Autoimmunity Reviews 13 (2014) 281–291

Contents lists available at ScienceDirect

Autoimmunity Reviews journal homepage: www.elsevier.com/locate/autrev

Review

Efficacy of aspirin for the primary prevention of thrombosis in patients with antiphospholipid antibodies: An international and collaborative meta-analysis Laurent Arnaud a,b,c,⁎,1, Alexis Mathian a,b,c,1, Amelia Ruffatti d, Doruk Erkan e, Maria Tektonidou f, Ricard Cervera g, Ricardo Forastiero h, Vittorio Pengo i, Marc Lambert j, Maria Angeles Martinez-Zamora k, Juan Balasch k, Stephane Zuily l, Denis Wahl l,2, Zahir Amoura a,b,c,2 a Department of Internal Medicine, French National Reference Center for Systemic Lupus and Antiphospholipid Syndrome, AP-HP, Groupe Hospitalier Pitié-Salpêtrière, 47-83 bd de l'hôpital, 75013 Paris, France b Institut National de la Santé et de la Recherche Médicale (INSERM), UMR-S 945, Paris, France c Université Pierre et Marie Curie, UPMC Univ Paris 06, Paris, France d Rheumatology Unit, Department of Clinical and Experimental Medicine, University of Padua, Padua, Italy e Hospital for Special Surgery, Weill Cornell Medical College, New York, USA f First Department of Internal Medicine, School of Medicine, National University of Athens, Athens, Greece g Department of Autoimmune Diseases, Hospital Clínic-Institut d'Investigacions Biomèdiques August Pi i Sunyer, University of Barcelona, Barcelona, Catalonia, Spain h Department of Physiology, Favaloro University, Buenos Aires, Argentina i Clinical Cardiology, Department of Cardiac Thoracic and Vascular Sciences, University of Padova, Padova, Italy j Internal Medicine Department, Universitary Hospital, Lille, France k Institut Clínic of Gynecology, Obstetrics and Neonatology, Hospital Clínic-Institut d'Investigacions Biomèdiques August Pi i Sunyer, Faculty of Medicine, University of Barcelona, Barcelona, Spain l Nancy University Hospital, Université de Lorraine, INSERM U961, Vandœuvre-Les-Nancy, France

a r t i c l e

i n f o

a b s t r a c t We performed a meta-analysis to determine whether aspirin has a significant protective effect on risk of first thrombosis among patients with antiphospholipid antibodies (aPL+). Observational and interventional studies identified from the Medline, Embase and Cochrane databases were selected if they assessed the incidence of first thrombosis in aPL+ patients treated with aspirin versus those without. Pooled effect estimates were obtained using a random-effects model. Of 1211 citation retrieved, 11 primary studies (10 observational and 1 interventional) met inclusion criteria, including a total of 1208 patients and 139 thrombotic events. The pooled odds ratio (OR) for the risk of first thrombosis in patients treated with aspirin (n = 601) was 0.50 (95%CI: 0.27 to 0.93) compared to those without aspirin (n = 607), with significant heterogeneity across studies (I2 = 46%, p = 0.05). Subgroup analysis showed a protective effect of aspirin against arterial (OR: 0.48 [95%CI: 0.28–0.82]) but not venous (OR: 0.58 [95% CI: 0.32–1.06]) thrombosis, as well as in retrospective (OR: 0.23 [0.13–0.42]) but not prospective studies (OR: 0.91 [0.52–1.59]). Subgroup analysis according to underlying disease revealed a significant protective effect of aspirin for asymptomatic aPL+ individuals (OR: 0.50 [0.25–0.99]), for systemic lupus erythematosus (SLE) (OR: 0.55 [0.31–0.98]) and obstetrical antiphospholipid syndrome (APS) (OR: 0.25 [0.10–0.62]). This meta-analysis shows that the risk of first thrombotic event is significantly decreased by low dose aspirin among asymptomatic aPL individuals, patients with SLE or obstetrical APS. Importantly, no significant risk reduction was observed when considering only prospective studies or those with the best methodological quality. © 2013 Elsevier B.V. All rights reserved.

Article history: Received 1 October 2013 Accepted 25 October 2013 Available online 2 November 2013 Keywords: Aspirin Thrombosis Meta-analysis Antiphospholipid antibodies

Contents 1. 2.

Introduction . . . . . . Methods . . . . . . . 2.1. Literature search 2.2. Study selection . 2.3. Data extraction .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

282 282 282 282 282

⁎ Corresponding author at: Service de Médecine Interne 2, Groupe Hospitalier Pitié-Salpêtrière, 47-83 bd de l'Hôpital, 75013 Paris, France. Tel.: +33 1 42 17 80 40; fax: +33 1 42 17 80 44. E-mail address: [email protected] (L. Arnaud). 1 Laurent Arnaud & Alexis Mathian contributed equally to this work. 2 Denis Wahl & Zahir Amoura contributed equally to this work. 1568-9972/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.autrev.2013.10.014

282

L. Arnaud et al. / Autoimmunity Reviews 13 (2014) 281–291

2.4. Disease definitions . . . . . . . . . . . . . . . . . . . . . . . . 2.5. Contact with authors and management of duplicate studies or patients 2.6. Quality assessment . . . . . . . . . . . . . . . . . . . . . . . . 2.7. Cumulative meta-analyses . . . . . . . . . . . . . . . . . . . . 2.8. Statistical analyses . . . . . . . . . . . . . . . . . . . . . . . . 3. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Literature search . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Publication bias . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Selected studies . . . . . . . . . . . . . . . . . . . . . . . . . 3.4. Assessment of methodological quality . . . . . . . . . . . . . . . 3.5. Risk of first thrombosis associated with aPL . . . . . . . . . . . . . 3.6. Risk of first thrombosis according to pathogenic backgrounds . . . . 3.7. Cumulative meta-analyses . . . . . . . . . . . . . . . . . . . . 3.8. Hemorrhagic risk . . . . . . . . . . . . . . . . . . . . . . . . 4. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Take-home messages . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix A. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix B. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix C. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix D. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

