Atrial fibrillation after patent foramen ovale closure in patients with embolic stroke of undetermined source

Despite diagnostic evaluation, 20–30% of embolic strokes remain unresolved being characterized as embolic strokes of undetermined source (ESUS).¹ For patients < 60 years, diagnostic evaluation includes search for relevant patent foramen ovale (PFO) as route for an embolus resulting in a stroke.² Several randomized clinical trials indicated reduced stroke recurrence after interventional PFO closure as opposed to medical therapy alone after ESUS.^3–5 Today, PFO closure is an established secondary stroke prevention strategy.

Undetected atrial fibrillation (AF) is another common underlying cause of ESUS.⁶ Importantly, the diagnosis of AF usually renders interventional PFO closure unnecessary as oral anticoagulation is considered to prevent stroke recurrence by both left atrial and crossed emboli. Hence, individual risk assessment for AF prior to interventional PFO closure is warranted, but particularly younger patients without overt AF risk factors are commonly considered for a prompt PFO closure in presence of a relevant shunt volume.

The occurrence of peri-interventional AF is a known consequence of PFO closure. In the pivotal PFO closure trials, the AF incidence ranged from 2–6%, but none of these studies performed long-term rhythm monitoring or systematic ECG-based follow-up.⁷ Such follow-up exists for general PFO closure cohorts only. Here, we first report results from a prospective, single-centre registry of ESUS patients who received both PFO closure and systematic rhythm monitoring using implantable cardiac monitors (ICMs).⁸ Importantly, continuous rhythm monitoring was available both before and beyond PFO closure, allowing to assess AF incidence peri-interventionally and during follow-up.

For the current analysis, we included patients ≥ 18 years who suffered an ESUS, were diagnosed with relevant PFO by transoesophageal echocardiography,^4,9 and had received PFO closure. Importantly, all patients were equipped with an ICM for continuous rhythm monitoring.

The primary clinical outcome was the occurrence of AF for a minimum of 30 s duration, diagnosed by ICM after PFO closure. The ICM remote monitoring system was routinely screened for AF episodes. Secondary analyses included duration of AF episodes, stroke recurrence, and antithrombotic regimes.

Between January 2018 and December 2021, we enrolled 604 patients with ESUS in our registry, of whom 91 patients (15.1%) presented with relevant PFO. Of these, 42 patients (46.2%) received PFO closure, of whom 35 patients (38.5%) with concomitant ICM monitoring were included into the final analysis. All included patients received a median duration of rhythm monitoring prior to PFO closure of 184 days (IQR: 98; 280 days) without documentation of AF.

We diagnosed 11 patients (31.4%) with AF after PFO closure during a mean follow-up of 1366 ± 410 days. Patients with AF were significantly older (55.3 ± 4.0 vs. 47.5 ± 11.8 years, P = 0.007) and had lower ROPE (Risk of Paradoxical Embolism) scores (5.0 (IQR: 4.5; 5.0) vs. 6.0 (IQR: 5.0; 7.0), P = 0.005), but did not differ in other parameters including cardiovascular risk factors, CHA₂DS₂-VASc score, or laboratory parameters like NT-pro B-type natriuretic peptide (Table 1).

All 11 patients with AF diagnosis showed at least one episode within the first 45 days of PFO closure. The median time to AF diagnosis was 13 days (IQR 10; 15.5), and patients with AF showed distinct episodes on a median of 6 days (IQR: 2.5; 7), cumulating to a median AF duration of 410 min (IQR: 88; 1847). Importantly, in four patients (11.4%), AF persisted beyond 45 days after PFO closure as revealed by long-term follow-up. Episodes after 45 days averaged 79 ± 105 minutes, and all were of paroxysmal character (Figure 1).

Once AF was diagnosed, 9 (81.8%) of 11 patients were switched to oral anticoagulation instead of antiplatelet therapy. Two patients opted to remain on acetylsalicylic acid. Two patients suffered cardioembolic stroke recurrence > 6 months after PFO closure, of whom one had also been diagnosed with post-closure AF.

