When results from a major trial with more than 900 patients failed to show that closure of a patent foramen ovale in patients who had a cryptogenic stroke worked better than medical therapy for preventing a future stroke, some experts suggested that the problem was patient selection. Researchers now have evidence that this explanation was right.

“Among patients with cryptogenic stroke, there is considerable variation in PFO [patent foramen ovale] prevalence, based on easily obtainable clinical characteristics.” Dr. David M. Kent said at the International Stroke Conference, which was sponsored by the American Heart Association. This wide variation suggests that the risk for cryptogenic stroke attributable to a PFO can vary from 0% to nearly 90%, said Dr. Kent, an internal medicine physician at Tufts Medical Center in Boston.

Mitchel L. Zoler/Elsevier Global Medical News

Dr. David M. Kent

“It’s very compelling that a patient has a stroke; you don’t know what caused it, and the patient has a hole in his heart and you have a safe way to close it. Do you close it? Of course, why not? The flaw is that the patient is 40 years old and had the hole in his heart for his entire life with no stroke. When a stroke then happens, maybe something else caused it,” Dr. Kent said in an interview.

His analysis showed that cryptogenic stroke patients who have a PFO often have one or more additional stroke risk factors – including diabetes, coronary artery disease, hypertension, smoking, and older age – that could as easily have been the culprit.

“There are cryptogenic stroke patients with a PFO who did not have paradoxical emboli; they had something else. The question is, can we identify the important PFOs? Can we find the PFOs that caused the first stroke? And, having found the PFOs that likely caused the first stroke, can we divide those patients into those with a high or low risk for having a second stroke? Not all PFOs are the same,” said Dr. David E. Thaler, a neurologist and director of the comprehensive stoke center at Tufts, as well as a collaborator with Dr. Kent on the ROPE (Risk of Paradoxical Embolism) project.

Results from the CLOSURE I (Evaluation of the STARFlex Septal Closure System in Patients With a Stroke or TIA [Transient Ischemic Attack] Due to the Possible Passage of a Clot of Unknown Origin Through a Patent Foramen Ovale) trial showcased the problem with presuming that when a patient has a cryptogenic stroke and also has a PFO the two are causally related. As Dr. Anthony J. Furlan reported at the American Heart Association scientific sessions in November 2010*, CLOSURE I randomized 909 patients who had a recent cryptogenic stroke or TIA as well as a PFO. Patients either received medical therapy with warfarin, aspirin, or a combination of both drugs, or underwent endovascular PFO closure using the device developed by the study’s sponsor. After 2 years of follow-up, the combined rate of new-onset stroke or TIA (the study’s primary end point) was 6% in the patients whose PFO had been closed and 8% in the control group, a difference that was not statistically significant. The study will be published in the New England Journal of Medicine on March 14. **

To explore what other stroke risk factors might exist in patients with a history of cryptogenic stroke, Dr. Kent and his associates pooled data for 3,023 cryptogenic strokes from 12 international data sets. All patients underwent assessment for a PFO, which existed in 1,274 patients and was not present in the remaining 1,749. About 30% of the patients were older than 65 years, and 59% were men.

The researchers analyzed all the patients by a list of other stroke risk factors, comparing the prevalence of each factor in the subgroups with and without PFOs.

They then performed a multivariate analysis, which identified six factors that each posed a significant, independent risk for a cryptogenic stroke: older age, diabetes, hypertension, current smoking, a history of a stroke or TIA, and a history of a deep stroke. The analysis showed that the presence of each of these risk factors reduced the likelihood by about 30% that the cryptogenic stroke resulting from the patient’s PFO.

Mitchel L. Zoler/Elsevier Global Medical News

Dr. David E. Thaler

Because the risk from each factor was roughly similar, Dr. Kent and his associates developed a point system to estimate the likelihood that a PFO caused a stroke in a cryptogenic stroke patient with a PFO. For example, in patients with none of these other risk factors, their PFO-attributable risk for their prior cryptogenic stroke was 88%, and their estimated 2-year risk for another stroke was about 5%. In contrast, among patients with all six of these risk factors, their PFO-attributable risk for their index cryptogenic stroke was 0%, and their estimated 2-year risk for having another stroke was 20%.

The researchers plan next to apply their new model for PFO-attributable stroke risk to the CLOSURE I data set. “We may be able to figure out why CLOSURE I didn’t work, and whether this approach could identify the patients who should have been in the trial” Dr. Thaler said. If the scoring method looks promising in the retrospective analysis, it might then be suitable for selecting patients to enroll into a new prospective study of PFO closure.

“The idea is that we’ll stratify patients in clinical studies, and – if all the stars align – we’ll be able to say that these are the patients who have the potential to benefit” from PFO closure, Dr. Kent said.

If these studies also find parameters that identify the types of PFOs that pose the greatest risk for causing strokes, the implications could extend beyond patients who already had a cryptogenic stroke.

“A quarter of the entire population has a PFO. Wouldn’t it be interesting if we could identify high-risk PFOs in the general population, among people who never had a stroke?” Dr. Thaler asked.

Dr. Kent said that he is a consultant to W.L. Gore & Associates. Dr. Thaler said that he is a consultant to W.L. Gore & Associates and AGA Medical.

*Correction, 3/12/2012: An earlier version of this article misstated the timing of the American Heart Association’s annual scientific sessions at which the data were presented.

**Correction, 3/12/012: An earlier version of this article incorrectly stated that the study had already been published.