The Guideline Development, Dissemination, and Implementation Subcommittee of the American Academy of Neurology (GDDI AAN) recently issued an updated practice advisory regarding the treatment of recurrent strokes in patients with patent foramen ovale (PFO).1 Since the last guideline published in 2004, 3 new Class I2–4 studies have been published guiding changes in recommendations regarding closure of PFO (all failing to meet primary end-points) as well as 1 Class II study addressing the efficacy of aspirin vs anticoagulation in the prevention of recurrent ischemic events.
The overall theme of analysis continues to highlight the uncertainty and complexity in medical decision making for a small and heterogeneous patient population. The GDDI AAN recommends against the use of routine PFO closure and found no evidence that anticoagulation provided any benefit over aspirin. In extraordinary cases where recurrent cryptogenic strokes (CS) occur despite adequate medical therapy, the AMPLATZER PFO Occluder device was recommended.
In an insightful editorial,5 stroke neurologist Anthony Furlan, MD, a CLOSURE investigator and Chairman of Neurology at University Hospitals Case Medical Center in Cleveland, OH, highlighted the statistical challenges experienced by all the Class I studies analyzed by the GDDI AAN, notably slow recruitment and heterogeneity of medical therapy arms. He pointed out that the driver of controversy is the per-protocol analysis of the RESPECT trial2 which suggested a small reduction in recurrent stroke. Subgroup analyses suggesting more benefit in patients with certain echocardiographic features (atrial septal aneurysm, large shunt) were entertained, but small numbers precluded any firm conclusions. With the available data, Dr Furlan suggested that young patients (age ≤50) with recurrent stroke on medical therapy without traditional risk factors and substantial shunts should be considered seriously for PFO closure. No study has been able to demonstrate the importance of atrial septal aneurysms or large shunts in the recurrence risk of stroke.
A meta-analyses pooling data from CLOSURE, PC, and RESPECT6 supported the notion that PFO closure may have a small benefit in a per-protocol analysis, although this may be offset by the incidence of new onset atrial fibrillation post-procedure. A portion of the analysis may also be affected by the inclusion of CLOSURE patients, where recurrent paradoxical events were frequently attributable to alternate mechanisms.7
When considering the value of an intention-to-treat analysis vs a per-protocol analysis, it may be prudent to sacrifice the value of controlling for dropout and crossover bias as a patient fitted with a closure device cannot drop out of the treatment group nor can a patient assigned to medical treatment easily cross over to the treatment arm. This line of thinking would suggest that the positive results seen in per-protocol analysis should have some weight when making medical decisions about these complex patients. The flaw in this logic is that it does not account for the failure rate of closure (in the RESPECT trial 72% met criteria for complete closure at 6 months).