Abstract
Objective
We evaluated which aPL combinations increase the risk of future thrombosis in patients with SLE.
Methods
This prospective cohort study consisted of SLE patients who had been tested for all seven aPL (LA, aCL isotypes IgM, IgG and IgA, and anti-β2-glycoprotein I isotypes IgM, IgG and IgA). Pooled logistic regression was used to assess the relationship between aPL and thrombosis.
Results
There were 821 SLE patients with a total of 75 048 person-months of follow-up. During the follow-up we observed 88 incident cases of thrombosis: 48 patients with arterial, 37 with venous and 3 with both arterial and venous thrombosis. In individual models, LA was the most predictive of any [age-adjusted rate ratio 3.56 (95% CI 2.01, 6.30), P < 0.0001], venous [4.89 (2.25, 10.64), P < 0.0001] and arterial [3.14 (1.41, 6.97), P = 0.005] thrombosis. Anti-β2-glycoprotein I IgA positivity was a significant risk factor for any [2.00 (1.22, 3.3), P = 0.0065] and venous [2.8 (1.42, 5.51), P = 0.0029] thrombosis. Only anti-β2-glycoprotein I IgA appeared to add significant risk to any [1.73 (1.04, 2.88), P = 0.0362] and venous [2.27 (1.13, 4.59), P = 0.0218] thrombosis among those with LA. We created an interaction model with four categories based on combinations of LA and other aPL to look at the relationships between combinations and the risk of thrombosis. In this model LA remained the best predictor of thrombosis.
Conclusion
Our study demonstrated that in SLE, LA remained the best predictor of thrombosis and adding additional aPL did not add to the risk, with the exception of anti-β2-glycoprotein I IgA.
Keywords: systemic lupus erythematosus, thrombosis, antiphospholipid antibodies, lupus anticoagulant, anticardiolipin, anti-β2-glycoprotein I, IgA isotype
Rheumatology key messages
In SLE, LA is still the best predictor of venous and arterial thrombosis.
Adding other aPL with IgG and IgM isotypes did not add to the risk.
The presence of anti-β2-glycoprotein I IgA is associated with thrombosis in SLE patients.
Introduction
APS has been classified as the development of venous and/or arterial thromboses, and/or pregnancy morbidity, in the presence of persistently raised levels of either the LA, aCL or anti-β2-glycoprotein I [1]. The classification of APS can only be made if at least one clinical and one persistent laboratory criterion are met.
aPL were first described in patients with SLE [2, 3]. They are present in 11–40% of patients with SLE [4–6]. APS is a significant cause of morbidity and mortality in patients with SLE. In a 10-year prospective study, thrombosis was found as the cause in 26.7% of SLE patients who died, and was always associated with the presence of aPL [7].
It is well known that LA positivity is more strongly associated with both arterial and venous thrombosis than either aCL or anti-β2-glycoprotein I antibodies [8, 9]. An unanswered question is which combinations of positive aPL add to the thrombosis risk.
Currently, the Sydney APS classification criteria include the IgG and IgM isotypes as a laboratory criterion [1]. However, there is controversy in the literature about the role of IgM isotypes. Large studies found that IgM aCL and IgM anti-β2-glycoprotein I are not associated with thrombotic events [9–13]. Moreover, elevated titres of IgA isotypes have been shown to be associated with thrombosis [10, 14–16]. Recent publications have confirmed that anti-β2-glycoprotein I IgA is a risk factor for the development of APS thrombosis [17–20].
The definition of persistent positivity in the Sapporo criteria (1999) [21] was 6 weeks, but was changed in the Sydney criteria (2006) [1] to 12 weeks. In our experience, aPL titres in SLE patients can fluctuate over time and contribute risk even at low and moderate titres [22]. Cross-sectional studies miss the true prevalence of aPL in SLE. In this prospective study, we evaluated which aPL combinations were associated with an increase in risk of future thrombosis in patients with SLE, using our longitudinal cohort in which patients were seen by protocol every 3 months.
Methods
Patient population
The Hopkins Lupus Cohort is a prospective longitudinal cohort of SLE patients ongoing since 1987. It was approved by the Johns Hopkins University School of Medicine Institutional Review Board on an annual basis. Informed written consent was obtained from all subjects. SLE patients were diagnosed according to revised ACR and SLICC criteria [23, 24]. At enrolment, a comprehensive medical history, including date of SLE diagnosis and information on prior thrombosis, was obtained from medical records and the patient. Visits were scheduled quarterly or more frequently, if medically necessary. At each clinic visit, laboratory tests were performed to complete SLE activity indices and for aPL (DRVVT and aCL at every visit; anti-β2-glycoprotein, most recently available, or cohort entry).
Measurement of aPL
This analysis included SLE patients who had been tested for all seven aPL: LA, aCL isotypes IgM, IgG and IgA, and anti-β2-glycoprotein I isotypes IgM, IgG and IgA. The DRVVT with confirmatory testing was performed as published [25]. Anti-β2-glycoprotein I testing became available after 2003. This analysis was based on cohort experience from the first measurement of anti-β2-glycoprotein I through October 2019. aCL and anti-β2-glycoprotein I protein (ELISA IgG, IgM, IgA; Inova Diagnostics, San Diego, CA, USA) were defined as positive when the antibody titre exceeded 20 units (medium titre, following the Sydney classification criteria). The LA was determined by DRVVT with mixing studies and confirmatory studies if prolonged [25]. It was defined as positive if a patient had a DRVVT of 45 s or more, and a positive confirm ratio of more than 1.4. We excluded DRVVT measures if the patients were taking anticoagulants (warfarin/heparin). Among those visits with DRVVT of 45 s or more, 22% had missing confirm ratios. These values were imputed using multiple imputations.
