Dr. Hunt served as a consultant for Novartis, and received speaking fees (less than $10,000 each) from Sanofi Aventis and Novo-Nordisk.
A systematic review of secondary thromboprophylaxis in patients with antiphospholipid antibodies
Article first published online: 29 NOV 2007
Copyright © 2007 by the American College of Rheumatology
Arthritis Care & Research
Volume 57, Issue 8, pages 1487–1495, 15 December 2007
How to Cite
Ruiz-Irastorza, G., Hunt, B. J. and Khamashta, M. A. (2007), A systematic review of secondary thromboprophylaxis in patients with antiphospholipid antibodies. Arthritis & Rheumatism, 57: 1487–1495. doi: 10.1002/art.23109
- Issue published online: 29 NOV 2007
- Article first published online: 29 NOV 2007
- Manuscript Accepted: 2 APR 2007
- Manuscript Received: 7 FEB 2007
- Venous thromboembolism;
- Antiphospholipid syndrome;
- Lupus anticoagulant
To systematically review the efficacy and safety data of different therapeutic approaches in patients with antiphospholipid antibodies (aPL) and thrombosis.
The Medline database and references from selected reports and review articles were used. Randomized controlled trials, prospective and retrospective cohort studies, and subgroup analysis (n > 15) that focused on the secondary thromboprophylaxis in patients with aPL were selected.
Sixteen studies were selected. Patients with venous events and a single test for aPL showed a low recurrence rate while receiving oral anticoagulation at a target international normalized ratio (INR) of 2.0–3.0. Patients with stroke and a single positive aPL test had no increased risk compared with those without aPL. Recurrence rates in patients with definite antiphospholipid syndrome (APS) and previous venous thromboembolism were lower than in patients with arterial and/or recurrent events, both with and without therapy. Only 3.8% of recurrent events occurred at an actual INR >3.0. Mortality due to recurrent thrombosis was higher than mortality due to bleeding (18 patients versus 1 patient reported).
For patients with definite APS, we recommend prolonged warfarin therapy at a target INR of 2.0–3.0 in APS patients with first venous events and >3.0 for those with recurrent and/or arterial events. For patients with venous thromboembolism or stroke and a single positive aPL test, we recommend further testing to determine if they have a persisting antibody. If they do not, the same therapy as for the general population should be used (warfarin at a target INR of 2.0–3.0 and low-dose aspirin, respectively).
Thrombosis in persons with antiphospholipid syndrome (APS) can occur in any vascular bed (arterial, microvascular, or venous), unlike genetic thrombophilias, which mainly cause venous thromboembolism (1). Deep vein thrombosis and stroke are the major causes of morbidity and mortality among individuals with APS (2). There are different patterns of disease between patients, but it is recognized that recurrent events tend to occur in the same vascular bed (3).
Long-term anticoagulation is widely used for secondary thromboprophylaxis due to the high risk of recurrent events after withdrawing treatment (4–8). However, the question of the intensity of oral anticoagulation has become an issue of great debate over the last 10 years. Consensus documents reflect the high degree of uncertainty (6, 7).
A recently published systematic review that focuses on the management of APS (8) includes grade-A recommendations for the long-term treatment of venous thrombosis and stroke in patients positive for antiphospholipid antibodies (aPL). However, due to stringent selection criteria (randomized controlled trials only), the authors analyzed data from a limited number of studies that, despite potentially optimal design, had limitations regarding the type of patients recruited (9). Therefore, we reviewed published data on the secondary prophylaxis of thrombosis in patients with aPL, expanding the focus to observational studies covering a wider range of patients.
MATERIALS AND METHODS
The literature search was performed using Medline on the PubMed Web page. The Web page was last accessed on October 14, 2006. Two complementary search strategies, entering the medical subject headings (MeSH) “antiphospholipid syndrome” and “antibodies, antiphospholipid,” were used to ensure the identification of articles in which APS criteria were not necessarily met: search strategy A: (“antiphospholipid syndrome/drug therapy” [MeSH] OR “antiphospholipid syndrome/therapy” [MeSH]) AND “thrombosis” [MeSH]; search strategy B: “antibodies, antiphospholipid” [MeSH] AND (“thrombosis/drug therapy” [MeSH] OR “thrombosis/therapy” [MeSH]). Additional articles were obtained by checking references from the selected studies, as well as from review articles and other sources known by the authors.
