Left atrial vorticity is independently associated with embolic brain infarcts and represents a promising imaging biomarker of cardioembolism in sinus rhythm and atrial fibrillation

Abstract

Abstract Background Atrial fibrillation (AF) is associated with embolic stroke, but risk scores such as CHA2DS2-VASc perform only modestly (C statistics 0.6–0.7). Meanwhile, up to 25% of embolic strokes in patients without AF have no identifiable cause, and occult left atrial (LA) thromboembolism may be a relevant mechanism in such cases. Purpose We hypothesised that imaging of left atrial blood flow could improve embolic risk prediction in patients with and without AF. We used 4D flow magnetic resonance imaging (MRI) to identify a biomarker that is: (a) independently associated with embolic brain infarction by brain MRI, (b) able to stratify blood flow characteristics both during AF and in sinus rhythm (SR), and (c) reproducible. Methods We recruited 3 patient cohorts to respectively address each aim. Firstly, to assess the association between LA flow parameters and embolic brain infarcts, we recruited cohort A, consisting of 134 patients (41% female; age 70±9 years) with a history of ischaemic stroke (N=44) or no history of stroke but with CHA2DS2VASc score ≥1 (N=90). Next, the sensitivity of 4D flow parameters to rhythm change was assessed in cohort B: 37 patients with persistent AF studied before and after cardioversion, whose results were compared with those of 23 healthy controls in SR [CHA2DS2-VASc = 0.0 (0.0–0.0)]. Finally, scan-rescan coefficients of variation (CV) and interval-scan CV at 30 days were determined in Cohort C (86 subjects; 64 in SR, 22 in AF). Brain MRI was used to identify large non-cortical or cortical brain infarcts (LNCCI) – i.e. infarcts likely to be embolic in origin. Results At least one LNCCI was present in 39 of 134 patients in cohort A. Lower LA vorticity was significantly associated with higher risk of prevalent LNCCIs (Figure 1), after adjustment for AF, age, and CHA2DS2VASc score [OR=2.10 (95% CI 1.12–3.92) per SD, P=0.02]. This association remained significant after further adjustment for other cardiac parameters (all P<0.05, Figure 1). By contrast, there was no significant association between peak velocity and LNCCIs (P=0.21). LA vorticity was sensitive to rhythm change, improving significantly in patients in cohort B in SR at ≥4 weeks after cardioversion (CV) of persistent AF (Figure 2A, paired P<0.001 vs pre-CV), but remained impaired compared to healthy controls (Figure 2B, P<0.01). Finally, reproducibility studies in cohort C showed that LA vorticity had a same-day scan-rescan CV of 7% without significant differences between SR and AF subjects (P>0.05), and also showed no significant temporal variability on interval scanning (P>0.05). Conclusions LA vorticity is reproducible, sensitive to changes in heart rhythm, and independently associated with embolic brain infarcts, suggesting a promising imaging biomarker of cardioembolism in SR and AF. LA blood flow imaging could improve stroke prediction and the personalisation of decisions about anticoagulation, regardless of heart rhythm. Funding Acknowledgement Type of funding sources: Public grant(s) – National budget only. Main funding source(s): Oxford BRC, BHF