INTRODUCTION: Antiphospholipid (aPL) antibody mediated interference with the anticoagulant effect of annexin A5 (A5) - designated "annexin A5 resistance" (A5R) - has been implicated in adverse clinical outcomes. In this study, we ask whether A5R is associated with anamnestic thrombotic complications and adverse pregnancy outcomes in patients with the lupus anticoagulant (LA).

PATIENTS & METHODS: We conducted a cross-sectional study including 143 patients (median age: 40.5 years, interquartile range (IQR): 31.8-60.0; female: n=119) who tested persistently positive for LA according to ISTH criteria (Miyakis et al. J Thromb Haemost 2006; 4: 295). Thrombotic complications were defined as a composite of arterial and/or venous thromboembolic events. Pregnancy complications were defined according to the modified Sapporo criteria. Out of the 143 patients, 94 (65.7%) had a history of a thrombembolic event (arterial: n=21, venous: n=80, both: n=7). Ninety-three women had at least one pregnancy, and 40 (43.0%) out of these women had at least one pregnancy complication as defined by the above mentioned criteria. It is important to mention that on average, these pregnancy complications had occurred more than ten years before study inclusion. A5R assays were done using phospholipid that was first exposed to patient plasmas as previously described; A5R results were expressed as the ratio of coagulation times with and without annexin A5. In a subsample (n=109), a blinded external validation using two separate measurements of the A5R was performed.

RESULTS: In blinded analysis, the concordance between the two A5R measurements was very high (Liu's concordance correlation coefficient=0.83, average difference=12.2 units (95% limits of agreement from Bland-Altman analysis: -37.0 - 61.3). The A5R was inversely correlated with levels of antibodies against cardiolipin (Spearman's rho for IgG and IgM: -0.65 and -0.24, respectively), and against beta2-glycoprotein I (Spearman's rho for IgG and IgM: -0.57 and -0.28). The median A5R level was 205 units (IQR: 205-250, range: 130-336). Patients with a prior history of a thrombotic complications had significantly lower median levels of A5R (193.1 vs. 238.7 units, mean difference: 26.7 units (95% CI: 11.0-42.4), p=0.001). This difference was independent of both IgG and IgM isotype antibody levels against cardiolipin and beta2-glycoprotein I (adjusted mean difference: 20.7 units (95% CI: 6.4-35.0), p=0.005), and of whether patients were on anticoagulation or not (adjusted mean difference: 24.8 units (95% CI: 4.3 - 45.2), p=0.018). In the 95 women with at least one pregnancy, average A5R were lower in women with (n=40) compared to those without (n=53) at least one pregnancy complication (median: 200.9 vs. 231.9 units), however, this difference was not statistically significant (t-test p=0.255).

CONCLUSION: A5R is associated with a history of thromboembolic complications in LA positive patients, suggesting that the ex vivo measurement of the A5R may reflect a prothrombotic phenotype in this patient population. Interestingly, we found no clear association between the A5R and anamnestic pregnancy complications was observed, however, the generalizability of this finding is hampered by the long time interval between anamnestic pregnancy complications and inclusion in this study.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.