Profiling of aPL antibodies by ELISA and MLDA
Furthermore, sera from patients with APS (
n = 85), DC (
n = 65), and NHS as controls (
n = 79) were assessed for comparison of aPL antibody in ELISA and the novel MLDA (Table
1).
Table 1
Number of aPL antibody positive sera investigating 85 APS patients, 65 DC patients, and 79 NHS in ELISA and in the MLDA
APS n = 85 | 45 | 32 | 27 | 33 | 50 | 28 | 17 | 7 | 3 | 39 | 33 | 31 | 27 | 57 | 63 |
DC n = 65 | 1 | 1 | 0 | 0 | 1 | 1 | 0 | 0 | 0 | 2 | 3 | 0 | 3 | 7 | 8 |
NHS n = 79 | 1 | 0* | 0 | 4 | 5 | 1 | 1 | 0 | 0 | 2 | 6* | 0 | 5 | 8 | 12 |
Patients suffering from APS demonstrated a significantly higher frequency of anti-CL IgG (P < 0.000001, respectively), anti-CL IgM (ELISA: P < 0.000001, respectively; MLDA: P < 0.000001, P = 0.0002, respectively), anti-β2 GPI IgG (P < 0.000001, respectively) and anti-β2 GPI IgM (ELISA: P < 0.000001, respectively; MLDA: P < 0.000001, P = 0.000049, respectively) compared to DC patients and NHS in both ELISA and MLDA. Anti-PS IgG and IgM detected by MLDA occurred significantly more frequently in patients with APS compared to the control groups (P < 0.000001, respectively and P = 0.000001, P = 0.000077, respectively). The number of anti-PI IgG positive patients was significantly elevated in the APS group in contrast to DC patients and NHS (P = 0.004573, P = 0.014091, respectively), whereas anti-PI IgM did not demonstrate a significant higher prevalence in this patient cohort in comparison with the control groups.
Remarkably, all anti-PI positive samples found in the APS patient cohort also showed a positive reactivity with the respective anti-PS isotype antibody. Furthermore, all anti-PS positive samples showed a positive anti-CL isotype antibody response, too. Regarding aPL IgG positive APS patients, all seven anti-PI IgG positive patients demonstrated positive anti-PS, anti-CL, and anti-β2 GPI IgG by MDLA either (P = 0.000447). Twenty (95.2%) of the remaining 21 anti-PS IgG positive APS patients revealed a similar pattern of positive anti-CL and anti-β2 GPI IgG by MDLA, while only one patient showed anti-CL IgG only (P < 0.000001). The two anti-PS IgG positive individuals in the control groups demonstrated anti-CL IgG only.
Comparing ELISA and MLDA data, there was no statistical difference in the frequencies of positive anti-CL and anti-β2 GPI IgG antbodies detected by either method (Table
2). The agreement between both methods was assessed as good for anti-CL IgG (kappa = 0.641, 95% confidence interval (CI): 0.541 to 0.767), moderate for anti-CL-IgM (kappa = 0.507, 95% CI: 0.357 to 0.657), very good for anti-β2 GPI IgG (kappa = 0.803, 95% CI: 0.685 to 0.921) and moderate for anti-β2 GPI IgM (kappa = 0.506, 95% CI: 0.352 to 0.659) according to inter-rater agreement statistics.
Table 2
Comparison of anti-CL and anti-β2 GPI antibodies detected in ELISA and MLDA
ELISA | positive | 32 | 15 | 47 | ELISA | positive | 22 | 11 | 33 |
| negative | 11 | 171 | 182 | | negative | 20 | 176 | 196 |
| n | 43 | 186 | 229 | | n | 42 | 187 | 229 |
anti-β2 GPI IgG
|
MLDA
| | |
anti-β2 GPI IgM
|
MLDA
| | |
| | positive | negative | n | | | positive | negative | n |
ELISA | positive | 24 | 3 | 27 | ELISA | positive | 21 | 16 | 37 |
| negative | 7 | 195 | 202 | | negative | 14 | 178 | 192 |
| n | 31 | 198 | 229 | | n | 35 | 194 | 229 |
It should be noted that comparing LAC with combined MDLA data in the group of patients with APS, there was no statistical difference according to McNemar's test (difference: 5.88%; 95% CI: -11.26% to 22.29%; P = 0.5677). In terms of MLDA, patient samples were scored positive when IgG or IgM antibodies to PI, PS, CL, or β2 GPI were detected above the cut-off for bands. Lupus anticoagulant and MLDA data were significantly related (contingency coefficient = 0.324, P = 0.0016). Strength of agreement between both methods was fair (kappa = 0.303, 95% CI: 0.152 to 0.455).
