MIAMI – The Randomized Evaluation of Recurrent Stroke Comparing PFO Closure to Established Current Standard of Care Treatment (RESPECT) and PC Trial were both presented last week at TCT 2012 after years of anticipation. Depending on one's point of view, the studies fail to support the continued treatment of patent foramen ovale (PFO) with the Amplatzer occluder (St Jude Medical, St Paul, MN) for the prevention of stroke recurrence or provide yet more evidence that closing PFOs is the right thing to do.

The conflict arises from the fact that both randomized, controlled clinical trials missed their primary end point. As reported by heartwire , the studies were negative, with RESPECT suggesting a benefit when evaluating the "as-treated" patients only, and this has left researchers, clinicians, and other experts trying to understand what went wrong and the potential reasons the trials don't line up with published observational reports. The only other randomized, controlled study to date, the CLOSURE I study, was also a negative trial.

Dr David Thaler (Tufts Medical Center, Boston, MA), one of the investigators of the RESPECT trial, agreed that the study did not meet the threshold of statistical significance but said that isn't the whole picture.

For the purists, it is what it is, and you can't get away from that.

"For the purists, it is what it is, and you can't get away from that," said Thaler. "However, the story is more complicated, because several people in the intention-to-treat analysis who had clinical outcomes never received the device. So, do we take that into consideration when we look at the totality of the evidence, or do we completely ignore any nuances about the data? I think more people are going to go away from these two trials with their prior opinions more fixed than there will be people whose minds are changed, although there will be some of that."

Dr Eric Horlick (University Health Network, Toronto, ON) points out that while the trials were small, with 980 patients in RESPECT and 414 patients in the PC Trial, they are the equivalent of a 5000-patient trial of patients with acute MI. "You have to be respectful, if you will, of the amount of energy and the amount of quality-adjudicated data," he told heartwire . "In the PFO world, adjudicated end points, where an events committee reviews what we call a stroke and where a core lab looks at the echo, these types of data are of the highest quality."

Horlick notes that three patients randomized to receive the Amplatzer PFO occluder didn't receive the device and had a clinical event. "So from the intention-to-treat [perspective], if you're a very religious clinical trialist, the headline is that this is a negative trial," he said. "But in this particular area, you just can't afford to throw the baby out with the bathwater, because there is so much high-quality information here. To me, when I look at the as-treated group, which is really reflective of what we do every day, these data are much more telling."

Taking a much more cautious interpretation, however, Dr S Claiborne Johnston (University of California, San Francisco) said that while he believes there is likely a subset of patients in whom PFO closure works, this still hasn't been proven in a randomized, controlled clinical trial.

"We don't have a positive trial, and it's a shame that we don't, but there is a reason that we set the p value that we do," he said. "There is a reason we don't trust as-treated and per-protocol analyses. They are inherently observational and biased. It doesn't make sense for us to throw those standards aside because it just feels like this thing should work. If this were a drug, there would be absolutely no question that it shouldn't be used. There is something about our acceptance of devices--maybe because it seems obvious about how they work--that we're more willing to tolerate imperfect evidence."

For Dr Jonathan Tobis (University of California, Los Angeles), himself a proponent of PFO closure, the patient is more likely to care about the as-treated analysis than the intention-to-treat analysis, but he suspects that the PC Trial and RESPECT are unlikely to alter the number of physicians performing the procedures each year.

"I think it's going to be status quo," he told heartwire . "I think the people who are against closure are going to find evidence in the intention-to-treat analysis and say the study didn't reach its primary end point, end of story. Those who are supportive of PFO closure, as I am, will find significant positive information from this study and believe that it's justified to close PFOs."

If this were a drug, there would be absolutely no question that it shouldn't be used.

As reported previously, in RESPECT, the 46.6% reduction in the risk of stroke was not statistically significant when assessed in the intention-to-treat analysis. However, when the investigators analyzed the data among patients treated per protocol, the 63.4% reduction in stroke was statistically significant, as was the 72.7% reduction in stroke when analyzed by patients who actually received the device.