283 283 283 283 283 283 283 283 283 283 284 285 285 287 287 287 288 288 288 288 289 288 290 291

1. Introduction

2.1. Literature search

Antiphospholipid antibodies (aPL) are a heterogeneous family of autoantibodies directed against phospholipid and/or phospholipid-binding proteins, which include, among others, lupus anticoagulants (LA), anticardiolipin antibodies (aCL) and anti-β2GPI antibodies (anti-β2GPI) [1,2]. aPL predispose to pregnancy morbidity and vascular thrombosis, which clinically define the antiphospholipid syndrome [3]. They can be detected in a variety of situations, including auto-immune diseases such as systemic lupus erythematosus (SLE), pre-surgical testing or blood donation in the general population, or during investigations for recurrent spontaneous abortion, fetal death or premature birth. They may also be detected during infections but in that case are mostly transient. The exact prevalence of aPL in the general population as well as in SLE patients is currently debated due to previous aPL assay standardization and threshold issues [4–8]. Based on a limited number of studies, the incidence of a first thrombotic event is estimated to range from 0 to 1 for 100 patient-years in asymptomatic aPL+ individuals, from 2 to 4 for 100 patient-years in aPL+ patients with SLE and up to 7 for 100 patient-years in those with obstetrical APS associated with SLE [9–11]. There is a general agreement on long term anticoagulation to prevent recurrences of thrombosis in patients with APS [12]. However, there is limited evidence regarding the adequate primary prevention of thrombosis in aPL carriers [13]. Aspirin has been shown effective for the primary prevention of arterial thrombotic events in high-risk individuals in the general population [14]. The use of low-dose aspirin has recently been advocated in aPL+ patients at high-risk for thrombosis such as those with SLE or positive LA or persistently positive aCL at medium-high titers or in the presence of other thrombotic risk factors [9]. Additionally, low dose aspirin may also prevent venous thrombosis as it has been recently reported in the secondary prevention of venous thromboembolism in the general population [15]. Yet, the prescription of low dose aspirin to prevent the first thrombosis in aPL carriers remains very controversial because of the conflicting results among the previous studies [13]. We therefore conducted a systematic review of the literature and metaanalysis to assess the efficacy of low-dose aspirin for the primary prevention of thrombosis in aPL+ patients.

Two main investigators (L.A. & A.M.) searched EMBASE (1974–July 2012), MEDLINE (1966–July 2012) and the Cochrane Database of Systematic Reviews (The Cochrane Library, 2012, issue 7) for original articles without language restrictions. Search strategy combined free text search, exploded MESH/EMTREE terms and all synonyms of the following Medical Subject Headings terms: antiphospholipid antibodies, systemic lupus erythematosus, obstetrical antiphospholipid syndrome, lupus-like syndrome, lupus coagulation inhibitor, anticardiolipin antibodies, beta 2-Glycoprotein I, aspirin, and thrombosis (see the detailed search strategy in Appendix A). We also searched for additional articles from the reference list of relevant papers obtained from the electronic search. In addition, the gray literature was explored by hand searching the conference abstracts of the American College of Rheumatology and the European League Against Rheumatism from January 1999 to July 2012.

2. Methods This meta-analysis was performed in accordance with the recommendations of the Meta-analysis of Observational Studies in Epidemiology (MOOSE) group [16].

2.2. Study selection Selection criteria were determined before data collection. Inclusion criteria included: (1) either observational and interventional studies; (2) reporting on primary prophylaxis (no prior thrombosis); (3) with continuous treatment with low-dose aspirin (4) in patients with aPL. For each study, a 2 × 2 table was constructed based on treatment with aspirin and occurrence of thrombosis. For the studies published by Forastiero et al. [17], Hereng et al. [18], and Cervera et al. [19], the 2 × 2 table was updated by personal communications of the authors. Studies with intermittent prophylaxis (i.e., only in high risk periods) were excluded [20]. Reviews, editorials, guidelines, letters, case reports, articles on different outcomes and sub-studies of eligible studies were also excluded. Whenever disagreement occurred, it was resolved by discussion between the 2 main investigators until a consensus was reached. A log of reasons for rejection of citations identified from the searches was kept. 2.3. Data extraction Data were simultaneously and independently extracted by the 2 main investigators. The recorded information for each selected study included: study design; patients' characteristics and underlying disease, if available (asymptomatic aPL carrier, SLE, obstetrical APS); test methods and thresholds for defining a positive aPL according or not to international standards [5,21–23]; methods that objectively verified the thromboembolic events in accordance with Sapporo's (before 2006)

L. Arnaud et al. / Autoimmunity Reviews 13 (2014) 281–291

[23] or Sydney's (after 2006) [5] consensus conference criteria; and the number of arterial or venous thrombotic events among individuals treated with aspirin versus those not treated. 2.4. Disease definitions SLE was defined according to the 1997 ACR criteria for SLE [24] and APS according to Sapporo's (before 2006) or Sydney's (after 2006) consensus conference criteria [5,23]. Asymptomatic aPL+ carriers were defined as individuals with aPL without history of thrombosis or obstetrical APS or defined SLE. 2.5. Contact with authors and management of duplicate studies or patients

283

venous) in aPL+ patients. Accordingly, we constructed 2 × 2 tables for each outcome from each study. We used the odds ratio (OR) as the measure of association in this meta-analysis and we obtained pooled-risk estimates by using a random-effects model, according to the method of DerSimonian and Laird (fixed-model effect was tested but not selected because of study heterogeneity, data not shown). Publication bias was examined using a funnel plot and Egger's regression test [33]. Heterogeneity (Cochran's chi-2 and I2 tests) was considered statistically significant at p b 0.10 and I2 N 50%. Separate a priori subgroup analyses were planned for arterial versus venous thrombotic events, for prospective versus retrospective studies, as well as for the different pathogenic backgrounds, when possible. Spearman's coefficient was used for testing correlations. The agreement between the two primary reviewers for quality score was measured using the intra-class correlation coefficient (ICC) with JMP 8 software (SAS Institute Inc., Cary, USA). The meta-analysis was performed with the use of the software RevMan version 5.0 (The Cochrane Collaboration, 2008, Copenhagen, Denmark).

