative. The European autoimmunity standardization initiative representing 15 European countries and the International Union of Immunologic Societies/World Health Organization/Arthritis Foundation/Centers for Disease Control and Prevention autoantibody standardising committee. A three-step process followed by a Delphi exercise with closed voting was applied.
Twenty-five recommendations for determining ANA (1–13), anti-double stranded DNA antibodies (14–18), specific antibodies (19–23) and validation of methods (24–25) were created. Significant differences between experts were observed regarding recommendations 24–25 (p<0.03). Here, we formulated recommendations for the assessment and interpretation of ANA and associated antibodies. Notably, the roles of IIFA as a reference method, and the importance of defining nuclear and cytoplasmic staining, were emphasised, while the need to incorporate alternative automated methods was acknowledged. Various approaches to overcome discrepancies between methods were suggested of which an improved bench-to-bedside communication is of the utmost importance. These recommendations are based on current knowledge and can enable harmonisation of local algorithms for testing and evaluation of ANA and related autoantibodies. Last but not least, new more appropriate terminologies have been suggested.
- Autoantibodies
- Autoimmunity
- Systemic Lupus Erythematosus
- Sjøgren's Syndrome
Statistics from Altmetric.com
Introduction
Autoantibodies are a hallmark of autoimmunity, of which anti-nuclear antibodies (ANA) have taken the centre stage for the past 60 years. The term ANA is now outdated and even confusing as this historical label has come to encompass antibodies directed at various cellular compartments including nuclear constituents, components of the nuclear envelope, mitotic spindle apparatus, cytosol, cytoplasmic organelles and cell membranes. Detection of anti-cellular antibodies of the ANA family is pivotal to the diagnosis of many autoimmune diseases.1 ,2 Moreover, specific antibodies of the ANA family may present years before the appearance of overt disease, and for some conditions serological assays can provide useful information on the likelihood of clinical course or complications (eg, inflammatory myopathies, systemic lupus erythematosus (SLE)). Hence, the determination of ANA may enable the prediction, diagnosis and activity determination of certain autoimmune diseases.3 ,4
ANA detection by the indirect immunofluorescence assay (IIFA) was first described in 1950 by Coons and Kaplan.5 This ‘gold’ standard technique is at present performed using HEp-2 cells or variants of this cell line (ie, HEp-2000).6 The IIFA requires a laborious process consisting of serial dilutions of positive sera, visual determination of the staining pattern, followed by a second test in which antigen specificity is determined.7 ,8 IIFA entails substantial technical expertise and while it is considered the ‘gold’ standard, it is only as good as the laboratory that performs this assay. Another limitation of IIFA is its lack of specificity. Indeed, depending on demographics, the population being studied, serum dilution, the cut-off used and other variables of this assay, up to 25% of sera from apparently healthy individuals can be ANA positive.9–14 Noteworthy, in the general population some individuals with a positive ANA test by IIFA do not have an autoimmune disease and are unlikely to develop one.
In past decades, the demand for ANA testing has increased remarkably, raising the need for high throughput in service laboratories. These changes resulted in a technical evolution and the development of novel diagnostic platforms including the ELISA, addressable laser bead immunoassay, chemiluminescence immunoassay, fluorescent-enzyme immunoassay, line immunoassays10 and others.8 ,15 ,16 The latter are based on a limited number of purified and/or synthetic autoantigens. For details on the methodology see the supplementary material (available online only). The test characteristics of these novel assays are strongly distinct from the ANA–IIFA. Debates regarding ANA determination by different tools have recently resulted in a position statement of the American College of Rheumatology (ACR), claiming that IIFA is the ‘gold’ standard for ANA detection, and that alternative methods should demonstrate performance as good as IIFA (http://www.rheumatology.org/practice/clinical/position/ana_position_stmt.pdf). This position statement is primarily based on the high sensitivity of IIFA for certain autoimmune diseases such as SLE and to a lesser extent systemic sclerosis (SSc). Moreover, it is strongly supported by the historical inclusion of the ANA, as performed by IIFA, as a criterion of certain diseases. Unfortunately, IIFA specificity and sensitivity is relatively low for other diseases (eg, inflammatory myopathies) and its technical limitations as detailed above are yet to be overcome.