Patients suffering from APS showed significantly more multiple positive samples detected by ELISA (41/85) and MLDA (44/85) compared with DC patients (ELISA: 0/65, MLDA: 2/65,
P < 0.000001, respectively) and NHS (ELISA: 0/79, MLDA: 6/79,
P < 0.000001, respectively) (Additional file
1, Table S1). The frequency of multiple positive samples detected in the APS patient cohort by ELISA was not significantly different from the frequency obtained by MLDA (
P = 0.759115). Interestingly, the MLDA detected a higher number of samples (39/85) with three and more positive aPL in the APS patient group compared to ELISA (29/85); however, this difference was not significant.
By comparing both detection methods in the DC group, there was also no significant difference in the frequency of multiple positive samples (P = 0.496124). In the NHS group; however, only the number of samples demonstrating three or more positive aPL antibodies was not significantly different (P = 1.0). There was only one NHS demonstrating three aPL IgM antibodies (anti- β2 GPI, anti-CL, and anti-PS) simultaneously in the MLDA.
The assay performance characteristics for the respective aPL detected by ELISA are summarized in Table
3. The specificities for all four detected aPL antibodies were remarkably high ranging from 97.2% (anti- β2 GPI IgM) to 100.0% (anti- β2 GPI IgG). The most sensitive aPL antibody assessed by ELISA was anti-CL IgG revealing a sensitivity of 52.9%. Taking into account the laboratory criteria for APS requiring at least one positive aPL antibody, combined ELISA data showed a sensitivity of 58.8% with a specificity of 95.8%. This resulted in a +LR of 14.1 and a -LR of 0.4.
Table 3
Performance characteristics of ELISA for IgG and IgM to CL and β2GPI
anti-CL IgG
| 52.9 | 41.8 to 63.9 | 98.6 | 95.1 to 99.8 | 38.1 | 9.5 to 153.2 | 0.5 | 0.4 to 0.6 |
anti-CL IgM
| 37.6 | 27.4 to 48.8 | 99.3 | 96.2 to 100.0 | 54.2 | 7.5 to 309.6 | 0.6 | 0.5 to 0.7 |
anti-β2 GPI IgG
| 31.8 | 22.1 to 42.2 | 100.0 | 97.5 to 100.0 | ∞ | | 0.7 | 0.6 to 0.8 |
anti-β2 GPI IgM
| 38.8 | 28.4 to 50.0 | 97.2 | 93.0 to 99.2 | 14.0 | 5.1 to 38.1 | 0.6 | 0.5 to 0.8 |
The MLDA performance characteristics are given in Table
4. Specificities for the respective aPL antibodies ranged from 93.8% (anti-CL IgM) to 100.0% (anti-PI IgG and IgM, anti- β2 GPI IgG). The most sensitive aPL antibody assessed by MLDA was also anti-CL IgG revealing a sensitivity of 45.9%. In contrast to the ELISA performance, the combined assessment of aPL by the MLDA revealed a remarkable sensitivity of 67.9% with a specificity of 89.6% (Table
5). This resulted in a low +LR of 6.5 and an acceptable -LR of 0.4. The moderate specificity of the MLDA was in particular due to a significantly higher number of false positive anti-CL IgM determined in the DC and NHS (9/144) groups compared to the ELISA data (1/144,
P = 0.019742).
