In this study we evaluated 20 "criteria" and "non-criteria" aPLs targeted at 10 different phospholipids or phospholipid-binding proteins in a cohort of well-defined "highly active serologically" APS patients (i.e. 84% were LAC positive and 55% "triple positive") aiming to correlate these aPLs with clinical phenotypes of APS.
APS is a unique acquired thrombotic condition, which causes both arterial and venous thrombosis that may reoccur despite anti-coagulation therapy. Arterial and recurrent thrombosis are both considered to have a worse outcome, thus are usually followed by a more forceful therapy9,26. Herein, we found that certain aPLs and/or aPLs profiles are associated with three APS thrombotic phenotypes: arterial thrombosis, CNS manifestations and recurrent thrombosis. Arterial thrombosis linked with the presence of any 7 or more aPLs compared to 6 or less. This stands in agreement with the notion that aPLs accumulation is an adverse marker of APS. The latter was put forward in 2007 by Bizzaro N. et al 27 and in 2011 by Pengo V. et al28 defining "triple positivity" as a risk factor for thrombosis. Later on Otomo K. et al 9 evaluated criteria (aCL, aβ2GP1) and non-criteria (aPS/PT complex) aPLs documented aPLs accumulation as a prognostic factor, so did Cervera R. et al 29 that demonstrated "non-criteria" aPLs (aPT, aPE, anti-vimentin etc,) relation to disease severity.
Additionally, we found a tie between arterial thrombosis and the existence of aPT. The thrombotic-predictive value of aPT was formerly suggested in a 15 year longitudinal study of SLE patients27 as well as in a prospective study conducted by Forastiero et al30. Lately Zhang S. et al31 found, similar to our data, that aPT relates to arterial thrombosis. In contrast, a review of 11 studies including 1440 patients concluded that aPT assessment was non-contributory for routine APS laboratory workout32. These contradicting conclusions may result from different methods used for detection of aPT or different cohorts of patients assessed. But perhaps the most striking difference is the role looked at as for routine APS workout aPT appears to render no benefit while a plausible role for risk stratification and phenotyping of "well-defined" APS and SLE patients may be suggested.
APS commonly affects the central nervous system (CNS) manifesting as stroke, seizures, dementia, cognitive dysfunction, chorea, migraine, psychosis, demyelination etc. 4,33. In this study, for the first time to the best of our knowledge, an association between APS-related CNS manifestations and a specific profile of four aPLs namely the co-presence of aPT aPG, aPI and aAN of the IgG isotype was observed, of which, like aPT also aAN and aPI were interrelated with thrombosis 34. Annexins are a group of 12 regulatory proteins that are involved in vesicle trafficking, calcium signaling, cell growth, division, and apoptosis. Annexin 2 and 5 have an affinity to phospholipids, and antibodies directed at these proteins were found in patients with either arterial or venous thrombosis. Likewise, anti-phosphatidylinositol antibodies were significantly associated with thrombosis among APS and SLE patients35.
One of the most difficult phenotypes of APS is the 'recurrent thrombosis' one. Approximately 20% of APS patients will experience recurrence within 3.4 to 16.3 years, depending on their treatments with antiplatelet, anticoagulant, or a combination of these therapies 36. Currently, there are no established risk factors for 'the thrombotic recurrence phenotype' though the plausible role of aPLs has been suggested, and may eventually allude to enhanced therapy for patients at risk37-39. Herein, we found that a low recurrence rate, defined as thrombosis free period of more than 3 years, links with a lower accumulative number of aPLs, further supporting the perception that more aPLs allied with a worst outcome.
