Closure of Patent Foramen Ovale Versus Medical Therapy in Patients With Cryptogenic Stroke or Transient Ischemic Attack
Updated Systematic Review and Meta-Analysis
Background and Purpose—Previous systematic reviews and meta-analyses compared the efficacy and safety of patent foramen ovale (PFO) closure versus medical treatment in patients with cryptogenic stroke or transient ischemic attack (TIA). Recently, new evidence from randomized trials became available.
Methods—We searched PubMed until September 24, 2017, for trials comparing PFO closure with medical treatment in patients with cryptogenic stroke/TIA using the items: stroke or cerebrovascular accident or TIA and patent foramen ovale or paradoxical embolism and trial or study.
Results—Among 851 identified articles, 5 were eligible. In 3627 patients with 3.7-year mean follow-up, there was significant difference in ischemic stroke recurrence (0.53 versus 1.1 per 100 patient-years, respectively; odds ratio [OR], 0.43; 95% confidence intervals (CI), 0.21–0.90; relative risk reduction, 50.5%; absolute risk reduction, 2.11%; and number needed to treat to prevent 1 event, 46.5 for 3.7 years). There was no significant difference in TIAs (0.78 versus 0.98 per 100 patient-years, respectively; OR, 0.80; 95% CI, 0.53–1.19) and all-cause mortality (0.18 versus 0.23 per 100 patient-years, respectively; OR, 0.73; 95% CI, 0.34–1.56). New-onset atrial fibrillation occurred more frequently in the PFO closure arm (1.3 versus 0.25 per 100 patient-years, respectively; OR, 5.15; 95% CI, 2.18–12.15) and resolved in 72% of cases within 45 days, whereas rates of myocardial infarction (0.12 versus 0.09 per 100 patient-years, respectively; OR, 1.22; 95% CI, 0.25–5.91) and any serious adverse events (7.3 versus 7.3 per 100 patient-years, respectively; OR, 1.07; 95% CI, 0.92–1.25) were similar.
Conclusions—In patients with cryptogenic stroke/TIA and PFO who have their PFO closed, ischemic stroke recurrence is less frequent compared with patients receiving medical treatment. Atrial fibrillation is more frequent but mostly transient. There is no difference in TIA, all-cause mortality, or myocardial infarction.
Despite thorough investigation, cause remains undetermined in a significant proportion of patients with ischemic stroke and terms like cryptogenic or embolic stroke of undetermined source are commonly used to describe these cases.1–3 Patent foramen ovale (PFO), a frequent finding in the general population,4,5 is considered a possible underlying mechanism in a proportion of embolic stroke of undetermined source patients,6 and its closure has been suggested as an efficacious intervention on top of medical treatment. Until recently, only 3 randomized controlled trials had tested this hypothesis yielding inconclusive results.7–9
In a previous systematic review and meta-analysis, we did not find evidence to support PFO closure in patients with cryptogenic stroke or transient ischemic attack (TIA) with unselected devices.10 However, in subgroup analysis, selected closure devices seemed to be superior to medical treatment,10 and an individual patient data meta-analysis revealed a reduced stroke risk with PFO closure.11
Recently, new evidence from randomized trials became available.12–14 The aim of this work is to update the results of the previous systematic reviews and meta-analyses with the newly available evidence.
The data that support the findings of this study are available from the corresponding author on reasonable request.
Search Strategy and Inclusion Criteria
We extended our previous search (originally performed on April 18, 2013, on MEDLINE) until September 20, 2017, to identify eligible articles in the English language. Abstracts were not considered for inclusion. The search items were stroke or cerebrovascular accident or transient ischemic attack and patent foramen ovale or PFO or paradoxical embolism and trial or study. In addition, we searched the references of related letters, reviews, and editorials to identify potentially eligible studies. To be eligible for the present analysis, the studies had to be randomized controlled trials of PFO closure versus medical therapy in patients with cryptogenic stroke or TIA. This work was performed according to the PRISMA statement (Preferred Reporting Items for Systematic Reviews and Meta-Analyses).15
Data Extraction and Outcomes
Data extraction was performed independently by 2 authors (V.P. and D.S.). Studies were screened to extract the number of patients in each treatment arm, the type of PFO closure device, the type of antithrombotic treatment, the underlying risk of the PFO, the length of follow-up, and the number of events for each outcome (ischemic stroke recurrence, all-cause mortality, TIA, new-onset atrial fibrillation, myocardial infarction, and any serious adverse event). PFOs were defined as high risk when there was moderate-to-large shunt and atrial septal aneurysm. Any discrepancy or uncertainty was resolved by consensus among the authors.
