Levels and avidities of antiphosphatidylethanolamine antibodies in patients with thrombotic events and immunologically-mediated diseases

Aims. Antiphosphatidylethanolamine antibodies (aPE) represent one type of antiphospholipid antibody (aPL) directed against the neutral phospholipids − phosphatidylethanolamines. The aim of this study was to evaluate levels and avidities of aPE in several groups of patients and compare them with conventional aPLs. Methods. aPE were analysed in a cohort consisting of 68 hospitalized patients. The other cohort comprised 22 patients with immunologically-mediated diseases. The control group consisted of 20 healthy persons. ELISA methods were used for determination of aPL. Avidities of aPE were tested by modified ELISA with urea as a chaotropic agent. Results. aPE IgG/IgM were significantly higher in the group of patients with venous thromboembolism than those with non-thrombotic internal disorders (P=0.02 for both Ig classes). aPE IgG/IgM elevated above cut-off values were found in 10.8% of patients with venous thromboembolism and as a single aPL in 6.5%. Levels of aPE IgG higher than our limit (>6 U/mL) were detected in 29% of patients with immunologically-mediated diseases with other positive aPL. Low-, intermediateand high-avidity aPE IgG were found in patients of both cohorts. The avidities of aPE IgG differed from those of anticardiolipin antibodies IgG. Neither aPE IgG levels nor avidity dynamics significantly changed during follow-up. Conclusion. aPE may be related to venous thromboembolism and may be part of the repertoire of aPL in immunologically-mediated diseases. There are patients with thrombosis negative for conventional aPL but positive for aPE. aPE IgG may have different avidities.


INTRODUCTION
Antiphospholipid antibodies (aPL) include a heterogeneous group of antibodies directed especially against negatively charged phospholipids alone and/or their complexes with various plasma proteins 1,2 . aPL contribute to various types of thrombotic and obstetric complications such as pregnancy loss, pre-eclampsia or placental insufficiency and are placed in context with valvular disease and some hematological disorders 1 . The mechanism of induction of these pathological conditions by aPL is complicated and involves the participation of various cells such as endothelial cells, monocytes, platelets, and proteins, e.g. complement 1 .
The persistently elevated levels of aPLs are laboratory criteria for the antiphospholipid syndrome (APS), which clinically manifests in thrombotic events and/or obstetric morbidity 3,4 .
Lupus anticoagulant, anticardiolipin antibodies (aCL) and anti-β2-glycoprotein I antibodies (aβ2GPI) IgG and IgM are listed in the current laboratory criteria for APS diagnosis 5 . Testing is recommended for all of these antibodies, as patients with triple aPL-positivity are at a higher risk of thrombosis or pregnancy complications than double or single-positive cases 5,6 .
In addition to "classical" aPL, antibodies against other phospholipids or their complexes with proteins are also produced 5 . Testing of other aPLs has not been suggested yet 5 . However, patients with clinical symptoms of APS but without an increase in any of recommended aPLs were identified. This group of patients was classified as seronegative APS (SN-APS) (ref. 7 ). aPL to other negatively charged phospholipids such as phosphatidylserine, phosphatidic acid or phosphatidylinositol known as non-criteria aPL have been detected in patients with SN-APS (ref. 8 ). Antibodies directed against the phosphatidylserine-prothrombin complex (aPS/PT) have been studied in more detail. Recent studies suggest that the addition of aPS/PT to the recommended aPL could improve the estimation of thrombotic risk 9,10 . Non-criteria aPL tested in patients with APS also include those reacting with phosphatidylethanolamine (aPE) (ref. 11 ).
Phosphatidylethanolamines (PE) are the second most abundant mammalian membrane phospholipids, which constitute 20-50% of all phospholipids in mammalian cell membranes 12 . PE detected at the luminal endothelial surface may be susceptible to aPE binding, which is consistent with an association between the presence of aPE and idiopathic thrombosis 13 . Although aPEs have been studied in a number of clinical trials in patients with thrombosis, and pregnancy morbidities and other diseases, their diagnostic benefit remains controversial 11,[14][15][16] . However, the pathogenicity of autoantibodies depends on various factors. One of them is their avidity. It is presumed that autoantibody avidity contributes to various clinical manifestations. There are studies on the avidity of common aPL, but to the best of our knowledge, the avidity of aPE has not been investigated yet 17 .
The aim of this study was to examine aPE in patients with thromboembolism, non-thrombotic internal and immunologically-mediated diseases. At the same time, we established the cut-off values for aPE IgG and IgM. We were interested not only in the levels of aPE, but also in their avidity. Therefore, another goal of the study was to introduce an ELISA method for determination of aPE IgG avidity using chaotropic agents and to evaluate aPE IgG avidity in selected groups of patients. We also monitored the dynamics of changes in the levels and avidities of aPE IgG. The last aim was to determine the relationship between aPE and conventional aPL.

