Impact of disease stage on the performance of strain markers in the prediction of atrial fibrillation.

BACKGROUND: Assessing atrial fibrillation (AF) risk may be useful in primary prevention (PP; people with risk factors) and secondary prevention (SP; eg. embolic stroke of unknown source). We sought whether disease stage influenced the prediction of AF by echocardiography. METHODS: We compared a PP cohort (351 community-based participants ≥65 years with ≥1 risk factor for AF) and a SP cohort (453 patients after transient ischemic attack or stroke). LV global longitudinal strain (GLS) and left atrial reservoir strain (LARS) were measured from DICOM images. AF was diagnosed by 12 lead ECG, Holter or by single lead monitor over median follow-up of 22 months (PP) and 35 months (SP). The clinical and echocardiographic characteristics of those with AF were compared to those in sinus rhythm. Nested Cox-regression models assessed for independent and incremental predictive value of LARS and GLS in both cohorts. RESULTS: AF developed in 42 PP (12%) and 60 SP (13%), and was associated with age, higher CHARGE-AF score, increased LA volume and LV mass (p < 0.05). Patients developing AF had reduced GLS (17 ± 3.5% vs. 20 ± 3%, p < 0.001) and LARS (28 ± 11% vs. 35 ± 8%, p < 0.001). However, the predictive value of both GLS (area under the ROC curve 0.83 vs 0.56, p < 0.001) and LARS (0.83 vs 0.57, p < 0.001) was greater in SP than PP. LARS was independently associated with AF in both cohorts (p < 0.05), but GLS was only independently associated in the SP cohort. CONCLUSION: AF risk assessment with LARS is suitable for different risk cohorts, but GLS is more useful in SP.