All patients enrolled in the selected studies had to have experienced at least 1 thrombotic event previous to inclusion in the study. The following types of studies were selected: randomized controlled trials (RCTs), observational prospective and retrospective cohort studies, and subgroup analysis of RCTs and of cohort studies. All patients, whether they were receiving therapy or not, were analyzed. Case reports, case series and small cohort studies (n < 15), and abstracts to medical meetings not published as full reports were excluded.
The following data were extracted or calculated from the selected sources by one of the authors (GR-I): type of study, number of patients, fulfillment of criteria for definite APS, type of initial thrombosis, treatment received, duration of followup, number/rate and type of recurrent thromboses and bleeding episodes, and international normalized ratio (INR) at the time of recurrent thrombosis and bleeding. Definite APS was defined according to the Sapporo criteria, i.e., persistent aPL, defined as repeated positivity for medium to high titers of anticardiolipin antibodies (aCL) and/or lupus anticoagulant (LAC) with assays more than 6 weeks apart (10).
Recommendations for therapy in the different clinical settings have been proposed. To grade the strength of the recommendations, the American College of Chest Physicians (ACCP) grading categories for antithrombotic and thrombolytic therapy have been used (11).
Search strategy A resulted in 279 articles, and search strategy B resulted in 234 articles (Figure 1). Seven articles were selected from search A (12–18) and 8 from search B, 5 of which were not identified by search A (19–23). A manual search found 4 additional articles (24–27), resulting in a total of 16 articles, which are summarized in Table 1. The different studies have been grouped under 3 categories: cohort studies, subgroup analysis, and RCTs.
|Author, year (ref.)||No.||Type of study||Sapporo criteria for aPL||Thrombotic events at diagnosis, arterial/venous|
|Rosove, 1992 (12)||70||Retrospective cohort||No||31/39|
|Derksen, 1993 (24)||19||Retrospective cohort||Yes||0/19|
|Khamashta, 1995 (13)||147||Retrospective cohort||Yes||67/80|
|Krnic-Barrie, 1997 (25)||61||Retrospective cohort||No||38/23|
|Muñoz-Rodriguez, 1999 (14)||47||Retrospective cohort||Yes||19/28|
|Ruiz-Irastorza, 2002 (15)||66||Retrospective cohort||Yes||51/32|
|Wittkowsky, 2006 (27)||36||Retrospective cohort||Yes||14/16†|
|Girón-Gonzàlez, 2004 (23)||158||Prospective cohort||Yes||70/106‡|
|Ames, 2005 (17)||67||Prospective cohort||Yes||17/50|
|Ginsberg, 1995 (19)||16||Prospective cohort subgroup analysis||No||0/16|
|Prandoni, 1996 (20)||15||Retrospective cohort subgroup analysis||Yes||0/15|
|Rance, 1997 (26)||27||Retrospective cohort subgroup analysis||No||0/27|
|Schulman, 1998 (21)||68||RCT subgroup analysis||No||0/68|
|Levine, 2004 (22)||720||RCT subgroup analysis||No||720/0|
|Crowther, 2003 (16)||114||RCT||Yes||27/87|
|Finazzi, 2005 (18)||109||RCT||Yes||44/75|
Nine cohort studies were included in this review, 7 of them retrospective (12–15, 24, 25, 27) and 2 prospective (17, 23). All but 2 cohort studies included patients who fulfilled laboratory criteria for definite APS (Table 1). This was not a requisite for entering the study by Rosove and Brewer (12), as 17% of patients had low titers of aCL. The study by Krnic-Barrie et al (25) only included patients with medium-high titers of aCL and/or LAC, but repeated determinations were not obtained.
Patients in the cohort studies had experienced both arterial and venous thrombotic events with similar frequency, except for patients in the studies by Derksen et al (24) and Ames et al (17), in which 100% and 75% of patients, respectively, had experienced venous events only.