All authors of selected studies were contacted by email to check and/ or update the event counts recorded and asked to provide additional information regarding the detailed number of arterial and venous thrombotic events; all but one responded. When we identified patients that had been included in multiple papers, the analysis was limited to the study with the largest number of patients in order to avoid duplications [25–28]. Because 10 duplicate patients were included in both studies by Pengo [29] and Ruffatti [25] we decided to include these two studies in the meta-analysis, as excluding one of those would have discarded an important number of unique patients. When it was impossible to extract data for a subgroup analysis (even after the authors were contacted), the study was excluded.

Our literature search identified 1211 citations, of which 11 unique studies [17–19,25,27,29,34–37] were included in this meta-analysis (Fig. 1). The reasons for the exclusion of some articles after a full text review [20,25,26,28,38–45] are shown in Fig. 1.

2.6. Quality assessment

3.2. Publication bias

The methodological quality of included articles was reviewed by the 2 main investigators. Any information about authors, journals, and institutions was deleted for this purpose. Studies were evaluated independently by means of an original scoring system adapted from the meta-analyses by Wahl et al. [30–32]. The quality assessment form encompassed 39 items distributed in 5 categories: clinical criteria (SLE patients classified or not according to the American College of Rheumatology criteria, obstetrical APS classified or not according to Sapporo's or Sidney's criteria); methods for outcome diagnosis (detailed description of clinical, laboratory and imaging tests); laboratory assessment of aPL (test methods and thresholds according or not to international standards); modalities of aspirin treatment (daily dose mentioned or not); quality of design (single-centre versus multicenter studies); randomized controlled trial versus prospective or retrospective cohort studies, and confounders (adequate description of cardiovascular risk factors, other thrombophilic disorders, co-prescription of chloroquine or hydroxychloroquine). We tested correlations (using Spearman's coefficient) between study quality scores and study effect estimates (expressed as log (OR)) as a way to assess relationships between study quality and magnitude of effect estimate.

Visual examination of the funnel plot (Appendix B) only revealed minor asymmetry and Egger's test (p = 0.38) did not show any statistical evidence for publication bias.

2.7. Cumulative meta-analyses To assess the contribution of individual studies to the pooled estimate, we performed cumulative meta-analyses according to: (i) descending quality score, therefore allowing examination of the impact of the studies of lesser quality on the pooled estimate (for this analysis, study names are blinded but available from main author upon request); (ii) increasing date of publication, to allow assessment of the changes in association between exposure to aspirin and occurrence of thrombosis in patients over time; and (iii) increasing study size to reveal how many patients are needed before statistical significance is obtained. 2.8. Statistical analyses In the primary analysis we considered the efficacy of aspirin for the primary prevention of all thrombotic events (i.e., both arterial and

3. Results 3.1. Literature search

3.3. Selected studies The characteristics of the 11 studies included in the meta-analysis [17–19,25,27,29,34–37] are shown in Table 1. Included studies were 6 prospective [17,19,25,27,29,34] and 4 retrospective [18,35–37] cohort studies and a single randomized controlled trial [34]. The article published by Erkan et al. in 2007 [34] reports on the results of both a prospective cohort study and a randomized controlled trial (APLASA study), and therefore accounts for two studies. The publication period runs from 2001 to 2012. Sample sizes varied from 65 to 238 aPL+ patients for a total of 1208. Six studies were multicentric [19,25,29,34,35]. All studies but one included SLE patients. Five studies excluded obstetrical APS. The length of follow-up ranged from 2.3 to 8.7 years. The percentage of female ranged from 68.1% to 100%. The mean/median age at study entry varied from 29.3 to 47.8 years. The presence of aCL, LA, and anti-β2GPI was assessed in 11, 11 and five [17,25,27,29,37] studies, respectively. The number of patients lost to follow-up was reported in 6 studies [18,19,25,34,36]. Confounding factors, such as cardiovascular risk factors and thrombophilic disorders were reported in seven [18,25,29,34,36,37] and three [17,29,36] studies, respectively. The prescription of chloroquine and hydroxychloroquine was mentioned in 5 studies [18,27,34,37] but the analysis of their independent effect on thrombosis rate was not performed by the authors. Only one trial monitored the compliance using participants' self-report [34]. 3.4. Assessment of methodological quality The average quality score of included studies was 64 of 100, ranging from 51 to 80. Inter-observer reliability between the two main investigators for quality scoring was very good (ICC: 0.87, 95%CI: 0.58–0.96, p b 0.001). We found a significant correlation between quality scores and log (OR) of primary studies (r = 0.74, p = 0.009).

284

L. Arnaud et al. / Autoimmunity Reviews 13 (2014) 281–291

Fig. 1. Flow-chart for study selection.

