We pooled available data on follow-up events in patients with patent foramen ovale and cryptogenic stroke to evaluate the net clinical benefit of different therapeutic strategies (percutaneous closure vs antiplatelet vs anticoagulant therapy). MEDLINE/PubMed and Cochrane databases and reviewed cited references to identify relevant studies were used; 3,311 patients from 21 clinical studies, both observational and randomized, with follow-up ≥12 months were overall included. Net clinical benefit was evaluated considering the cumulative incidence of both stroke and/or transient ischemic attack and major bleeding events. Anticoagulant therapy was more effective than antiplatelet therapy in preventing recurrent stroke and/or transient ischemic attack (event rates: 7.7% vs 9.8%, respectively, p = 0.03), but at the price of more than sixfold greater risk of major bleeding (7.1% vs 1.3%; odds ratio 6.49, 95% confidence interval 3.25 to 12.99, p <0.00001). Patent foramen ovale closure was associated over the long term with significant net clinical benefit versus both antiplatelet and anticoagulant therapy; such benefit was driven by 50% relative reduction of stroke and/or transient ischemic attack versus antiplatelet therapy and by 82% relative reduction of major bleeding versus anticoagulant therapy. In conclusion, results of this large study-level meta-analysis may influence practice patterns in patients with patent foramen ovale and cryptogenic stroke; an individualized approach tailored on both the risk of recurrent cerebral events and the bleeding risk should be used to identify the best therapeutic option (percutaneous closure vs antiplatelet therapy vs anticoagulant therapy).
The optimal pharmacologic strategy (anticoagulant vs antiplatelet therapy) of medically treated patients with patent foramen ovale (PFO) and previous cryptogenic stroke (CS) remains unclear. Available data on the recurrence of cerebral events during follow-up do not support a routine strategy with catheter-based PFO closure versus medical therapy, and this has been included in recent guidelines on the topic. Moreover, in previous studies and meta-analyses, the long-term bleeding risk associated with percutaneous PFO closure versus medical treatment has not been accurately evaluated. As a consequence, the role for secondary thromboembolic prevention of any therapeutic option has not been weighted against the “iatrogenic” risk of bleeding; thus, the net clinical benefit of different treatment strategies remains unknown. In this meta-analysis of studies on patients with PFO and CS, we have (1) pooled available data on medically treated patients to evaluate efficacy and safety of anticoagulant versus antiplatelet therapy during follow-up and (2) compared the long-term ischemic and bleeding outcome of PFO closure versus a specific medical therapy (antiplatelet or anticoagulant) to assess the net clinical benefit of those different strategies.
Methods
For the various outcome measures, we have considered the following: (1) observational studies describing the clinical outcome of patients with PFO and CS receiving medical therapy (antiplatelet or anticoagulant) or undergoing percutaneous PFO closure and (2) randomized trials comparing percutaneous PFO closure versus medical therapy in patients with PFO and CS. All studies had to report results with a follow-up duration of ≥12 months.
Clinical outcomes analyzed during follow-up were as follows: (1) recurrence of stroke and/or transient ischemic attack (TIA) according to the definition used in each investigation; (2) overall incidence of major bleeding, as defined by the investigators in each study; and (3) net clinical benefit, including in the analysis the cumulative incidence of recurrent stroke and/or TIA and major bleeding. When information about an outcome of interest was not available, the study was not used for this end point.
MEDLINE/PubMed and Cochrane databases up to July 31, 2014, and reviewed cited references to identify eligible studies were used. Search keywords were “patent foramen ovale” and “cryptogenic stroke” combined with the words “observational,” “randomized,” “antiplatelet,” “anticoagulant,” “percutaneous closure,” “transient ischemic attack,” and “bleeding.” Published abstracts from meetings of the American College of Cardiology, American Heart Association, and European Society of Cardiology were also reviewed. Studies were excluded when raw data were not available or in case of duplicate publication of results.
This meta-analysis was conducted following the guidelines of the Cochrane Handbook for Systematic Reviews of Interventions 4.2.5. Data from each study were extracted by 2 independent reviewers and entered into a structured spreadsheet. Disagreements were resolved by consensus. Absolute numbers were recalculated when percentages were reported. Analysis was performed at the study level. Databases of the individual studies were not obtained from the investigators or sponsoring institutions.
All analyses were performed using the Review Manager 5.2 software (available from The Cochrane Collaboration at http//www.cochrane.org ). We used the Mantel-Haenszel method to report fixed-effects estimates as they are more robust in meta-analysis calculations when there are small numbers of events. We tested heterogeneity of the included studies with Q statistics and the extent of inconsistency between results with I 2 statistics. The possibility of publication bias was assessed by funnel plot analysis. Data are presented as odds ratio (OR) with 95% confidence intervals (CIs) and 2-tailed p values.