Table 4
Performance characteristics of MLDA for IgG and IgM to PS, PI, CL, and β2GPI
anti-PS IgG
| 32.9 | 23.1 to 44.0 | 98.6 | 95.1 to 99.8 | 23.7 | 5.8 to 97.1 | 0.7 | 0.6 to 0.8 |
anti-PS IgM
| 20.0 | 12.1 to 30.1 | 99.3 | 96.2 to 100.0 | 28.8 | 3.9 to 212.6 | 0.8 | 0.7 to 0.9 |
anti-PI IgG
| 8.2 | 3.4 to 16.2 | 100.0 | 97.5 to 100.0 | ∞ | | 0.9 | 0.9 to 1.0 |
anti-PI IgM
| 3.5 | 0.7 to 10.0 | 100.0 | 97.5 to 100.0 | ∞ | | 1.0 | 0.9 to 1.0 |
anti-CL IgG
| 45.9 | 35.0 to 57.0 | 97.2 | 93.0 to 99.3 | 16.5 | 6.1 to 44.6 | 0.6 | 0.5 to 0.7 |
anti-CL IgM
| 38.8 | 28.4 to 50.0 | 93.8 | 88.5 to 97.1 | 6.2 | 3.1 to 12.3 | 0.6 | 0.6 to 0.8 |
anti-β2 GPI IgG
| 36.5 | 26.7 to 47.6 | 100.0 | 97.5 to 100.0 | ∞ | | 0.6 | 0.5 to 0.8 |
anti-β2 GPI IgM
| 31.8 | 22.1 to 42.8 | 94.4 | 89.4 to 97.6 | 5.7 | 2.7 to 12.1 | 0.7 | 0.6 to 0.8 |
Table 5
Comparison of the performance characteristics of ELISA and MLDAf
at least one aPL antibody by ELISA
| 58.8 | 47.6 to 69.4 | 95.8 | 91.1 to 98.5 | 14.1 | 6.3 to 31.5 | 0.4 | 0.3 to 0.6 |
at least one aPL antibody by MLDA
| 67.9 | 56.8 to 77.6 | 89.6 | 83.4 to 94.0 | 6.5 | 4.0 to 10.8 | 0.4 | 0.3 to 0.5 |
at least one aPL IgG or at least three aPL IgM by MLDA
| 67.1 | 56.0 to 76.9 | 96.5 | 92.1 to 98.9 | 19.3 | 8.1 to 46.3 | 0.3 | 0.2 to 0.5 |
Clinical association of MLDA findings
Regarding the number of patients with clinical symptoms and additional disease in the cohort of APS patients, 57/85 patients suffered from DVT and 18/85 patients fulfilled the diagnostic criteria of SLE, which was significantly higher in comparison with the DC group (17/65, P < 0.000001; 3/65, P = 0.003918; respectively). In contrast, the number of patients suffering from TIA and/or ischemic stroke (13/85) and recurrent miscarriages (8/85) did not differ significantly to the DC group (6/65, 10/65, respectively).
Assessing all aPL antibodies detected by ELISA and MLDA, only anti-PI IgM (3/10), anti-PS IgM (5/10), and anti-CL IgM (7/10) antibodies detected by the MLDA demonstrated a significant higher prevalence in the APS patients suffering from TIAs compared with the remaining APS patients (0/75, P = 0.001215; 12/75, P = 0.024165; 26/75, P = 0.041781; respectively). Interestingly, the detection of three or more aPL IgM antibodies by MLDA also revealed a significant higher prevalence in this APS patient cohort (5/10 vs 11/75; P = 0.018102). More detailed analyses of the 13 APS patients with ischemic stroke and/or TIA also revealed a significant higher prevalence of anti-PI IgM (3/13) and anti-CL IgM (9/13) antibodies compared with the remaining APS patients (0/72, P = 0.002826; 24/72, P = 0.027352; respectively). Again, the detection of three or more aPL IgM antibodies by MLDA demonstrated a significant higher prevalence in this APS patient group (6/13 vs 10/72, P = 0.01376).
Remarkably, the absence of anti-CL IgM in the eight APS patients with pregnancy morbidity assessed by ELISA was significantly different to its occurrence in the remaining 77 APS patients (P = 0.022343). Only one patient of this group demonstrated positive anti-CL IgM assessed by MLDA alone and one patient anti- β2 GPI IgM detected by ELISA together with several aPL IgG by both methods. All other aPL IgM antibodies assessed by the MLDA were negative, whereby the absence of anti-β2 GPI IgM detected by this technique almost reached statistical significance (P = 0.051114).
The prevalence of LAC was significantly higher in the 18 patients with APS and SLE compared to the remaining 67 APS patients without SLE (18/18 vs 44/67, P = 0.002159), which was not seen for any other aPL antibody investigated.