The primary outcome of this meta-analysis was recurrence of ischemic stroke and the secondary end points were all-cause mortality, transient ischemic attack, new-onset atrial fibrillation, and myocardial infarction.
Data were analyzed on an intention-to-treat basis. Odds ratios (OR) and 95% confidence intervals (CI) were calculated for each outcome using the Mantel–Haenszel random-effects method. Heterogeneity between trials was assessed by measuring inconsistency using the I2 index, which measures the proportion of the variability in effect estimates that can be attributed to heterogeneity rather than chance. I2 is calculated as follows: I2=100%×(Q−df)/Q, where Q is Cochran heterogeneity statistic and df the degrees of freedom. A value of 0% indicates no observed heterogeneity, and larger values show increasing heterogeneity.16 Mean follow-up duration was calculated by the follow-up duration reported in the trials.
In case of statistically significant associations, absolute risk reduction (ARR) or absolute risk increase was calculated as (ME/MS)−(IE/IS), where ME is the number of events in the medical therapy group, MS the number of patients randomized to medical therapy, IE is the number of events in the PFO closure (intervention) group, and IE the number of patients randomized to PFO closure; relative risk reduction (RRR) or relative risk increase as [(ME/MS)−(IE/IS)]/(ME/MS); number needed to treat (NNT) to prevent/cause 1 event as 1/[(ME/MS)−(IE/IS)].
All analyses were performed with the Review Manager (RevMan) version 5.1 (Copenhagen: The Nordic Cochrane Center, The Cochrane Collaboration, 2011).
This article adheres to the American Heart Association Journals’implementation of the Transparency and Openness Promotion guidelines.
Among 851 potentially eligible articles identified in the search, we identified 2 new eligible studies13,14 and the long-term follow-up results of 1 study included in the original meta-analysis.12 The flow diagram of the study selection and the main characteristics of the 5 trials,8,9,12–14 which were finally included in the analysis, are presented in the online-only Data Supplement.
Among 3627 patients with cryptogenic stroke or TIA and PFO in these 5 trials, 1829 patients were allocated to PFO closure and 1798 patients were allocated to medical treatment. The mean follow-up was 3.7 years and ranged between 2.0 and 5.9 years between trials.
There was statistically significant difference in the rate of ischemic stroke recurrence between the PFO closure arm and the medical treatment arm (0.53 versus 1.1 per 100 patient-years, respectively; OR, 0.43; 95% CI, 0.21–0.90; RRR, 51.5%; ARR, 2.1%; NNT, 46.5 for 3.7 years) with significant heterogeneity (I2=53%; Figure 1). There was no statistically significant difference in the rate of all-cause mortality (0.18 versus 0.23 per 100 patient-years, respectively; OR, 0.73; 95% CI, 0.34–1.56; Figure 2) or TIA events (0.78 versus 0.98 per 100 patient-years, respectively; OR, 0.80; 95% CI, 0.53–1.19; Figure 3).
New-onset atrial fibrillation occurred more frequently in the PFO closure arm than in the medical treatment arm (1.3 versus 0.25 per 100 patient-years, respectively; OR, 5.15; 95% CI, 2.18–12.15; relative risk increase, 403.1%; absolute risk increase, 3.8%; NNT, 26.2 for 3.7 years; Figure 4). In the PFO occlusion arms, 56.6% of the events of atrial fibrillation were transient and 72% resolved within 30 to 45 days.
The rates of myocardial infarction (0.12 versus 0.09 per 100 patient-years, respectively; OR, 1.22; 95% CI, 0.25–5.91; Figure 5) and any serious adverse events (7.3 versus 7.3 per 100 patient-years, respectively; OR, 1.07; 95% CI, 0.92–1.25) were similar (online-only Data Supplement).
Analysis According to the Type of PFO Closure Device
In subgroup analyses of the studies reporting outcomes according to the type of PFO closure device used, compared with the medical treatment, the rate of ischemic stroke recurrence was lower in the AMPLATZER PFO Occluder arm (1394 patients in 2 studies; 0.50 versus 0.89 per 100 patient-years, respectively; OR, 0.56; 95% CI, 0.31–1.00; RRR, 43.4%; ARR, 2.1%; NNT, 48.2 for 3.7 years, without heterogeneity between the 2 trials; Figure 1), in the Septal Occluder arm (Helex or Cardioform) (664 patients in 1 study; 0.42 versus 1.68 per 100 patient-years, respectively; OR, 0.24; 95% CI, 0.09–0.66; RRR, 74.7%; ARR, 4.0%; NNT, 24.9 for 3.7 years; Figure 1), and in 1 study in which several devices were used (663 patients in 1 study; 0 versus 1.15 per 100 patient-years, respectively; OR, 0.05; 95% CI, 0.00–0.81; RRR, 100.0%; ARR, 4.0%; NNT, 24.8 for 3.7 years; Figure 1) but not in the STARFlex device (1.34 versus 1.40 per 100 patient-years in the medical treatment arm; OR, 0.95; 95% CI, 0.43–2.11; Figure 1). On the contrary, there was no difference in all-cause mortality, TIA events, or any serious adverse event between the medical treatment arm and the PFO closure arm in any type of PFO closure device (Figures 2 and 3; online-only Data Supplement, respectively).