3.5. Risk of first thrombosis associated with aPL Among the 1208 aPL+ patients from the 11 included studies, 601 (49.8%) received prophylaxis with aspirin. Forty-seven thrombotic events (7.8%) occurred in those receiving aspirin versus 92 events (15.2%) in those who did not (p b 0.0001). Three out of eleven included studies (27%) reported a significant association between aspirin prophylaxis and the overall risk of first thrombosis. By pooling the 11 included studies

(Fig. 2), we found an overall OR of 0.50 for the risk of first thrombosis in aPL+ patients treated with aspirin versus those who were not treated (95%CI: 0.27 to 0.93). We found no significant correlation between the publication year and the risk of thrombosis among individual studies (r = −0.03, p = 0.93). We found significant but limited statistical heterogeneity across studies (I2 = 46%, p = 0.05), as also suggested by the visual inspection of the L'Abbé plot (Appendix C). Subgroup analysis according to the type of thrombotic event (Fig. 3) showed a protective

Table 1 Studies addressing primary prophylaxis of thrombosis with low dose aspirin in patients with antiphospholipid antibodies. Author, year, (Reference)

Study type

Erkan 2001 [26]

Number enrolled

Population studied Obstetrical APS nb. (%)

SLE nb. (%)

Ageayr, mean ± SD or median (range)

Patients treated with aspirin nb. (%)

Follow-up duration yr, mean ± SD or median (range)

Retro

65

65 (100)

NA

29.3 ± 5

31 (47.7)

8.1 ± 3.5

Forastiero 2005 [8]

Pros

108

33 (30.6)

42 (3–85)

31 (28.7)

3.8 (0.3–10)

Tarr 2007 [18]

Pros

81

NA

32.6 ± 11.0

52 (64.2)

5.0

Erkan 2007 [25]h

RCT

98

0 (0.0)

48 (49.0)

2.30 ± 0.95

h

Erkan 2007 [25]

Pros

74

0 (0.0)

61 (82.4)

2.46 ± 0.76

Hereng 2008 [9]h

Retro

103

0 (0.0)

75 (72.8)

5.3 ± 2.1

121

121 (100)

20 (18.5) 81 (100) 64 (65.3) 26 (35.1) 37 (35.9) NA

40 (0–82)

50 (41.3)

5.0

31.0 ± 13

72 (55.8)

8.7

45 ± 15.0

37 (35.6)

4.5

h

Cervera 2009 [10] h Pros Tektonidou 2009 [28]h Pengo 2011 [20]h

Retro

129

0 (0.0)

Pros

104

0 (0.0)

Ruffatti 2011 [16]h

Pros

238

32 (13.4)

Martinez–Zamora 2012 [27]h

Retro

87

57 (65.5)

129 (100) 14 (13.5) 69 (29.0) 0 (0.0)

43.1 42.7 47.8 46.2 41.7

± ± ± ± ±

12.8d 14.0e 14.8d 14.1e 16.7

40.9 ± 11.1

123 (51.7)

2.92 ± 1.0

32.8 ± 3.5f 31.9 ± 5.6g

21 (24.1)

7.3 ± 3.5f 8.2 ± 4.3g

Thrombotic events nb. (%) Total

Arterialb Venousc

21 (32.3) 12 (11.1) 3 (3.7) 6 (6.1) 4 (5.4) 18 (17.5) 3 (2.5) 26 (20.2) 25 (24.0) 10 (4.2) 11 (12.6)

14 (21.5) 6 (5.6)

7 (10.8)

3 (3.7)

0 (0.0)

4 (4.1)

2 (2.0)

1 (1.4)

3 (4.1)

9 (8.7)

9 (8.7)

2 (1.7)

1 (0.8)

16 (12.4) 13 (12.5) 3 (1.3)

10 (7.8) 12 (11.5) 7 (2.9)

7 (8.0)

4 (4.6)

6 (5.6)

APS, antiphospholipid syndrome; SLE, Systemic Lupus Erythemathosus; Retro, retrospective cohort study; RCT, randomized controlled trial; Pros, prospective cohort study; NA, not available. a At study entry b Including transient ischemic attack and catastrophic antiphospholipid syndrome c Including pulmonary embolism d For the group treated with aspirin e For the group not treated with aspirin f For obstetric APS g For asymptomatic carrier h Some data were reported through personal communications with the first author in 2011 or 2012, except for ref [9] for which data were provided by Dr Marc Lambert.

L. Arnaud et al. / Autoimmunity Reviews 13 (2014) 281–291

285

Fig. 2. Forest plot for main analysis.

effect of aspirin when considering arterial (OR: 0.48 [95%CI: 0.28–0.82]) but not venous (OR: 0.58 [95%CI: 0.32–1.06]) thrombosis separately. Finally, we observed a significant decreased in the risk of first thrombosis among aPL+ individuals treated with aspirin in the subgroup analysis considering only retrospective studies (OR: 0.23 [95%CI: 0.13–0.42]), but not prospective studies (OR: 0.91 [95%CI: 0.52–1.59]) (Fig. 4).

3.6. Risk of first thrombosis according to pathogenic backgrounds The underlying disease associated with the presence of aPL may influence the risk of thrombosis and explain heterogeneity among studies. Subgroup analysis according to the pathogenic background revealed a significant protective effect of aspirin for asymptomatic aPL+ patients (OR: 0.50 [0.25–0.99]), for those with systemic lupus erythematosus

(SLE) (OR: 0.55 [0.31–0.98]), as well as for those with obstetrical antiphospholipid syndrome (APS) (OR: 0.25 [0.10–0.62]) (Fig. 5).

3.7. Cumulative meta-analyses Additionally, we performed cumulative meta-analysis (i.e., adding the studies one at a time according to a specified criterion [e.g., date of publication, size or quality] and updating the OR as each new study is added), as a manner to assess the influence of this criterion on the overall risk. When arranged by ascending publication date, we found that the risk estimate of first thrombosis was stable from the year 2009 (Appendix D, panel A). When ordered by increasing study size, we found that the impact of the largest studies was minor when the total number of included patients reached 500 (Appendix D, panel B). The cumulative meta-

Fig. 3. Forest plot for artery versus venous subgroup analysis.