Definition of the thromboses
The past history of thrombotic events was determined at cohort entry by review of all historical records and patient interviews, and was updated at subsequent visits. Any thrombosis was defined as any of the following: venous thrombosis [deep venous thrombosis (DVT), pulmonary embolus (PE) or other venous thromboses] or any arterial thrombosis [myocardial infarction (MI), cerebrovascular accident (CVA), digital ischaemia, or other arterial thromboses]. Venous thromboses were defined by ultrasound or venogram. Arterial thrombosis, in case of stroke, was identified by brain MRI or CT and, in case of MI, by appropriate electrocardiographic changes, creatine kinase or troponin change, or cardiac imaging. Other arterial thrombosis was identified as appropriate for the site involved. Patients who had a history of thrombosis before their first measurement of anti-β2-glycoprotein I were excluded. Follow-up for each patient was either the first incident thrombosis or end of follow-up, which was the last recorded visit in the database.
Statistical analyses
For this prospective analysis, we constructed a dataset with one record for each month of follow-up for each patient. At each person-month, it contained a variable indicating whether the patient had experienced any thrombosis in that month. In addition, each record contained the aPL information of the patient supplied at the most recent prior clinic visit. In some instances, anti-β2-glycoprotein I was not assessed at each visit, due to cost. Of those patients who had anti-β2-glycoprotein I assessments, 62% had it measured only once. We used the most recent evaluation of this variable at a prior visit in our analysis.
Rates of thrombosis were calculated as the number of events, divided by the number of person-months at risk, and results were converted to rates per 1000 person-years. To assess the relationship between aPL and thrombosis, we used pooled logistic regression [26]. For analyses, including LA, we ran it for each imputed dataset. ‘PROC MIANALYZE’ in SAS 9.4 (SAS Institute, Cary, NC, USA) was then used to pool the results to obtain summary estimates. All rate ratios (RR) were adjusted for age at the person month.
Results
There were 821 patients with a complete profile of 7 aPL with a total of 75 048 person-months of follow-up. Of the 821 patients, 94% were female, 51% were Caucasian, 40% were African-American and 50% were <30 years old at the time of SLE diagnosis. Forty-two percent of the patients entered the cohort within 1 year of their SLE diagnosis. The cumulative classification criteria were 48% malar rash, 18% discoid rash, 50% photosensitivity, 53% oral ulcer, 72% arthritis, 47% serositis, 43% renal disorder, 8% neurological disorder, 70% hematological disorder, 87% immunological disorder and 98% ANA positivity, based on revised ACR classification criteria [24]. Additional SLICC classification criteria included 19% direct Coombs test, 58% low C3, 50% low C4 and 12% low CH50 [23].
Among these patients, there were 88 incident cases of thrombosis: 48 patients with arterial [CVA (n = 20), MI (n = 12), digital ischaemia (n = 8) or other arterial thromboses (n = 8)]; 37 patients with venous thrombosis [DVT/PE (n = 33) or other venous thromboses (n = 4)]; and 3 patients with both arterial and venous thrombosis (1 patient with DVT/PE and other arterial thrombosis, 1 patient with CVA and other venous thrombosis, and 1 patient with DVT and MI). The mean (s.d.) disease duration of SLE at the time of thrombosis for the 88 patients was 15 (9.3) years.
Tables 1–3 show the relationship between any, venous and arterial thrombotic events, and the most recent past assessment of each aPL. In individual models, LA was the most predictive of any [RR 3.56 (95% CI 2.01, 6.30), P < 0.0001], venous [4.89 (2.25, 10.64), P < 0.0001] and arterial [3.14 (1.41, 6.97), P = 0.005] thrombosis. Anti-β2-glycoprotein I IgA positivity was a significant risk factor for any [2.00 (1.22, 3.30), P = 0.0065] and venous [2.8 (1.42, 5.51), P = 0.0029] thrombosis. There was no strong evidence that any aPL other than LA significantly associated with arterial thrombosis. aCL IgA showed a higher RR [3.08 (0.75, 12.65), P = 0.1195)] for any thrombosis, but it did not reach statistical significance, likely due to the small number of events and the small number with aCL IgA positivity.
Table 1.
Relationship between any thrombosis and the most recent past assessment of each aPL
| aPL | No. of any thrombotic events | Person-years | Rate per 1000 person-years | Age-adj. RR (95% CI) | P-value | |
|---|---|---|---|---|---|---|
| LA | (−) | 74 | 5936 | 12.5 | 1.00 (ref) | |
| (+) | 14 | 318 | 44.0 | 3.56 (2.01, 6.30) | <.0001 | |
| aCL-G | (−) | 84 | 5986 | 14.0 | 1.00 (ref) | |
| (+) | 4 | 268 | 14.9 | 1.1 (0.4, 3) | 0.859 | |
| aCL-M | (−) | 85 | 6000 | 14.2 | 1.00 (ref) | |
| (+) | 3 | 254 | 11.8 | 0.83 (0.26, 2.61) | 0.7446 | |
| aCL-A | (−) | 86 | 6209 | 13.9 | 1.00 (ref) | |
| (+) | 2 | 45 | 44.4 | 3.08 (0.75, 12.65) | 0.1195 | |
| aB2GPI-G | (−) | 84 | 6056 | 13.9 | 1.00 (ref) | |
| (+) | 4 | 198 | 20.2 | 1.48 (0.54, 4.05) | 0.4442 | |
| aB2GPI-M | (−) | 79 | 5557 | 14.2 | 1.00 (ref) | |
| (+) | 9 | 697 | 12.9 | 0.91 (0.46, 1.81) | 0.7797 | |
| aB2GPI-A | (−) | 68 | 5447 | 12.5 | 1.00 (ref) | |
| (+) | 20 | 807 | 24.8 | 2.00 (1.22, 3.3) | 0.0065 | |
aCL-G: aCL IgG; aCL-M: aCL IgM; aCL-A: aCL IgA; aB2GPI-G: anti-β2-glycoprotein I IgG; aB2GPI-M: anti-β2-glycoprotein I IgM; aB2GPI-A: anti-β2-glycoprotein I IgA; age-adj. RR: age-adjusted rate ratio. Bold values: LA was the most predictive of any [3.56 (2.01, 6.30), P < 0.0001] thrombosis. Anti-β2-glycoprotein I IgA positivity was also a significant risk factor for any [2.00 (1.22, 3.3), P = 0.0065] thrombosis.
