Introduction

Antiphospholipid syndrome (APS) is a systemic autoimmune disease distinguished by two primary hallmarks: the persistent existence of antiphospholipid antibodies (aPL) and several clinical manifestations such as thrombosis and obstetrical manifestations linked to these antibodies. The 2023 ACR/EULAR classification criteria have expanded to include thrombocytopenia, microangiopathy, and cardiac valve disease1,2,3.

APL encompass a remarkably heterogeneous spectrum of autoantibodies. This includes lupus anticoagulants, which were first identified in the recognition of APS, based on phospholipid-dependent coagulation functional tests4. Subsequently, antiphospholipid autoantibodies, such as anticardiolipin antibodies (aCL) with IgG and IgM isotypes, were incorporated into classification criteria in 19995, and anti-β2 glycoprotein I antibodies (anti-β2GPI) with IgG and IgM isotypes were added to revised classification criteria in 20066. Additionally, several studies have found autoantibodies closely associated with the clinical manifestations of APS, including anti-phosphatidylserine/prothrombin (anti-PS/PT)7, anti-β2GPI domain I8, anti-plasmin9, anti-annexin A510, anti-annexin A211 and other recently discovered autoantibodies like anti-neutrophil extracellular trap antibodies12. To date, both aCL and anti-β2GPI antibodies have served as pivotal benchmarks in the diagnostic framework for APS.

The 2023 criteria have updated the content framework, incorporating entry criteria and a scoring system to enhance the specificity of classifying APS, aiming to increase the homogeneity in clinical research. They establish clinically relevant thresholds (40 and 80 units) for aCL/anti-β2GPI positivity, with a focus on quantifying aPL positivity across distinct domains and assigning different weights accordingly. Notably, moderate (40–79 units) or high positive (≥ 80 units) results for IgM of aCL and/or aβ2GPI contribute only 1 point, whereas IgG positivity holds a minimum value of 4 points. With laboratory domains mandating a minimum of 3 points, these updates prevent the inclusion of isolated IgM positivity in research studies. Furthermore, the 2023 ACR/EULAR criteria continue to recommend enzyme-linked immunosorbent assay (ELISA) as the exclusive method for detection, sparking extensive discussions following the release of criteria13,14,15.

The clinical significance of IgM isotype of aCL and anti-β2GPI in APS has been a subject of controversy16. Several studies have suggested that IgM aCL/anti-β2GPI, similar to IgG aCL/anti-β2GPI, is associated with an increased susceptibility to thrombosis17, whereas conflicting perspectives argue that IgM aCL/anti-β2GPI might not offer additional predictive value for the occurrence of thrombotic events18,19. This debate extends to pregnancy morbidity, with a meta-analysis demonstrating a correlation between both isotypes (IgM and IgG) of aCL/anti-β2GPI and miscarriage20. Conversely, a case-control study found a correlation only between the anti-β2GPI IgG isotype and fetal death21. These findings underscore the complexity surrounding the clinical implications of IgM isotypes in the context of APS.

Recently, there has been a growing adoption of alternative solid phase assays such as the addressable multiplex flow immunoassay and the chemiluminescent immunoassay (CIA) in laboratories22,23. In practice, ELISA has been found to present problems such as being time-consuming, dependent on the test operator, and lacking harmonization between laboratories. Values varied considerably among different kits, and even when the same kit was used, high data variability still existed24,25,26. Several studies have reported that CIA exhibits diagnostic efficiency comparable to ELISA27,28,29,30,31. In our previous study, we have also compared the clinical performance of CIA and ELISA for aCL and anti-β2GPI. The results suggested that CIA had a good agreement and correlation with ELISA based on the manufacturer’s cutoff reference values32. Besides technological advantages in terms of expansive dynamic range, lack of interfering emissions and elevated signal intensity, one study also indicated that CIA improved the risk assessment of thrombosis33,34.

