Validation of overestimation ratio and TL-SVS as imaging biomarker of cardioembolic stroke and time from onset to MRI.

Abstract

We aimed to determine in the "THRACE" trial, the clinical and MRI technical parameters associated with the two-layered susceptibility vessel sign (TL-SVS) and the overestimation ratio (overR). Patients with pre-treatment brain gradient echo (GRE) sequence and an etiological work-up were identified. Two readers reviewed TL-SVS, i.e., a SVS with a linear low-intense signal core surrounded by a higher intensity and measured the overR as the width of SVS divided by the width of the artery. Binomial and ordinal logistic regression respectively tested the association between TL-SVS and quartiles of overR with patient characteristics, cardioembolic stroke (CES), time from onset to imaging, and GRE sequence parameters (inter slice gap, slice thickness, echo time, flip angle, voxel size, and field strength). Among 258 included patients, 102 patients were examined by 3 Tesla MRI and 156 by 1.5 Tesla MRI. Intra- and inter-reader agreements for quartiles of overR and TL-SVS were good to excellent. The median overR was 1.59 (IQR, 1.30 to 1.86). TL-SVS was present in 101 patients (39.2%, 95%CI, 33.1 to 45.1%). In multivariate analysis, only CES was associated with overR quartiles (OR, 1.83; 95%CI, 1.11 to 2.99), and every 60min increase from onset to MRI time was associated with TL-SVS (OR, 1.72; 95%CI, 1.10 to 2.67). MRI technical parameters were statistically associated with neither overR nor TL-SVS. Independent of GRE sequence parameters, an increased overR was associated to CES, while the TL-SVS is independently related to a longer time from onset to MRI. • An imaging biomarker would be useful to predict the etiology of stroke in order to adapt secondary prevention of stroke. • The two-layered susceptibility vessel sign and the overestimation ratio are paramagnetic effect derived markers that vary according to the MRI machines and sequence parameters. • Independent of sequence parameters, an increased overestimation ratio was associated to cardioembolic stroke, while the two-layered susceptibility vessel sign is independently related to a longer time from onset to MRI.