Recurrence rates among untreated patients were high in the cohort studies, ranging from 19% to 29% per year (Table 2). The only apparent exception was the study by Ames et al (17), in which no patient treated with warfarin at an INR <2.0 had recurrent events. However, this study was weakened by the short duration of followup in this subgroup (median 9 weeks).
|Author, year (ref.)||Treatment||Followup||Thrombosis rates||Bleeding rates (M/m)|
|Rosove, 1992 (12)||None||161.2 pt-yrs||0.19/pt-yr||0|
|Low-dose aspirin||37.8 pt-yrs||0.32/pt-yr||0|
|Warfarin INR <2.0||11.3 pt-yrs||0.57/pt-yr|
|Warfarin INR 2.0–2.9||40.9 pt-yrs||0.07/pt-yr||0.031/pt-yr (whole warfarin group)|
|Warfarin INR ≥3.0||110.2 pt-yrs||0|
|Derksen, 1993 (24)||None||8–248 months (whole group)||NA||NA (2 patients, 2M)|
|Warfarin INR 2.5–4.0|
|Khamashta, 1995 (13)||None||280.6 pt-yrs||0.29/pt-yr|
|Low-dose aspirin||240.3 pt-yrs||0.18/pt-yr|
|Warfarin INR <3.0||141.3 pt-yrs||0.23/pt-yr||0.071 (all warfarin, M+m)|
|Warfarin INR ≥3.0||197.3 pt-yrs||0.015/pt-yr||0.017 (all warfarin, M)|
|Krnic-Barrie, 1997 (25)||None||124.9 pt-yrs||0.192/pt-yr (A) 0.11/pt-yr (V)||NA (4 patients)|
|Low-dose aspirin||36.6 pt-yrs||0.082/pt-yr (A) 0.027/pt-yr (V)|
|Warfarin||63.0 pt-yrs||0.048/pt-yr (A) 0 (V)|
|Warfarin + low-dose aspirin||30.6 pt-yrs||0 (A) 0 (V)|
|Muñoz-Rodriguez, 1999 (14)||None Low-dose aspirin||Median 49 months (4–50)||91% 41%||NA (4 patients, 4 M)|
|Warfarin INR 2.5–3.5||19%|
|Ruiz-Irastorza, 2002 (15)||Warfarin INR 3.0–4.0||66 pt-yrs||0.09/pt-yr||0.06/pt-yr (M)|
|Wittkowsky, 2006 (27)||Warfarin INR 2.0–3.0 (n = 10)||62.5 pt-yrs (whole group)||0.096/pt-yr (whole group)||0.032/pt-yr (M) (whole group)|
|Warfarin INR >3.0 (n = 26)|
|Girón-Gonzàlez, 2004 (23)||Warfarin INR 2.5–3.5||624 pt-yrs||0.005/pt-yr||0.006/pt-yr (M)|
|Ames, 2005 (17)||Warfarin INR <2.0||Median 9 weeks||0||0 (M) 0 (m)|
|Warfarin INR 2.0–3.0||Median 122 weeks||0.04/pt-yr||0.0057/pt-yr (M) 0.109/pt-yr (m)|
|Warfarin INR 3.1–4.0||Median 9 weeks||0.1/pt-yr||0.10/pt-yr (M) 1.42/pt-yr (m)|
|Warfarin INR >4.0||Median 5 weeks||0||0 ? (m)|
|Ginsberg, 1995 (19)||None||Mean 8.7 months||18%||NA|
|Warfarin INR 2.0–3.0||Median 3 months (3 weeks to 6 months)||0|
|Prandoni, 1996 (20)||None||Median 8 yrs (1–10) (whole group)||0.038/pt-yr||NA (3 patients, 3m)|
|Warfarin INR 2.0–3.0||0|
|Rance, 1997 (26)||None||1–4 yrs||NA (4 patients)||NA|
|Warfarin INR 2.0–3.0||0|
|Schulman, 1998 (21)||None||4 yrs (whole group)||0.1/pt-yr||NA|
|Warfarin INR 2.0–2.85||0|
|Levine, 2004 (22)||Low-dose aspirin||2 yrs||22.18%||NA|
|Warfarin INR 1.4–2.8||26.15%|
|Crowther, 2003 (16)||Warfarin INR 2.0–3.0||Mean 2.7 yrs||0.013/pt-yr||0.03/pt-yr (M) 0.14/pt-yr (m)|
|Warfarin INR 3.1–4.0||0.032/pt-yr||0.027/pt-yr (M) 0.087/pt-yr (m)|
|Finazzi, 2005 (18)||Warfarin INR 2.0–3.0||Mean 3.3 yrs||0.016/pt-yr||0.016/pt-yr (M) 0.033 pt-yr (m)|
|Warfarin INR 3.0–4.5||Mean 3.5 yrs||0.031/pt-yr||0.010/pt-yr (M) 0.079 pt-yr (m)|
Another consistent finding in the cohort studies was a clear dose effect of oral anticoagulation, with fewer recurrent thrombotic events among patients treated with high-intensity anticoagulation (INR of 3.0–4.0) as compared with those with a target INR of 2.0–3.0 (Table 2). Again, the exception was the study by Ames et al (17). In this cohort, the risk for recurrent events was highest among patients receiving high-intensity oral anticoagulation. This was the result of 2 recurrent events during a short median followup of 9 weeks.