Patients
We analysed samples from two cohorts of patients − one consisted of 68 patients hospitalized in Thomayer Hospital in Prague (TH cohort; age: 58±18 years, mean±SD; sex: 32 females, 36 males) and the other cohort of patients examined at the Immunological Department of the Institute of Medical Biochemistry and Laboratory Diagnostics in Prague (ID cohort; n = 22; 38±15 years, mean±SD; sex: 17 females, 5 males).
In the TH cohort, patients were recruited randomly over a period of 3 years. We divided patients in the TH cohort into two groups. The first consisted of patients who were hospitalised for venous thromboembolism (n=46; 59±19 years; sex: 22 females, 24 males). Patients with venous thrombosis in the lower limbs predominated (n=42) and in addition, bilateral pulmonary embolism was concurrently diagnosed in 10 patients. Four patients were hospitalized for ischemic stroke. The diagnoses were performed by experienced internists based on valid diagnostic criteria using case history, clinical examination and Doppler-sonography in particular. The second group of non-thrombotic patients with various internal diseases included mainly those hospitalised with gastrointestinal or cardiometabolic disorders (n=22; 56±16 years; sex: 10 females, 12 males).
The ID cohort of patients with immunologically-mediated diseases comprised aCL IgG positive patients with systemic immunological diseases predominantly with systemic lupus erythematosus (SLE) or systemic connective tissue disease and infertility or kidney disorders.
We determined the levels of aCL, aPE, and the avidities of aPE in serum samples of patients. We also tested aPS/PT levels in patients of the TH cohort. In some patients not all parameters due to the lack of serum could be determined.
In addition to these two cohorts of patients, we examined a control group of 20 healthy subjects with the aim of establishing the limit values for aPE IgG/IgM and aPS/PT Ig G/ IgM levels (age: 50±7 years, mean±SD; sex: 10 females, 10 males).
All subjects gave written informed consent regarding participation in the study. The Ethics Committees of the General University Hospital, Prague and Thomayer Hospital, Prague, approved the study.

Levels of various types of antiphospholipid antibodies
aPE IgG/IgM and aPS/PT IgG/IgM were analysed by ELISA kits (AESKULISA Ethanolamin-GM and AESKULISA Serine-Prothrombin-GM, AESKU. DIAGNOSTICS, Wendelsheim, Germany). aCL IgG/ IgM and aβ2GPI IgG levels were determined by the ELISA kits (ELISA Anti-cardiolipin antibodies, Orgentec, Mainz, Germany). Used ELISA kits have CE IVD (In Vitro Diagnostic Medical Devices) certifications. ELISA assays were performed according to the manufacturer's instructions. Serum samples were diluted 1:101 and 1:100 for aPE IgG/IgM or aPS/PT IgG/IgM and aCL or aβ2GPI analyses, respectively.
Phosphatidylethanolamine or cardiolipin plus native human ß2-glycoprotein I and native phosphatidylserine plus human native prothrombin were used for coating on microwells in aPE or aCL ELISAs and aPS/PT ELISA, respectively. The ELISA kits for aCL IgG/IgM and aβ2GPI IgG (Orgentec) are calibrated against the internationally recognised reference sera from E.N. Harris, Louisville and the specific reference material IRP 97/656 (IgG) and HCAL (IgG) / EY2C9 (IgM). The results were expressed in units GPL (aCL IgG) and MPL (aCL IgM) or arbitrary units U/mL (aβ2GPI IgG). The standards in the ELISA kits for aPE IgG/IgM and aPS/PT IgG/IgM (AESKU. DIAGNOSTICS) contained human serum. The levels of aPE IgG/IgM and aPS/PT IgG/IgM were expressed in U/mL.

