Multiparametric CMR for detection of cardiac amyloidosis in patients with undifferentiated concentric left ventricular hypertrophy

Multiparametric CMR can detect cardiac amyloidosis (CA) in patients with undifferentiated concentric left ventricular hypertrophy (LVH). A retrospective patient search was performed for patients ≥ 18 years of age who received 1.5T CMR from January 2018 to August 2022 to evaluate for myocardial infiltration in the setting of known concentric LVH at a tertiary medical center. Clinical records were reviewed for positive diagnosis or exclusion of CA. CMR were post-processed to evaluate ventricular volumes and function, native T1 mapping, and extracellular volume (ECV). Late gadolinium enhancement (LGE) was qualitatively evaluated and each segment categorized into vascular or non-vascular LGE using the American Heart Association 16 segment model. Feature tracking strain (FTS) was performed on a subset of CA positive and CA negative patients. Group comparisons were made using one-way ANOVA (parametric) or Kruskal-Wallis (non-parametric) tests. Receiver operator characteristic (ROC) analysis with area under curve (AUC) values were generated for both individual and combined parameters using binary logistic regression (IBM SPSS Statistics V26.0, and DATATab 2025) to determine optimal cut-off parameters for detection of CA. CMR were performed in 278 patients for myocardial infiltration evaluation in the setting of known concentric LVH (mean age 63.2 ± 14.9, 46% female). Diagnostic groups were determined as follows: CA positive (n = 60), CA negative (n = 100) and CA unknown (n = 118). CA positive patients, when compared to both CA negative and CA unknown groups, respectively, demonstrated significantly higher age (69.9 ± 10.3 vs. 59.7 ± 14.2 and 62.9 ± 16.4 years), native T1 (1122.4 ± 64.6 vs. 1056.8 ± 69.7 and 1051.4 ± 54.0 ms), ECV (46.4 ± 11.5 vs. 32.1 ± 7.2 and 32.1 ± 7.6%) and number of segments with infarct-atypical LGE (10.2 ± 7.3 vs. 2.7 ± 4.7 and 2.0 ± 4.3). ROC AUC values were calculated for native T1 (0.80), ECV (0.88), and number of infarct-atypical LGE segments (0.76). A 4 parameter model including age, native T1, ECV, number of segments with non-vascular LGE demonstrated an AUC of 0.91 for detection of CA, with a sensitivity of 92% and specificity of 81%, which when applied to the CA unknown group, indicated 13 patients (11%) in this group may have CA. CA positive patients demonstrated reduced basal peak systolic strain and diastolic strain rates when compared to CA negative patients; a model combining these parameters with patient age demonstrated an AUC of 0.79 for detection of CA. Multiparametric CMR can discriminate CA positive patients from CA negative and undifferentiated LVH patients in a real-world tertiary center population. These findings demonstrate that CMR has significant diagnostic potential for detection of CA in patients with undifferentiated LVH.