Table 2.
Relationship between venous thrombosis and the most recent past assessment of each aPL
| aPL | No. of venous thrombotic events | Person-years | Rate per 1000 person-years | Age-adj. RR (95% CI) | P-value | |
|---|---|---|---|---|---|---|
| LA | (−) | 32 | 5660 | 5.7 | 1.00 (ref) | |
| (+) | 8 | 284 | 28.2 | 4.89 (2.25, 10.64) | <0.0001 | |
| aCL-G | (−) | 38 | 5688 | 6.7 | 1.00 (ref) | |
| (+) | 2 | 256 | 7.8 | 1.06 (0.26, 4.43) | 0.9334 | |
| aCL-M | (−) | 38 | 5704 | 6.7 | 1.00 (ref) | |
| (+) | 2 | 239 | 8.4 | 1.31 (0.32, 5.43) | 0.7123 | |
| aCL-A | (−) | 39 | 5900 | 6.6 | 1.00 (ref) | |
| (+) | 1 | 44 | 23.0 | 4.2 (0.57, 30.97) | 0.1592 | |
| aB2GPI-G | (−) | 37 | 5745 | 6.4 | 1.00 (ref) | |
| (+) | 3 | 198 | 15.1 | 2.22 (0.68, 7.23) | 0.1845 | |
| aB2GPI-M | (−) | 35 | 5267 | 6.6 | 1.00 (ref) | |
| (+) | 5 | 677 | 7.4 | 1.13 (0.44, 2.88) | 0.803 | |
| aB2GPI-A | (−) | 28 | 5168 | 5.4 | 1.00 (ref) | |
| (+) | 12 | 776 | 15.5 | 2.8 (1.42, 5.51) | 0.0029 | |
aCL-G: aCL IgG; aCL-M: aCL IgM; aCL-A: aCL IgA; aB2GPI-G: anti-β2-glycoprotein I IgG; aB2GPI-M: anti-β2-glycoprotein I IgM; aB2GPI-A: anti-β2-glycoprotein I IgA; age-adj. RR: age-adjusted rate ratio. Bold values: LA was the most predictive of venous [4.89 (2.25, 10.64), P < 0.0001] thrombosis. Anti-β2-glycoprotein I IgA positivity was also a significant risk factor for venous [2.8 (1.42, 5.51), P = 0.0029] thrombosis.
Table 3.
Relationship between arterial thrombosis and the most recent past assessment of each aPL
| Antiphospholipid antibody | No. of arterial thrombotic events | Person-years | Rate per 1000 person-years | Age-adj. RR (95% CI) | P-value | |
|---|---|---|---|---|---|---|
| LA | (-) | 44 | 5770 | 7.6 | 1.00 (ref) | |
| (+) | 7 | 299 | 23.4 | 3.14 (1.41, 6.97) | 0.005 | |
| aCL-G | (−) | 48 | 5811 | 8.3 | 1.00 (ref) | |
| (+) | 3 | 258 | 11.6 | 1.6 (0.5, 5.18) | 0.4322 | |
| aCL-M | (−) | 49 | 5821 | 8.4 | 1.00 (ref) | |
| (+) | 2 | 248 | 8.1 | 0.92 (0.22, 3.79) | 0.9078 | |
| aCL-A | (−) | 50 | 6029 | 8.3 | 1.00 (ref) | |
| (+) | 1 | 41 | 24.6 | 2.32 (0.31, 17.12) | 0.4095 | |
| aB2GPI-G | (−) | 49 | 5878 | 8.3 | 1.00 (ref) | |
| (+) | 2 | 191 | 10.5 | 1.35 (0.33, 5.58) | 0.6751 | |
| aB2GPI-M | (−) | 46 | 5383 | 8.5 | 1.00 (ref) | |
| (+) | 5 | 687 | 7.3 | 0.85 (0.34, 2.14) | 0.727 | |
| aB2GPI-A | (−) | 42 | 5305 | 7.9 | 1.00 (ref) | |
| (+) | 9 | 764 | 11.8 | 1.53 (0.74, 3.14) | 0.2499 | |
aCL-G: aCL IgG; aCL-M: aCL IgM; aCL-A: aCL IgA; aB2GPI-G: anti-β2-glycoprotein I IgG; aB2GPI-M: anti-β2-glycoprotein I IgM; aB2GPI-A: anti-β2-glycoprotein I IgA; age-adj. RR: age-adjusted rate ratio. Bold value: LA was the most predictive of arterial [3.14 (1.41, 6.97), P = 0.005] thrombosis.