286

L. Arnaud et al. / Autoimmunity Reviews 13 (2014) 281–291

Fig. 4. Forest plot for subgroup analysis according to study design.

analysis of studies arranged by descending quality scores demonstrated a decreasing pooled OR for the risk of first thrombosis, as lower-quality score studies were successively added to the higher-quality score studies

(Appendix D, panel C). Sensitivity analysis performed by including the 50% of studies with the highest methodological quality did not demonstrate a protective effect of aspirin (OR: 0.92 [95%CI: 0.50–1.69]).

Fig. 5. Forest plot for subgroup analysis according to the pathogenic background.

L. Arnaud et al. / Autoimmunity Reviews 13 (2014) 281–291

3.8. Hemorrhagic risk No major bleeding occurred in the 4 studies for which hemorrhagic events were reported [18,34,37]. Additionally, three patients treated with aspirin experienced minor bleeding in the APLASA study [34]. 4. Discussion Our meta-analysis of 11 studies including 1208 aPL+ patients and 139 thrombotic events revealed that aPL+ patients treated with long term low-dose aspirin have a 50% risk reduction for the occurrence of a first thrombotic event than those who did not receive this treatment. Additionally, subgroup analysis according to the pathogenic backgrounds revealed that the risk reduction was significant among asymptomatic aPL+ individuals as well as among aPL+ patients with SLE or obstetrical APS. Up to now, a single randomized, double-blind, placebo-controlled clinical trial, namely the Antiphospholipid Antibody Acetylsalicylic Acid (APLASA) study [34] addressed our question of interest. This study, which showed no protective effect of aspirin, was terminated early because no thrombotic event occurred in the placebo group, which was attributed to the relatively short follow-up duration, the good control of additional vascular risk factors and the inclusion of a significant proportion of individuals with a low risk aPL profile [9]. Additionally, 10 observational studies [17–19,25,27,29,35–37] also addressed our question of interest by examining asymptomatic aPL+ individuals, patients with SLE and aPL, or patients with obstetrical APS. Three of these studies showed a significant decrease in the risk of first thrombosis among patients treated with aspirin [18,35,37] while a non-significant decrease was observed in five studies [19,25,28,29,34,36] and a non-significant increase in two studies [17,34] (Figs. 3 & 4). As expected, significant statistical heterogeneity was observed across included studies. We therefore used a random-effects model and had planned a priori subgroup analyses to investigate the sources of this heterogeneity. First, we grouped studies according to their design. We found that the risk of first thrombosis was significantly decreased when considering retrospective but not prospective studies (Fig. 4), which can have different meanings. On the one hand, retrospective studies are more susceptible to biases than prospective ones. Therefore, selection or measurement biases may account for the difference observed between study designs. In the other hand, aPL may be transient. We therefore cannot exclude that the status of aPL may have changed at some time for the prospective studies, even though repeated aPL measurements were performed in most of these studies. Importantly, we observed no significant publication bias, but we found a strong correlation between the methodological quality scores and the risk estimates of included studies. Additionally, we did not observe a significant protective effect of low-dose aspirin in the cumulative meta-analysis according to quality scores and the sensitivity analysis, when considering only studies with the best methodology. Importantly, the methodological quality scores were assessed based on the methodology reported in the articles. Therefore, the quality scores may have been underestimated for some studies as the authors may not have fully reported on the methodology that was used. Altogether, these results strongly suggest that additional studies with better methodology are urgently needed. Another important feature is that the laboratory methods employed for the measurement of aPL have been heterogeneous across included studies, and enrolled subjects have been classified according to updating laboratory criteria. Additionally, concomitant pro- and antithrombotic risk factors have been infrequently taken into account across these studies. For instance, hypertension was found to be an independent predictor of thrombosis in the retrospective study by Ruffatti [25], whereas arterial events have been associated with the coexistence of metabolic syndrome [46] or smoking and oral contraceptives in primary APS [47]. On the contrary, hydroxychloroquine use may be protective for thrombosis in SLE as it is shown in several observational studies