Validation results from various laboratories have revealed notable discrepancies in cut-off values. This underscores the importance of conducting validation across multiple laboratories35. The new criteria include a defined threshold for a moderate-high titer in ELISA, with the 40/80 serving as the designated moderate-high titer for the ELISA method. It’s important to note that this particular threshold does not translate to CIA, given the substantial numerical variations observed in CIA values36. There is a paucity of research aimed at identifying semi-quantitative thresholds for CIA, kin to those established for ELISA, making it difficult to achieve standardized interpretation across laboratories. Moreover, in the context of the CIA method, the absence of a precisely defined numerical value corresponding to the moderate-to-high titer threshold in ELISA introduces challenges in effectively stratifying risks.

In the present study, we delineated precise semi-quantitative thresholds for CIA analyses targeting aCL/anti-β2GPI, classifying them into discrete categories of moderate and high concentrations. The establishment of these thresholds was guided by their alignment with ROC specificity values observed in ELISA corresponding to moderate and high levels. Moreover, we calculated likelihood ratios at different thresholds and juxtaposed them with ELISA results to substantiate the effectiveness of the CIA in identifying APS36,37. Our findings substantiated that CIA exhibited clinical performance on par with ELISA, suggesting its prospective broad implementation in classification for APS and risk stratification of thrombosis, pregnancy morbidity, and thrombocytopenia. Furthermore, we investigated the value of IgM in the diagnosis and risk assessment of antiphospholipid syndrome and found that IgM possessed relatively lower yet still considerable clinical significance.

Methods

Study cohorts

The study adhered to the principles set forth in the Declaration of Helsinki and obtained approval from the Institutional Review Board of Ruijin Hospital (ID: 2016-61), affiliated with Shanghai Jiao Tong University School of Medicine in Shanghai, China. A cohort comprising 166 individuals meeting the 2023 ACR/EULAR classification criteria for APS was identified from the APS-Shanghai (APS-SH) database, established by expert rheumatologists and statisticians at the Shanghai Jiao Tong University School of Medicine (Shanghai, China)38,39. Clinical and laboratory test results were retrieved for analysis, with exclusions for individuals testing positive solely for IgM of aCL and/or aβ2GPI to adhere to the laboratory standards requiring a minimum of three points. Definitions for thrombosis, pregnancy morbidity, thrombocytopenia, microvascular disease, and cardiac valve involvement were in accordance with the 2023 ACR/EULAR criteria2,3. Additionally, disease controls included 30 individuals with ankylosing spondylitis (AS), 30 with Sjögren’s syndrome (SS), 31 with rheumatoid arthritis (RA), and 103 with systemic lupus erythematosus (SLE). Moreover, 120 individuals without any autoimmune diseases were selected as healthy controls. Demographic and clinical characteristics of the cohort are detailed in Table S1.

Laboratory assays

All samples were analyzed for aCL IgG/IgM and anti-β2GPI IgG/IgM at Ruijin Hospital with two commercially available immunoassays: ELISA (QUANTA Lite®, Inova Diagnostics) and CIA (QUANTA Flash®, Inova Diagnostics). The ELISA assays used conform to the Harris standard, ensuring standardized and reliable measurements of antiphospholipid antibodies.

The manufacturer’s recommended cut-off value for aCL and β2GPI IgG/IgM CIA stands at 20 luminescence units (CU). Quantification of aPL titers utilized GPL/MPL units for ELISA aCL IgG/IgM isotypes, SMU/SGU for ELISA aβ2GPI IgM/IgG, and chemiluminescent units (CU) for both CIA aCL and aβ2GPI isotypes. The detection of LAC was conducted in accordance with established international protocols outlined by the International Society of Thrombosis and Hemostasis (ISTH). This involved employing screening, mixing studies, and confirmation tests utilizing the diluted Russell’s viper venom time (dRVVT) methodology40. LA positivity was assessed based on persistent positivity, as defined by the 2023 ACR/EULAR classification criteria.