The high incidence of AF by ICM monitoring after PFO closure warrants discussion as the occurrence of AF after PFO closure is not a new finding. The pivotal PFO closure studies reported an AF incidence of 2–6%.^3–5 Yet, no systematic rhythm follow-up was applied. Two other studies in general PFO closure patients relied on ICM monitoring and reported AF in 28.9%¹⁰ and 37%,¹¹ respectively. Importantly, ours is the first systematic ICM-based analysis in patients undergoing PFO closure for secondary stroke prevention post-ESUS for whom the diagnosis of AF results in an immediate therapeutic consequence.

A large meta-analysis suggested that AF risk was only increased in the first 45 days after PFO closure, postulating a mechanically and temporarily induced sub-entity of AF as opposed to idiopathic AF.^7,12,13 Another study by Skibsted et al.¹⁴ also showed an increased AF risk only during an extended peri-interventional period of 90 days. Our results confirm that all patients with incident AF experienced at least one episode of AF within the first 45 days, most even within the first 20 days. However, in more than one-third of patients, AF persisted beyond the periprocedural period, with AF documented up to 962 days after the intervention. It may thus be inferred that rhythm monitoring is particularly favourable during the initial 45 days as AF episodes recorded longer after the intervention occurred in patients with a periprocedural AF incidence.

Figure 1

Atrial fibrillation in patients after PFO closure. Central study details shown in upper panel. Antithrombotic regimens were at the discretion of the treating physicians; all patients on oral anticoagulation had been diagnosed with atrial fibrillation (AF). Lower panel illustrates time course of AF diagnoses for all individuals with AF after patent foramen ovale (PFO) closure. Days until PFO closure represent the implantable cardiac monitor (ICM) monitoring duration without an AF diagnosis prior to PFO closure. Every AF episode is presented by a dot in relation to the timeline after PFO closure. AF episodes 6 months after PFO closure are cumulated (#: AF on Day 261 after PFO closure, ##: AF on Days 388, 455, 650, and 962 after PFO closure). The total time in AF is presented in days, hours, and minutes for each individual. The total time of follow-up represents the duration of home monitoring in total (*until ICM explantation, **until ICM end of service, ***until ICM recommended replacement time).

Open in new tabDownload slide

AF detection begets the question of relevance of this finding. Previous results suggested that PFO closure-related AF episodes are usually short and self-limiting.¹² Indeed, our data revealed no case of persistent AF. However, episodes typically lasted several hours, hence not fulfilling the criteria of very short, potentially not therapeutically relevant episodes. The conclusion that this sub-entity of AF is clearly temporal and thus benign may hence be not valid. Different underlying pathophysiologies may result in this AF entity including undetected AF before PFO closure despite prolonged rhythm monitoring or induced AF due to haemodynamic and mechanical changes of PFO closure.

Most recently, the NOAH-AFNET 6 and ARTESIA trials suggested that anticoagulation in patients with pacemaker-detected, subclinical AF does not yield the same benefit compared to clinically diagnosed AF.^15,16 These novel results inform decision-making on the antithrombotic strategy. Yet, the immediate transferability to ESUS patients after PFO closure is limited, given that here AF may also be the cause of the initial stroke. Therefore, the necessity for oral anticoagulation in these patients should remain to be discussed most carefully once AF is documented. For consensus on the most appropriate antithrombotic regimen for AF in this specific group, additional research is warranted.

In conclusion, we demonstrate that incident AF after PFO closure constitutes a common clinical problem, which may not be restricted to a peri-interventional period. Of note, our results are based on a single-centre registry, which limit generalizability. Importantly, the implications of such an AF diagnosis in ESUS patients are insufficiently understood and require further investigation. This includes the identification of patients at an increased risk of AF in the first place. Subsequently, criteria need to be established for the appropriate duration and type of rhythm monitoring after PFO closure to identify those in whom AF persists and requires tailored secondary stroke prevention.

Funding

A.S.v.F. was supported by the German Research Foundation (413635475) and the Munich Clinician Scientist Program (MCSP) through the Clinician Scientist PRogram In Vascular MEdicine (PRIME).

Data availability

Full data can be obtained from the corresponding author upon reasonable request.

References

Author notes