Therefore, an international group of experts (laboratory specialists, scientists and clinicians) have developed a set of recommendations for the appropriate assessment and interpretation of anticellular antibodies of the ANA family determined by different methods.
Methods
Expert committees
Two groups of experts participated in this initiative. The European autoimmunity standardization initiative (EASI) was formed a decade ago. This forum was assembled with the aims of improving the diagnosis of systemic autoimmune-rheumatic diseases (SARD), the rationale of autoantibody testing and to stimulate interactions between clinical and laboratory specialist while considering cost-effectiveness and novel techniques (http://www.easi-network.com/). In this study expert teams from 15 European countries, namely Italy, Germany, The Netherlands, France, UK, Sweden, Belgium, Finland, Norway, Switzerland, Austria, Spain, Ukraine, Portugal and Israel participated. Each national EASI team consists of clinicians, laboratory specialists and scientists, dedicated to the field of SARD. The final vote for each team was the average score of all participants from the same country.
The International Union of Immunologic Societies/World Health Organization/Arthritis Foundation/Centers for Disease Control and Prevention (IUIS/WHO/AF/CDC) committee for the standardisation of autoantibodies in rheumatic and related diseases (http://www.autoab.org) was established in the early 1980s, and is also affiliated with the ACR. The establishment of this group was primarily based on the recognised needs for reference human autoimmune sera that were critical for academic, clinical and industrial laboratories, as well as the need to communicate to the wider stakeholder community and to facilitate ongoing activities in continuing the mission in autoantibody standardisation.17
Development of recommendations
A three-step process was applied for creating these recommendations for the determination and interpretation of ANA and related antigen-specific antibody testing. First, 31 preliminary statements were generated by the EASI forum, based on previous algorithms from Italy, Germany, The Netherlands and the ACR. Following review of the literature, these recommendations were subjected to open comments by the EASI teams and grading by each EASI team on a scale of 1–5 (ie, full agreement to no agreement, respectively). Thereafter, based on the results of the first stage, the EASI forum reformulated a new set of 25 recommendations. These were subjected to review and modifications by both the EASI teams and the IUIS/WHO/AF/CDC committee members.
In the third stage a Delphi exercise with closed voting followed. During this exercise the final 25 recommendations were separately voted on and given a score from 0 (absolutely no agreement with the recommendation) to 10 (maximal support for the recommendation). The means and SD scores of all participants were calculated to determine the level of agreement for each recommendation. Therefore, the highest agreement is defined by the highest mean score accompanied by the lowest SD. Separate scores for the voting by the EASI groups versus the IUIS members were assessed using the t test (p<0.05 was considered significant).
Results
Twenty-five recommendations were formulated in this study and divided into four subgroups. The first 13 recommendations were dedicated to the ANA test (1–13), five to the determination of anti-double stranded DNA antibodies (14–18) and five to the determination of specific antibodies including those directed to extractable nuclear antigens (ENA; 19–23). Recommendations 24–25 addressed the need for local validation of different methods. Each recommendation was graded by a Delphi voting score composed of voting by the EASI and the IUIS groups. Scores are presented as the mean±SD of the entire forum (table 1).
The recommendations for detection of autoantibodies to cellular components commonly referred to as ANA
Significant statistical differences between the voting of the EASI teams and the IUIS members were observed only regarding recommendations 24 and 25, which received the grades 8.3±3 vs 10±0 and 7.4±2.9 vs10±0, respectively (p<0.03).
The broad spectrum of nuclear and cytoplasmic staining patterns, in relation to antibody specificity and associated disease, is specified in table 2.
IIFA nuclear/cytoplasmic patterns detected on HEp-2 substrates and related antigens/diagnosis
Discussion
In this study we formulated 25 recommendations intended to clarify and standardise clinical and technical aspects pertaining to the determination of ANA and related antigen-specific antibodies. In addition, key principles regarding terminology, discrepancies between methods and communication between laboratory and clinical specialists were addressed.