Avidity of aPE IgG
aPE IgG avidity was determined by the modified ELISA method using chaotropic agents described in details in previous studies 18,19 . The wells of microtiter strips were coated with 50 µL of phosphatidylethanolamine 50 µg/mL (Merck KGaA, Darmstadt, Germany). We utilized urea as a chaotropic agent, similarly as in our avidity assay for aCL IgG. Initially, we tested different urea concentrations − 2 mol/L, 4 mol/L, 6 mol/L and 8 mol/L in sera diluted 1:50, 1:100 and 1:200 to find the optimal conditions for aPE IgG avidity determination.
The results of aPE IgG avidity were expressed as the ratio of the antibodies bound in the wells with chaotrope to the total antibodies bound in the absence of a chaotrope.
absorbance in the wells with chaotrope absorbance in the wells without chaotrope We classified the antibodies with AI>0.6 as high-avidity, while those characterised by AI<0.4 were low-avidity. The range of AI from 0.4 to 0.6 indicated intermediateavidity 20,21 .

Statistical analysis
The Shapiro-Wilk test was used to test the normality of the data. Since this was not normal, nonparametric tests were used. The comparison of two groups we used the Mann-Whitney test. The Spearman correlation coefficient or Passing-Bablok regression was applied to analyse the relationship of variables. The paired measurements were evaluated by the Wilcoxon matched pairs test. Fischer´s exact test was performed for comparison of the qualitative data. The level of statistical significance was set to 0.05. Statistical analyses were performed in the Medcalc (MedCalc, Ostend, Belgium).
To express reference limits for aPE IgG/IgM and PS IgG/IgM, we applied the robust method according to the document of Clinical and Laboratory Standards Institute (CLSI) C28-A3 recommended for the evaluation of smaller number of reference subjects 22 using MedCalc software.

Cut-off values (upper limits) for aPE IgG/IgM and aPS/PT IgG/IgM
We established the cut-off values for aPE IgG/ IgM and aPS/PT IgG/IgM in 20 healthy persons using 95 % reference intervals. The clinically important cut-off values (upper limits) are shown in Table 1. We used these values in further evaluation of aPL in our patients.
The values recommended by the manufacturer of the ELISA kit were utilized in aCL (>10 GPL for aCL IgG, >7 MPL for aCL IgM) and aβ2GPI IgG (>8 U/mL).

Levels of tested aPL in patients with thromboembolism
The comparison of non-criteria aPL antibodies showed significant differences for aPE IgG and IgM between groups of patients with venous thromboembolism and those with non-thrombotic internal diseases (Fig. 1A−F). These antibodies were significantly higher in patients with venous thromboembolism (aPE IgG and IgM, P=0.02 for both Ig classes). However, no significant differences between these patients´ groups and the control group were observed with the exception of significantly higher aPE IgM in controls than in non-thrombotic internal disease (P=0.04). Levels of aPS/PT IgG or IgM were similar in both groups of patients. No significant differences were found for aCL IgG or IgM levels and aβ2GPI IgG. Fig. 2A shows the levels of aPL tested in patients with pulmonary embolism. aPL levels above the upper limits of our normative values were found only in two patients with pulmonary embolism (one patient had increased aPE IgM, the other patient had increased other aPLs). Once, aCL IgG and aPS/PT IgG values were higher than the corresponding cut-off limits in patients with ischemic stroke (Fig. 2B).
Patients with thromboembolism and increased aPLs upper cut-off limits are listed in Table 2. 32.6% of patients with thromboembolic diseases had at least one of the tested aPL antibodies increased. aCL IgG/IgM were elevated in 19.5%, and similarly also aPS/PT IgG/IgM, followed by increased aPE IgG/IgM in 10.8% of patients. An isolated increase in aCL IgG/IgM as well as aPE IgG/IgM was observed in 6.5% of patients for each of the antibodies. aPS/PT IgG/IgM were elevated as a single aPL in the 8.6% of patients (Table 3). aβ2GPIs were not increased in any patient with thromboembolic diseases. In the patients with non-thrombotic internal diseases, only one patient had an increased aPL, namely aCL IgM. No significant differences were found in aPE IgG/IgM and aPS/PT IgG/ IgM positive patients with thromboembolism and control.

