Background: Paroxysmal atrial fibrillation (PAF) is often asymptomatic and is more prevalent than persistent AF in patients with ischemic stroke (IS). Long-term monitoring with insertable cardiac monitors (ICM) yields an up to 30% PAF detection rate in cryptogenic IS patients. Yet, the prevalence of PAF in patients with a presumed stroke etiology other than AF remains unknown.
Methods: Prospective cohort study of non-cryptogenic IS patients implanted within 10 days of stroke onset with the Reveal LINQ ICM (n=47 enrolled, 45/47 (95.7%) implanted). Patients were monitored until PAF detection (adjudicated by study cardiologist) or minimum of 12 months. Inclusion required a defined stroke etiology other than AF based on standard stroke evaluation (including ≥24 hour cardiac telemetry), age≥40, and either a virtual CHADS2 score ≥3, or ≥2 PAF related comorbidities (COPD, hyperthyroidism, obesity, prior MI, PR interval >175ms, or renal impairment). Patients with high-risk cardiac sources, recent MI or cardiac bypass surgery, pacemaker or defibrillator, permanent anticoagulation indication or contraindication, and pregnancy were excluded.
Results: Mean age 64.8 years (range 40-88 years), 15/47 (31.9%) female, mean virtual CHADS2 score 3.5 (range 3 to 5), mean CHA2DS2-VASc score 4.5 (range 3 to 8), and 8/47 (17.0%) had ≥2 PAF comorbidities. Lacunar stroke etiology predominated (n=33), followed by extracranial atherosclerotic stenosis (n=4), cardioembolic (n=3), arch atheroembolic (n=3), intracranial atherosclerosis (n=4), and extracranial dissection (n=2). Five had multiple potential sources. Interim analysis (mean monitoring 264 days, range 0 to 642 days) demonstrates 3/45 (6.7%) implanted patients have PAF (range of longest AF episode: 0.3 to 18.7 hours). Mean time to PAF diagnosis 162.7 days (range 9-356 days). All were started on anticoagulation. There was 1 device related complication (skin erosion) and 1 patient had recurrent stroke.
Conclusions: Occult PAF in non-cryptogenic IS patients is infrequent and may reflect the background prevalence of PAF in this demographic. Despite this, long-term cardiac monitoring may promote optimization of stroke prevention therapy for a small but unanticipated proportion of these patients.