Defining semi-quantitative thresholds

To establish the moderate to high titer thresholds for the detection of aCL/anti-β2GPI antibodies using the CIA method, we separately calculated the manufacturer’s threshold for the CIA method, the 99th percentile value from the reference population following established guidelines, and conducted a receiver operating characteristic (ROC) analysis to identify the threshold associated with the largest Youden index41,42,43. Crucially, we carefully chose the cut-off values on the ROC curve of CIA to closely match the specificity observed in ELISA at titers of 40 and 80, respectively. After a comprehensive assessment of their clinical performance, these values were ultimately assigned as the moderate/high thresholds for the CIA method. Importantly, both thresholds demonstrated a likelihood ratio comparable to the moderate/high titer thresholds observed in ELISA. In instances where the specificity of ELISA at 80 GPL/MPL was 1, we selected the minimum value from the ROC results that met the criterion of matching the same specificity in the CIA method.

Statistical analysis

Spearman’s correlation test was applied to explore the quantitative correlation between diverse types of antibodies and different detection methods. A non-parametric approach was utilized to determine the 99th percentile values. The DeLong test was used to evaluate the potential presence of a statistically significant difference in the Area Under the Curve (AUC) calculated from ROC curves. Assessing the coherence among distinct detection methods and diverse thresholds, we computed Cohen’s kappa coefficient along with its corresponding 95% confidence intervals. Extensive calculations were conducted to ascertain sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), positive likelihood ratio (PLR), negative likelihood ratio (NLR), and odds ratios (OR). Further stratifying the antibody titer into three layers according to set thresholds, logistic regression analysis was employed to compute OR values for high and moderate titers relative to the low titer, accompanied by their respective 95% confidence intervals. For combinations involving two or more antibodies, regression analyses were employed to evaluate their performance in risk analysis, respectively. We utilized the McNemar’s test for 2 × 2 contingency tables, while the Wilcoxon test was employed to compare continuous variables. Statistical significance was set at a two-tailed α value of 0.05. All data analyses were conducted using R (Version 4.1.3).

Results

Quantitative correlation between IgM and IgG of aCL and aβ2GPI detected by ELISA and CIA

The Spearman’s correlation coefficient matrix between various antibodies and different detection methods was depicted in Fig. 1. A comparative analysis of CIA and ELISA assays demonstrated strong correlations when assessing four identical antibodies, with correlation coefficients for aCL IgG at 0.71, aCL IgM at 0.62, aβ2GPI IgG at 0.73, and aβ2GPI IgM at 0.61 and all the p-values below 0.05. These findings unveiled strong correlations between ELISA and CIA methods for detection of aPL. Additionally, under the identical method, the correlation coefficients between aCL and aβ2GPI within the same isotype all exceeded 0.6 while those between IgG and IgM isotype within the same antibody remained below 0.45. These outcomes underscored the rationale for scoring aCL/aβ2GPI IgG and aCL/aβ2GPI IgM as distinct groups.

Fig. 1
Fig. 1The alternative text for this image may have been generated using AI.
Full size image

Matrix of correlation (Spearman correlation coefficient) between ELISA and CIA methods. The correlation matrix portrayed the pairwise Spearman correlation coefficients among IgG and IgM isotypes of aCL/aβ2GP1 detected using both ELISA and CIA methods. Circles in the upper right corner represented the strength of correlations, with numerical values presented in the lower left corner. All values were above zero, indicating positive correlations between the detection results. Robust correlations were observed for different antibodies and diverse detection methods within the same isotype, while varying antibody isotypes showed comparatively weaker associations.

Qualitative agreement between ELISA and CIA for detection of aCL and aβ2GPI

The 2023 criteria have defined cut-off values for detecting moderate to high titers of aCL and aβ2GPI through the ELISA method, establishing them at 40U and 80U respectively. To establish the appropriate moderate thresholds for CIA method, we examined several strategies including the cut-off recommended by the manufacturer, the 99th percentile of healthy individuals, the cut-off determined by the largest Youden index, and the calculated threshold considering equal specificity of 40U in ELISA for identifying APS using ROC curves. Notably, thresholds derived from the largest Youden index, depicted in Fig. 2, significantly deviated from those obtained through the other three methods. This discrepancy arose from the Youden index’s simultaneous evaluation of sensitivity and specificity, contrasting the APS classification criteria that prioritize specificity over sensitivity44.