In the PC Trial, closing the PFO failed to reduce the primary end point of death from any cause, nonfatal stroke, transient ischemic attack, and peripheral embolism. There was an 80% reduction in the risk of stroke, but this reduction did not reach statistical significance.

Fragile Data

Speaking with heartwire , Dr Sanjay Kaul (Cedars-Sinai Medical Center, Los Angeles, CA) said that for the big picture, especially with regard to RESPECT, the evidence does not end with the p value. Although he was not involved in the studies, Kaul participated in a "deep-dive" session highlighting the trials at TCT 2012 and expressed his concerns about the evidence base. Tongue-in-cheek, Kaul said that when asked if the results from the PC Trial and RESPECT moved the needle in terms of his support for PFO closure to prevent stroke recurrence, he said they did: "From not, to maybe not."

That said, Kaul acknowledges the complexities of the trials.

"Both trials missed their primary end point, but that would be a rather uncharitable interpretation, because you have such low event rates--it took nine years to accrue 25 events--and just one more event here and one more event there make all the difference. In fact, I did a back-of-the-envelope calculation and showed that if you had one fewer event in the device arm and one more event in the control arm, [RESPECT] would have reached the primary end point."

Johnston said that practitioners and researchers performing PFO closures have a personal bias, in that many of them believe the device works and prevents future strokes. "It's a nice, easy procedure that reimburses well," said Johnston. "That tends to drive decision making and the behavior of physicians, unfortunately. My take on this is that it might work. The benefits are going to be small, and I suspect there are going to be populations in whom this is particularly useful, but I haven't seen that clearly defined yet."

Just one more event here and one more event there make all the difference.

To heartwire , Kaul said the randomized, controlled trial data are "fragile," meaning one or two events in the treatment arms can swing the results from negative to positive and vice versa, whereas the published observational data are highly positive. Such observational studies, however, are also significantly flawed with confounding, ascertainment biases, and publication biases. It remains difficult to collect more evidence, including in the ongoing REDUCE study, given the small number of clinical events.

"We have the data that we have, and there's not a lot we can do with it," said Kaul. "From the science-of-medicine perspective, there is insufficient evidence to adjudicate efficacy for PFO closure for cryptogenic stroke, despite the fact that it's relatively safe. From an art-of-medicine perspective, however, you could argue that one could consider this in high-risk patients." Such patients include those with large shunts and atrial septal aneurysm without conventional risk factors for stroke and/or who are unwilling to take lifelong oral anticoagulant therapy.

"I personally think that there are circumstances where it would be reasonable to recommend such an intervention, but the strength of the evidence is not definitive," Kaul told heartwire . "It's not conclusive."

But I would be very happy to refer them into a randomized trial that has a realistic possibility of getting done in the next five years or so.

For Johnston, he said there is a need for more clinical-trial data, despite the inherent difficulties the field has faced enrolling patients. "I think we need a trial of adequate power," said Johnston. "The field has to commit to it. As a stroke neurologist, I am not comfortable, unless I'm really convinced that I should be referring patients for PFO closure at this point. But I would be very happy to refer them into a randomized trial that has a realistic possibility of getting done in the next five years or so."

A Heterogenous Patient Population

To heartwire , Thaler said the PC Trial and RESPECT highlight the importance of understanding cryptogenic-stroke patients as a heterogenous population. Despite having a PFO and a cryptogenic stroke, other mechanisms might be at play, and the PFO might be incidental and unrelated to the initial stroke. In addition, within the subset of patients with PFO-attributable cryptogenic stroke, the recurrence rate of stroke is very low, said Thaler.

"It's not zero, and it's higher than anybody would want--it's somewhere between 1% and 2% per year," he said. "If you're 46 years old, which is what the population was in our study, then that's 40 years of 1% or 2% per year of risk. The question then becomes how to ameliorate that risk as safely as possible. The PC Trial and the RESPECT trial showed us that stroke outcomes are reduced in people who get the device, and depending on how carefully and rigorously you apply the rules of clinical trials, you will be more or less convinced by that reduction."

