Regional Analysis of Myocardial Strain to Wall Thickness Ratio in Cardiac Amyloidosis and Hypertrophic Cardiomyopathy
Michael P Gannon,Cristina P Sison,Shahryar G Saba,Michael P. Gannon,Cristina P. Sison,Shahryar G. Saba
DOI: https://doi.org/10.1097/rti.0000000000000772
IF: 5.528
2024-01-09
Journal of Thoracic Imaging
Abstract:Background: Increased left ventricular wall thickness is a hallmark of cardiac amyloidosis (CA). Several other disease states, including hypertrophic cardiomyopathy (HCM), share this common feature. Myocardial strain has emerged as a diagnostic and prognostic tool to differentiate causes of increased left ventricular wall thickness. We sought to determine if regional strain differences were present in CA when compared with HCM when indexed to wall thickness as well as adjusting for important factors such as ejection fraction (EF), age, sex, and hypertension. Methods: We performed a multicenter, retrospective analysis of 122 patients in 3 groups: CA (n=40), HCM (n=44), and controls (n=38). Using commercially available software, we determined peak systolic strain measurements in the base, mid, and apical segments in all 3 cardinal directions of radial strain, circumferential strain, and longitudinal strain. The regional strain was indexed to wall thickness to create a strain to wall thickness (STT) ratio. Analysis of Variance was performed to examine the association of each strain parameter with the disease group, adjusting for age, sex, hypertension, and EF. Multinomial logistic regression was performed to determine which combination of variables can potentially be used to best model the disease group. Results: Ratios of STT at all 3 levels were significantly different with respect to the cardinal directions of radial, circumferential, and longitudinal strain in a multivariable analysis adjusting for age, sex, and hypertension. Specifically, with respect to the basal segments, the STT ratio across CA, HCM, and normal were significantly different in radial (1.13±0.34 vs. 3.79±0.22 vs. 4.12±0.38; P <0.0001), circumferential (−0.79±0.10 vs. −1.62±0.07 vs. −2.25±0.11; P <0.0001), and longitudinal directions (−0.41±0.09 vs. −1.03±0.06 vs. −1.41±0.10; P <0.0001). When adjusting for age, sex, hypertension and EF, only the base was significantly different between the CA and HCM groups in the radial (1.49±0.37 vs. 3.53±0.24; P <0.0001), circumferential −1.04±0.10 vs. −1.44±0.06; P <0.005), and longitudinal (−0.55±0.10 vs −0.94±0.06; P =0.007) directions. Using multinomial logistic regression, the use of age, left ventricular EF, global longitudinal strain, and basal radial strain yielded a diagnostic model with an area under the receiver operating characteristic curve (AUC) of 0.98. A model excluding age, despite being likely an independent predictor in our cohort, yielded an overall AUC of 0.90. When excluding age, the overall AUC was 0.91 and specifically when discriminating CA from HCM was 0.95. Conclusions: Regional myocardial strain indexed to wall thickness with an STT ratio can differentiate between etiologies of increased left ventricular wall thickness. Differences in myocardial deformation may be independent of wall thickness. Differences in basal strain when indexed to wall thickness in all 3 cardinal directions between CA and HCM are independent of EF. Multinomial logistic regression analysis using strain parameters differentiates CA and HCM with excellent diagnostic accuracy.
radiology, nuclear medicine & medical imaging