Terminology
The terms ‘anti-nuclear antibodies’ (ANA) and ‘extractable nuclear antigens’ (ENA) are no longer technically correct and do not cover the entire spectrum of relevant autoantibodies. ‘ANA’ may now detect antibodies directed against nuclear and non-nuclear elements (see recommendation 13), while ‘ENA’ may refer to antigens that are neither extractable nor nuclear. Therefore, one may suggest changing these outdated terms to appropriate ones, such as anticellular antibodies and specific antibodies, respectively. Such a change in nomenclature requires broad agreement within the medical community and an adjustment period, as most manuscripts and textbooks utilise these ‘classic terms’. In the current manuscript we have used the traditional term ANA and ENA, with the appreciation that they refer to a broader antigen specificity.
Discrepancy between platforms
The increasing demand for ANA tests led to the development of new automated methods for their detection, some of which probably represent the future of diagnostic serology.18 However, the inconsistencies between methods are currently a burden on those who perform and interpret these tests. For this reason, we have addressed the concepts of discordant results and communication between laboratory and clinical specialists. As a general rule, if discrepancy between methods is observed, especially in the setting of high clinical suspicion, another approved and validated diagnostic platform should be utilised. Notably, in certain cases, choosing another platform or deciding which method is preferred may require a collaborative assessment of clinicians and laboratory specialists.19 For example, the use of anti-dsDNA antibody titration for monitoring SLE activity is better achieved by using a quantitative assay such as ELISA or the Farr assay, whereas at the time of the diagnosis, the immunofluorescence on Crithidia luciliae immunofluorescence test (CLIFT) or the Farr assays may offer the best specificity.20 Discrepancies between clinical assessment and laboratory results have also been reported. It is our recommendation that in cases of high clinical suspicion a physician request for the determination of specific antibodies should be granted, irrespective of the result of previous tests. For instance, specific antibodies directed against Jo-1, ribosomal P or SS-A/Ro may be detected in patients who are ANA negative by IIFA. In this regard, one may consider the possibility that for an individual patient serology may change over time. Another important issue at present concerns the use of ANA/anti-DNA testing as inclusion criteria for clinical trials that may require the use of several platforms balancing specificity versus sensitivity.
Communication between the laboratory bench and the clinic
This is an all-embracing issue detailed in many of our recommendations (ie, 7, 11, 12, 13, 14, 16, 19, 20 and 21). Notably, many clinicians are unaware of the specific method or the differences between methods used for ANA detection, and may erroneously apply the test characteristics of IIFA to other methods. Therefore, by and large the autoimmune laboratory should specify each method used. In addition, in the case of ANA detection by IIFA, titres and patterns obtained should be reported. If multiple methods are being used either for ANA, anti-dsDNA or other antigen-specific autoantibodies, the results obtained by each method should be reported separately. In addition, if further testing is advised (eg, the determination of anti-dsDNA in samples with high titre homogenous ANA by IIFA) a laboratory comment may be included as detailed below.
The recommendations
The first subgroup of recommendations (1–13) is dedicated to ANA screening at the first level of diagnosis.1 Most of our knowledge regarding ANA screening is based on data obtained by IIFA. As such, the prevalence of ANA–IIFA among SLE patients is expected to be above 95%, while a lower prevalence may be detected using methods that rely on a limited number of antigens.7 ,8 Furthermore, ANA can be detected in patients with autoimmune, malignant or infectious diseases as well as in healthy subjects. Hence, the performance of any ANA assay is highly dependent on its pre-test probability, and interpretation of results should be governed by clinical circumstances.1 ,18 In these recommendations, like others, we have recognised IIFA on HEp-2 (or HEp-2000) to be the reference method of choice for ANA screening.6 ,21 However, we have also acknowledged the limitations of IIFA, including the time required for its performance and the necessity of highly qualified laboratory personnel. The latter raises the urgent need for training programmes on IIFA performing and autoimmune serology interpretation.22 In contrast, the alternative automated methods (see supplementary material, available online only) have come into common use, especially in high-throughput laboratories. These methods are simple to operate and may have the added value of better recognition of specific antibodies. Alas, they often lack sufficient sensitivity for certain diseases (ie, SLE and SSc)23–27 (recommendation 5). Considering the pro and cons, we have included the alternative assays in our recommendations as long as the platform being used is clearly specified (recommendation 6) and the result is not referred to as ‘ANA test’ or ‘ANA screen’, which are typically associated with ANA testing by IIFA (recommendation 7).
