Modality-dependent variability in left atrial strain between echocardiography and cardiac MRI after pulsed field ablation

Background

Left atrial (LA) strain is central to assessing mechanical recovery after pulsed field ablation (PFA), yet the agreement between echocardiography and cardiac magnetic resonance imaging (MRI) for its quantification is unestablished.

Methods

We prospectively enrolled 27 PFA patients undergoing echocardiography and MRI at baseline and 3 months. LA reservoir, conduit, and contractile strain were quantified by speckle-tracking echocardiography and MRI feature-tracking. Paired tests compared two modalities, Spearman’s correlation assessed their association, and agreement for each strain component was evaluated by intraclass correlation coefficients (ICC) and Bland–Altman analysis.

Results

Baseline echo-derived reservoir strain was significantly higher than MRI-derived values (Mean Diff. 4.09%, p<0.001), and remained elevated at 3-month follow-up (3.63%, p=0.009). An initially moderate-to-strong correlation (r = 0.66, p<0.001) almost disappeared at follow-up (r = 0.05, p = 0.82), paralleled by a substantial decline in ICC from 0.435 (95% CI: −0.023 to 0.724) to 0.031 (95% CI: −0.256 to 0.370). Bland–Altman analysis showed a baseline bias of −4.09% (LoA −12.91% to +4.74%), widening to −15.54% to +8.28% at 3 months, with significant proportional bias at both baseline (β = −0.606, p = 0.001) and follow-up (β = −1.103, p = 0.004).

For conduit strain, MRI values were slightly higher at baseline and follow-up but don’t have significant differences (P>0.05). Correlation declined from r=0.58 (p=0.001) to r=0.31 (p=0.15), with ICC decreasing from 0.523 (95% CI: 0.195, 0.748) to 0.141 (95% CI: -0.243, 0.503). Bland–Altman biases and LoAs were 0.80% (−5.32% to +6.93%) at baseline and 1.71% (−8.13% to +11.55%) at follow-up, with significant proportional bias at both timepoints (β=0.592, p=0.004 baseline; β=1.119, p=0.001 follow-up).

Contractile strain measurements from MRI were consistently higher than Echo at baseline (Mean Diff. 2.91%, p<0.001) and follow-up (1.63%, p=0.004). In contrast to other components, correlation improved slightly (r=0.55 to r=0.58), with ICC rising from 0.437 (95% CI: −0.036, 0.730) to 0.478 (95% CI: 0.068, 0.745). Bland–Altman biases decreased from 2.91% (LoA −3.07% to +8.90%) to 1.63% (LoA −3.17% to +6.44%), with no proportional bias (p>0.05).

Conclusion

PFA induces modality-dependent discrepancies in left atrial strain. Reservoir and conduit strains showed loss of inter-modality correlation and widening agreement limits after ablation, whereas contractile strain demonstrated modest convergence between echocardiography and MRI. These results indicate PFA alters electromechanical coupling in ways differentially captured by each modality, while the improved consistency in contractile strain may reflect a regularization of atrial contraction. Overall, these findings challenge the assumption that imaging modalities are interchangeable in PFA outcome evaluation and highlight the need for modality-specific interpretation.