Localizing myocardial scar on echocardiography. How good does it work in the presence of conduction delays?

Abstract

Abstract Funding Acknowledgements Type of funding sources: Public grant(s) – National budget only. Main funding source(s): Research Foundation - Flanders (FWO) Introduction Myocardial scar detection with echocardiography in patients with ischemic heart disease typically relies on semi-quantitative evaluation of regional systolic wall thickening. In patients scheduled for cardiac resynchronization therapy (CRT) however, such echo scar estimation is complicated by the presence of dyssynchronous contraction and differential regional remodelling. Visual assessment of myocardial shortening during systole may be an alternative approach. We tested this against cardiac magnetic resonance (CMR) with late gadolinium enhancement (LGE) in patients without and with conduction delay. Methods 122 patients with ischemic heart disease were included (n = 58 without, and n = 64 with conduction delay). Scar burden of the LV was determined in all patients on a segmental level in both CMR and echo. Reading of echo was blinded for CMR data and vice versa. Myocardial scar was defined as LGE > 50% of transmural thickness. On echo, scar was assessed visually, and defined as thin, echogenic myocardium with no visible shortening during systole. Analysis was performed per segment (18 segment model), and per region (6 walls with basal and mid segment and the apex region consisting of all apical segments). An additional analysis was performed with a tolerance of one adjacent segment in order to account for potential image misalignment between modalities. Results 2196 segments were available for comparison between echo and CMR. On CMR, 548 of those segments were defined as having >50% transmural scar. In echo, 565 segments were detected as having scar. On a segmental level, no difference was found for the correct assignment of segments by echo as having scar or not between patients without or with conduction delay (AUC 0.79 vs. 0.79; p = 0.968) (Figure, top panels). See Figure for sensitivity and specificity. If one segment tolerance was allowed, segments were correctly assigned with equal accuracy in both patient groups (AUC 0.98 vs. 0.96; p = 0.999) (see Figure; w. tolerance). Agreement on the level of LV regions was comparable. 295 regions had a scar on CMR while 286 regions were identified by echo. Echo correctly identified a scar in the same LV wall or apex as compared to CMR similarly in patients without or with conduction delay (AUC 0.79 vs. 0.77; p = 0.698). If one segment tolerance was allowed, correct identification improved further and was not different between both groups (AUC 0.93 vs. 0.91; p = 0.999). The extent of a scar was slightly underestimated (9%) by echocardiography in comparison to CMR in patients without, and slightly overestimated (3%) in patients with conduction delays. Conclusions Scars can be localized on echocardiography with good agreement to CMR-LGE as gold standard. The match between echo and CMR was similar for patients with and without conduction delay. Our findings demonstrate that echo can provide a valid impression of localization and extent of myocardial scar, even in the presence of conduction delays. Abstract Figure.