In summary, cohort studies, mainly retrospective, included patients with definite or probable APS with both arterial and venous thromboses. The rate of recurrent events was highest among untreated patients and lowest among those given warfarin at an INR ≥3.0 as compared with those receiving lower-intensity warfarin or aspirin. Patients with arterial events were at a higher risk of recurrences.
Five studies were classified as subgroup analyses (Table 1). These studies comprised 3 subgroup analyses of cohort studies (2 retrospective [20, 26] and 1 prospective ) and 2 subgroup analyses of RCTs (21, 22). Two studies were published as letters to the editor (20, 26). All of these studies were designed to investigate the relationship between aPL positivity and recurrent thrombotic events and were not focused on therapy. Four studies reported the outcome of patients receiving warfarin at an INR of 2.0–3.0 and those receiving no treatment (19–21, 26) and 1 study compared warfarin at an INR of 1.4–2.8 with aspirin (22).
Only 1 study in this group recruited patients fulfilling laboratory criteria for definite APS (20). In the remaining articles, only single aPL tests were performed. Four studies included patients presenting with venous thromboembolism only (19–21, 26), most of whom had experienced 1 event. The frequencies of recurrent events seen in both treated and untreated patients (Table 2) were lower than those found in cohort studies.
The most recent article, the Antiphospholipid Antibodies and Stroke Study (APASS), focused on patients with cerebral arterial events (22), and was a subgroup analysis of the Warfarin versus Aspirin Stroke Study, an RCT comparing low-dose warfarin with aspirin as secondary prophylaxis of stroke in the general population (28).
Patients were labeled as aPL positive, despite only a single test, if they had aCL (including IgA isotype) at any level. Moreover, low titers of aCL were included and LAC testing was not performed according to international recommendations. Therefore, not surprisingly 41% of the 1,770 individuals tested were aPL positive. However, only 6.7% of patients were positive for both aCL and LAC, and aCL at high titers was present in only 0.2%.
The APASS study did not find any association between the presence of one positive aPL test and the risk of recurrent events (22). Furthermore, the subgroup of patients positive for aPL did not show a different risk of recurrent thrombosis whether treated with aspirin or warfarin (median INR 1.9). The rate of recurrent events was lower than in the cohort studies (see Table 2).
In summary, subgroup analyses included patients with either venous or arterial-only (mostly first) events, most of them not fulfilling criteria for the classification of definite APS. The rate of recurrent events in both treated and untreated patients was lower than that reported in the cohort studies.
Randomized controlled trials.
Two RCTs have been recently published (16, 18), both comparing standard anticoagulant treatment (target INR of 2.5) with high-intensity anticoagulant treatment (target INR of 3.5) for the secondary prophylaxis of thrombosis in APS (Table 1).
Both the Canadian study (16) and the European study (18) required persistently positive aCL at medium-high levels and/or LAC, thus fulfilling classification criteria for APS. Patients with venous thromboembolism represented 76% and 63%, respectively, of the individuals enrolled. Patients with recent stroke were excluded from the Canadian trial. Also, in both studies, patients with recurrent thrombosis while receiving oral anticoagulant treatment were excluded.
