Prevention of PFO-related stroke: an evolving concept

The relationship between the presence of a patent foramen ovale (PFO) and an increased risk of ischemic stroke was first recognized in 1988, when two case-control studies reported on a significantly increased PFO frequency in young patients who had experienced a stroke of otherwise unknown cause compared with stroke-free controls.^{1, 2} An increased risk, although of lesser magnitude, was then reported even in older patients.³ The purported stroke mechanism was paradoxical embolism, or the embolization to the systemic circulation of a thrombus originating in the venous circulation via the right-to-left shunt allowed by the PFO. Since a PFO is present in approximately one in four adults, the PFO-related risk of stroke could have important health implications at a population level; however, population-based studies failed to detect a significant increase in stroke risk,^{4, 5} possibly because the increased risk present in a minority of individuals with PFO was diluted by the lack thereof in the majority of them. Primary prevention strategies for PFO-related first stroke were therefore unfeasible, given the virtual impossibility to identify individuals at high risk of stroke, other than those with relatively infrequent conditions that increase thrombus presence/formation (hypercoagulability; deep venous thrombosis) or facilitate a right-to-left shunt (pulmonary hypertension). On the other hand, effective secondary prevention strategies were clearly needed for patients with PFO who had already suffered a stroke. Since paradoxical embolism was the purported stroke mechanism, antithrombotic treatment was tested first. In a randomized trial of warfarin versus aspirin, both treatments were shown to decrease the frequency of recurrent stroke to a similar extent, bringing it down to rates very similar to that observed in patients without PFO;⁶ however, younger patients, who are at the greatest risk of PFO-related stroke recurrence, were underrepresented in the study.

The introduction of devices that could be deployed via a transvenous catheter provided a simpler option for PFO closure, which had theretofore required open heart surgery, and opened the way to a new possible approach for preventing stroke recurrence and, shortly thereafter, to clinical trials comparing the efficacy of PFO closure versus medical therapy. While the earlier trials failed to show a significant difference,^7–9 the most recent ones suggested a superiority of PFO closure;^10–12 in a meta-analysis including four of the trials for a total 2531 patients followed from 3 to almost 6 years, PFO closure was associated with a stroke recurrence rate of 1.8%, compared with 5.1% for medical therapy.¹³ The benefit was observed especially, but not exclusively, in patients with particular PFO features (large shunts; coexisting atrial aneurysm).^{11, 12} PFO closure was associated with peri-procedural complications (ranging from 3.2% to 5.9% across the trials) and with an increased frequency of atrial fibrillation (ranging from 2.9% to 7.3%, the majority occurring within the first 30 days after the procedure). The trials had limitations, including the relatively small number of primary outcome events and the open-label ascertainment of endpoints, which may have affected the results because of ascertainment bias. In the present issue of Stroke, Messe’ and colleagues report on additional results from one of the trials, the Gore REDUCE trial, describing in detail the neuroimaging findings observed during the follow-up in the two treatment arms (PFO closure and antiplatelet therapy).¹⁴ Brain MRI was performed at enrollment and during follow-up, either after the detection of stroke symptoms through a validated questionnaire, or at 2 years in patients that remained symptoms-free. PFO closure was associated with a reduced frequency of combined strokes and silent brain infarcts resulting from the reduced frequency of the former, but not of the latter; infarct number, size and location did not differ significantly between treatment arms, or between clinically manifest and silent infarcts. The study provides an important objective validation of the clinical observation of a decreased stroke incidence in the PFO-closure group, thus minimizing the possibility that the ascertainment bias associated with the open label study design may have affected the results; it also provides interesting data on the neuroimaging features of the newly developed brain lesions observed with each treatment. At the same time, it raises questions related to some unexpected, or not immediately clear, findings. The observation that PFO closure was superior to antiplatelet therapy alone in preventing infarcts associated with clinical symptoms but not silent ones is surprising, and not easy to explain. The possibility that clinically manifest and silent infarcts may represent somewhat different entities, with the latter being less frequently embolic in origin, seems to be negated by the similar number, size and location of both infarct types. As the Authors mentioned, some mild periprocedural strokes may have gone undetected, or their symptoms been considered secondary to anesthesia effects; this circumstance would decrease the treatment difference observed for clinical strokes and increase that for silent infarcts, while leaving the combined endpoint, and therefore the main conclusion of the study, unchanged. Finally, the minimum MRI infarct size considered (3 mm), although established in the literature, may have led to the exclusion of smaller and more often asymptomatic lesions, whose frequency might have differed between treatments; in fact, the total number of clinically manifest and of silent infarcts in the study was very similar (17 and 20, respectively), which may suggest a certain underestimation of silent lesions. These considerations also serve as a reminder that the study results were based on a small number of endpoints, which suggests caution in their interpretation, especially when subgroup comparisons are concerned; this applies to the reported results of the lack of a relationship between shunt size or incident atrial fibrillation and the outcomes of interest, or of similar neuroimaging features between symptomatic and silent infarcts.

Like for other similar trials, the results of the Gore REDUCE trial do not represent a comparison of PFO closure versus antiplatelet treatment, but rather of PFO closure plus antiplatelet treatment (which was continued for the entire duration of the follow-up) versus antiplatelet treatment alone; therefore, the real clinical decision is whether or not to add PFO closure to an already effective preventive treatment (cryptogenic stroke recurrence rate on medical therapy is estimated at 1% per year),¹⁵ weighing possible benefit (further increase in protection) against risk (periprocedural complications, including atrial fibrillation, which was observed in 7.3% of patients in this trial; unestablished long-term safety record). This decision must involve a multidisciplinary team (neurologist and cardiologist) to assess the appropriateness of PFO closure, taking into account the patient’s clinical characteristics (embolic-appearing index stroke; absence of other possible etiologies; suitable PFO morphology) and informed preferences, an approach that was also supported in the most recent practice advisory of the American Academy of Neurology (AAN)¹⁵ and in the recently published AHA/ASA guideline for the secondary prevention of stroke;¹⁶ under these conditions, the AAN advisory supported PFO closure in patients with cryptogenic stroke, bearing in mind the peri-procedural risk and a possible over 2.5-fold increase in serious non-periprocedural atrial fibrillation,¹⁵ whereas the AHA/ASA guideline also supported PFO closure in patients 18 to 60 years of age with a nonlacunar ischemic stroke of undetermined cause and high-risk PFO anatomic features, but considered the indication as not well established when those features are absent.¹⁶ Both guidelines agreed that there is not enough evidence on whether PFO closure is superior to systemic anticoagulation in this setting.

Some aspects of the prevention of PFO-related stroke recurrence remain unresolved. Since almost all clinical trials were conducted in patients younger than 60 years, limited data on treatment comparisons are available in older patients. Also, as systemic anticoagulation was not systematically tested in most trials, it is unclear whether patients who need anticoagulation because of coexisting conditions might benefit from adding PFO closure to it, or just be exposed to possible adverse effects from both treatments. Trials comparing new oral anticoagulants versus aspirin may be warranted in older patients, currently excluded from PFO closure recommendations, and in younger patients unwilling to undergo the procedure.¹⁵ Even with these open issues, the field of prevention of PFO-related recurrent stroke has come a long way in recent years, with several effective preventative options being now available for application to patients with different risk profiles. Besides further refinement of these options, future research will hopefully investigate novel and better ways to identify individuals with a PFO who, because of individual characteristics or associated conditions, may be at increased risk of suffering a first stroke, an important topic that has received comparatively limited attention in recent years.