Since the aPL are correlated, we examined whether an association between each specific aPL and thrombosis was seen after adjustment for the presence of LA. We found that the association between anti-β2-glycoprotein I IgA any [1.73 (1.04, 2.88), P = 0.0362] and venous [2.27 (1.13, 4.59), P = 0.0218] thrombosis persisted after adjustment for LA. aCL IgG, IgM and IgA, and anti-β2-glycoprotein I IgG and IgM positivity were not associated with increased risk of any, venous and arterial thrombosis after adjusting for LA. For aCL IgA, although the point estimate showed a positive association, the number of patients was relatively low, and this relationship did not reach statistical significance (Table 4).
Table 4.
Additive effect of other aPL adjusting for LA
| Antibody combinations | ANY thrombosis |
VENOUS thrombosis |
ARTERIAL thrombosis |
|||
|---|---|---|---|---|---|---|
| Age-adj. RR (95% CI) | P-value | Age-adj. RR (95% CI) | P-value | Age-adj. RR (95% CI) | P-value | |
| Model 1: LA + aCL-G | ||||||
| LA (+) vs (−) | 3.9 (2.14, 7.09) | <0.0001 | 5.8 (2.56, 13.15) | <0.0001 | 3.11 (1.34, 7.23) | 0.0084 |
| aCL-G (+) vs (−) | 0.63 (0.22, 1.81) | 0.3910 | 0.47 (0.1, 2.13) | 0.3289 | 1.04 (0.3, 3.62) | 0.9449 |
| Model 2: LA + aCL-M | ||||||
| LA (+) vs (−) | 3.88 (2.16, 6.98) | <0.0001 | 5.09 (2.28, 11.36) | <0.0001 | 3.41 (1.5, 7.78) | 0.0035 |
| aCL-M (+) vs (−) | 0.53 (0.16, 1.73) | 0.2938 | 0.76 (0.17, 3.31) | 0.7157 | 0.61 (0.14, 2.63) | 0.5083 |
| Model 3: LA + aCL-A | ||||||
| LA (+) vs (−) | 3.42 (1.9, 6.13) | <0.0001 | 4.67 (2.11, 10.34) | 0.0001 | 3.06 (1.36, 6.88) | 0.007 |
| aCL-A (+) vs (−) | 1.86 (0.44, 7.89) | 0.3981 | 2.14 (0.28, 16.56) | 0.4659 | 1.57 (0.21, 11.97) | 0.6623 |
| Model 4: LA + aB2GPI-G | ||||||
| LA (+) vs (−) | 3.68 (2.01, 6.71) | <0.0001 | 4.85 (2.09, 11.28) | 0.0002 | 3.23 (1.41, 7.44) | 0.0058 |
| aB2GPI-G (+) vs (−) | 0.84 (0.29, 2.41) | 0.7408 | 1.03 (0.29, 3.73) | 0.9617 | 0.84 (0.19, 3.69) | 0.8172 |
| Model 5: LA + aB2GPI-M | ||||||
| LA (+) vs (−) | 3.71 (2.08, 6.64) | <0.0001 | 4.99 (2.26, 11.02) | <0.0001 | 3.32 (1.47, 7.47) | 0.0038 |
| aB2GPI-M (+) vs (−) | 0.76 (0.38, 1.52) | 0.4341 | 0.88 (0.34, 2.3) | 0.8008 | 0.72 (0.28, 1.83) | 0.4875 |
| Model 6: LA + aB2GPI-A | ||||||
| LA (+) vs (−) | 3.16 (1.76, 5.68) | 0.0001 | 3.95 (1.77, 8.83) | 0.0008 | 2.96 (1.31, 6.68) | 0.0091 |
| aB2GPI-A (+) vs (−) | 1.73 (1.04, 2.88) | 0.0362 | 2.27 (1.13, 4.59) | 0.0218 | 1.33 (0.64, 2.78) | 0.4469 |
aCL-G: aCL IgG; aCL-M: aCL IgM; aCL-A: aCL IgA; aB2GPI-G: anti-β2-glycoprotein I IgG; aB2GPI-M: anti-β2-glycoprotein I IgM; aB2GPI-A: anti-β2-glycoprotein I IgA; age-adj. RR: age-adjusted rate ratio.
Next, we fitted a model to look at the relationships between combinations of LA and other aPL and the risk of any, venous and arterial thrombosis. We created four categories based on combinations of LA and other aPL: both negative; both positive; LA positive, aPL (i) negative; LA negative, aPL (i) positive; where aPL (i) could be aCL IgG/M/A or anti-β2-glycoprotein I IgG/M/A. In this interaction model, LA remained the best predictor of thrombosis. The risk of any thrombosis after adding anti-β2-glycoprotein I IgA to LA was 2.3 times the risk for those who had LA without anti-β2-glycoprotein I IgA. However, it did not reach statistical significance (P = 0.1253). Adding other aPL with different isotypes did not seem to increase the risk of any thrombosis (Table 5).
Table 5.