From the laboratory perspective, standardisation of IIFA is vital, and several variables of this method have been specified. The recommended substrate for ANA–IIFA derives from a human epithelial cell line (HEp-2 or HEp2000) and should contain a sufficient number of mitotic cells to enable adequate pattern recognition. Low positive control samples as well as an internal cut-off control, randomly distributed in distinct runs, may enable continuous training of technicians. The secondary detection antibody (conjugate) should consist of fluorochrome-labelled anti-human IgG-specific antiserum (recommendation 9), either FITC or another validated new-generation fluorochrome. A special consideration, at a regular interval, to the fluorochrome potency and ultraviolet bulb strength was suggested. Another variable of ANA testing is the screening dilution of sera.6 ,21 Following review of the literature and many discussions among the committee experts, it was agreed that the optimal screening dilution must be defined locally, and the titre related to the 95th percentile of healthy controls considered the optimal screening dilution. In cases of uncertainty, while taking into consideration sensitivity, specificity and cost (ie, wells on slides, technician time, other reagents), we have pointed out that a screening dilution of 1 : 160 was often found to be the most suitable for the evaluation of adult patients.6 ,21 In an international study, ANA was documented in 31.7% of healthy subjects at 1 : 40 serum dilution, 13.3% at 1 : 80 and 5.0% at 1 : 160.21 Conversely, at 1 : 160 dilution the sensitivity for SARD is not perfect and a negative result does not exclude disease. Notably, no consensus could be achieved among experts regarding the screening dilution for children (<16 years), for which some use a dilution of 1 : 40, or the need for further dilution once a positive result was obtained. Previously, it was generally accepted that individuals without autoimmune disease would present with lower serum levels while high-titre ANA are clinically more significant.12 However, there is ample evidence that low-titre ANA may be significant and that titres following the screening threshold of 1 : 80 or 1 : 160 have no bearing on diagnosis or disease activity.7 ,15
Patterns of ANA by IIFA reflect the topographic distribution of target autoantigens (recommendations 11, 12)28 and may convey significant information about antibody specificity (table 2).29 The role of cell cycle dynamics (ie, interphase, mitotic) in defining patterns is of great importance, especially for complex patterns related to autoantigens that exhibit a dynamic behaviour along the cell cycle, such as centromere protein (CENP)-F, NuMA-1 and topoisomerase-1.30–32 The recognition of both nuclear and cytoplasmic patterns enables ‘reflex testing’ (ie, testing for specific antibodies depending of the pattern) and improves the utility of the serological evaluation (recommendation 13).1 ,28
Antibodies directed against dsDNA are a key marker for diagnosis and follow-up of patients with SLE33 ,34 (recommendations 14–18). Theoretically, if ANA–IIFA is negative one should not proceed to defining anti-dsDNA antibodies,35 ,36 although if clinical suspicion of SLE is substantial anti-dsDNA antibodies assessment may be requested by the clinician. On the other hand, certain IIFA patterns (eg, homogenous) are suggestive of anti-dsDNA antibody (table 2).37 Regrettably, in most laboratories ‘reflex testing’ cannot be performed unless requested; thereby the addition of a note to the ANA–IIFA result should be considered (eg, ‘This pattern may represent antibodies directed to DNA or chromatin components. Further tests are advised if there is a clinical suspicion of systemic lupus erythematosus’). The appropriate method of anti-dsDNA antibodies determination is still a matter of debate. The Farr assay and the CLIFT offer high clinical specificity (recommendation 15).35 ,38 In point of fact, the Farr assay is the one ‘specific’ for anti-dsDNA related to SLE diagnosis, owing to its ability to detect high avidity antibodies. However, both assays have substantial drawbacks. The Farr assay requires the use of radioisotopes, and may detect anti-dsDNA antibodies induced by drugs (ie, anti-tumour necrosis factors)39 while the CLIFT is not quantitative and is accompanied by relatively low sensitivity. Alternative methods for anti-dsDNA detection have recently been developed. Most of them provide high sensitivity but are hampered by lower specificity and lack of correlation with each other.38 ,40 Reflecting on these aspects, we have recommended that positive results obtained by these alternative methods be confirmed by CLIFT or Farr assay. Nonetheless, if discrepancy between methods is observed both results should be reported. In such cases a careful explanation should be provided by the laboratory and the final interpretation should take into consideration the clinical scenario and the ANA–IIFA pattern. Last but not least, for follow-up of anti-dsDNA titres a quantitative method (ie, not CLIFT) (recommendations 16, 17), preferably by the same method and performed by the same laboratory, is recommended (recommendation 18).