Both studies demonstrated no advantage of high-intensity anticoagulation (target INR of 3.0–4.0) versus standard anticoagulation (target INR of 2.0–3.0) in terms of preventing recurrent events. In fact, a combined analysis of both trials performed by Finazzi et al (18) showed a borderline significant increase of recurrent thrombosis among patients in the high-intensity arm (odds ratio 2.49, 95% confidence interval 0.93–6.67, P = 0.07). However, overall the event rates were much lower than expected, and lower than those seen in cohort studies (see Table 2).
It is also important to note that patients randomized to high-intensity anticoagulation frequently failed to achieve the target INR. In the Canadian study, this group was below the therapeutic range 43% of the time (16). Finazzi et al did not report on this issue (18); however, the mean INR in the high-intensity group was only 3.2 (implying that an important number of measurements were below the threshold of 3.0) versus 2.5 in the standard-intensity group.
In summary, RCTs recruited patients fulfilling APS criteria, mostly with first venous events and with a low incidence of recurrences. No advantage of high- versus standard-intensity anticoagulation was demonstrated, although patients in the INR >3.0 groups were frequently below the intended intensity of anticoagulation.
Patient INRs at the time of recurrent thrombosis.
Ten studies reported the INR measured at the time of recurrent thrombotic events (12, 14–17, 20, 23–25, 27). In 3 additional studies thromboses only occurred among patients who were not receiving anticoagulant treatment (19, 21, 26) (Table 3). The authors were blinded to the treatment received at the time of evaluating bleeding in only 1 RCT (16).
|Author, year (ref.)||INR at thrombosis/type of thrombosis (A/V)||INR at bleeding/type of bleeding (M/m)|
|Rosove, 1992 (12)||1.42, 1.57, 1.65, 1.72, 1.88, 1.94, 2.22, 2.33, 2.60 (5A/4V); low-dose aspirin (n = 12, 10A/2V); no treatment (n = 31, 14A, 17V)||26 (M), 6.4 (M), ? (M)|
|Derksen, 1993 (24)||3.6 (A), 2.6 (A), 2.2 (V); no treatment (n = 15, 1A, 14V)||7.5 (M), 7.3 (M)|
|Khamashta, 1995 (13)||NA||≥3.0 (n = 29, 7M, 22m)|
|Krnic-Barrie, 1997 (25)||<2.1 (n = 3), 2.2–3.8 (n = 2)||>3.5 (n = 2), <3.5 (n = 2)|
|Muñoz-Rodriguez, 1999 (14)||1.8 (A), 1.8 (V), 2.46 (V), 2.5–3.5 (n = 2, 2A); low-dose aspirin (n = 15); no treatment (n = 23)||4.32 (M)|
|Ruiz-Irastorza, 2002 (15)||2.1 (V), 2.3 (A), 2.5 (A), 2.5 (V), 2.6 (A), ? (A)||>20 (M), 5.2 (M), 2.5 (M), 1.9 (M)|
|Wittkowsky, 2006 (27)||1.4, 1.8, 2.4, 2.8, 3.8, 5.2 (3A, 3V)||2.5 (M), 2.6 (M)|
|Girón-Gonzàlez, 2004 (23)||1.5–2.4 (n = 3, 3V)||3.0–3.6 (n = 4, 4M)|
|Ames, 2005 (17)||2.5 (V), 2.5 (V), 2.7 (V), 2.8 (A), 2.8 (V), 2.9 (A, V), 3.8 (A), 3.9 (A)||2.0–3.0 (n = 20, 1M, 19m), 3.0–4.0 (n = 29, 2M, 27m)|
|Ginsberg, 1995 (19)||No treatment (n = 2, 2V); no thrombosis seen on warfarin||NA|
|Prandoni, 1996 (20)||2.0–3.0 (n = 2, 1V, 1A); no treatment (n = 9, 1A, 8V)||NA|
|Rance, 1997 (26)||No treatment (n = 4, 4V); no thrombosis seen on warfarin||NA|
|Schulman, 1998 (21)||No treatment (n = 20)||NA|
|Levine, 2004 (22)||NA||NA|
|Crowther, 2003 (16)||0.9 (V), 1.0 (A), 1.6 (A), 1.9 (V), 2.8 (V), 3.1 (V), 3.9 (A); off warfarin (n = 1, 1 V)||NA|
|Finazzi, 2005 (18)||NA||NA|
Of the 180 events reported, 104 (57%) occurred when patients were not taking any anticoagulant or antiaggregant drug. An additional 27 events (15%) occurred among patients treated with aspirin only; interestingly, most of these events were arterial. The remaining 49 recurrences (27%) were seen in patients treated with warfarin. Within this group, the actual INR at the time of the event was <3.0 in 42 cases (86%) (Table 4).