Relationship between aPL with LA and risk of ANY thrombosis
| Antibody combinations | No. of events | Person- years | Rate (per 1000 person-years) | Age-adj. RR (95% CI) | P-value | |
|---|---|---|---|---|---|---|
| (1) LA and aCL-G | LA (+), aCL-G (−) | 11 | 232 | 47.4 | 1.00 (ref) | |
| LA (+), aCL-G (+) | 3 | 86 | 34.7 | 0.76 (0.21, 2.74) | 0.6715 | |
| LA (-), aCL-G (+) | 1 | 181 | 5.5 | 0.12 (0.02, 0.93) | 0.042 | |
| LA (-), aCL-G (−) | 73 | 5754 | 12.7 | 0.27 (0.14, 0.51) | <0.0001 | |
| (2) LA and aCL-M | LA (+), aCL-M (−) | 12 | 254 | 47.2 | 1.00 (ref) | |
| LA (+), aCL-M (+) | 2 | 64 | 31.2 | 0.63 (0.14, 2.85) | 0.5537 | |
| LA (−), aCL-M (+) | 1 | 190 | 5.3 | 0.11 (0.01, 0.84) | 0.0334 | |
| LA (−), aCL-M (−) | 73 | 5746 | 12.7 | 0.26 (0.14, 0.49) | <0.0001 | |
| (3) LA and aCL-A | LA (+), aCL-A (−) | 13 | 303 | 42.9 | 1.00 (ref) | |
| LA (+), aCL-A (+) | 1 | 16 | 64.1 | 1.42 (0.18, 11) | 0.7352 | |
| LA (−), aCL-A (+) | 1 | 29 | 33.9 | 0.74 (0.1, 5.74) | 0.7739 | |
| LA (−), aCL-A (−) | 73 | 5906 | 12.4 | 0.29 (0.16, 0.52) | <0.0001 | |
| (4) LA and aB2GPI-G | LA (+), aB2GPI-G (−) | 11 | 246 | 44.7 | 1.00 (ref) | |
| LA (+), aB2GPI-G (+) | 3 | 72 | 41.4 | 0.96 (0.27, 3.46) | 0.9481 | |
| LA (−), aB2GPI-G (+) | 1 | 126 | 7.9 | 0.18 (0.02, 1.39) | 0.1000 | |
| LA (−), aB2GPI-G (−) | 73 | 5810 | 12.6 | 0.28 (0.15, 0.53) | <0.0001 | |
| (5) LA and aB2GPI-M | LA (+), aB2GPI-M (−) | 11 | 234 | 47.1 | 1.00 (ref) | |
| LA (+), aB2GPI-M (+) | 3 | 85 | 35.4 | 0.73 (0.2, 2.64) | 0.6333 | |
| LA (−), aB2GPI-M (+) | 6 | 612 | 9.8 | 0.2 (0.08, 0.56) | 0.0018 | |
| LA (−), aB2GPI-M (−) | 68 | 5324 | 12.8 | 0.27 (0.14, 0.51) | <0.0001 | |
| (6) LA and aB2GPI-A | LA (+), aB2GPI-A (−) | 7 | 220 | 31.8 | 1.00 (ref) | |
| LA (+), aB2GPI-A (+) | 7 | 98 | 71.4 | 2.28 (0.8, 6.52) | 0.1253 | |
| LA (−), aB2GPI-A (+) | 13 | 709 | 18.3 | 0.58 (0.23, 1.45) | 0.2438 | |
| LA (−), aB2GPI-A (−) | 61 | 5227 | 11.7 | 0.37 (0.17, 0.8) | 0.0119 | |
aCL-G: aCL IgG; aCL-M: aCL IgM; aCL-A: aCL IgA; aB2GPI-G: anti-β2-glycoprotein I IgG; aB2GPI-M: anti-β2-glycoprotein I IgM; aB2GPI-A: anti-β2-glycoprotein I IgA; age-adj. RR: age-adjusted rate ratio.
Incident cases of thrombosis were reduced when we looked at venous and arterial thrombosis separately. In combinations with aCL, we did not find any cases with venous thrombosis. Adding other aPL (anti-β2-glycoprotein I IgG/IgM) with different isotypes did not increase the risk of venous thrombosis. Anti-β2-glycoprotein I IgA numerically increased the risk of venous thrombosis among those with LA, but statistically it was not significant (P = 0.1081). In arterial thrombosis, we did not find any additive risk of anti-β2-glycoprotein I IgA to LA. In combinations with aCL IgA or anti-β2-glycoprotein I IgG, we did not find any cases with arterial thrombosis.
Discussion
Based on the present prospective analysis of a large number of SLE patients, our study demonstrated that out of three aPL, LA remained the best predictor of risk of any, venous and arterial thrombosis. The presence of anti-β2-glycoprotein I IgA, which is not a part of APS classification criteria, was also independently associated with thrombosis in SLE patients. Only anti-β2-glycoprotein I IgA provides additional predictive value for thrombosis among SLE patients with LA positivity.
In most studies, it has been shown that in SLE, LA is the main predictor of thrombosis [9, 27–29]. In contrast, a few studies have suggested that the association of single LA positivity with thrombosis was weaker [30, 31]. In a population-based case–control study, LA with the concomitant positivity of anti-β2-glycoprotein I or anti-PT showed the highest thrombosis risk. In contrast, single LA positivity without anti-β2-glycoprotein I or anti-PT did not carry an increased risk for DVT [32].
In the present study having anti-β2-glycoprotein I IgA appeared to add significant risk to any thrombosis risk, and venous thrombosis risk in patients who were also LA positive. However, in arterial thrombosis, there was no strong evidence that any other aPL had additive risk to LA. Other groups have suggested that different antibody profiles were better predictors of thrombosis risk status. The best known profile is called ‘triple positivity’, which requires simultaneous positivity of LA, aCL and anti-β2-glycoprotein I [33]. Sciascia et al. [34] evaluated 23 possible combinations of aPL profiles in SLE patients, and found that combining LA, anti-β2-glycoprotein I and aPS-PT had the best diagnostic accuracy for both thrombosis and pregnancy loss. A Japanese group devised an antiphospholipid score (aPL-S) by testing multiple aPL, including three tests for LA (aPTT, kaolin clotting time and DRVVT) and six antiphospholipid assays (IgG and IgM aCL, IgG and IgM anti-β2-glycoprotein I, IgG and IgM aPS-PT) to diagnose APS and to predict the risk of thrombosis [35].
