CHU Hotel-Dieu, Immunology Laboratory, Nantes and 1CHU Hotel-Dieu, Internal Medicine, Nantes, France.
Correspondence to: Marie Audrain, CHU Hotel-Dieu, Immunology Laboratory, Nantes, France. E-mail: marie.audrain{at}chu-nantes.fr
![]() |
Abstract |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Methods. We determined the presence of aCL and anti-ß2GP1 IgG and IgM antibodies in 67 sera from 62 patients and reviewed the data separately. Each serum sample was tested with six commercial aCL determination kits and with four commercial and one in-house anti-ß2GP1 determination kit. We then analysed the operating characteristics of each kit (sensitivity, specificity, positive and negative predictive values) and we analysed the absolute and 2x2 agreements.
Results. The 62 patients included had primary antiphospholipid syndrome (APS) in 10 cases, secondary APS for eight, systemic lupus (SLE) for 23 and other diagnoses for the remaining 21. Operating characteristics differed from one kit to another. Good agreement was found using sensitive aCL determination kit and specific anti-ß2GP1 determination kit. Agreement between kits was medium for IgG aCL. 2x2 concordance studies showed a group of three aCL kits which were quite homogenous and showed that all anti-ß2GP1 kits formed quite a homogenous group.
Conclusion. A high degree of variability still persists for aCL antibody determination posing the question of the qualification of commercial or in-house kits and the question of standardization of results. A better concordance is found for high positive results. Good agreement exists for anti-ß2GP1 kits. aCL determination is still needed and should be complemented by anti-ß2GP1 determination.
KEY WORDS: aCL, anti-ß2GP1, antiphospholipid antibodies.
![]() |
Introduction |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
It is now known that aPL antibodies are not only directed against phospholipids but also against a complex of phospholipid and phospholipid-binding plasma proteins called cofactors such as ß2glycoprotein 1 (ß2GP1) [2, 3] or prothrombin [4]. Furthermore, antibodies directed at these cofactors can be detected in ELISA in the absence of phospholipids [5, 6].
Tests for quantification of aCL or anti-ß2GP1 antibodies are performed with in-house or with commercial kits, but major difficulties still remain and the lack of standardization persists, which renders the inter-laboratory comparison, and sometimes the results, hazardous. International workshops have been organized and calibrated standards and international units (GPL and MPL) have been defined. Nevertheless, there are still some differences in results [711]. Determination of aPL antibodies is, however, important for diagnosis and for preventive or curative therapy.
The aim of this study was to estimate the performance characteristics of 10 commercial kits and one in-house kit for the detection and quantification of aCL (six kits) and anti-ß2GP1 (five kits) antibodies and to evaluate the degree of variability between different kits.
![]() |
Patients and methods |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
LAC was determined in the haematology department according to the guidelines of the International Society on Thrombosis and Haemostasis, subcommittee for the standardization of lupus anticoagulants [14].
Sera
We selected 67 sera from 62 patients sent to the laboratory for routine aCL determination; a diagnosis was available for all of them. Sera were collected by venipuncture and stored at -20°C in aliquots until use in order to avoid freezingdefreezing cycles. Medical records of the studied subjects were retrospectively reviewed by one physician (hospital referent for systemic disease) without knowledge of the biological data. The clinical criteria for primary APS (PAPS) diagnosis were as follows [1]: vascular thrombosis (venous or arterial) or pregnancy morbidity (one or more unexplained fetal deaths beyond the 10th week of gestation; one or more premature births following severe pre-eclampsia or severe placenta insufficiency; three or more unexplained consecutive spontaneous abortions before the 10th week of gestation). The clinical criteria for secondary APS (SAPS) diagnosis were established as the association of APS as well as the criteria for autoimmune disease according to the American Rheumatology Association criteria [15]. SLE diagnoses were established according to the American Rheumatology Association criteria [15].
Clinical and biological data were registered and diagnosis was made associating at least one clinical manifestation and one biological criterion. The biological criteria were the presence of LAC or aCL antibodies on two or more occasions at least 6 weeks apart. aCL antibodies were determined with the aCL kit used for routine testing in our laboratory (kit 1). Anti-ß2GP1 were also determined for each serum but not taken into account for classification. The kit used for this determination was an in-house kit as previously described [12].
