Aetiology and classification of ischaemic stroke with positron emission tomography thrombus imaging

Background Ascertaining the role and origin of thromboembolism in ischaemic stroke remains challenging in clinical practice because of the absence of robust imaging to identify the presence of in vivo thrombosis. Flourine-18 GP1 ([18F]GP1) radiotracer is highly specific for the activated glycoprotein IIb/IIIa receptor expressed on activated platelets and has high sensitivity for detecting in vivo thrombosis. We aimed to identify the presence of thrombosis in patients with ischaemic stroke to inform the aetiology and classification of stroke. Methods In a single centre prospective cross-sectional observational study, patients with acute ischaemic stroke underwent ambulatory electrocardiogram, bubble-contrast echocardiography and [18F]GP1 positron emission tomography and computed tomography (PET/CT) angiography within 21 days of symptom onset. [18F]GP1 uptake was qualitatively and quantitively assessed in the heart, brain and systemic vasculature including intracranial, aortic, carotid, vertebral arteries and pelvic and lower limb veins. An expert stroke panel classified stroke with the Causative Classification System with and without (18F)GP1 PET/CT results. Results Seventy patients with acute ischaemic stroke (28 stroke of undetermined aetiology, 24 cardioembolic, 9 small artery occlusion, 7 large artery atherosclerosis and two other causes) participated. Forty-four had at least one source of [18F]GP1 uptake: 22 carotid, 7 vertebral, 4 aorta, 3 left atrium, 2 left ventricle, 2 lower limb deep veins, 8 native valves (4 aortic and 4 mitral) and two prosthetic valves. [18F]GP1 uptake led to stroke reclassification in 19 (27%) patients which included 14 out of 28 (50%) cases of stroke of undetermined aetiology. [18F]GP1 uptake was identified on native valves (4 aortic and 4 mitral valves) including one case of endocarditis and one bicuspid aortic valve with moderate stenosis. The remaining cases of valvular [18F]GP1 uptake occurred on thickened and/or calcified valves on echocardiography or CT angiography (Figure 1). Only 3/6 patients with atrial fibrillation and presumed cardioembolism had [18F]GP1 uptake in the atrial appendage. There were two cases of paradoxical embolism from a deep vein thrombosis through a patent foramen ovale (Figure 2). Conclusion Hybrid [18F]GP1 PET/CT imaging in patients with acute ischaemic stroke identified the source of thrombosis across stroke subgroups and reclassified a large proportion of cases, especially in those with an undetermined aetiology. Importantly, the role of native mitral and aortic valve thrombosis appears to have been underappreciated, and conversely, the role of left atrial appendage thrombosis in atrial fibrillation appears to have been over emphasised. [18F]GP1 positron emission tomography holds major promise in exploring the role and origin of thrombosis in stroke that could help guide management including the use of anti-thrombotic therapies and patent foramen ovale closure devices.