Analysis According to the Risk of the PFO
The outcomes included in the PFO risk–related analysis were occurrence of stroke in the CLOSE trial (Patent Foramen Ovale Closure or Anticoagulants Versus Antiplatelet Therapy to Prevent Stroke Recurrence); the composite of stroke or transient ischemic attack during 2-year follow-up, all-cause mortality during the first 30 days, or death from neurological causes between 31 days and 2 years in the CLOSURE-1 trial (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); recurrent stroke in the Gore-REDUCE trial (GORE® HELEX® Septal Occluder / GORE® CARDIOFORM Septal Occluder and Antiplatelet Medical Management for Reduction of Recurrent Stroke or Imaging-Confirmed TIA in Patients With Patent Foramen Ovale); composite of all-cause mortality, nonfatal stroke, TIA, or peripheral embolism in the PC trial (Clinical Trial Comparing Percutaneous Closure of Patent Foramen Ovale Using the Amplatzer PFO Occluder with Medical Treatment in Patients With Cryptogenic Embolism); and recurrent ischemic stroke in the RESPECT trial (Randomized Evaluation of Recurrent Stroke Comparing PFO Closure to Established Current Standard of Care Treatment).
Outcomes in the PFO closure arm were significantly better compared with the medical therapy arm in patients with high-risk PFO (Figure 6, top; 0.51 versus 1.43 per 100 patient-years, respectively; OR, 0.39; 95% CI, 0.16–0.96; RRR, 61.0%; ARR, 3.3%; NNT, 30 for 3.7 years). On the contrary, there was no difference in outcomes between the 2 groups in low-risk PFOs (Figure 6, 0.96 versus 1.3 per 100 patient-years, respectively; OR, 0.79; 95% CI, 0.43–1.43; Figure 6).
This updated meta-analysis of randomized controlled trials of PFO closure versus medical treatment shows that ischemic stroke recurrence, but not TIA or all-cause mortality or myocardial infarction or any serious adverse event, is less frequent in patients with cryptogenic stroke/TIA and PFO who have their PFO closed compared with patients receiving medical treatment. On the contrary, new-onset atrial fibrillation was more frequent in the PFO closure group but mostly transient.
The effect of PFO closure on the risk of ischemic stroke recurrence was not evident in our previous meta-analysis. The inclusion of 2 more randomized trials in the present analysis,13,14 as well as the recent publication of the long-term follow-up of one of the previously included trials,12 increased the statistical power of the meta-analysis to detect a significant effect.
In our previous meta-analysis, the effect of PFO closure on the risk of ischemic stroke recurrence was seen only in the device-type subgroup analysis in 1 of the 2 PFO closure devices that had been tested in the included trials.7,12 In particular, we had shown that the AMPLATZER PFO Occluder, but not the STARflex device, was superior to medical treatment for the prevention of ischemic stroke recurrence. Similar results were reported recently also by a pooled analysis of these trials.11 The present meta-analysis confirms these findings for AMPLATZER PFO Occluder and extends it also to other devices, which have been used in the recent trials like the Helex and Cardioform Septal Occluders.13,14
There was an important difference in the rate of new-onset atrial fibrillation between the PFO closure group and the medical treatment group, which was confirmed for all devices tested. These episodes were frequently terminated soon after their initiation, and it is likely that they did not greatly increase the risk of stroke12–14; still, their clinical significance with regard to the risk of embolic events or to the need for initiation of anticoagulant treatment need to be analyzed further.
Strengths of the present meta-analysis include the calculation of ARR, absolute risk increase, and NNT for significant associations, the conduction of the analysis according to the PRISMA guidelines15 and on an intention-to-treat basis, and the subgroup analyses according to the device used for PFO closure.