Patients were classified into four groups corresponding to diagnosis. Group I included 10 PAPS (11 sera), all positive for aCL and/or LAC. Group II included eight SAPS (nine sera), seven associated with SLE and one in a context of myositis, all positive for aCL and/or LAC. Group III included 23 SLE (24 sera), only one of whom showed clinical criteria of APS, although this patient did not have the biological criteria. It should be noticed that in this group, five women were under 25 and might not yet have shown episodes of thrombosis or obstetrical complications. Group IV included 21 other diseases (23 sera): clinical manifestations were thrombosis (six), recurrent fetal abortions (one), intra-uterine fetal growth retardation (one), coagulation disorders (four), medical history of thrombosis (three), renal graft (one), CREST syndrome (one), amyotrophic lateral sclerosis (one), cutaneous lesions (one), Horton (one) and autoimmune haemolytic anaemia (one).
Data analysis
For every kit, results were first reported as designed in the manufacturer's instructions. Quantitative results were considered negative, equivocal or indeterminate, weakly positive, positive or highly positive, depending on the manufacturer. Then, for determination of operating characteristics (sensitivity, specificity, positive predictive value and negative predictive value), we reported results as either negative or positive, using the cut-off values proposed by the manufacturers. It was decided to consider positive those that were called weakly positive, positive or highly positive and to consider negative those that were called negative, indeterminate or equivocal.
Sensitivity, specificity, positive predictive value and negative predictive value were defined as follows: sensitivity (Se), calculated as the frequency of a positive test in a population of patients with the disease, Se = TP/TP+FN; specificity (Sp), calculated as the frequency of a negative test in a control population, Sp = TN/TN+FP; positive predictive value (PPV), calculated as the frequency of patients with the disease in a population positive for the test as, PPV = TP/TP+FP; and negative predictive value (NPV), calculated as the frequency of disease-free subjects in a population negative for the test, NPV = TN/TN+FN. Where TP represents true positive, i.e. positive serum from a patient with APS defined as described above; TN represents true negative, i.e. negative serum from a patient without APS; FP represents false positive, i.e. positive serum from a patient without APS; and FN represents false negative, i.e. negative serum from a patient with APS.
Absolute concordance between kits was defined for IgG and IgM aCL and anti-ß2GP1 antibodies when all kits reported the same interpretation (negative or positive). Results of absolute agreement were expressed as number of sera where all kits agreed/number of sera tested with all kits. Concordance was also evaluated in 2x2 tables, and expressed as number of sera where two kits agreed/number of sera tested with these two kits.
![]() |
Results |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Frequency of positive results
The frequency of positive IgG aCL results varied considerably from one kit to another: 1864% (Table 1). The frequency of positive results in each of the four groups also showed great differences, more important in groups II, III and IV than in group I. Kits 1 and 6 showed 58 and 75%, respectively, of IgG aCL positive sera in group III, and 52 and 53%, respectively, of positive sera in group IV. However, kit 1 identified 82% IgG aCL positive sera in PAPS and 89% in SAPS. The other kits identified 55% of positive sera in PAPS and 4450% in SAPS.
|
|
|
|
|
|
![]() |
Discussion and conclusion |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
We planned our study in order to evaluate the agreement between several kits of aCL and anti-ß2GP1 antibodies and to approach the operating characteristics of each. As cut-off levels were not similar between kits, we decided to report results as positive or negative according to each manufacturer's instructions.
There was a medium agreement between all aCL kits especially for IgG isotype (51%). Three kits showed good agreement. In the PAPS group, the agreement was greater (64%). We could not focus on differences within each commercial kit manufacturer because all details of manufacturing were not available. However, it was clear that cut-off levels of positivity and interpretation of titres were different. The low level of agreement could be first explained by the fact that a lot of sera were weak positive for one manufacturer and would be indeterminate or equivocal for another, leading to a completely different interpretation: positive for one and negative for the other. Discrepancies were often observed with these low positive sera. A second explanation was that, with low positive values inter- and intra-assay reproducibilities were usually worse than with high values and the same serum testing in another run could be negative. A better agreement was observed in high positive values, emphasizing the fact that the choice of a commercial kit was critical for sera which were low positive or borderline.
Cut-off levels and interpretation were also different for anti-ß2GP1 antibodies, but the agreement between kits was better and even correct for IgG and IgM anti-ß2GP1 antibody determination considering all sera together or taking each group separately; this could be explained by the fact that the antigen was better characterized and no cofactor needed. In conclusion, a high degree of variability still persisted for aCL antibody determination kits and a good agreement was observed for the anti-ß2GP1 antibody kits.