Fig. 2
Fig. 2The alternative text for this image may have been generated using AI.
Full size image

ROC curves of distinct antibody isotypes detected by ELISA or CIA. ROC analysis was conducted to evaluate the diagnostic performance of aCL IgG/IgM and anti-β2GPI IgG/IgM antibodies using both CIA and ELISA methods, with the moderate cutoff value applied. The AUC for all antibodies identified through ELISA and CIA were displayed. Notably, the AUC value for anti-β2GPI IgG/IgM was higher for CIA compared to ELISA. Conversely, the AUC values for aCL IgG/IgM showed similarity between CIA and ELISA. The ROC curve included black points denoting the threshold, sensitivity, and specificity corresponding to the maximum Youden’s index under the CIA method.

As illustrated in Table 1, the threshold determined by aligning specificity with that of ELISA at 40U for APS identification showed the closest alignment with the likelihood ratio compared to other approaches, thereby emerging as our preferred choice. Hence, we separately calculated the thresholds for moderate levels of four antibodies. The threshold values for aCL IgG and IgM were 116.70 CU and 39.75 CU, respectively, while for aβ2GPI IgG and IgM, they were 409.60 CU and 25.75 CU, respectively. The corresponding Kappa values provided additional support (Table 1). For aCL IgG and IgM, the Kappa values were 0.84 and 0.80, while for aβ2GPI IgG and IgM, the values stood at 0.71 and 0.67, respectively.

Table 1 Clinical performance of aPL for identifying APS.
Full size table

Subsequently, we determined the corresponding high threshold in the CIA method, aligning it with the equal specificity of 80U obtained through ELISA in the ROC curve. The values exhibited notable variability across various antibodies, particularly in the case of aβ2GPI, where IgG registered 2733.00 CU while IgM marked 36.00 CU. The thresholds for aCL IgG and IgM were calculated as 446.20 and 332.60, respectively. Such variances potentially stemmed from differences in concentration distribution, and the CIA method seemed to amplify these distinctions. Nevertheless, the agreement between CIA and ELISA remained strong, confirming the reliability of established values. In summary, the moderate and high thresholds for CIA demonstrated a harmonious alignment with ELISA.

Clinical performances of aCL and aβ2GPI detected by ELISA and CIA in identifying APS

The ELISA and CIA methods demonstrated comparable specificity and sensitivity in identifying APS at moderate titers. While IgM isotype of aCL and aβ2GPI showed slightly higher specificity than IgG isotype of aCL and aβ2GPI, IgG isotype exhibited much greater sensitivity (Table 1). The specificity at high thresholds was close to or equal to 1, with considerably lower sensitivity. The lower sensitivity of the IgM isotype of aCL and aβ2GP1 at both moderate and high level suggests its restricted capability to independently identify APS patients, thereby elucidating the lower scores in the 2023 classification criteria. CIA exhibited performance comparable to ELISA in ROC analysis (Fig. 2). The AUC values of CIA were significantly higher than those of ELISA for aβ2GPI IgG (P = 0.014) and IgM (P = 0.047). No statistically significant differences emerged between the two detection methods for aCL IgG and aCL IgM. In both ELISA and CIA methods, IgG isotype of aCL and aβ2GP1 exhibited higher AUC values than IgM isotypes. Overall, CIA served as a viable alternative to ELISA owing to similar clinical performance. Additionally, IgM isotype of aCL and aβ2GP1 might not be suitable as a stand-alone diagnostic tool considering its low sensitivity.

