IJCS | Volume 33, Nº4, July and August 2020

314 Figure 4 - Risk Ratios for atrial fibrillation in six major trials. CI: confidence interval. For references and trial acronyms, see text. Pereira et al. Patent foramen ovale in cryptogenic stroke Int J Cardiovasc Sci. 2020; 33(4):307-317 Original Article patients with high-risk anatomic PFO features. Therefore, better and stricter patient selection in more recent RCTs may have increased the probability of having a stroke due to PFO and consequently may have increased the likelihood that PFO closure would be effective. Patent foramen ovale presumably provides an anatomic substrate for paradoxical embolism, which may be the cause of most of the cryptogenic strokes. 21 Our findings confirm that PFO closure significantly decreased the rate of recurrent ischemic stroke. The risk of TIA, however, was unaltered, pointing in the direction of a different pathophysiology of TIA in this setting (possibly unrelated to paradoxical embolism) and the potential misclassification of non-ischemic events as TIA. Each study demonstrated low frequency of device and procedure-related complications but a significant increase of AF in the interventional group was seen, which could in theory increase the risk of recurrent stroke. However, most cases of atrial fibrillation occurred early after the procedure with no recurrence during follow-up. The key to an appropriate treatment strategy could be to detect which patients may derive more benefit from PFO closure. Recent studies have shown some characteristics that increase the potential benefit for the patient, but more studies are needed to clarify this issue. 22 The decision to choose a given type of treatment should be multidisciplinary and shared with the patient, considering the preferences of each person. The major sources of data heterogeneity are presented in Table 1 – differences in inclusion criteria, in device used, inmedical therapy, and inmean follow-up. Patients requiring long-termanticoagulation therapyweremostly excluded from the clinical trials. Thus, the population of patients under anticoagulation therapy does not seem to have a proven benefit with PFO closure for the time being. Limitations The studies included were all open label and not double blind, which might impact the results with differential evaluation of suspected events and unequal referral of those events to the adjudication committees. As stated above, there was non-uniformity in the follow- up period, patients’ characteristics, inclusion criteria and closure device used between the studies included. TIA was only a primary endpoint in two of the clinical trials, namely, PC trial and CLOSURE I trial. Preference of some patients and physicians prompted a differential dropout of studies and crossovers between the two treatment groups that may have biased the trials results. Thus, PFO closure was not performed in all patients initially assigned and not all patients who underwent the procedure had a successful closure.