|Thrombotic recurrences†||Hemorrhagic complications|
|INR < 3||42 (13A, 16V)||23||INR < 3||24 (5M, 19m)||26|
|INR > 3||7 (4A, 1V)||3.8||INR > 3||69 (20M, 49m)||74|
|Low-dose aspirin||27 (10A, 2V)||15|
|No treatment||104 (16A, 42V)||57|
Data were consistent throughout all the studies in which they were reported. This was also true for the RCT by Crowther et al (16): 6 of 8 events occurred while the INR was <3.0 and only 2 of the 6 thromboses seen in patients randomized to high-intensity warfarin occurred while the actual INR was 3.0–4.0 (Table 3).
In conclusion, the data show that most recurrent events took place in patients not receiving any anticoagulant or antiplatelet treatment alone. One of every 4 thrombotic events occurred in patients receiving oral anticoagulation, whereas the proportion of events occurring while the INR was ≥3.0 was low at 14% of events in patients treated with anticoagulants.
Bleeding rates were reported in 8 studies (12, 13, 15–18, 23, 27) (Table 2). Major bleeding rates ranged from 0.57% to 10% per year. The relationship between the intensity of warfarin therapy and bleeding was not investigated by most authors. The RCTs by Crowther et al (16) and Finazzi et al (18) did not obtain differences in the rates of major bleeding between high- and standard-intensity groups (limitations regarding the actual intensity achieved in the groups with INRs of 3.1–4.0 have already been discussed). Ames et al (17) found an increased risk of major bleeding in patients with INRs >3.0. In contrast, Ruiz-Irastorza et al (15) found rates of intracranial and fatal hemorrhages among patients with target INRs of 3.0–4.0 similar to those seen in patients with conditions other than APS treated with lower-intensity anticoagulation. However, the actual level of anticoagulation in this cohort was also lower than desired.
Seven studies reported the INR measured at the time of bleeding (12, 14, 15, 17, 23, 25, 27). Of 93 episodes, 24 (26%) occurred with INRs <3.0, whereas the remaining 69 (74%) occurred with INRs ≥3.0; the proportion of major bleeding within both groups was similar: 20% versus 29% (Table 4).
In summary, bleeding rates in patients with APS treated with oral anticoagulants varied widely. Three of every 4 episodes in which the INR was documented at the time of bleeding occurred with INRs ≥3.0; however, the proportion of major/minor bleeding was similar at standard- or high-intensity anticoagulation.
Mortality due to thrombosis and bleeding.
Four studies reported a total of 18 deaths directly related to recurrent thromboses: 12 arterial, 5 venous, and 1 multiple (12, 13, 17, 21). In addition, 10 patients in the cohort by Giron-Gonzalez et al died as a consequence of the presenting thrombotic event (23). In contrast, only 1 patient died due to bleeding (23).
APS can be a complex condition with a wide spectrum of clinical manifestations. Thrombotic complications are the most common cause of death and serious morbidity in APS (2). Among these, stroke is particularly important due to the high mortality and frequent permanent disability. Studies of patients with systemic lupus erythematosus have demonstrated that definite secondary APS is a very serious condition with prognostic implications (29). Primary prevention is obviously a clinical goal; however, many patients, especially those with primary APS, present with a thrombotic event, so that secondary prophylaxis is the only possible option.
Transient aPL, usually low-level aCL, can be found in a number of conditions such as infections and connective tissue diseases, and even in normal individuals without known pathologic significance (30). Those few studies comparing single with repeated aPL tests have demonstrated that the relationship with thrombosis is stronger when aPL are repeatedly positive (31–33). Therefore, persistently positive aPL at significant titers is a necessity for the diagnosis of APS, both from the classification and from the clinical viewpoint (10, 34). Unfortunately, many studies focusing on the therapeutic aspects of APS have included patients with only 1 test for aPL, usually aCL. As a result, as many as 6 of the 16 studies included in this review did not fulfill criteria for definite APS (Table 1), in all cases due to inappropriate laboratory criteria.