Levels of tested aPL in patients with immunologically-mediated disorders
Patients selected to the ID cohort had elevated aPL IgG levels determined by ELISA antiphospholipid screen including aCL, aPS, phoshatidylinositol (aPI) and phospha tidic acid antibodies (aPA), but not aPE. All patients of ID cohort were also positive for aCL IgG (>10 GPL). Only aPE IgG was examined in 17 patients from the ID cohort of patients. Some patients were examined repeatedly, so that we examined aPE IgG in 26 samples in total. The levels of aPE IgG were significantly higher than in controls (P<0.001). The levels of aPE IgG higher than our cut-off value (>6 U/mL) were detected at least once in five patients (29%). Three of these patients were followed for SLE or for systemic connective tissue disease, one patient for interstitial nephritis and one woman was treated for infertility. Table 4 summarizes the results of the aPL IgG tests against various phospholipids / proteins performed as part of an immunological examination in five patients with elevated aPE IgG. All patients were also positive for other aPL IgG against other phospholipids / proteins. No significant differences between the 1 st and 2 nd measurement of APE IgG levels were found in repeatedly examined patients (n=8). Two patients experienced a change with an overlap to above limit level.

Avidity of aPE IgG
Initial experiments to establish the optimal conditions for the determination of aPE IgG avidity involved analyses of variously diluted sera at different urea concentrations (Fig. 3). Higher urea concentrations dissociated interactions in immune complexes formed between aPE IgG and PE bound on the surface of the microtiter well more effectively than lower ones. Higher serum dilutions (1:100, 1:200) resulted in some samples in too low absorbances and thus to lower accuracy of results. Urea of 6 mol/L or 8 mol/L at a 1:50 dilution of sera appeared to be suitable, because these conditions allowed us to sufficiently disrupt immune complexes 23 . Comparison of AI using urea 6 mol/L and 8 mol/L showed a significant correla-tion (n=48, r=0.71, P<0.0001) determined Passing-Bablok regression.
We examined the aPE IgG avidity using urea 6 mol/L at 1:50 dilution in 43 patients with thromboembolism in the TH cohort and in 38 sera in the ID cohort (22 patients). Patients with venous thromboembolism synthesised low-, intermediate and high-avidity aPE IgG, AI values ranging from 0.23-0.87. The low-avidity aPE IgG (AI less than 0.4) was observed in 25%, intermediateavidity aPE IgG (AI 0.4-0.6) in 35% and those with highavidity (AI more than 0.6) in 40% of patients (Fig. 4A). AI values in immunological patients varied from 0.13 to 0.78. Low-avidity aPE IgG was similar as those in the immunological patients (27 %), intermediate avidity in 46% and high-avidity again in 27% of patients (Fig. 4B). In repeatedly examined patients, we evaluated avidity only in the first sample. The aPE IgG avidity in patients examined a second time did not differ significantly from the previous examination.  AI values for aCL IgG avidity determined by a similar procedure to a previous study 24 were significantly higher than those for aPE IgG in the group of patients with thromboembolism (P<0.0001) as well as with immunologically-mediated diseases (P=0.04). APE IgG avidities in immunological patients did not correlate with those of aCL IgG, while in those with thromboembolism the avidities of aPE and aCL correlated (r=0.36, P=0.02).

Relationship between aPE and other aPL
TH cohort − aPE IgG levels significantly correlated with aCL IgG only in a group of patients with thromboembolism (r=0.31, P=0.04). Another significant correlation was demonstrated between aPE IgM and aCL IgM (r=0.54, P=0.0001) or aPS/PT IgM (r=0.58, P<0.0001) in the same patient group. In addition, aPE IgM significantly correlated with aPS/PT IgM (r=0.79, P=0.002) also in the group of patients with non-thrombotic internal diseases and controls.
Concerning the ID cohort, there was no significant correlation between aPE IgG and aCL IgG or other noncriteria aPLs (aPS, aPI, and aPA) examined during followup in the group of patients with immunologically-mediated diseases. A significant correlation was also found between aPE IgM and aPS/PT IgM (r=0.51, P=0.02) in controls.