The association between aCL/aβ2GPI and clinical manifestations

We conducted an analysis of the OR values for individuals with high and moderate antibody titers in comparison to those with low titers, regarding the occurrence of thrombosis, pathological pregnancy, and thrombocytopenia events (Fig. 3). In a comprehensive overview, except aCL IgM detected by CIA, higher concentrations of aCL and aβ2GPI IgG/IgM were generally associated with an increased OR value for all three clinical events, whether identified via ELISA or CIA. This observation supported the rationale for establishing two thresholds in the semi-quantitative interpretation of antibodies. While the IgG isotype of aCL and aβ2GPI demonstrated stronger associations with all three clinical events, the IgM isotype also showed significant yet reduced correlations, particularly with thrombosis and pregnancy morbidity. For instance, aCL IgM exhibited associations with thrombosis and pregnancy morbidity, albeit with lower OR values compared to IgG. However, the correlation between aβ2GPI IgM and clinical manifestations was less consistent. At moderate titers, the confidence interval for aβ2GPI IgM spanned 1, indicating inconclusive risk assessment. Yet, with higher titers, aβ2GPI IgM demonstrated a clearer association. Additionally, CIA yielded higher OR values for the IgG isotype of aCL and aβ2GPI compared to ELISA across all three clinical manifestations, although this discrepancy was not evident for the IgM isotype. In conclusion, employing CIA method and establishing two thresholds in semi-quantitative result interpretation could assist clinicians in APS risk assessment. Although IgM antibodies exhibit lower OR values compared to IgG, their role in risk assessment should not be overlooked, particularly in the context of higher titers.

Fig. 3
Fig. 3The alternative text for this image may have been generated using AI.
Full size image

Odds ratios for moderate and high titers of aCL and aβ2GP1 in association with thrombosis, pregnancy morbidity and thrombocytopenia. The results were presented in the form of odds ratios, comparing high and moderate titers to low titers in relation to thrombosis, pregnancy morbidity and thrombocytopenia, along with their respective 95% confidence intervals. High titers for the IgG and IgM isotypes of aCL and aβ2GP1 were defined as values exceeding 80 U in ELISA. In CIA, the thresholds varied among antibodies: 446.20 CU for aCL IgG, 332.60 CU for aCL IgM, 2733.00 CU for anti-β2GPI IgG, and 36.00 CU for anti-β2GPI IgM, respectively. Moderate titers are defined as values falling below the previously mentioned high thresholds and ranging above 40 U in ELISA. In CIA, the thresholds differed among antibodies: 116.70 CU for aCL IgG, 39.75 CU for aCL IgM, 409.60 CU for anti-β2GPI IgG, and 25.75 CU for anti-β2GPI IgM, respectively.

Combined values of aPLs in identifying APS patients

The clinical performance of combined antibodies detected by ELISA and CIA was exhibited in Table 2. When examining the combined detection of IgG for both aCL and aβ2GPI, IgM for both aCL and aβ2GPI, and both IgG and IgM together, the specificity demonstrated a substantial increase compared to individual antibodies, surpassing 0.99. Taking into account the intrinsically high specificity of LAC, its combination with antibodies achieved a flawless specificity score of 1. The combined aCL IgG and aβ2GPI IgG exhibited the higher sensitivity at 42.17% in ELISA and 56.63% in CIA compared to the combination of aCL IgM and aβ2GPI IgM. Moreover, simultaneous detection of IgG and IgM contributed to improving the homogeneity of the study cohort.

Table 2 Clinical performance of combined aPL and LAC for APS diagnosis.
Full size table

According to the latest 2023 classification criteria, particularly for IgG isotypes of aCL and aβ2GPI, higher antibody levels and simultaneous positivity for both antibodies can yield higher scores. Consequently, the adopted scoring system reflected consideration for enhancing the specificity of the included study population.

Combined values of aPLs in risk stratification

To further investigate the risk assessment value of combined antibodies at moderate to high titers for clinical events, we analyzed the OR values of these combined antibodies in relation to thrombosis, pregnancy morbidity, and thrombocytopenia (Fig. 4). There was a consistent trend where IgG isotype of aCL/aβ2GPI generally surpassed corresponding IgM isotype in OR values, whether tested independently or in combination by ELISA and CIA. While not consistently matching the risk stratification strength of the IgG isotype of aCL/aβ2GPI, the IgM isotype of aCL/aβ2GPI still demonstrated significant value in risk assessment for thrombosis and pregnancy morbidity. Regarding thrombocytopenia, the lower limit of the confidence interval for OR values involving IgM isotype of aCL/aβ2GPI was always lower than or slightly higher than 1, indicating that IgM antibodies did not establish themselves as reliable predictors whether evaluated separately or collectively.