Results
Overall, 1,199 screened citations met the search criteria. The progress through the different steps of the search leading to the final number of 21 included studies is illustrated in Figure 1 . A total of 3,311 patients were included. Table 1 summarizes the main descriptors of the studies and the patients’ characteristics. Most investigations were prospective, with only 3 being retrospective ; 17 studies were observational and 4 randomized. No study but CLOSURE (Evaluation of the STARFlex Septal Closure System in Patients with a Stroke and/or Transient Ischemic Attack due to Presumed Paradoxical Embolism through a Patent Foramen Ovale) used the association of aspirin plus warfarin in medically treated patients. The type of antiplatelet and anticoagulant therapy in each investigation is reported in Table 1 , when available. Definition of major bleeding in each study, when available, is reported in Table 2 .
| Study | Prospective | Type of study | Inclusion criteria | Age (yrs) Mean±SD | N. patients with anti- platelet therapy included in the analysis ∗ | N. patients with anti-coagulant therapy included in the analysis | Type of anticoagulant therapy | N. patients with closure | Type of device used | Follow-up duration (mo.) Mean±SD |
|---|---|---|---|---|---|---|---|---|---|---|
| Hanna | No | Observational | CS | 43 | 6 | 5 | Warfarin | 2 | NA | 28±13 |
| Mas | Yes | Observational | CS | 39±10 | 48 | 22 | NA | – | – | 22±16 |
| Haussmann | Yes | Observational | CS, TIA | 46±13 | 17 | 15 | Warfarin | – | – | 59±12 |
| Bougousslavsky | Yes | Observational | CS, TIA | 44±14 | 92 | 37 | Acenocoumarol | 11 | NA | 36 |
| Cujec | No | Observational | CS, TIA | 38±11 | 36 | 38 | Warfarin | 14 | NA | 42±26 |
| Mas | Yes | Observational | CS | 40 | 255 | 6 | NA | – | – | 37±10 |
| Homma | Yes | Randomized | CS | 57±13 | 106 | 97 | Warfarin | – | – | 24 |
| Windecker | No | Observational | CS, TIA | 51±13 | 79 | 79 | Warfarin | 150 | Amplatzer, PFOStar, Sideris, Angel Wing, CardioSEAL | 29±22 |
| Schuchlenz | Yes | Observational | CS, TIA | 48 | 66 | 47 | Warfarin | 167 | Amplatzer, CardioSEAL, STARflex | 32 |
| Harrer | Yes | Observational | CS, TIA | 52±15 | 59 | 54 | Warfarin | – | Rashkind, Amplatzer, ASDOS, Sideris, CardioSEAL, PFOStar | 52±32 |
| Thanapoulos | Yes | Observational | CS, TIA | 40±12 | 44 † | – | NA | 48 | Amplatzer | 24 |
| Cerrato | Yes | Observational | CS, TIA | 44±13 | 48 | 17 | Warfarin | 21 | NA | 63±28 |
| Casaubon | Yes | Observational | CS, TIA | 50 | 41 ‡ | 20 | Warfarin | 60 | CardioSEAL, Amplatzer | 32 |
| Serena | Yes | Observational | CS, TIA | 53 | 234 | 63 | Acenocoumarol | – | – | 24±13 |
| Lee | Yes | Observational | CS | 53±13 | 99 | 60 | Warfarin | 22 | CardioSEAL, Amplatzer | 42 |
| Paciaroni | Yes | Observational | CS, TIA | 40±10 | 93 | 24 | NA | 121 | Amplatzer, PFOStar, CardioSEAL/ STARflex | 24 |
| Mazzucco | Yes | Observational | CS | 42±9 | 49 | 3 | NA | 51 | Amplatzer, BioSTAR | 27±9 |
| Wahl | Yes | Observational | CS, TIA | 50±13 | 57 | 46 | Warfarin | 150 | Amplatzer, PFOStar, Sideris, Angel Wing, CardioSEAL | 132 |
| CLOSURE I | Yes | Randomized | CS, TIA | 45±9 | 252 | 151 | Warfarin | 447 | STARflex | 44 |
| RESPECT | Yes | Randomized | CS | 46±10 | 359 ‡ | 121 | Warfarin | 499 | Amplatzer | 31±24 |
| PC TRIAL | Yes | Randomized | CS, TIA, Periph. embolism | 44±10 | 155 | 64 | NA | 204 | Amplatzer | 48 |
∗ Aspirin or clopidogrel was used in all studies, unless otherwise specified.