Other than LA, we found that only anti-β2-glycoprotein I IgA was an independent significant risk factor for any and venous thrombosis. IgA isotypes of aPL were not included in the Sydney APS criteria, due to conflicting data on the association with APS manifestations [1]. Although acceptance of the IgA isotype in APS remains controversial, its diagnostic and clinical significance in SLE had been well demonstrated. Since 2012, the IgA isotype of the aPL has been accepted as a laboratory criterion for SLE in the SLICC SLE Classification Criteria [23]. Anti-β2-glycoprotein I IgA antibodies are frequently seen in SLE patients, with a high association with thrombosis [36]. In a study of patients with isolated IgA anti-β2-glycoprotein I positivity, the increased risk for thromboembolic events was observed only in patients with SLE [37]. A 5-year follow-up of 248 asymptomatic individuals with isolated IgA anti-β2-glycoprotein I positivity showed that the presence of anti-β2-glycoprotein I IgA was an independent risk factor for APS events [18]. The pathogenic role of anti-β2-glycoprotein I IgA in APS patients has also been demonstrated in an experimental mouse model. Purified IgA anti-β2-glycoprotein I isolated from APS patients were injected into mice, and femoral vein clotting was observed. IgA anti-β2-glycoprotein I antibodies were thrombogenic and upregulated tissue factor in mice [38]. These data support that IgA isotypes play a role in the pathogenesis of both primary or secondary APS.
There is controversy over the clinical significance of IgM isotypes, which are part of the Sydney APS classification criteria [12, 39, 40]. In SLE, IgM isotypes were not found to be associated with thrombosis [11, 13]. Similarly, in a systematic review, aCL IgM was not predictive for thrombosis [9]. Moreover, it has been shown that IgM aCL is not associated with lifetime thrombosis risk in SLE [10, 11]. In addition, IgA anti-β2-glycoprotein I was more strongly associated with DVT and with stroke than was IgM [19]. IgA anti-β2-glycoprotein I antibodies in secondary APS patients showed better utility than that provided by the IgM isotype [20]. In the present study, neither IgG nor IgM isotypes provide additional predictive value for thrombosis among SLE patients who were LA positive.
aPL thrombosis in SLE patients is thought to be greater with persistent aPL positivity. The definition of ‘persistent positivity’ was 6 weeks in the Sapporo criteria (1999) [21] and revised to 12 weeks in the Sydney criteria (2006) [1]. However, in SLE, aPL often fluctuate and are often at low or moderate titres, which still confer risk [41–43]. In Hopkins Lupus Cohort, SLE patients were seen quarterly and were tested for aPL (LA and aCL) at every visit. This allowed us to observe this fluctuation during routine clinical practice. The clinical significance of this fluctuation is still under debate. Some studies reported that the increased risk of thrombosis in SLE was associated with the presence of LA and with a persistently positive aCL [41, 44]. Male et al. [45] found, in multivariate analyses, that LA was the strongest predictor of thrombosis, even if grouping patients with one or two positive tests. For transiently positive antibodies, the strength of the association with thrombosis decreased for aCL and aPT, but remained the same for LA and anti-β2-glycoprotein I. Therefore, they suggested serial testing is required only for aCL and aPT. We have shown that LA also fluctuates in SLE 20 years after diagnosis [46].
The main advantages of our study are a large number of patients with SLE, and the prospective assessment of both aPL and the thrombotic events. Retrospectively designed and cross-sectional studies on aPL were unable to assess the temporal relationship with APS manifestations in SLE patients.
This study is not without limitations. Although patients in this cohort were tested for aCL and LAat each visit, 60% of patients were measured for anti-β2-glycoprotein I only once, due to cost. Thus, we could not compare the risk of thrombosis between persistently and transiently anti-β2-glycoprotein I-positive patients. A second limitation is that even given the large number of patients, the number of thrombotic events in same categories was small.
In conclusion, LA remained the strongest predictor of the risk of thrombosis, and in individual models, anti-β2-glycoprotein I IgA showed a significant association with thrombosis. Moreover, anti-β2-glycoprotein I IgA was the only aPL that increased the risk for future thrombosis among patients who were also LA positive. This is the first US SLE cohort study to examine categories of combinations of aPL and thrombosis risk. Particularly for SLE, the ‘triple positive’ category did not confer additional risk over LA alone; however, anti-β2-glycoprotein I IgA (an isotype not considered in ‘triple positivity’ [33] did.
Funding: The Hopkins Lupus Cohort is funded by National Institutes of Health (NIH) AR 43727 and NIH AR 69572.
Disclosure statement: The authors have declared no conflicts of interest.
Data availability statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.