Horlick and Tobis both expressed similar concerns, telling heartwire that patients in the medical-therapy arm of RESPECT continued to have recurrent events, not just in the first six months but over time. This highlights the importance of long-term data. "What we're really missing in this group of people is 10- to 20-year follow-up," said Horlick. "What happens to someone who has had a stroke, is 20 years old, and has a PFO and an atrial septal aneurysm? I don't care what happens to them in two years. They are going to be my patient for 20 years."

I don't care what happens to them in two years. They are going to be my patient for 20 years.

For Thaler, he believes that PFO closure is a useful treatment as long as patients are properly selected. RESPECT had a strict set of inclusion criteria, and he worries that even if approved by the Food and Drug Administration (FDA), which is not a given, physicians would be less rigorous in selecting and treating patients with PFOs who had a stroke. He cited the creation of a set of criteria for identifying PFO-related stroke in the ROPE study, such as the absence of vascular risk factors like hypertension, diabetes, smoking, and older age. In addition, Thaler said the PFO-related stroke needs to be confirmed by a neurologist to ascertain that no other possible mechanisms are at play.

Horlick agreed. "I know that when one of my stroke neurologists sees somebody, they are going to be looked at with a microscope and turned upside down," he said. "If there is anything that looks to them like it might be the cause of that stroke, they are not getting sent my way. That's very important."

Rampant Off-Label Use of PFO Closures

To heartwire , Kaul noted that patients are routinely treated with PFO-closure devices. With more than 8000 PFO-closure devices implanted each year, he believes the sheer number of closures is not justified by the evidence, and nothing can stop physicians from implanting PFO-closure devices off-label. With more than 80 000 patients having received the device, the 2000-plus patients studied in three clinical trials pales in comparison. Experts have previously lamented that off-label use has compromised the ability for PFO-closure devices to be studied in clinical trials.

In one US study, investigators reported a 50-fold increase in the number of percutaneous PFO/atrial septal defect closures between 1998 and 2004. Some physicians have questioned whether it is even ethical to enroll patients in a randomized trial of PFO closure vs medical therapy because there is no longer clinical equipoise. Both Kaul and Johnston disagree with such an interpretation of the data.

I think physicians need to curb their enthusiasm.

"I think physicians need to curb their enthusiasm," Kaul said. "Patients need to be a little bit more skeptical, and perhaps the payers need to tighten their purse strings and not reimburse it based on some willy-nilly indication. The sponsors also need to be responsible to ensure that more patients are studied under rigorously controlled conditions."

The number of PFO procedures performed in the US might slow if guideline writers take a strong opposing stand or if the FDA declines to approve the Amplatzer PFO occluder. "The other way to do this would be if insurance stopped reimbursing PFO closure for secondary stroke prevention, except in the case of very specific instances," said Johnston. "This would dramatically reduce the rates outside clinical trials."

Dr John Carroll (University of Colorado, Denver), the lead investigator of RESPECT, said that all concerned and interested parties need to wait and not evaluate the results of the studies based on 12-minute presentations. He believes the rigorous FDA review that will be part of moving the device from its current investigational status to a potential approval will be helpful in fully assessing benefit and safety. Currently, the PFO-closure field has had 10 years of false starts--including humanitarian device exemption (HDE) approvals that were withdrawn--and an ongoing period of off-label use with little regulatory control.

"RESPECT studied a device that remains investigative in the US, and we have had clear demonstration that the kind of device is a determinant of outcomes," Carroll told heartwire . "We have had a decade of off-label closure of PFOs using numerous devices not designed for PFOs, not rigorously tested as a secondary-prevention strategy, and with no collection of data in thousands of patients to assess outcomes or move the field forward. Long term, this is unacceptable for our patients, who want the best recommendations we can provide to optimize their chances of never having another stroke."

The PC Trial and RESPECT studies were sponsored by AGA Medical/St Jude Medical. Carroll and Thaler were on the RESPECT steering committee, Horlick was an investigator in the CLOSURE I and RESPECT studies, and Tobis was an investigator in RESPECT. Kaul and Johnston report no conflicts of interest related to PFO-closure-device therapy.