CMR-derived left atrial strain predicts incident atrial fibrillation in transthyretin cardiac amyloidosis
P Lustig, L Schmid, M Poledniczek, C Kronberger, J Lee, N Ermolaev, C Binder, M Sponder, R Schoenbauer, A Kammerlander, C Hengstenberg, F Duca, J Bergler-Klein, S StojkovicAbstract
Background
Patients with cardiac amyloidosis (ATTR-CM) are at high risk of developing atrial fibrillation (AF) which often precipitates hemodynamic deterioration and heart failure decompensation. Left atrial (LA) strain parameters derived from cardiac MRI (CMR) reflect atrial mechanical function and may aid in identifying patients at-risk before clinical AF onset.
Purpose
To evaluate the prognostic value of CMR-derived LA strain parameters for the development of incident AF in patients with ATTR-CM.
Methods
We retrospectively analyzed a total of 86 patients with ATTR-CM and no prior AF (median age 77 years, 80% male). All patients underwent baseline CMR with feature-tracking–based LA strain analysis. LA peak global longitudinal strain, reservoir strain, conduit strain, booster strain, and peak global radial strain were quantified. Incident AF during follow-up was assessed. Kaplan–Meier survival analyses compared AF-free survival between the highest respective strain quartile (Q1, best strain) and the combined lower three quartiles (Q2-Q4). Univariable and multivariable Cox regression models evaluated the association between LA strain and AF onset. Receiver-operating characteristic (ROC) analyses assessed discriminative performance.
Results
During a median follow-up of 26.6 months, 37 (43%) developed AF. Kaplan–Meier curves demonstrated a markedly lower AF-free survival in patients with reduced LA strain (Q2-Q4) compared with better strain values (Q1) across all strain parameters (all log-rank p<0.006). In contrast, LA volume alone showed a non-significant trend toward predicting AF, emerging only beyond two years of follow-up (p = 0.087). In univariable Cox regression, all LA strain parameters predicted incident AF (all p < 0.01), with the strongest association for LA conduit (HR 0.715, 95% CI 0.596–0.856) and reservoir strain (HR 0.829, 95% CI 0.749–0.918). In multivariable models adjusted for age, sex, NT-proBNP and LA volume, each strain parameter remained an independent predictor of AF. The strongest associations were observed for LA reservoir strain (HR 0.83, 95% CI 0.73–0.93; p = 0.002) and LA conduit strain (HR 0.74, 95% CI 0.60–0.90; p = 0.003). ROC analysis confirmed significant discriminative ability across all strain parameters (p ≤ 0.001) with LA conduit (AUC 0.724) and reservoir strain (AUC 0.722) performing best.
Conclusion
CMR-derived LA strain parameters independently predict incident AF in ATTR-CM. Impairment of LA reservoir and conduit strain demonstrated the highest predictive value, indicating that LA mechanical dysfunction precedes arrhythmia onset. Incorporating LA strain assessment thus may enhance risk stratification and guide early AF surveillance strategies in patients with ATTR-CM.AF-free survival by LA StrainROC analysis LA strain