Reference
- 1.Miyakis S, Lockshin MD, Atsumi T. et al. International consensus statement on an update of the classification criteria for definite antiphospholipid syndrome (APS). J Thromb Haemost 2006;4:295–306. [DOI] [PubMed] [Google Scholar]
- 2.Boey ML, Colaco CB, Gharavi AE. et al. Thrombosis in systemic lupus erythematosus: striking association with the presence of circulating lupus anticoagulant. Br Med J (Clin Res Ed). 1983;287:1021–3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Harris EN, Gharavi AE, Boey ML. et al. Anticardiolipin antibodies: detection by radioimmunoassay and association with thrombosis in systemic lupus erythematosus. Lancet 1983;2:1211–4. [DOI] [PubMed] [Google Scholar]
- 4.Petri M. Epidemiology of the antiphospholipid antibody syndrome. J Autoimmun 2000;15:145–51. [DOI] [PubMed] [Google Scholar]
- 5.Merrill JT, Buyon JP, Utset T.. A 2014 update on the management of patients with systemic lupus erythematosus. Semin Arthritis Rheum 2014;44:e1–2. [DOI] [PubMed] [Google Scholar]
- 6.Unlu O, Zuily S, Erkan D.. The clinical significance of antiphospholipid antibodies in systemic lupus erythematosus. Eur J Rheumatol 2016;3:75–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Cervera R, Khamashta MA, Font J. et al. Morbidity and mortality in systemic lupus erythematosus during a 10-year period: a comparison of early and late manifestations in a cohort of 1,000 patients. Medicine (Baltimore) 2003;82:299–308. [DOI] [PubMed] [Google Scholar]
- 8.Asherson RA. The catastrophic antiphospholipid syndrome, 1998. A review of the clinical features, possible pathogenesis and treatment. Lupus 1998;7:S55–62. [DOI] [PubMed] [Google Scholar]
- 9.Galli M, Luciani D, Bertolini G, Barbui T.. Lupus anticoagulants are stronger risk factors for thrombosis than anticardiolipin antibodies in the antiphospholipid syndrome: a systematic review of the literature. Blood 2003;101:1827–32. [DOI] [PubMed] [Google Scholar]
- 10.Domingues V, Magder LS, Petri M.. Assessment of the independent associations of IgG, IgM and IgA isotypes of anticardiolipin with thrombosis in SLE. Lupus Sci Med 2016;3:e000107. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Mehrani T, Petri M.. IgM anti-β2 glycoprotein I is protective against lupus nephritis and renal damage in systemic lupus erythematosus. J Rheumatol 2011;38:450–3. [DOI] [PubMed] [Google Scholar]
- 12.Galli M, Reber G, de Moerloose P, de Groot PG.. Invitation to a debate on the serological criteria that define the antiphospholipid syndrome. J Thromb Haemost 2008;6:399–401. [DOI] [PubMed] [Google Scholar]
- 13.Swadźba J, Iwaniec T, Szczeklik A, Musial J.. Revised classification criteria for antiphospholipid syndrome and the thrombotic risk in patients with autoimmune diseases. J Thromb Haemost 2007;5:1883–9. [DOI] [PubMed] [Google Scholar]
- 14.Pierangeli SS, Liu XW, Barker JH, Anderson G, Harris EN.. Induction of thrombosis in a mouse model by IgG, IgM and IgA immunoglobulins from patients with the antiphospholipid syndrome. Thromb Haemost 1995;74:1361–7. [PubMed] [Google Scholar]
- 15.Shen YM, Lee R, Frenkel E, Sarode R.. IgA antiphospholipid antibodies are an independent risk factor for thromboses. Lupus 2008;17:996–1003. [DOI] [PubMed] [Google Scholar]
- 16.Tsutsumi A, Matsuura E, Ichikawa K. et al. IgA class anti-beta2-glycoprotein I in patients with systemic lupus erythematosus. J Rheumatol 1998;25:74–8. [PubMed] [Google Scholar]
- 17.Murthy V, Willis R, Romay-Penabad Z. et al. Value of isolated IgA anti-beta2-glycoprotein I positivity in the diagnosis of the antiphospholipid syndrome. Arthritis Rheum 2013;65:3186–93. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Tortosa C, Cabrera-Marante O, Serrano M. et al. Incidence of thromboembolic events in asymptomatic carriers of IgA anti β2 glycoprotein-I antibodies. PLoS One 2017;12:e0178889. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Mehrani T, Petri M.. Association of IgA Anti-beta2 glycoprotein I with clinical and laboratory manifestations of systemic lupus erythematosus. J Rheumatol 2011;38:64–8. [DOI] [PubMed] [Google Scholar]
- 20.Ruiz-García R, Serrano M, Martínez-Flores JA. et al. Isolated IgA anti-β2 glycoprotein I antibodies in patients with clinical criteria for antiphospholipid syndrome. J Immunol Res 2014;2014:704395. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Wilson WA, Gharavi AE, Koike T. et al. International consensus statement on preliminary classification criteria for definite antiphospholipid syndrome: report of an international workshop. Arthritis Rheum 1999;42:1309–11. [DOI] [PubMed] [Google Scholar]
- 22.Petri MA, Avci MMagder LS.. Evaluation of different ways to identify persistent positivity of lupus anticoagulant in systemic lupus erythematosus. Lupus Sci Med 2020;7:e000406. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Petri M, Orbai AM, Alarcón GS. et al. Derivation and validation of the Systemic Lupus International Collaborating Clinics classification criteria for systemic lupus erythematosus. Arthritis Rheum 2012;64:2677–86. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Tan EM, Cohen AS, Fries JF. et al. The 1982 revised criteria for the classification of systemic lupus erythematosus. Arthritis Rheum 1982;25:1271–7. [DOI] [PubMed] [Google Scholar]
