What Will the FDA Do?

Ziyad M. Hijazi, MD, MPH


J Invasive Cardiol. 2012;24(12):625-626. 

One of the many shortcomings of comparative observational studies examining patent foramen ovale (PFO) closure versus medical management included in meta-analysis is failure to adjust for potential confounders; therefore, the need for randomized studies is very obvious since they provide the best scientific evidence.

The only published randomized, controlled trial (RCT) comparing medical management with percutaneous PFO closure is the CLOSURE 1 (Evaluation of the STARFlex septal closure system in patients with a stroke and/or transient ischemic attack due to presumed paroxysmal embolism through the patent foramen ovale) trial. With much anticipation, the trial was first presented in the fall of 2010 during the Annual Scientific Sessions of the American Heart Association in Chicago with final results published in 2012.[1] Briefly, the trial enrolled 909 patients at 87 trial sites between the ages of 18 and 60 years. The inclusion criteria included prior ischemic stroke or transischemic attack (TIA; radiologic evidence was not a requisite) within the previous 6 months and presence of a PFO on transesophageal echocardiography. The trial randomized patients to either PFO closure or medical therapy in 1:1 fashion. The primary endpoints were a composite stroke or TIA during 2 years of follow-up, as well as death. An atrial septal aneurysm, defined as septal excursion of over 10 mm, was present in 35%-37% of patients. Of the 447 patients assigned to the closure group, 405 underwent the actual procedure. Of those, successful closure, defined as implantation of a STARFlex device with no procedural complications, was achieved in 362 participants (89.4% success rate). At 6 months of follow-up, effective closure, defined as grade 0 or 1 residual shunt, was documented in 315 patients (86.1% closure rate). After 2 years of follow-up, a statistically non-significant difference in the incidence of the primary endpoint was found (5.5% closure group vs 6.2% medical therapy group [adjusted hazard ratio (HR), 0.78; 95% confidence interval [CI], 0.45–1.35; P=.37). The insignificant trend toward better outcome with PFO closure was driven by less TIAs in the closure group (3.1% closure group vs 4.1%; adjusted HR, 0.75; 95% CI, 0.36–1.55; P=.44). Stroke occurrence was identical among patients undergoing PFO closure compared to medically treated patients (2.9% vs 3.1%; adjusted HR, 0.90; 95% CI, 0.41–1.98; P=.79). Further, no differences were found comparing the treatment modalities "per protocol" as compared to "intention to treat." Unexpectedly, potential alternative explanations for recurrent neurologic events (as opposed to PFO-mediated) were found in 80% of patients. Not unexpectedly, adverse events were more common in the PFO closure group. Vascular complication rate was 3.2% in the PFO closure group, with none in the medical treatment arm. Incidence of atrial fibrillation was higher in the closure group as compared to the medical treatment group (5.7 vs 0.7%; P<.001). Prespecified subgroup analysis did not demonstrate any increased benefit from closure in subgroups such as patients with atrial septal aneurysm or substantial right-to-left shunt.

The second randomized trial that was completed and presented in October at the Transcatheter Cardiovascular Therapeutics 2012 conference in Miami is the RESPECT (Randomized evaluation of recurrent stroke comparing PFO closure to established current standard of care) trial. Publication of the data is pending.[2] This was a multicenter trial randomizing PFO closure with the Amplatzer PFO occluder versus medical management, which could be either therapy with antiplatelets or anticoagulants. The trial enrolled 980 patients with prior cryptogenic stroke and a PFO between the ages of 18 and 60 years. TIAs were not included. Primary endpoints were recurrence of stroke or death. Device implantation was attempted in 464 patients. Atrial septal aneurysms were present in one-third of patients. Procedural success was 96.1% and effective closure was achieved in 93.5% of patients. One important distinction from CLOSURE 1 is the very low incidence of procedural complications, with no device-related thrombus formation or device embolization. Major bleeding occurred in 1.6% and major vascular complications in 0.8% of cases. Median follow-up time was 2.2 years (range, 0–8.1 years). There were fewer patient dropouts in the device group (n = 48) as compared to the medical treatment group (n = 90). During follow-up, recurrent strokes occurred in 9 patients in the PFO closure group compared to 16 events in the medical treatment group. Of the 9 patients in the PFO closure group, 3 patients suffered a recurrent stroke following randomization, but prior to PFO closure. As a consequence, the primary endpoint was not reached in the intention-to-treat analysis (relative risk reduction, 46.6%; P=.157), while the "as-treated analysis" (classifying patients into treatment groups according to the treatment they actually received) was statistically significant (relative risk reduction, 72.7%; P=.007). Overall, recurrent event rate was low; at 5 years, recurrent strokes occurred in 2.21% of patients as compared to 6.4% in the medically treated patients. Important findings of the study included two groups of patients with PFO who benefited the most from PFO closure: patients with substantial shunt size (recurrent event rate 0.8% for PFO closure vs 4.3% for medical management; HR, 0.178; 95% CI, 0.039–0.813), as well as patients with atrial septal aneurysm (recurrent event rate 1.1% for PFO closure vs 5.3% for medical management; HR, 0.187; 95% CI, 0.04–0.867). The latter group was substantiated in other non-randomized studies.