287

[40,48]. ORs provided in this meta-analysis are generally unadjusted for other thrombotic risk factors as well as for potential protective factors. Patients had heterogeneous pathogenic backgrounds, including asymptomatic aPL, SLE, and obstetrical APS. Importantly, subgroup analyses according to these disease subsets revealed a significant protective effect of aspirin across those three pathogenic backgrounds (Fig. 5). While this meta-analysis strengthens the recently published recommendations of the 13th aPL task force about the use of low-dose aspirin for the primary prevention of thrombosis in SLE patients and aPL + individuals, one should keep in mind that these recommendations were rated weak recommendations by the task-force itself, because these were only based on low- to moderate-quality evidence [9]. Recommendations from the 13th aPL task force [9] have also highlighted the importance of considering both the risks for recurrence and the risk for major bleeding. Because bleeding events were scarcely reported in included studies our meta-analysis could not formally assess the benefit/risk ratio of aspirin, while the latter carries a small, but noninsignificant risk of bleeding [49]. Yet, no major bleeding occurred in the four studies for which these data were reported, and minor bleeding occurred only in 3 patients treated with aspirin during the APLASA study [34]. Studying primary prevention trials with aspirin and the risk of bleeding, aspirin prophylaxis has been proposed only for patients who are at high risk for future thrombosis (e.g., N3% per year, concomitant cardiovascular disease risk factors) [50,51]. Regarding this issue, the 13th aPL task force recommends that asymptomatic aPL carriers should be divided in patients with high risk profile (LA positivity or triple positivity [LA + aCL + anti-β2-GPI] or isolated persistently positive aCL at medium-high titers) and patients with low risk profile (isolated, intermittently positive aCL or anti-β2-GPI at low-medium titers) [9], but we were not able to describe which proportion of high risk versus low risk asymptomatic carriers was studied in the papers included in the meta-analysis. Regarding the hemorrhagic risk of aspirin in SLE, Wahl et al. have performed a numerical simulation showing that prophylactic aspirin could be effective in reducing the number of thrombotic events compared to inducing bleeding episodes [44]. A similar analysis remains to be conducted in obstetrical APS, as the benefit/risk ratio of continuous administration of low dose aspirin remains unknown in this specific population. Finally we would like to balance the conclusion of this meta-analysis, as limited data suggest that intermittent prophylaxis with aspirin or heparin may be effective in some cases and should therefore be formally evaluated. In the APLASA study [34], thrombotic events mostly occurred when high-risk aPL profile patients were exposed to additional risk factors for thrombosis, while receiving low dose aspirin. In a 3-year prospective study [20] of 178 asymptomatic aPL carriers receiving low molecular weight heparin or aspirin only during high risk periods, Giron-Gonzales et al. recorded no episode of thrombosis. Together, these observations advocate the need for a formal evaluation of an intermittent prophylaxis with aspirin or heparin only during high risk periods, in asymptomatic aPL carriers. Taking into account all published studies irrespective of the study design, we did observe a decreased risk for a first thrombotic event among individuals treated with low-dose aspirin. However, no significant risk reduction was observed when considering only prospective studies or those with the best methodological quality. We therefore believe that high-quality randomized controlled trials, stratifying aPL+ patients according to their thrombotic risk, taking into account additional pro-thrombotic and protective confounding factors, and comparing continuous and intermittent prophylaxis during high-risk periods are needed to derive further high-grade evidence-based guidelines. Take-home messages • The use of low dose aspirin to prevent a first thrombosis in patients with antiphospholipid antibodies (aPL) remains controversial. • Our meta-analysis of 11 primary studies including a total of 1208

288

L. Arnaud et al. / Autoimmunity Reviews 13 (2014) 281–291

patients compared the occurrence of a first thrombotic event in aPL patients with versus without low-dose aspirin. • We observed a significant decrease in the risk of arterial but not venous thrombosis in aPL patients treated with low-dose aspirin. • The risk of first thrombotic event was significantly decreased by low dose aspirin among asymptomatic aPL individuals, patients with systemic lupus or obstetrical antiphospholipid syndrome. • No significant risk reduction was observed when considering only prospective studies or those with the best methodological quality.

Acknowledgments None. Competing interests No author has any conflict of interest to disclose. Contributions Laurent ARNAUD, Alexis MATHIAN, Stephane ZUILY, Denis WAHL & Zahir AMOURA designed the research. Amelia RUFFATTI, Doruk ERKAN, Maria TEKTONIDOU, Ricard CERVERA, Ricardo FORASTIERO, Vittorio PENGO, Marc LAMBERT, Maria Angeles MARTINEZ-ZAMORA, Juan BALASCH, Laurent ARNAUD, Alexis MATHIAN, Stephane ZUILY, Denis WAHL & Zahir AMOURA performed the research. Laurent ARNAUD, Alexis MATHIAN, Stephane ZUILY, Denis WAHL & Zahir AMOURA analyzed the data. Laurent ARNAUD, Alexis MATHIAN, Doruk ERKAN, Maria TEKTONIDOU, Denis WAHL & Zahir AMOURA wrote the paper. All authors approved the final manuscript. Funding None.

Appendix A Search strategy Two main investigators (L.A. & A.M.) searched EMBASE (1974– July 2012), MEDLINE (1966–July 2012) and the Cochrane Database of Systematic Reviews (The Cochrane Library, 2012, issue 7) for original articles without language restrictions. Search strategy combined free text search, exploded MESH/EMTREE terms and all synonyms of the following Medical Subject Headings terms: antiphospholipid antibodies, systemic lupus erythematosus, obstetrical antiphospholipid syndrome, lupus-like syndrome, lupus coagulation inhibitor, anticardiolipin antibodies, beta 2-Glycoprotein I, aspirin, and thrombosis. The specific search strategy for PubMed was: ((“Lupus Erythematosus, Systemic”[Mesh]) OR (“Lupus”[all fields]) OR (“Antiphospholipid Syndrome”[Mesh]) OR (“Antiphospholipid Syndrome”[all fields]) OR (“lupus-like”[all fields]) OR (“beta 2-Glycoprotein I”[Mesh]) OR (“Antibodies, Antiphospholipid”[Mesh]) OR (“Antiphospholipid”[all fields])) AND ((“thrombosis”[all fields]) OR (“thrombosis”[Mesh])) AND ((“aspirin”[Mesh]) OR (“aspirin”[all fields])). The specific search strategy for Embase was: (((‘systemic lupus erythematosus’/exp OR ‘systemic lupus erythematosus’) OR (‘antiphospholipid syndrome’/exp OR ‘antiphospholipid syndrome’) OR (‘lupus like syndrome’/exp OR ‘lupus like syndrome’) OR (‘beta2 glycoprotein 1 antibody’/exp OR ‘beta2 glycoprotein 1 antibody’) OR (‘phospholipid antibody’/exp OR ‘phospholipid antibody’)) AND (‘thrombosis’/exp OR ‘thrombosis’) AND ((‘aspirin’/exp OR ‘aspirin’) OR (‘acetylsalicylic acid’/exp OR ‘acetylsalicylic acid’))).

Appendix B Funnel-Plot for all studies included in meta-analysis Visual inspection of the funnel-plot for all studies included in the meta-analysis does not suggest significant publication bias. The dashed line shows the pooled estimate of all included studies.

L. Arnaud et al. / Autoimmunity Reviews 13 (2014) 281–291

Appendix C L'Abbé Plot of all studies included in meta-analysis The observed risk (proportion) for first thrombotic events in the aspirin group is plotted against the observed risk in the control group.