DISCUSSION
The aim of the study was to examine aPE IgG and/or IgM in patients with thromboembolism, non-thrombotic internal and immunologically-mediated diseases. The evaluation of aPE was not limited to levels, it included their avidities. The dynamics of aPE IgG levels and avidities were monitored in a subset of patients. Our study demonstrated significantly higher levels of aPE IgG and IgM in a cohort of patients with venous thromboembolisms compared to those with other internal diseases and in the cohort of immunologically-mediated diseases than in controls. Increases in aPE IgG and / or IgM levels above our cut-off values (95th percentile of the control group) were observed only in 10.8% of patients with thrombosis while in patients with immunologically-mediated diseases in 29%. APE IgG levels appear to be stable. It seems that aPE IgG/IgM may be related to other aPLs, but only in some patient groups. The avidities of aPE IgG can range from low-avidity to high-avidity antibodies in various diseases. As in the case of aPE IgG levels, their avidities did not show significant dynamics upon repeated examination. aCL -anticardiolipin antibodies, aPL -antiphospholipid antibodies, aPE -antiphosphatidylethanolamine antibodies, aPS/PT -antibodies directed against the phosphatidylserine-prothrombin complex, IgG -immunoglobulin G, IgM -immunoglobulin M  Several studies have already described the relationship between the main clinical manifestations of APS and the presence of antibodies to the neutral phospholipidphosphatidylethanolamine 11 . The positivity of aPE in our patients with thromboembolism was similar to that in the multicentre study of Sanmarco et al. 25 . The proportion of elevated aPE IgG/IgM levels was the lowest among the examined aPLs, but it was the only positive aPL in 6.5 % of patients. Therefore, we supported the opinion that aPE may be the only type of aPL in patients with thrombosis 26 .
As expected, the proportion of positive aPE IgG was higher, more than two-fold (29%) in patients with immunologically-mediated diseases compared to patients with thrombosis. However, it should be taken into account that the cohort of patients with immunologically-mediated diseases included patients who were aCL IgG positive. The result is close to that of a study by Bertolaccini et al. 27 , which showed aPE IgG in 46% of patients with SLE and at the same time aPL positive. In contrast, Zhang et al. 28 reported aPE positivity in only 8.1% of Chinese patients with APS. Either commercial kits or in-house ELISA methods can be used to determine aPE. Unfortunately, there are no international standards for aPE as for conventional aPLs. In addition, the impact of variously modified ELISA methods on the results has been demonstrated 29,30 . This makes it difficult to compare different studies and may lead to ambiguous results.
The observation of stable aPE levels in patients followed over time was consistent with Sanmarco et al. 25 but this focused on patients with thrombosis.
The association between aPE and other aPLs is unclear. Consistently with Bertolaccini et al. 27 we also found no relationship between aPE IgG and aPL IgG against negatively charged phospholipids in patients with immu- nologically-mediated diseases. However, aPE IgG/IgM levels correlated with aCL IgG/IgM in the thromboembolism group which may indicate a common involvement of both aPL antibodies in this pathological manifestation. Not only the levels of antibodies, but also their avidities have an impact on the severity of autoimmune disease 17 . To our knowledge, no previous studies have investigated this aspect of aPE. There are various methodological approaches for determining the avidity of autoantibodies. Commonly used assays are ELISA in the presence of various chaotropic agents which disrupt the bonds between the antigen coated on the surface of the well and the corresponding antibody. These ELISA methods have already been described for testing the avidity of aPL IgG, specifically aCL or a aβ2GPI (ref. 31,32 ). For this reason, we also chose a modified ELISA using urea as a chaotropic agent to test the avidity of aPE.
A pilot study of avidity determination showed that the avidity of aPE IgG can vary over a wide range. Interestingly, we did not demonstrate any relationship between the avidity of aPE IgG and aCL IgG in patients with immunologically-mediated diseases. While low-and medium-avidity antibodies were mainly represented among APEs, higher avidity antibodies predominated in aCL. A similar finding was made by Totland et al. 33 , who examined the avidity of common onconeural antibodies anti-Hu and anti-Yo and observed that anti-Yo generally had higher avidity than anti-Hu. It can be assumed that different aPLs may be involved in the pathogenesis of the disease in various ways, not only with respect to different binding targets. This is known from the studies on the avidity (affinity) of other autoantibodies 17 . For example, high-avidity IgG antinuclear antibodies could be a potential biomarker for the assessment of the prognosis of SLE activity 34 . Conversely, in another study, the levels as well as the avidity of antibodies to amyloid β were lower in patients with Alzheimer's disease than in healthy controls 35 .
The avidities of aPE IgG expressed as AI in most patients with immunologically-mediated diseases did not differ significantly during patient follow-up, i.e. there was no switching between low-avidity and high-avidity antibody categories. It can be assumed that the examination of avidity took place in the period when the value of avidity was already differentiated.

CONCLUSION
In conclusion, we confirmed that aPE may be related to venous thromboembolism and may be part of the repertoire of aPL in immunologically-mediated diseases. aPE IgG may have different avidities, but low and medium avidities predominate. The clinical relevance of avidity testing will need to be verified in a larger cohort of patients.