The identification of antibody specificity to nuclear and cytoplasmic antigens may further differentiate between diseases41–44 (table 2; recommendation 19). Although there is a clear correlation between IIFA patterns and antigen specificity (table 2), this association is not perfect and antigen specificity cannot be deduced from the IIFA pattern alone. Furthermore, disagreements between methods to detect specific antibodies and/or IIFA patterns were reported.44–46 In these cases an alternative method can be employed (recommendations 1, 20, 21). In addition, unlike anti-dsDNA, the coexistence of negative ANA by IIFA and seropositivity for specific antibodies by solid phase assay is not rare (ie, anti-SSA/Ro).47 Thus, a physician request to determine specific antibodies should be granted, irrespective of ANA–IIFA (recommendation 23). Notably, another aspect concerning anti-SSA/Ro antibody is the detection of its variants directed at Ro60 and Ro52/TRIM21, as some tests currently in use include only the Ro60 antigen. Detection of anti-SSA/Ro is recommended in different scenarios including counselling of patients with autoimmune disease who desire a pregnancy. The latter are mainly related to anti-Ro52/TRIM21 antibodies. Therefore, specifying the antigen used and evaluation of both anti-Ro52 and anti-Ro60 should be considered. The importance of specific antibody levels is still a matter of debate. In particular, the significance of anti- ribonucleoprotein levels (recommendation 22) data are lacking regarding levels determined by current methods in comparison to the historical haemaglutination test used for the mixed connective tissue disease criterion.48
The fourth subgroup of our recommendations addresses the issue of local validation of methods (24, 25). Both recommendations were fully accepted by the IUIS group and significantly less by the EASI teams. This divergence between experts may be explained by the different regulations applied in the different countries, namely in America and Canada local validation is frequently required by the authorities. Nevertheless, as the list of available methods is bound to increase, it is highly likely that in the years to come verification of assays will be an important feature of autoimmune laboratories worldwide.
Conclusions
Defining anticellular antibodies is an essential component of the evaluation of patients with autoimmune diseases. The wide array of methods for their determination prompted questions and debate among laboratory personnel, clinicians, as well as co-authors of this manuscript. Therefore, in this study we have offered recommendations that are applicable and may facilitate the design of local algorithms for ANA detection. We have pointed out that the changes in the field of autoimmune diagnostics necessitate dedicated training of those who demand, perform and interpret the results. Therefore, in parallel to defining new autoantibodies and the development of new platforms an urgent need for training programmes for clinician, technicians and the industry has been recognised. Last but not least, further studies are required to address issues such as the local validation of platforms, significance of titres and patterns among different populations (ie, children), and different stages of diseases (ie, pre-clinical vs fully established; active vs quiescent) and the role of novel automated IIFA methods that may overcome many limitations of the manual form of this assay.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online supplement
Footnotes
-
Handling editor Tore K Kvien
-
NA-L and JD contributed equally.
-
Correction notice This article has been corrected since it was published Online First. Additional collaborators have been included.