289

Visual inspection of the L'Abbé plot suggests moderate heterogeneity between included studied. Circle size represents study sample size. Studies located to the right of the diagonal line (OR = 1.0) suggest a protective effect of aspirin. The dotted line represents the estimated overall aspirin treatment effect (OR = 0.50).

290

L. Arnaud et al. / Autoimmunity Reviews 13 (2014) 281–291

Appendix D Cumulative meta-analyses according to year of publication (A), study size (B) and methodological quality (C).

L. Arnaud et al. / Autoimmunity Reviews 13 (2014) 281–291

References [1] Staub HL, Dal Ben ER, Bauer ME. The antiphospholipid syndrome and Tregs. Autoimmun Rev 2013 Aug 31. http://dx.doi.org/10.1016/j.autrev.2013.08.004. [2] Mahler M, Norman GL, Meroni PL, et al. Autoantibodies to domain 1 of beta 2 glycoprotein 1: a promising candidate biomarker for risk management in antiphospholipid syndrome. Autoimmun Rev 2012;12:313–7. [3] Ruiz-Irastorza G, Crowther M, Branch W, et al. Antiphospholipid syndrome. Lancet 2010;376:1498–509. [4] Pierangeli SS, Favaloro EJ, Lakos G, et al. Standards and reference materials for the anticardiolipin and anti-beta2glycoprotein I assays: a report of recommendations from the APL task force at the 13th International Congress on antiphospholipid antibodies. Clin Chim Acta 2011;413:358–60. [5] 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. [6] Tozzoli R, Villalta D. Autoantibody profiling of patients with antiphospholipid syndrome using an automated multiplexed immunoassay system. Autoimmun Rev 2014 Jan 13;13(1):59–63. [7] Pengo V, Banzato A, Denas G, et al. Correct laboratory approach to APS diagnosis and monitoring. Autoimmun Rev 2013;12:832–4. [8] Roggenbuck D, Egerer K, von Landenberg P, et al. Antiphospholipid antibody profiling: time for a new technical approach? Autoimmun Rev 2012;11:821–6. [9] Ruiz-Irastorza G, Cuadrado MJ, Ruiz-Arruza I, et al. Evidence-based recommendations for the prevention and long-term management of thrombosis in antiphospholipid antibody-positive patients: report of a task force at the 13th International Congress on antiphospholipid antibodies. Lupus 2011;20: 206–18. [10] Staub HL, Bertolaccini ML, Khamashta MA. Anti-phosphatidylethanolamine antibody, thromboembolic events and the antiphospholipid syndrome. Autoimmun Rev 2012;12:230–4. [11] Galarza-Maldonado C, Kourilovitch MR, Perez-Fernandez OM, et al. Obstetric antiphospholipid syndrome. Autoimmun Rev 2012;11:288–95. [12] Pengo V, Ruiz-Irastorza G, Denas G, et al. High intensity anticoagulation in the prevention of the recurrence of arterial thrombosis in antiphospholipid syndrome: ‘PROS’ and ‘CONS’. Autoimmun Rev 2012;11:577–80. [13] Ceccarelli F, Chighizola C, Finazzi G, et al. Thromboprophylaxis in carriers of antiphospholipid antibodies (APL) without previous thrombosis: “pros” and “cons”. Autoimmun Rev 2012;11:568–71. [14] Berger JS, Krantz MJ, Kittelson JM, et al. Aspirin for the prevention of cardiovascular events in patients with peripheral artery disease: a meta-analysis of randomized trials. JAMA 2009;301:1909–19. [15] Becattini C, Agnelli G, Schenone A, et al. Aspirin for preventing the recurrence of venous thromboembolism. N Engl J Med 2012;366:1959–67. [16] Stroup DF, Berlin JA, Morton SC, et al. Meta-analysis of observational studies in epidemiology: a proposal for reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA 2000;283:2008–12. [17] Forastiero R, Martinuzzo M, Pombo G, et al. A prospective study of antibodies to beta2-glycoprotein I and prothrombin, and risk of thrombosis. J Thromb Haemost 2005;3:1231–8. [18] Hereng T, Lambert M, Hachulla E, et al. Influence of aspirin on the clinical outcomes of 103 anti-phospholipid antibodies-positive patients. Lupus 2008;17:11–5. [19] Cervera R, Khamashta MA, Shoenfeld Y, et al. Morbidity and mortality in the antiphospholipid syndrome during a 5-year period: a multicentre prospective study of 1000 patients. Ann Rheum Dis 2009;68:1428–32. [20] Giron-Gonzalez JA, Garcia del Rio E, Rodriguez C, et al. Antiphospholipid syndrome and asymptomatic carriers of antiphospholipid antibody: prospective analysis of 404 individuals. J Rheumatol 2004;31:1560–7. [21] Brandt JT, Triplett DA, Alving B, et al. Criteria for the diagnosis of lupus anticoagulants: an update. On behalf of the Subcommittee on Lupus Anticoagulant/ Antiphospholipid Antibody of the Scientific and Standardisation Committee of the ISTH. Thromb Haemost 1995;74:1185–90. [22] Pengo V, Tripodi A, Reber G, et al. Update of the guidelines for lupus anticoagulant detection. Subcommittee on Lupus Anticoagulant/Antiphospholipid Antibody of the Scientific and Standardisation Committee of the International Society on Thrombosis and Haemostasis. J Thromb Haemost 2009;7:1737–40. [23] 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. [24] Hochberg MC. Updating the American College of Rheumatology revised criteria for the classification of systemic lupus erythematosus. Arthritis Rheum 1997; 40:1725. [25] Ruffatti A, Del Ross T, Ciprian M, et al. Risk factors for a first thrombotic event in antiphospholipid antibody carriers: a prospective multicentre follow-up study. Ann Rheum Dis 2011;70:1083–6.

