Blaming the PFO in Patients With Cryptogenic Ischaemic Stroke: Round 2

Mohamad Alkhouli; David R. Holmes


Eur Heart J. 2019;40(11):925-927. 

Patent foramen ovale (PFO) has been the subject of great interest and controversy, and has captivated the attention of specialists across multiple disciplines. The last 10 years (Round 1) have focused on assessing the relationship between PFOs and cryptogenic strokes in ambulatory patients with a history of stroke or transient ischaemic attack (TIA). The major morbidity and mortality associated with stroke, the availability of percutaneous closure technology, and the high prevalence of PFOs in the general population prompted a flurry of investigations in the field.[1–3] However, the large body of evidence that emerged from case control series, institutional registries, and the first three PFO randomized clinical trials (RCTs) often led to conflicting conclusions, discordant societal guidelines, and different practice patterns.[4–7] Invasive cardiologists, among others, embraced the concept of device closure for secondary stroke prevention at least in younger patients (<65 years), while neurologists remained doubtful of the cause–effect relationship and favoured medical treatment as a first-line therapy in most patients. Nonetheless, the last 2 years witnessed two tipping points for PFO closure. First, in October 2016, the Food and Drug Administration (FDA) approved the AMPLATZER PFO Occluder (St. Jude, Minneapolis, MN, USA), although the panel vote reflected unsettled debates, with an overwhelming majority approving the safety of the device (15–1) but only a slight majority approving of the device's efficacy (9–7). Secondly, three RCTs were published in September 2017, the results of which unequivocally suggested the superiority of device closure over medical therapy for secondary stroke prevention.[8–10] The FDA approval, along with these emerging positive data, led to an increasing utilization of PFO device closure, but also to a growing interest in multidisciplinary collaboration and additional investigations to identify the optimal candidates for the procedure.[11,12]

What did we learn from Round 1 of PFO closure for stroke prevention? (i) There is now compelling evidence that PFOs can be blamed for 'some' cryptogenic strokes, at least in a certain subset of patients. (ii) Anatomic features of PFOs vary significantly and the magnitude of benefit of PFO closure appears to be highest in patients with high-risk PFO anatomical features (e.g. those with large right to left shunt, or with significant septal aneurysm). (iii) Controversies remain regarding the optimal use of PFO closure devices, partially due to the strikingly high prevalence of PFOs, the relative infrequent incidence of strokes/TIAs that are felt to be cryptogenic, and the non-negligible risk of incident atrial fibrillation following PFO device closure.

We now move to Round 2 where the emerging science aims to confirm the findings of the RCTs, to ascertain the pathological impact of PFO in various clinical settings, and to guide optimal patient selection for device closure or medical therapy. In this issue of the European Heart Journal, Friedrich et al. bring to the forefront the topic of PFO-attributable ischaemic strokes in patients undergoing non-cardiac surgery (Take home figure).[13] The study cohort is derived from a retrospective institutional registry of 182 393 patients who were undergoing non-cardiac surgery with general anaesthesia at three hospitals in the USA between 2007 and 2015. The first analysis of this database compared peri-operative (30-day) ischaemic stroke rates between patients with and those without a diagnosis of PFO, and found that the estimated risks of stroke were 5.9/1000 patients with PFO and 2.2/1000 patients without PFO [adjusted odds ratio (aOR) 2.66; 95% confidence interval (CI) 1.96–3.63; P <0.001].[14] These findings provided a unique 'proof-of-concept' of the cause–effect relationship between PFO and strokes in a different cohort of patients from those who have been studied in RCTs and large registries.

Take Home Figure.

Potential association between patent foramen ovale and post-operative strokes in patients undergoing non-cardiac surgery. PFO, patent foramen ovale; RA, right atrium.