APS is the most frequently acquired risk factor for recurrent pregnancy losses, ischemic placental dysfunction, fetal growth restriction, preeclampsia, premature birth and intrauterine death14,40. Apart from the thrombotic variants of APS, obstetric APS (OAPS) is probably the most common phenotype. Obstetric morbidity may be the only presentation of APS or may co-exist with thrombotic and non-thrombotic phenotypes41. In the current study, we found that obstetric morbidity was linked with CNS and arterial thrombotic phenotypes, which stands in agreement with other reports such as a recent study by Gris JC et al 42 documenting OAPS association with mental disorders. Besides, we found OAPS to be linked with a profile of two aPLs: aCL and aPS or the sole presence of aAN of the IgG isotype. Recently, aCL was found to be the most common aPL present in a large cohort of 750 pregnancies43. Equally, aPS was linked with OAPS in several studies and in particular, monoclonal aPS antibodies were found to reduce yolk sac growth in animal models as well as placental trophoblastic cell growth and proliferation in humans44 . In contrast to aCL and aPS, the role of aAN in OAPS is yet controversial. The latter was linked with recurrent pregnancy morbidity and losses in some studies35,45 but this association was not ascertained in others41,46,47. Still, the prediction of aPL related pregnancy morbidity is an issue of great debate, especially among women defined as "only aPLs carriers" or those with less than three early miscarriages. Thus, although our results require further studies, the idea that certain aPLs or aPLs profile may be used as a marker of pregnancy morbidity is intriguing.
There is a strong link between aPLs and venous thrombosis as was described in numerous studies as well as the current one while comparing APS patients to healthy subjects and patients diagnosed with sepsis. However, none of the aPLs studied herein was specifically linked with venous thrombosis among APS patients. In other words the vast majority of patients with venous thrombosis were aPL sero-positive and no differences were documented compared to APS patients that did no exhibit venous thrombosis. In contrast, a striking observation in this study was the inverse association of anti P-acid antibody of the IgM isotype with venous thrombosis. The anti P-acid antibody was rarely studied as a single antibody and for the best of our knowledge this is the first report of such inverse association. Of note anti P-acid was linked with thrombosis in several studies which evaluated mostly IgG antibodies and regarded thrombosis in general both arterial and venous. In a recent study of sero-negative APS patients, anti P-acid was linked with fetal losses and not with thrombosis 48. Further studies are required to verify such an inverse correlation. Interestingly in this study the anti-P-acid IgM, as well as other IgM antibodies were linked with older age.
Last but not least, from the diagnostic perspective, the LIA was easy, efficient and with good sensitivity and specificity while employing cutoffs ascertained by healthy controls. This method enabled the discrimination of aPL found in APS patients from those in asymptomatic carriers as reported previously24 .Within our controls, some aPLs were numerically more prevalent among septic control patients compared to healthy subjects, but a significant difference was observed only for anti-β2GP1 IgG. The latter transient appearance during infection was formerly reported49,50. Furthermore, we observed interactions between different aPLs of the same isotype. Similar observations have been reported and led to the hypothesis that broad aPLs positivity may be due to epitope spreading51.
Our study has several limitations, as our cohort included "serologically active" APS patients (84% were LAC positive and 55% triple positive) that fulfill the criteria for APS , which could cast doubt on the implication of our results to patients with "lower serological activity" (e.g. single low titer aPL). A selection bias was inevitable as all patients had a least one clinical manifestation of APS to fulfill its classification criteria. Noteworthy, this study aimed to define subtypes of APS rather than use aPL for diagnosis or classification of disease. Additionally, aPLs were evaluated in this study using a single blood sample from each subject and a single method of detection. The former do not allow estimations of aPL profiles at the time of events, but rather a retrospective clinical correlation, as well as the lack of long term follow up, nor the role of other aPLs or aPL-complexes as anti-PS/PT (only PS and PT separately) or anti- domain-1 of β 2GPI. The latter two aPL may add value to the assessment of APS patients requiring further studies. Using multiple comparative methods for detection of aPLs is of significance, hence the use of the line dot blot (LIA) semi quantitative method as a single method is a limitation. However, all patients included in this study were originally criteria aPL positive and recently Thaler et al. demonstrated similar results for detection of non-criteria aPL by the LIA method compared to ELISA52. Thus, we assume that there will be no significant differences by ELISA to our results by LIA. Moreover, Thaler el al 52 reported a higher sensitivity of the LIA technique for aPL recognizing anionic phospholipids. Lastly, the aim of our study was to evaluate APS profiles, therefore correlation with a single criteria or non-criteria manifestation (e.g. intra uterine fetal death of thrombocytopenia) was not evaluated.