† Clopidogrel plus aspirin in all patients.
| Study | Major bleeding definition |
|---|---|
| Hanna | No definition provided |
| Mas | No definition provided |
| Haussmann | No definition provided |
| Bougousslavsky | No definition provided |
| Cujec | No definition provided |
| Mas | No definition provided |
| Homma | Intracranial, intraspinal, intracerebral, subarachnoid, subdural or epidural hemorrhage or any other bleeding requiring transfusion |
| Windecker | No definition provided |
| Schuchlenz | Fatal, intracranial, retroperitoneal bleeding or any bleeding requiring transfusion or emergency procedures |
| Harrer | No definition provided |
| Thanapoulos | Gastrointestinal bleeding |
| Cerrato | No definition provided |
| Casaubon | No definition provided |
| Serena | No definition provided |
| Lee | Bleeding episodes causing interruption of therapy |
| Paciaroni | Hemorrhagic stroke |
| Mazzucco | Fatal, symptomatic intracranial, intraocular or retroperitoneal hemorrhage or overt bleeding resulting in a decrease in haemoglobin >3 gr/dL |
| Wahl | Life-threatening or disabling bleeding |
| CLOSURE | Intracranial, intraocular or retroperitoneal hemorrhage or any hemorrhage requiring a transfusion or resulting in a hematocrit decrease >15% or hemoglobin decrease >5 g/dL |
| RESPECT | No strict definition provided |
| PC TRIAL | Bleeding requiring any blood transfusion |
To evaluate the natural history of patients with PFO and CS maintained on different medical treatments, we have considered the “conservative therapy” cohorts of PFO studies (i.e., patients undergoing percutaneous closure were excluded) and performed a comparison between antiplatelet and anticoagulant treatment on the recurrence of stroke and/or TIA. A total of 18 studies, 15 observational and 3 randomized, enrolling 2,798 patients, were included in the analysis. Of those, 1,939 patients had received antiplatelet therapy and 859 oral anticoagulant therapy. Mean follow-up was 36 months (range 22 to 63). Occurrence of recurrent stroke and/or TIA during follow-up was 9.8% in patients receiving antiplatelet treatment versus 7.7% in those receiving anticoagulant therapy, corresponding in the former to a significant 53% higher relative risk (OR 1.53, 95% CI 1.04 to 2.23, p = 0.03; Figure 2 ), without heterogeneity among the studies and without evidence of publication bias. The absolute excess risk for recurrent stroke and/or TIA with antiplatelet therapy was 0.7%/year (3.3%/year vs 2.6%/year with anticoagulant treatment). We then investigated the long-term bleeding risk of antiplatelet versus anticoagulant therapy. Again, patients undergoing percutaneous PFO closure were excluded, and 15 studies, 14 observational and 1 randomized, were analyzed, for a total of 1,407 medically treated patients (930 on antiplatelet and 477 on oral anticoagulant therapy) analyzed. Mean follow-up duration was 42 months, range 22 to 132 months. Rates of major bleeding were 7.1% in patients receiving oral anticoagulant therapy versus 1.3% in those on antiplatelet treatment (OR 6.49, 95% CI 3.25 to 12.99, p <0.00001; Figure 2 ), without heterogeneity among the studies and without evidence of publication bias. The absolute excess risk for major bleeding events with anticoagulant therapy was 1.6%/year (2%/year vs 0.4%/year with antiplatelet therapy).
We specifically compared long-term efficacy and safety of percutaneous PFO closure versus antiplatelet medical therapy and versus anticoagulant medical therapy to evaluate the net clinical benefit of those 3 strategies. For this purpose, data were available from a total of 11 studies, all observational, with a mean follow-up length of 36 months (range 22 to 63). Incidence of recurrent stroke and/or TIA was significantly lower in patients undergoing percutaneous closure versus those receiving antiplatelet therapy (4.3% vs 9.2%, respectively; OR 0.50, 95% CI 0.35 to 0.71, p <0.0001; Figure 3 ), without difference among the 2 groups in the occurrence of major bleeding (1% vs 1.3%, p = 0.25; Figure 3 ). As a consequence, a significant net clinical benefit with percutaneous closure was observed (OR 0.30, 95% CI 0.18 to 0.51, p <0.00001; Figure 3 ), without heterogeneity among the studies and without evidence of publication bias; in particular, the absolute excess risk for recurrent stroke and/or TIA with antiplatelet therapy was 1.7%/year (3.1%/year vs 1.4%/year with PFO closure).