In the current study, the authors extended their previous 30-day analysis to (i) compare the long-term risk of ischaemic stroke after surgery between patients with and without a PFO and (ii) assess the impact of antithrombotic therapy on the incidence of post-operative stroke in patients with a PFO. The presence of a PFO was associated with a significant increase in post-operative acute ischaemic stroke at 1 year; 54 (4.7%) in patients with a PFO vs. 1588 (1.1%) in patients without a PFO (aOR 2.01; 95% CI 1.51–2.69; P < 0.001). This association persisted after multiple sensitivity analyses that (i) accounted for competing risks of death or loss to follow up; (ii) excluded strokes that occurred in the first 30 days; (iii) included only patients with >2-year follow-up; and (iv) excluded patients in whom echocardiographic evaluation was not performed. In terms of antithrombotic regimen, only combination antithrombotic therapy (dual antiplatelet agents or an antiplatelet and an anticoagulant) mitigated the risk of post-operative stroke attributable to PFO (OR 0.41; 95% CI 0.22–0.75; P for interaction = 0.004). Based on these data, the authors concluded that post-operative PFO patients are considered a 'high-risk phenotype', who are subject to increased short- and long-term risks of right to left ischaemic events via their PFOs following surgery. Two additional exploratory findings in this study supported the authors' conclusion: (i) patients who had prior PFO closure, albeit a small cohort (n = 131), had no increase in the post-operative ischaemic stroke rate within 1 year compared with patients without PFO (0.8% vs. 0.4%, P = 0.97) and (ii) PFO was associated with large-vessel territory infarct rather than lacunar strokes, a distribution similar to that seen in ambulatory patients with cryptogenic strokes studied in the early PFO literature.

The intriguing results of this study should be interpreted in the context of its important limitations despite the commendable efforts by the authors to address these limitations with multiple risk adjustments, and numerous sensitivity and exploratory analyses. (i) The PFO diagnosis was established by retrospective querying of medical records using administrative International Classification of Diseases 9 (ICD-9) and ICD-10 codes, and not by systematic survey with cardiac imaging modalities. These codes: (a) are not specific for PFO (also used to code atrial septal defects); (b) have low sensitivity [evident by the very low rate of PFO diagnosis overall (0.8%)]; and (c) are subject to screening bias (patients with post-operative strokes are more likely to undergo screening studies and hence receive a PFO diagnosis). (ii) Patients in the PFO group were older and had a wide range of higher risk features including hypertension, diabetes, hyperlipidaemia, smoking, atrial fibrillation, heart failure, and chronic kidney disease. In addition, they underwent more procedures that were considered higher risk or emergent compared with the no-PFO group. These higher risk baseline characteristics of the PFO patients, and the retrospective nature of this study, make the association between post-operative strokes (especially long-term ones) and the PFO problematic.

The findings of this study, although intriguing, raise several key issues that need to be explored in further studies.

  1. This analysis focuses on the risk of PFO-attributable stroke in patients undergoing non-cardiac surgery. Only one study examined the same issue in patients undergoing cardiac surgery, and found PFO not to be associated with an increased risk of post-operative ischaemic stroke.[15] However, given that strokes after cardiovascular operations constitute the vast majority of post-operative strokes, re-examination of the role of PFO in strokes following these procedures is warranted.

  2. Do we screen every patient undergoing non-cardiac surgery for the presence of a PFO, or can we identify high-risk procedures (e.g. orthopaedic surgery) and high-risk patients (e.g. those with a cancer diagnosis) in whom screening will probably be high-yield/beneficial?

  3. If such high-risk scenarios were identified, what would be the best management of the PFO: peri-operative venous duplex screening ± antithrombotic prophylaxis vs. pre-operative device closure?

  4. Single antiplatelet or anticoagulation therapy alone did not mitigate the risk of post-operative PFO-attributable strokes in this study, while a combined regimen did. Nonetheless, long-term combination therapy may be associated with significant risk of bleeding, non-compliance, and other issues, and hence studies testing various prophylactic medical regiments will also be needed.

  5. If the main proposed mechanism of cryptogenic stroke in PFO patients undergoing surgery is thrombo-embolism associated with debilitation and immobility following surgery, do we need to examine the risk of PFO-attributable strokes in other cohorts of non-surgical patients who share the same underlying substrates for this mechanism (e.g. patients with limited mobility, those with recurrent venous thrombo-embolism, or those with an established diagnosis of hypercoagulopathic state, etc.)?

The current investigation explores the pathological implications of PFOs in patients undergoing surgery with or without a prior stroke or TIA. Despite its provocative findings, this study leaves many issues unresolved. While randomized clinical trials will be needed to address these remaining issues, conducting such trials will be likely to face the same challenges that faced 'Round 1' RCTs, including the need for a large number of participants due to the low event rates and the large number of confounding variables, etc. In the interim, collaborative multicentre studies may aid in bridging the knowledge gap and addressing a potential unmet clinical need in preventing post-operative ischaemic strokes.