Hemorrhagic events are the most frequent complication of anticoagulant therapy. Different studies have shown a large range of variation in the actual frequency of major bleeding among patients receiving long-term oral anticoagulation (35). Intensity of oral anticoagulation and age of the patient are the 2 factors most consistently cited as increasing the risk of bleeding (35, 36). However, the consensus document of the ACCP makes it clear that in the risk-benefit analysis of deciding to treat with oral anticoagulation, the bleeding risk must always be taken in the context of the potential decrease of thromboembolic events (35).
Many uncertainties surround the decision of starting anticoagulation in a patient with APS and thrombosis. The most critical message of this systematic review is that important limitations affect all the studies that focused on the secondary thromboprophylaxis in APS. Therefore, it seems clear that any recommendations in this area must be taken with caution and put into the specific clinical context.
Limitations are much related to the type of study. The main biases of cohort studies are nonrandom allocation of treatment and nonblinded interpretation of outcomes. As a consequence, patients with different clinical severity could have been unevenly distributed among groups. Moreover, authors aware of the treatment received by a given patient might have interpreted confusing outcomes in one or another sense. In addition, retrospective studies (7 of 9 in this group; see Table 1) are at risk for missing information and patient recall bias. Also, all of the cohort studies except for the study by Ames at al (17) classified patients according to intention to treat rather than actual achieved INR. Lastly, arterial and venous events were only analyzed separately in the study by Krnic-Barrie et al (25). However, despite these potential flaws, the cohort studies included patients who, in most cases, fulfilled APS criteria and were unselected in terms of severity, and included a good mix of patients with venous and arterial thromboses.
On the contrary, all but one of the subgroup analyses (20) studied patients who did not achieve APS criteria due to the lack of repeated tests for aPL. In fact, none of these studies was designed to investigate the effects of different therapies. Four of the subgroup analyses only recruited patients with venous thromboembolism (19–21, 26) (Table 1), most of them with first events. Low rates of recurrence were seen in these cohorts.
Within this group of studies, the APASS merits special comment. It was the only study that actually focused on arterial cerebral events, and it was large and prospective. Due to these unique characteristics, the APASS was the only basis for the grade-A recommendation of either aspirin or warfarin at an INR of 1.4–2.8 in patients with first stroke and positive aPL given in the systematic review by Lim et al (8). Unfortunately, the definition criteria for aPL positivity in this study were inappropriate, resulting in a low specificity, so that 41% of positive patients were found in an unselected population with age at stroke averaging 63 years. High positive aCL and concomitant detection of aCL and LAC together were infrequent, suggesting that most aPL detected were transient findings without any clinical significance. Thus, this study was performed in a population whose individuals, in a great majority, did not have APS. In fact, the authors themselves acknowledged that their results could not be generalized to younger patients or those with other manifestations of APS (22). Consequently, no results of the APASS should be invoked to make any recommendations for patients with definite APS and stroke (37).
The last group of studies comprises the articles that had the optimal design of RCTs, including intent-to-treat analysis (16, 18). However, both studies were also subject to limitations. Patients with arterial events were a minority in both studies (Table 1). In fact, due to stringent exclusion criteria, a substantial number of high-risk patients were not enrolled (see previous section). As a consequence, the event rates were much lower than expected. In addition, the study by Finazzi et al was terminated early, well below achieving the estimated sample size, due to difficulties in enrolling patients (18). All of the above resulted in a significant increase in beta error. The most relevant drawback of both trials was the lack of achievement of target INRs in patients randomized to a target >3.0. Thus the intent-to-treat analysis can be misleading; in fact, most patients with recurrent events had an INR <3.0 at the time of the recurrence, whatever the therapeutic arm (16).
A valid criticism of our review is that it is also at risk for a number of biases, due to the heterogeneity of the selected studies in terms of populations, end points, therapeutic schemes, presentation of data, and, especially, regarding the diversity of results. Such a diversity of results precluded any statistical analysis of pooled data. In addition, a reporting bias is possible because not all of the studies contained data for every area of analysis. On the positive side, however, it seems clear that differences among studies were more related to the type of patients than to the type of study itself. A number of results were consistent enough throughout several studies, both within and without subgroups.
