Antiphospholipid antibodies in pulmonary embolism treated with direct oral anticoagulants: Prevalence data from unselected consecutive patients

Background Direct oral anticoagulants (DOACs) are the first-choice treatment option for the prevention of the recurrence of venous thrombosis in patients with pulmonary embolism (PE); however, their effect in patients with antiphospholipid syndrome (APS) is challenged. Therefore, the prevalence of antiphospholipid autoantibodies in patients with PE is noteworthy. Objectives To determine the prevalence of unselected patients presenting with PE who meet the criteria for APS based on elevated levels of anticardiolipin (aCL) and anti-β2-glycoprotein I (aβ2GPI) antibodies. Methods Consecutive patients with PE, in whom DOACs were primarily initiated, were tested for aCL and aβ2GPI. If the levels were elevated, the tests were repeated after 12 weeks for APS diagnosis. Laboratory results and patient characteristics were retrospectively collected from a laboratory information system and electronic patient journal entries over a 2-year period. Results The prevalence of APS based on consistently elevated levels of aCL or aβ2GPI was 3.7% (10 of 267 patients). Three patients were double positive. In 11 out of 21 patients (52%) with initially elevated values, the levels of the antibodies normalized after 12 weeks. The patient characteristics were largely similar in those with APS and those without APS; however, patients with APS tended to be older and more likely to receive antithrombotic treatment at the time of PE. Conclusion We found a relatively low prevalence of APS based on aCL or aβ2GPI. The high rate of normalized levels of the antibodies after 12 weeks reaffirms the need for repeated tests for APS diagnosis. Older patients more frequently met the criteria for APS. Determining the effectiveness of DOACs in non–triple-positive patients with APS following venous thromboembolism is important to further determine the feasibility of unselected tests in patients with PE.


| I N T R O D U C T I O N
Pulmonary embolism (PE) ranks high among cardiovascular causes of mortality, and treatment with anticoagulants after the event is a cornerstone in the prevention of long-term recurrence and morbidity [1]. Current guidelines recommend direct oral anticoagulants (DOACs) as first-choice anticoagulation therapy after PE in most patients [1,2]. tests separated by at least 12 weeks [3]. In thrombotic APS, vitamin K antagonists (VKAs) are generally indicated in place of DOACs as secondary prophylaxis [4,5]. Unselected testing for aPL antibodies after PE is not the usual practice. However, the potential treatment implications of APS make attention to the frequency of elevated levels of aPL antibodies in patients presenting with PE relevant.
This retrospective report examined the prevalence of APS based on consistently elevated levels of aβ2GPI and aCL antibodies in unselected patients with PE in whom DOACs were primarily initiated.

| Patient population
Patients diagnosed with PE at the University Hospital of Southwestern Denmark are referred to a thrombosis and anticoagulation clinic for follow-up within a month after their acute hospital course. As part of the diagnostic evaluation and to support decision making regarding thrombosis prophylaxis, all patients are tested for aβ2GPI and aCL antibodies at the first visit to the clinic as a routine protocol. If the levels are elevated, they are retested after a minimum period of 12 weeks to diagnose APS. LAs are not tested because of the risk of confounded results in these patients, in whom treatment has begun, primarily with DOACs, during their hospital stay. While evaluating whether patients should undergo further thrombophilia tests, in addition to those for aβ2GPI and aCL, the team reviews the presentation of the patient with adherence to Danish recommendations for thrombophilia workup.

| Blood sampling and analysis
The methods for blood sampling, sample prehandling, and biochemical analyses are described in detail elsewhere [6]. aCL and aβ2GPI analyses were performed using EliA (Thermo Fisher Scientific), fully automated random access fluorescence enzyme immunoassays on the Phadia 250 instrument (Phadia GmbH), with a coefficient of variation of <10% for both the antibodies. A 99th percentile cutoff was used, as is the recommended practice [3,7], determined by in-house validation in keeping with the approach suggested by the International Society on Thrombosis and Haemostasis (ISTH) [6]. The results of aβ2GPI and aCL tests were obtained from the laboratory information system. Relevant patient characteristics and risk factors, as presented in Table 1, were obtained via manual review by the main author of electronic patient journal entries on admission to the emergency department, based on description of PE diagnosis by a radiologist, and at the first visit to the thrombosis and anticoagulation clinic. Finally, conclusions from team conferences were obtained.

| Data collection and presentation
The regional council approved the study for access to data records (Journal Number 21/49334). According to Danish guidelines, this registry-based study was not subject to review by an ethics board.
Continuous variables are presented as mean ± SD when normally distributed, geometric mean (95% CI) when natural logarithmic transformation was used, and otherwise as median (interquartile range) when skewed. Categorical variables are reported as numbers and proportions (%).