Fig. 4
Fig. 4The alternative text for this image may have been generated using AI.
Full size image

Odds ratios of combined aCL, aβ2GPI and LAC for risk stratification. The results were presented as odds ratios correlating the individual and combined positivity of biomarkers with the occurrence of thrombosis, pregnancy morbidity and thrombocytopenia, along with their respective 95% confidence intervals. A positive outcome for IgG and IgM isotypes of aCL/aβ2GPI was defined as exceeding 40 U in ELISA, while a positive outcome in CIA was determined by surpassing established thresholds (116.70 CU for aCL IgG, 39.75 CU for aCL IgM, 409.60 CU for anti-β2GPI IgG, and 25.75 CU for anti-β2GPI IgM).

To explore potential variations in aPL antibody profiles among individuals with different clinical manifestations, we created a heatmap (Fig. S1). Our analysis revealed a lack of substantial disparity in both antibody profiles and titers among patients presenting with thrombosis, pregnancy morbidity and thrombocytopenia.

Discussion

The APS is a systemic autoimmune disease characterized by the persistent prevalence of aPL. The testing for aPL plays a crucial role in identifying APS patients, with ongoing expansion of the aPL profile and updates in testing standards. Despite the ongoing exploration and application of alternative methods like CIA, ELISA remains the sole method endorsed in international classification criteria for detecting aPL45. Limitations of ELISA include its lengthy process, reliance on manual labor, high background noise, and variability between different experiments or laboratories24,25,26,33. The high level of automation, along with the clinical implications related to its efficiency and the ability to conduct numerous antibody tests, including aCL and aβ2GPI of different isotypes (IgA/IgG/IgM), and aβ2GPI-domain 1 (classified as non-criteria aPL), positions CIA as the most advanced in the clinical laboratory33,38,39 Studies have proved that the CIA had a parallel diagnostic efficiency compared to ELISA27,28,29,30,31,32,34, yet its broader utilization remains constrained due to inter-platform discrepancies46. The global expansion of CIA methods across multiple clinical centers requires standardized CIA methods and comprehensive validation, along with a comparison with ELISA within the same cohort.

ACL IgG/IgM present in titers > 40 GPL/MPL or exceeding the 99th percentile, and aβ2GPI IgG/IgM in titers surpassing the 99th percentile, measured by a standardized ELISA, were categorized as moderate-high aPL titers. The definition of moderate to high titers of aPL guides the risk stratification of patients and the formulation of treatment strategies47. The 2023 criteria stratify ELISA titers into three tiers, delineating 40 U as the moderate threshold and 80 U as the high threshold2,3. However, analogous stratification for CIA, aligning with ELISA thresholds, lacks adequate research focus. Previous study reported that the utilization of 40U/80U as moderate/high thresholds is deemed acceptable for aCL/aβ2GPI IgG ELISA, but not for CIA36. In this study, we analyzed various thresholds including manufacturer’s recommendations, the 99th percentile of healthy individuals, the largest Youden index, and the corresponding cut-off calculated for ELISA at 40U with equal specificity for identifying APS. The implementation of classification criteria in clinical research seeks to minimize heterogeneity among enrolled patients. Hence, prioritizing consideration is given to the specificity of the relevant indicators within the criteria. Consequently, we established a moderate threshold for CIA, which exhibited the same specificity as ELISA at 40U. Similarly, the high threshold in the CIA method was determined based on the equal specificity of 80U by ELISA in the ROC curve. Subsequent Kappa consistency test, Spearman correlation analysis, and likelihood ratio validated the chosen CIA thresholds. It is worth noting that the numerical values of the cut-offs for different isotypes of each antibody vary, and they are comparable in magnitude to those calculated based on ROC sensitivity in the aforementioned study36. This discrepancy is partly attributed to the wide dynamic range of the results obtained through the CIA method.