The following conclusions can be drawn. First, the absolute risk of recurrent events while receiving oral anticoagulation in patients with venous thrombosis and aPL who do not fulfill laboratory criteria for definite APS (10) is low. This finding is supported by the results of subgroup analyses (19, 21, 26).
Second, patients presenting with stroke with a low titer of aPL on one occasion do not have an increased risk of recurrent events compared with other stroke patients. This is supported by the APASS (22).
Third, among patients with definite APS, those presenting with a first venous event could be considered to be at a relatively low risk for recurrences with oral anticoagulation. This finding is supported by the results of 1 subgroup analysis (20) and 2 RCTs (16, 18).
Fourth, patients with definite APS and arterial events or recurrent events are at a high risk for recurrences, even when treated with oral anticoagulation at a target INR of 2.0–3.0. This is supported by the results of cohort studies (12–15, 25, 27).
Fifth, recurrences are infrequent among patients effectively receiving oral anticoagulation at an INR of 3.0–4.0. This is supported by cohort studies (12, 14, 15, 17, 23–25, 27). Most thrombotic events in patients receiving warfarin take place at INRs <3.0. This is supported by cohort studies (12, 14, 15, 17, 23–25, 27) and 1 RCT (16).
Sixth, among patients with APS, the risk of recurrent thrombosis was higher than the risk of major bleeding. Also, the mortality rate associated with thrombosis was much higher than that secondary to hemorrhages. This is supported by cohort studies (12–15, 17, 23, 25, 27), 1 subgroup analysis (20), and 2 RCTs (16, 18).
Finally, no data are available to clarify the role of aspirin combined with warfarin in resistant cases or the impact of correcting cardiovascular risk factors.
Accordingly, the following recommendations for the secondary prophylaxis of thrombosis in patients with aPL can be made. First, patients with APS at low risk (first venous event) should be treated with warfarin at an INR of 2.0–3.0. This recommendation is supported by RCTs (16, 18). Due to the methodologic problems already discussed, the final grade of recommendation is 1B.
Second, patients with APS with arterial thrombosis and/or recurrent venous events should be treated with warfarin at an INR >3.0. This 1C recommendation is based on cohort studies (12–15, 25, 27), because RCTs included few patients with this profile.
Third, patients with venous thromboembolism or stroke with a single positive test for aPL should undergo further aPL testing. If they have only 1 positive aPL test, then they should be treated no differently than the general population (warfarin at a target INR of 2.0–3.0 and low-dose aspirin, respectively). This 1C grade recommendation is based on subgroup analyses (19, 21, 22, 26).
Finally, there are no data to recommend additional antithrombotic treatment such as aspirin for patients who experience recurrent events while receiving oral anticoagulants achieving a 3.0–4.0 target INR. Likewise, the impact of correcting cardiovascular risk factors has not been established.
It is important to acknowledge the high degree of uncertainty due to the limitations already discussed and to the low power of one of the RCTs (18). Therefore, future studies may change these recommendations.
Dr. Ruiz-Irastorza had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study design. Ruiz-Irastorza, Hunt, Khamashta.
Acquisition of data. Ruiz-Irastorza.
Analysis and interpretation of data. Ruiz-Irastorza, Hunt, Khamashta.
Manuscript preparation. Ruiz-Irastorza, Hunt, Khamashta.
Statistical analysis. Ruiz-Irastorza.
- 3Management of thrombosis in antiphospholipid syndrome. In: KhamashtaMA, editor. Hughes syndrome. 2nd ed. London: Springer-Verlag; 2006. p. 547–54., .
- 18A randomized clinical trial of high-intensity warfarin vs. conventional antithrombotic therapy for the prevention of recurrent thrombosis in patients with the antiphospholipid syndrome (WAPS). J Thromb Haemost 2005; 3: 848–53., , , , , , et al.
- 19Antiphospholipid antibodies and venous thromboembolism. Blood 1995; 10: 3685–91., , , , , , et al.
- 30Anticardiolipin testing. In: KhamashtaMA, editor. Hughes syndrome. 2nd ed. London: Springer-Verlag; 2006. p. 275–90..