Essentials
• The rate of antiphospholipid syndrome in patients with pulmonary embolism (PE) is ill defined.
• This was a retrospective report of consecutive patients with PE from a thrombosis treatment clinic.
• The elevated levels of anticardiolipin or anti-β2-glycoprotein I antibodies normalized in 11 of 21 patients.
• A relatively low prevalence of antiphospholipid syndrome (3.7%) was detected in unselected patients with PE.
We identified 287 consecutive patients with PE for review, either using the diagnostic codes (248 patients) or via cross-reference to booking lists (39 patients). A total of 18 patients were excluded because of incorrect registering of diagnostic codes (n = 4), because of reversal of diagnosis after review of uncertain radiologic findings (n = 3), or because they were not tested for the aPL antibodies (n = 11).
In sum, 269 patients were tested for the aPL antibodies. Two patients who were positive for the aPL antibodies did not undergo repeat measurement at 12 weeks, both because of comorbidities (severe dementia and disseminated cancer) and a lack of a clinical consequence of repeat measurement. They were excluded from the APS versus non-APS comparisons. Of the resulting patients, a total of c CT included PE found using CT pulmonary angiography and incidentally using abdominal CT (n = 2) or positron emissions tomography CT (n = 5). All others were detected using scintigraphy. The patient with APS was heterozygous for factor V Leiden mutation. Four patients without APS had high-risk congenital thrombophilia (1 had compound heterozygous for factor V Leiden and prothrombin mutations, 1 had antithrombin deficiency, and 2 had protein S deficiency), and 6 were heterozygous for factor V Leiden mutation.
JUSTINUSSEN ET AL. Overall, those who met the criteria for APS and those who did not were similar, apart from patients with APS being older and more frequently on antithrombotic treatment at the time of PE (Table 1).
Although the latter finding may be driven by greater thrombotic potential in older patients, previous studies have also indicated a relatively high frequency of recurrence in patients with APS despite antithrombotic treatment [8]. DOACs were initiated in most patients (34% on apixaban, 44% on rivaroxaban, and 7% on edoxaban vs 11% on low-molecular-weight heparin and 4% on VKAs). The treatment of thrombotic APS with DOACs remains a debated topic [9]. In a recent meta-analysis by Khairani et al. [10], which included 4 randomized controlled trials, with 472 patients, a 5-fold increased odds ratio for subsequent arterial thrombosis was reported in patients randomized to DOACs versus that in those randomized to treatment with VKAs after thrombotic  Cancer workup was done in parallel with APS testing, and the management of coexistent APS and cancer was performed in a case-by-case approach.
APS have noted in their guidance statements that discussion and shared decision making could be applied to patients with non-"highrisk" APS, who have already had several months of good adherence to treatment with DOACs after the first episode of venous thromboembolism [4]. Because of these key knowledge gaps, further research has been encouraged in patients with "lower-risk" APS [4]. In our report, all patients with APS switched to long-term treatment with VKAs following a team conference-based consensus and dialogue with the patients.
In a review of the literature on APS in 2013, Andreoli et al. [11] reported the frequency of aPL antibodies in 9.5% of patients with deep venous thrombosis across 21 studies, which resembles our report of 8.6% of patients with initial positivity for aPL antibodies. In 3 more recent retrospective studies, definitive APS was reported in 4% to 10% of patients with venous thromboembolism selected for thrombophilia workup [12][13][14]. With the exclusion of patients who were single positive for LAs in these studies, the results adjust down to 3% to 8% of patients with APS based on elevated levels of aCL and aβ2GPI antibodies, more comparable with our findings.
Six out of 10 patients who met the criteria for APS were aged >80 years (  [15][16][17][18]. However, the findings on clinical presentation in older versus younger patients with APS differ because some have found higher rates of primary APS and triple aPL positivity in older patients [16,17], and others have reported higher rates of single-positive APS [18]. Male predominance has also been reported to be more prevalent in older patients [16,18]. Because of the lack of evidence on APS in this subgroup, the optimal treatment option is unknown for older patients with consistently elevated values. Our cutoffs were determined in a local healthy population aged between 18 and 66 years. The possibility that an added positive correlation between age and normal aPL titers could play a role in our findings cannot be ruled out. There are no age-specific cutoffs, and interestingly, those used by our laboratory were more rigorous than In clinical guidelines, it is suggested to classify elevated titers of aCL and aβ2GPI determined using enzyme-linked immunosorbent assay into "low range" (20-40 "units") or "moderate-to-high range" (>40 "units" or >99th percentile) [19,20]. Our 99th percentile Phadia or EliA cutoffs, excluding that for aCL IgM, were lower in absolute terms than the 40-unit thresholds; however, as previously reported by Vandevelde et al. [21], the agreement of these thresholds, when applied to automated systems, as is used in this report, can be poor.
Furthermore, while taking into consideration the potential variability in titers among platforms, our locally derived 99th percentile, which has also been suggested by the ISTH guidelines, should provide the best estimate for our population.
Although unselected patients with a confirmed diagnosis of PE, repeated aPL tests after 12 weeks, and a locally derived 99th percentile cutoff value determined with adherence to recommended guidelines [3,6,7]  gested that they play a role [23,24]. All our participants were from a single center in Denmark and were expected to be predominately Caucasian. Thus, our findings may not be generalizable across all ethnicities.
In conclusion, we reported a prevalence of 3.7% for APS based on elevated levels of aCL or aβ2GPI antibodies in unselected patients with PE, in whom DOACs were primarily initiated. Older patients more frequently met the criteria for APS. The high rate of normalized antibody levels after 12 weeks (52%) reaffirms the need for repeated tests for APS diagnosis.

FUNDING
The authors received no funding for this study.

AUTHOR CONTRIBUTIONS
T.J. was involved in designing, data collection, analysis, writing of the original draft, review, and editing. J.G. was involved in designing, supervision, review, and editing. V.B. planned the study and was involved in designing, supervision, review, and editing.