Potentially large yet uncertain benefits: a meta-analysis of patent foramen ovale closure trials

Abstract

Background and Purpose: Despite widespread clinical use, the efficacy of percutaneous closure of patent foramen ovale (PFO) for secondary prevention in patients with cryptogenic stroke has been formally tested only recently by 3 randomized clinical trials (RCTs) with null results in their primary analyses. We aimed to draw more precise conclusions from these RCTs via meta-analysis.

Methods: Meta-analysis of incidence rates (IRs) of stroke in treatment arms and hazard ratios (HRs) for the efficacy of closure with random-effects model. Sensitivity analyses were explored for additional outcomes and types of analyses.

Results: Two different types of closure devices were used (StarFLEX in CLOSURE-I trial and Amplatzer PFO occluder in RESPECT and PC-trial). Summary IRs of stroke were low in both arms: 0.76 (95% confidence interval, 0.30-1.96) per 100 person-years in the device group and 1.30 (95%CI, 0.94-1.81) in the medical group. Meta-analysis for HR of stroke showed a non-significant 45% risk reduction (summary HR=0.55 (0.26-1.18)) and results remained non-significant when the two Amplatzer-device trials were combined. However, statistically significant and stronger results were obtained in sensitivity analyses, when composite outcomes or fixed-effects models were considered. Subgroup meta-analyses failed to show any detectable closure-effect modifiers.

Conclusions: Meta-analysis of 3 RCTs for PFO closure does not remedy the uncertainty of individual RCTs. Different inclusion and analysis criteria can lead to mutually conflicting summary conclusions on the benefits of closure. Our findings underscore the importance of generating more randomized evidence for informed decisionmaking on the risks and benefits of closure versus medical therapy.

Keywords: foramen ovale, patent; meta-analysis; patent foramen ovale closure; secondary prevention; stroke.

Figure 1

Incidence rates of stroke (95% confidence intervals) in the device and medical treatment groups of 3 randomized clinical trials with their corresponding meta-analytic summaries.

Figure 2

Forest plot for the meta-analysis of hazards ratios of stroke of mechanical closure versus medical treatment from 3 randomized clinical clinical trials.

Figure 3

Forest plot of subgroup meta-analyses results for the primary component outcomes of the 3 randomized clinical trials. Subgroup meta-analytic summaries (hazard ratios with corresponding 95% confidence intervals) and p-values for the tests of interaction are displayed in the twocolumns on the right side of the graph, respectively. Abbreviations: ASA, atrial septal aneurysm; TIA, transient ischemic attack.