Comparative analysis demonstrated significantly higher CIA area under the curve (AUC) values for aβ2GPI IgG and IgM compared to ELISA, whereas no substantial distinction was observed for aCL. The masking of crucial β2GPI epitopes during ELISA plate coating48,49 might account for this disparity, indicating that the presentation of antigens on magnetic particles30 in CIA enhanced diagnostic efficacy. Furthermore, CIA demonstrated comparable performance to ELISA in risk stratification, notably in the case of IgG isotype of aCL/aβ2GPI, where CIA generally yielded higher OR values than ELISA. Taken together, these findings underscored CIA as a viable alternative diagnostic method. Evaluation of antibody stratification OR values concerning thrombosis, pregnancy morbidity, and thrombocytopenia revealed a consistent trend: elevated antibody titers, whether by ELISA or CIA, corresponded to heightened OR values, indicating a stronger correlation with clinical symptoms. This substantiated the significance of establishing dual thresholds.

The 2023 criteria assign varied scores to distinct antibody subtypes and titers, allocating only 1 point for IgM positivity2. This has sparked extensive debate regarding the exclusion of individuals solely positive for IgM from the criteria13,14,15. Previous investigations into the significance of IgM yielded inconclusive outcomes16,17,18,19,20,21. Our study demonstrates that IgM antibodies, though exhibiting lower sensitivity compared to IgG, still contribute to the identification of APS patients. Specifically, IgM antibodies showed a higher specificity for APS diagnosis, which was crucial for minimizing false-positive results. For instance, the combination of IgM with IgG antibodies enhanced specificity, supporting its utility as a complementary marker in APS diagnostics. However, the lower sensitivity of IgM antibodies underscored its limited ability to independently identify APS patients, aligning with the 2023 criteria’s decision to assign only 1 point to IgM positivity. In terms of risk stratification, IgM antibodies demonstrated predictive value for thrombosis and pregnancy morbidity, albeit with lower odds ratios compared to IgG. This suggests that while IgM may not be as strongly associated with clinical outcomes as IgG, it still plays a role in risk assessment, and the complete exclusion of IgM from the classification criteria may not be justified.

Currently, LA detection still adheres to the guidelines of the ISTH40, and our results demonstrate that LA exhibits the highest odds ratio for clinical outcomes such as thrombosis and pregnancy morbidity. This highlights the strong diagnostic and prognostic value of LA in APS. In contrast, while aCL and anti-β2GPI titers also show significant associations with clinical manifestations, their odds ratios are comparatively lower. However, caution is warranted when testing patients receiving anticoagulant therapy due to the potential for false-positive results caused by the influence of anticoagulant drugs. The introduction of new testing systems, such as the prothrombin-activating Taipan snake venom time (TSVT) screening test and ecarin time (ET) confirmatory test, has shown reduced susceptibility to the effects of vitamin K antagonists (VKA) and direct factor Xa inhibitors (DFXaI)50. Moving forward, we will advocate for the validation of TSVT/ET in both clinical and experimental settings to enhance the reliability of LAC results.

The constraints of our study stem from its singular focus on a retrospective design within a single center. To bolster the robustness of our findings, a pivotal step involves conducting an international multicenter study. The results obtained from the CIA method reveal a considerable range in magnitude, and due to variations among manufacturers, these thresholds may not universally apply when extended to other reagent vendors or platforms. Importantly, we have undertaken a pragmatic methodological analysis using our large cohort, providing semiquantitative thresholds customized for non-ELISA platforms based on clinical approach.

In conclusion, our study defined dual thresholds for IgG/IgM isotypes of aCL/aβ2GPI in CIA corresponding to the moderate/high levels of ELISA. Importantly, likelihood ratios for dual thresholds, defined by levels of specificity, were utilized to achieve harmonization across antibody detection platforms. Further analysis highlighted its similarity in clinical diagnostic accuracy and risk stratification when compared to ELISA. Based on the analysis of this study, we do not currently advocate for the complete exclusion of IgM from the classification criteria. Moreover, careful consideration is necessary, as the assignment of only one point, even when both aCL IgM and aβ2GPI IgM are positive simultaneously, may lead to a substantial portion of patients with only the IgM isotype consistently being excluded from APS-related research.