-
Collaborators JW Cohen Tervaert (Maastricht, The Netherlands); D Hamann (Amsterdam, The Netherlands); RHWM Derksen (Utrecht, The Netherlands); H Hooijkaas (Rotterdam, The Netherlands); I Klasen (Nijmegen, The Netherlands); P Limburg (Groningen, The Netherlands); R Smeenk (Amsterdam, The Netherlands); P van Daele (Rotterdam, The Netherlands); Kerstin Elvin (Stockholm, Sweden); Charlotte Dahle (Linköping, Sweden); Thomas Skogh (Linköping, Sweden); Jaakko Antonen (Tampere, Finland); Anna-Maija Haapala (Tampere, Finland); Marja Hietarinta (Turku, Finland); Teija Häkkinen (Helksinki, Finland); Arno Hänninen (Turku, Finland); Sakari Jokiranta (Helsinki, Finland); Anna Karjalainen (Oulu, Finland); Aino Laatikainen (Kuopio, Finland); Marjatta Leirisalo-Repo (Helsinki, Finland); Riitta Luosujärvi (Helsinki, Finland); Aaro Miettinen (Helsinki, Finland); Jari Siuro (Vantaa, Finland); Sune Sved (Vantaa, Finland); Tamara Tuuminen (Mikkeli, Finland); Timo Walle (Helsinki, Finland); Fátima Coelho (S. João Portugal); Maria José Santos (Barreiro Portugal); Mónica Bogas (Alto Minho Portugal); João P Ramos (Porto Portugal); João Faro-Viana (Lisboa Portugal); Nils Olivier Olsson (Dijon, France); Georges Chyderiotis (Lyon, France); Bach-Nga (Reims, France); Dr. Vincent Aubert (Lausanne, Switzerland); Clemens Dahinden (Bern, Switzerland); Eric Dayer (Sion, Switzerland); Françoise Mascart (Bruxelles, Belgium); Jean-Paul Tomasi (Louvain, Belgium); Martine Vercammen (Brussel, Belgium); Patrick Verschueren (Leuven, Belgium); Fréderic Houssiau (Louvain, Belgium); Bernard Lauwerys (Louvain, Belgium); Filip De Keyser (Gent, Belgium); Walter Fierz (Liechtenstein Switzerland); Ingmar Heijnen (Basel, Switzerland); Franco Keller (Bellinzona, Switzerland); Gerhard Müllner (Luzern, Switzerland); Stephan Regenass (Zürich, Switzerland); Pascale Roux-Lombard (Genève, Switzerland); Urs Wirthmüller (Bern, Switzerland); Boris Gilburde (Ramat-Gan, Israel); Rachel Levi (Beer Sheba, Israel); Daphna Lipinsky (Rehovot, Israel); Mira Barak (Haifa, Israel); Abu-Shakra Mahmud (Beer Sheva, Israel), Batia Donnenfeld (Tel Aviv, Israel); Ruth Fishel (Jerusalem, Israel), Bleiber Margalit (Tel Aviv, Israel); Mina Foguel (Rehovot, Israel); Kivity Shay (Ramat-Gan, Israel); Elena Csernok (Germany); Bernhard Schlüter (Germany); Bettina Becker (Germany); Rudolf Mierau (Germany); Torsten Witte (Hannover, Germany); Wolfram Seidel (Germany); Ingrid Frank (Germany); Arno Kromminga (Germany); Falk Hiepe (Berlin, Germany); Karsten Conrad (Dresden, Germany); Michael Sticherling (Germany); Guenter Steiner (Vienna, Austria), Chopyak Valentyna (Ukraine), Inmaculada Alcalá (Tenerife, Spain), Montserrat Alsina (Barcelona, Spain), María José Amengual (Barcelona, Spain), Ángela Carrasco (Madrid, Spain), Luis Fernández (Cáceres, Spain), María Ángeles Figueredo (Madrid, Spain), Carmen Gelpí (Barcelona, Spain), Juana Jiménez (Madrid, Spain), María Rosa Juliá (Palma de Mallorca, Spain), Marcos López-Hoyos (Santander, Spain), Lourdes Mozo (Oviedo, Spain), Cecilia Muñoz (Madrid, Spain), Francisco Javier Muñoz (Almería, Spain), Pilar Palomino (Madrid, Spain), Álvaro Prada (San Sebastián, Spain), Carmen Rodríguez Hernandez (Cádiz, Spain), Garbiñe Roy (Madrid, Spain), Estibaliz Ruiz (Barcelona, Spain), Luisa Vargas (Badajoz, Spain)..
-
Contributors All authors have actively participated in this study as well as in writing the manuscript.
-
Competing interests None.
-
Provenance and peer review Not commissioned; externally peer reviewed.