Introduction: Relative apical longitudinal strain (RALS) is defined as (average apical LS/(average basal & mid-ventricular LS)). A threshold of 2 has been found to have high sensitivity and specificity for differentiating cardiac amyloidosis (CA) from other causes of left ventricular hypertrophy (LVH). This threshold was developed using General Electric (GE) software, and its reproducibility among different software vendors is unknown. Hypothesis: In patients with CA, regional segmental LS patterns and relative apical longitudinal strain will vary among software vendors. Methods: Speckle-tracking echocardiography was retroactively performed by an experienced technician on two patient cohorts, CA (n=52) and LVH (n=52), using software from two independent vendors: EchoPAC (GE Medical Systems) and TomTEC (TOMTEC Imaging Systems GMBH). For each vendor and patient, strain values for the basal, mid, and apical segments were averaged to obtain three regional LS values which were then used to calculate global longitudinal strain (GLS) and RALS. Results: EchoPAC demonstrated greater average apical LS (-16.5±5.7 vs -13.1±6.6, p<0.001) and RALS (2.1±0.9 vs 1.7±0.7, p<0.001) compared to TomTEC. Bland-Altman analysis yielded a mean bias of -0.4 with limit of agreement 2.2 (p<0.001) in RALS between the two vendors. ROC curve analysis using a RALS cutoff of 2 to differentiate CA from the overall control group showed similarly high specificity (EchoPAC 85%, TomTEC 83%) between vendors but lower sensitivity for TomTEC (23% vs 45%) (Figure 1). LVH subgroup analysis showed similar comparisons. Overall difference in area-under-curve (AUC) was significant (AUC = 0.78 EchoPAC vs AUC = 0.52 TomTEC, p < 0.001). Conclusions: Software measurements of regional LS and thus RALS vary between vendors. Further efforts are needed for intervendor regional strain fidelity. For now, different RALS thresholds to diagnose CA may be needed for various vendors.
Read full abstract