291

[26] Ruffatti A, Del Ross T, Ciprian M, et al. Risk factors for a first thrombotic event in antiphospholipid antibody carriers. A multicentre, retrospective follow-up study. Ann Rheum Dis 2009;68:397–9. [27] Tarr T, Lakos G, Bhattoa HP, et al. Analysis of risk factors for the development of thrombotic complications in antiphospholipid antibody positive lupus patients. Lupus 2007;16:39–45. [28] Tarr T, Lakos G, Bhattoa HP, et al. Clinical thrombotic manifestations in SLE patients with and without antiphospholipid antibodies: a 5-year follow-up. Clin Rev Allergy Immunol 2007;32:131–7. [29] Pengo V, Ruffatti A, Legnani C, et al. Incidence of a first thromboembolic event in asymptomatic carriers of high-risk antiphospholipid antibody profile: a multicenter prospective study. Blood 2011;118:4714–8. [30] Wahl DG, Guillemin F, de Maistre E, et al. Risk for venous thrombosis related to antiphospholipid antibodies in systemic lupus erythematosus—a meta-analysis. Lupus 1997;6:467–73. [31] Wahl DG, Guillemin F, de Maistre E, et al. Meta-analysis of the risk of venous thrombosis in individuals with antiphospholipid antibodies without underlying autoimmune disease or previous thrombosis. Lupus 1998;7:15–22. [32] Zuily S, Regnault V, Selton-Suty C, et al. Increased risk for heart valve disease associated with antiphospholipid antibodies in patients with systemic lupus erythematosus: meta-analysis of echocardiographic studies. Circulation 2012;124:215–24. [33] Egger M, Davey Smith G, Schneider M, et al. Bias in meta-analysis detected by a simple, graphical test. BMJ 1997;315:629–34. [34] Erkan D, Harrison MJ, Levy R, et al. Aspirin for primary thrombosis prevention in the antiphospholipid syndrome: a randomized, double-blind, placebo-controlled trial in asymptomatic antiphospholipid antibody-positive individuals. Arthritis Rheum 2007;56:2382–91. [35] Erkan D, Merrill JT, Yazici Y, et al. High thrombosis rate after fetal loss in antiphospholipid syndrome: effective prophylaxis with aspirin. Arthritis Rheum 2001;44:1466–7. [36] Martinez-Zamora MA, Cervera R, Balasch J. Recurrent miscarriage, antiphospholipid antibodies and the risk of thromboembolic disease. Clin Rev Allergy Immunol 2012;43(3):265–74. [37] Tektonidou MG, Laskari K, Panagiotakos DB, et al. 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. [38] Erkan D, Yazici Y, Peterson MG, et al. A cross-sectional study of clinical thrombotic risk factors and preventive treatments in antiphospholipid syndrome. Rheumatology (Oxford) 2002;41:924–9. [39] Ginsburg KS, Liang MH, Newcomer L, et al. Anticardiolipin antibodies and the risk for ischemic stroke and venous thrombosis. Ann Intern Med 1992;117:997–1002. [40] Kaiser R, Cleveland CM, Criswell LA. Risk and protective factors for thrombosis in systemic lupus erythematosus: results from a large, multi-ethnic cohort. Ann Rheum Dis 2009;68:238–41. [41] Mok MY, Chan EY, Fong DY, et al. Antiphospholipid antibody profiles and their clinical associations in Chinese patients with systemic lupus erythematosus. J Rheumatol 2005;32:622–8. [42] Shah NM, Khamashta MA, Atsumi T, et al. Outcome of patients with anticardiolipin antibodies: a 10 year follow-up of 52 patients. Lupus 1998;7:3–6. [43] Tincani A, Taglietti M, Biasini C, et al. Thromboembolic events after fetal loss in patients with antiphospholipid syndrome: comment on the article by Erkan et al. Arthritis Rheum 2002;46:1126–7 [author reply 1127]. [44] Wahl DG, Bounameaux H, de Moerloose P, et al. Prophylactic antithrombotic therapy for patients with systemic lupus erythematosus with or without antiphospholipid antibodies: do the benefits outweigh the risks? A decision analysis. Arch Intern Med 2000;160:2042–8. [45] Reiner P, Galanaud D, Leroux G, et al. Long-term outcome of 32 patients with chorea and systemic lupus erythematosus or antiphospholipid antibodies. Mov Disord 2011;26:2422–7. [46] Rodrigues CE, Bonfa E, Caleiro MT, et al. Association of arterial events with the coexistence of metabolic syndrome and primary antiphospholipid syndrome. Arthritis Care Res (Hoboken) 2012;64:1576–83. [47] 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. [48] Ruiz-Irastorza G, Ramos-Casals M, Brito-Zeron P, et al. Clinical efficacy and side effects of antimalarials in systemic lupus erythematosus: a systematic review. Ann Rheum Dis 2008;69:20–8. [49] Seshasai SR, Wijesuriya S, Sivakumaran R, et al. Effect of aspirin on vascular and nonvascular outcomes: meta-analysis of randomized controlled trials. Arch Intern Med 2012;172:209–16. [50] Finazzi G. Aspirin in asymptomatic patients with confirmed positivity of antiphospholipid antibodies? No. Intern Emerg Med 2008;3:197–200. [51] Barbhaiya M, Erkan D. Primary thrombosis prophylaxis in antiphospholipid antibodypositive patients: where do we stand? Curr Rheumatol Rep 2011;13:59–69.

Efficacy of aspirin for the primary prevention of thrombosis in patients with antiphospholipid antibodies: an international and collaborative meta-analysis.

We performed a meta-analysis to determine whether aspirin has a significant protective effect on risk of first thrombosis among patients with antiphos...
2MB Sizes 0 Downloads 0 Views