Another randomized trial that was presented at the same meeting was the PC-Trial (Percutaneous closure of PFO versus medical management in patients with cryptogenic stroke). Again, the data were not published yet.[3] This European trial randomized 414 patients to the two treatment strategies. Patients had to be less than 60 years of age and strokes and TIAs were allowed as index events. The primary endpoints were a composite of death, stroke, TIA, or peripheral embolism. The trial documented no statistically significant benefit of PFO closure compared to medical management (HR, 0.63; 95% CI, 0.24–1.62; P=.34). Because of a low event rate of recurrent events, the study was underpowered, questioning the validity of subgroup analyses. A little less than a quarter of patients enrolled had an atrial septal aneurysm. Presence or absence of this septal abnormality did not influence treatment effect of PFO closure.

For comparison of two different treatment modalities, RCTs are considered the gold standard. Hence, CLOSURE 1, RESPECT, and the PC-Trial are the best data available to guide treatment recommendations for our patients. We all know the significant shortcomings of CLOSURE 1, and based on that we continued to enroll patients in other trials examining this issue (RESPECT and REDUCE). However, we were actually enthusiastic hearing the results of the RESPECT trial. Although the trial did not meet the primary endpoints, we believe with appropriate analysis of the data, something good may come out of this trial when the FDA panel and Circulatory Device System examine the data. If we analyze the data "as treated" and take off the 3 patients who sustained strokes after randomization to device but prior to them receiving the device and add them to the medical arm of the treatment, then no question, the primary endpoints would have been met.

So, it is all up to the FDA on how they will analyze the data. It will be unfair and perhaps unethical if the 3 patients who sustained strokes in the device arm prior to receiving the device were kept in that arm for the sake of statistical analysis. Furthermore, subgroup analysis of "patients with PFO and either atrial septal aneurysm or large shunt" clearly demonstrated the best benefit from device closure. The RESPECT trial added tremendous data to the field of PFO closure. Several limitations of CLOSURE 1 do not apply to this well-conducted study. Even though off-label PFO closure may have resulted in exclusion of the highest risk patients, the RESPECT trial, according to "as treated," documented a statistically significant reduction in recurrent stroke in patients undergoing PFO closure. Of utmost importance is also the documented safety and rate of complete closure of percutaneous PFO closure with the Amplatzer PFO occluder used in the RESPECT trial, which is clearly superior to the CLOSURE 1 data.

Several preliminary conclusions can be drawn that are also supported by the smaller PC-Trial, which failed its primary endpoint potentially because it was underpowered. First, recurrent event rates of stroke in patients with cryptogenic stroke and PFO are low. Second, the event rates seem to be lower and of less severity (on MRI) following percutaneous PFO closure as compared to medical management. Third, PFO closure with contemporary devices is very safe. Fourth, patients with atrial septal aneurysms coexisting with PFOs, identified as a high-risk subgroup in a prior landmark trial,[4] statistically have significantly fewer recurrent strokes when undergoing PFO closure.

So, we hope the FDA will look into the data favorably. Until then, the cardiology and neurology communities face a conundrum on how to treat and council these patients.