The inherent limitations of all meta-analyses apply also to the present meta-analysis: differences in the selection criteria among trials, variations in the definitions of comorbidities used in the trials, and differences in the length of follow-up among trials. In addition, the diagnostic approach of cryptogenic stroke was prespecified for each study but not uniform across trials, for example, prolonged heart rhythm monitoring was not required.7 Moreover, different devices were used in the trials; we tried to overcome this by performing subgroup analyses according to the device used in each trial. Furthermore, the primary outcomes differed across trials; we chose the recurrence of ischemic stroke as the primary end point in this meta-analysis as it is perhaps the clinically most relevant in this patient group. Also, the rate of drop-outs differed between groups, which resulted in unequal exposure to the risk of stroke recurrence.8,17 Moreover, the mean duration of the follow-up was calculated based on the mean follow-up duration reported in each study, which may deviate slightly from the true mean follow-up duration. In addition, referral for adjudication of end points was not blinded.17 Finally, these trials included patients between 18 and 60 years of age. It is unknown whether the results of this meta-analysis apply also to the elderly patients.
Although the overall effect of PFO closure on the risk of ischemic stroke recurrence is evident by the present meta-analysis, an individual patient data pooled analysis of all randomized trials of PFO closure would provide further insight into the role of atrial septal aneurysm, the size of the shunt, the age of the patient, the type of antithrombotic treatment, and other patient or PFO parameters. For example, the RESPECT study reported a significantly increased rate of venous thromboembolism in patients who underwent PFO closure, presumably related to the anticoagulation used in the medical arm12; given that venous thromboembolism seems to be the main issue in stroke related to PFO and the important role of nonvitamin K antagonists in the prevention of venous thromboembolism,18 this could have implications for the use of nonvitamin K antagonists as the comparator of the PFO closure.
In conclusion, the updated meta-analysis of randomized controlled trials of PFO closure versus medical treatment shows that ischemic stroke recurrence, but not TIA or all-cause mortality or myocardial infarction or any serious adverse event, is less frequent in patients with cryptogenic stroke/TIA and PFO who have their PFO closed compared with patients receiving medical treatment. On the contrary, new-onset atrial fibrillation was more frequent in the PFO closure group but mostly transient.
Dr Ntaios is responsible for study concept and design, acquisition of data, analysis and interpretation, preparation of the article, and study supervision. Dr Papavasileiou helped in acquisition of data, analysis and interpretation, and critical revision of the article for important intellectual content. Dr Sagris contributed toward acquisition of data, analysis and interpretation, and critical revision of the article for important intellectual content. Dr Makaritsis helped in analysis and interpretation and critical revision of the article for important intellectual content. Dr Vemmos assisted in analysis and interpretation and critical revision of the article for important intellectual content. Dr Steiner assisted with analysis and interpretation and critical revision of the article for important intellectual content. Dr Michel contributed toward analysis and interpretation and critical revision of the article for important intellectual acquisition of data.
Dr Michel has received within the last 3 years through his institution research grants from the Swiss Heart Foundation, Boehringer, and BMS; speaker fees from Boehringer-Ingelheim, Bayer, Daiichi-Sankyo, Medtronic, and Amgen; honoraria from scientific advisory boards from Boehringer-Ingelheim, Bayer, Pfizer, and BMS; consulting fees from Medtronic, Astra-Zeneca, and Amgen. All this support is used for stroke education and research. The other authors report no conflicts.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.117.020030/-/DC1.
- Received November 6, 2017.
- Revision received December 13, 2017.
- Accepted December 15, 2017.
- © 2018 American Heart Association, Inc.
- Hart RG,
- Catanese L,
- Perera KS,
- Ntaios G,
- Connolly SJ
- Michel P,
- Odier C,
- Rutgers M,
- Reichhart M,
- Maeder P,
- Meuli R,
- et al
- Ntaios G,
- Papavasileiou V,
- Milionis H,
- Makaritsis K,
- Manios E,
- Spengos K,
- et al
- Kent DM,
- Dahabreh IJ,
- Ruthazer R,
- Furlan AJ,
- Reisman M,
- Carroll JD,
- et al
- Saver JL,
- Carroll JD,
- Thaler DE,
- Smalling RW,
- MacDonald LA,
- Marks DS,
- et al
- Mas JL,
- Derumeaux G,
- Guillon B,
- Massardier E,
- Hosseini H,
- Mechtouff L,
- et al
- Søndergaard L,
- Kasner SE,
- Rhodes JF,
- Andersen G,
- Iversen HK,
- Nielsen-Kudsk JE,
- et al
- Higgins JP,
- Thompson SG,
- Deeks JJ,
- Altman DG
- Becattini C,
- Agnelli G