- 25.Petri M, Nelson L, Weimer F. et al. The automated modified Russell viper venom time test for the lupus anticoagulant. J Rheumatol 1991;18:1823–5. [PubMed] [Google Scholar]
- 26.Ngwa JS, Cabral HJ, Cheng DM. et al. A comparison of time dependent Cox regression, pooled logistic regression and cross sectional pooling with simulations and an application to the Framingham Heart Study. BMC Med Res Methodol 2016;16:148. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Reynaud Q, Lega JC, Mismetti P. et al. Risk of venous and arterial thrombosis according to type of antiphospholipid antibodies in adults without systemic lupus erythematosus: a systematic review and meta-analysis. Autoimmun Rev 2014;13:595–608. [DOI] [PubMed] [Google Scholar]
- 28.Urbanus RT, Siegerink B, Roest M. et al. Antiphospholipid antibodies and risk of myocardial infarction and ischaemic stroke in young women in the RATIO study: a case-control study. Lancet Neurol 2009;8:998–1005. [DOI] [PubMed] [Google Scholar]
- 29.Gebhart J, Posch F, Koder S. et al. Increased mortality in patients with the lupus anticoagulant: the Vienna Lupus Anticoagulant and Thrombosis Study (LATS). Blood 2015;125:3477–83. [DOI] [PubMed] [Google Scholar]
- 30.Pengo V, Biasiolo A, Gresele P. et al. A comparison of lupus anticoagulant-positive patients with clinical picture of antiphospholipid syndrome and those without. Arterioscler Thromb Vasc Biol 2007;27:e309–10. [DOI] [PubMed] [Google Scholar]
- 31.Pengo V, Testa S, Martinelli I. et al. Incidence of a first thromboembolic event in carriers of isolated lupus anticoagulant. Thromb Res 2015;135:46–9. [DOI] [PubMed] [Google Scholar]
- 32.de Groot PG, Lutters B, Derksen RH. et al. Lupus anticoagulants and the risk of a first episode of deep venous thrombosis. J Thromb Haemost 2005;3:1993–7. [DOI] [PubMed] [Google Scholar]
- 33.Pengo V, Biasiolo A, Pegoraro C. et al. Antibody profiles for the diagnosis of antiphospholipid syndrome. Thromb Haemost 2005;93:1147–52. [DOI] [PubMed] [Google Scholar]
- 34.Sciascia S, Murru V, Sanna G. et al. Clinical accuracy for diagnosis of antiphospholipid syndrome in systemic lupus erythematosus: evaluation of 23 possible combinations of antiphospholipid antibody specificities. J Thromb Haemost 2012;10:2512–8. [DOI] [PubMed] [Google Scholar]
- 35.Otomo K, Atsumi T, Amengual O. et al. Efficacy of the antiphospholipid score for the diagnosis of antiphospholipid syndrome and its predictive value for thrombotic events. Arthritis Rheum 2012;64:504–12. [DOI] [PubMed] [Google Scholar]
- 36.Fanopoulos D, Teodorescu MR, Varga J, Teodorescu M.. High frequency of abnormal levels of IgA anti-beta2-glycoprotein I antibodies in patients with systemic lupus erythematosus: relationship with antiphospholipid syndrome. J Rheumatol 1998;25:675–80. [PubMed] [Google Scholar]
- 37.Sweiss NJ, Bo R, Kapadia R. et al. IgA anti-β2-glycoprotein I autoantibodies are associated with an increased risk of thromboembolic events in patients with systemic lupus erythematosus. PLoS One 2010;5:e12280. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Ruiz-Limon PR-PZ, Carrera Marin AL. et al. IgA anti-β2glycoprotein I antibodies are pathogenetic in a mouse model of APS. Arthritis Rheum 2012;64:S742 [abstract 1731]. [Google Scholar]
- 39.Galli M, Borrelli G, Jacobsen EM. et al. Clinical significance of different antiphospholipid antibodies in the WAPS (warfarin in the antiphospholipid syndrome) study. Blood 2007;110:1178–83. [DOI] [PubMed] [Google Scholar]
- 40.Urbanski G, Yelnik CM, Maillard H. et al. Antiphospholipid syndrome with isolated isotype M anticardiolipin and/or anti-B2GPI antibody is associated with stroke. Stroke 2018;49:2770–2. [DOI] [PubMed] [Google Scholar]
- 41.Martinez-Berriotxoa A, Ruiz-Irastorza G, Egurbide MV. et al. Transiently positive anticardiolipin antibodies and risk of thrombosis in patients with systemic lupus erythematosus. Lupus 2007;16:810–6. [DOI] [PubMed] [Google Scholar]
- 42.Out HJ, van Vliet M, de Groot PG, Derksen RH.. Prospective study of fluctuations of lupus anticoagulant activity and anticardiolipin antibody titre in patients with systemic lupus erythematosus. Ann Rheum Dis 1992;51:353–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Inanc M, Donohoe S, Ravirajan CT. et al. Anti-beta2-glycoprotein I, anti-prothrombin and anticardiolipin antibodies in a longitudinal study of patients with systemic lupus erythematosus and the antiphospholipid syndrome. Br J Rheumatol 1998;37:1089–94. [PubMed] [Google Scholar]
- 44.Tektonidou MG, Laskari K, Panagiotakos DB, Moutsopoulos HM.. Risk factors for thrombosis and primary thrombosis prevention in patients with systemic lupus erythematosus with or without antiphospholipid antibodies. Arthritis Rheum 2009;61:29–36. [DOI] [PubMed] [Google Scholar]
- 45.Male C, Foulon D, Hoogendoorn H. et al. Predictive value of persistent versus transient antiphospholipid antibody subtypes for the risk of thrombotic events in pediatric patients with systemic lupus erythematosus. Blood 2005;106:4152–8. [DOI] [PubMed] [Google Scholar]
- 46.Somers E, Magder LS, Petri M.. Antiphospholipid antibodies and incidence of venous thrombosis in a cohort of patients with systemic lupus erythematosus. J Rheumatol 2002;29:2531–6. [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.