With regards to sensitivity, specificity, positive and negative predictive values, results showed few differences from our previous study [12]. Several explanations could be made. First, in the previous study only one kit for aCL and one kit for anti-ß2GP1 antibody determination were tested. Secondly, this study was conducted with 191 consecutive sera with a lot of negative sera and 184 patients who were diagnosed as having other diseases, and only seven as having APS. However, the aim of the study was to focus on the role of anti-ß2GP1 antibody determination for diagnosis of APS. In this previous study, the number of APS patients was too low to calculate sensitivity. However, we could approach specificity. In the present study, the sera were not consecutive, but were included when having a positive aCL determination or a suspicion of APS (presence of aCL was confirmed on a second sample). The aim of this study was to compare several kits for aCL and anti-ß2GP1 antibody determination. We observed that the percentage of positivity for aCL IgG ranged from 18 to 64% for the six kits. For IgG and IgM anti-ß2GP1 antibodies the percentage of positivity ranged from 8 to 15% and 3 to 19%, respectively. We observed that three IgG aCL kits had low sensitivity, leading to a high specificity, one was highly sensitive and poorly specific and one poorly sensitive and poorly specific. A poor sensitivity was observed for aCL IgM determination. Sensitivity was, however, very questionable since the notion of a false negative was a matter of concern with some patients having APS without aCL but with a LAC or another aPL. In the international consensus [1], the other aPLs are not considered as a biological criterion, but we may not exclude formally a diagnosis of APS without testing all the aPL antibodies. In another way, specificity was also questionable since one cannot predict if a patient, who was aCL positive without clinical criteria, would not develop clinical APS later, especially if this patient was a young woman who could benefit from preventive treatment. Were they really FP sera since they were almost anti-ß2GP1 negative? Could we identify these patients as patients who would develop clinical APS later or could we identify them as patients who had true aCL antibodies without clinical criteria? Young patients included at the time of the study in groups III or IV might switch in the future and perhaps might be authentic APS without clinical criteria because of preventive therapy established as soon as aCL or LAC was detected. However, classification of patients according to aCL criteria, based on the other kits rather than our reference kit, did not show large discrepancies since only one patient switched from group I to group IV and possibly two from group IV to group I.
Anti-ß2GP1 antibody determination did not seem to be sensitive but was very specific. If we associated firstly a screening for aCL and a screening for anti-ß2GP1 antibodies in aCL positive patients, results in terms of sensitivity, specificity and predictive values were greater. We did not observe patients with isolated anti-ß2GP1 antibodies.
In conclusion, we observed that a high degree of variability still persists for aCL antibody determination, putting into question testing of sera by one or several kits. However, we observed that a better concordance is found for high positive results. This is in agreement with the Sapporo criteria for diagnosis of PAPS. Only medium or high values had to be considered as biological criteria. There were indications suggesting that a good agreement exists for anti-ß2GP1 antibodies.
As we could not establish parametric correlations, and as concordance was not perfect, the question of qualification and cut-off values of commercial or in-house kits is still being asked. Which sera could be used to participate in a reference panel in order to validate kits? High positive sera would favour specific but not sensitive kits, low positive sera would favour sensitive but not specific kits. Another question is the standardization of the different kits. They all used international standards and GPL and MPL units, but cut-off and results are different. The lack of standardization is easily demonstrated by national or international external quality programs. It would be interesting to define an international reference serum which could be included in each assay and which would allow us to recalculate result values using this reference, as is sometimes proposed.
Testing for anti-ß2GP1 antibodies in all aCL positive sera could help diagnosis and increase the specificity of a highly sensitive but low-specific aCL method. But, these anti-ß2GP1 antibodies are not yet considered as biological criteria, are of low sensitivity and a lot of patients had PAPS with aCL without anti-ß2GP1 antibodies.
We concluded that (i) aCL determination is still needed and could not be replaced by anti-ß2GP1 determination; (ii) it might be completed with anti-ß2GP1 determination and if all is negative with other aPL antibody determinations; (iii) as a lot of sera are borderline, some sera have to be tested on another kit; and (iv) a good working relationship between biologists and physicians is needed.
The authors have declared no conflicts of interest.
![]() |
References |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|