We retrospectively enrolled 134 suspicious CA patients who obtained CMR assessment in our institution from January 2017 to February 2021 (Fig.1). Inclusion criteria were: (1) CMR indicated suspicious CA; and (2) peripheral tissue biopsy confirmed amyloidosis (kidney, subcutaneous fat, tongue, or bone marrow), demonstrating Congo red staining with typical birefringence under polarized light; and (3) immunofluorescence or immunohistochemistry showed positive light-chain staining.
And we also included 38 HCM patients as well as 20 healthy controls with matched sex and age who underwent CMR scanning from May 2018 to January 2020. HCM patients’ inclusion followed these criteria: (1) normal LV cavity size with unexplained LV wall thickness ≥ 15 mm, or ≥ 13 mm with family history; and (2) without any other disease could lead to myocardial hypertrophy (such as amyloidosis, aortic stenosis, or Fabry disease) and (3) the myocardial thickness was not be explained by elevated blood pressure [23, 24]. Controls had neither history nor symptom of cardiovascular diseases with normal CMR findings.
Subjects were excluded if (1) with cardiac systolic dysfunction (LVEF ≥ 50%); or (2) atrial fibrillation was observed; or (3) mitral regurgitation occurred; or (4) CMR image quality was suboptimal for analysis. Clinical history, serological examinations, and electrocardiograph results of all subjects around the time of CMR scanning were collected.
The Institutional Ethics Committee approval of this study was obtained. The need for written informed consent was waived due to the retrospective design.
Cardiac magnetic resonance examination
All CMR examinations were performed on a 3T MR scanner (MAGNETOM Skyra, Siemens Healthcare) with supine position. Cine images were scanned using ECG-gated steady-state free precession during expiratory breath-holds. Standard 2-chamber, 3-chamber, 4-chamber, and short-axis views were acquired. The number of phases was 25. Typical image parameters were: field of view (FOV): 360 × 360mm2; repetition time (TR): 37.7ms; time to echo (TE): 1.4ms; flip angle: 55°; slice thickness: 8mm; voxel size: 1.9 × 1.9 × 8.0mm3; bandwidth: 965Hz / pixel.
Image analysis was performed offline using CVI42 (Circle Cardiovascular Imaging, Calgary, Canada). Patients’ CMR data in random order were analysis by two double-blinded observers (J.Y.L. and W.Z., with 2 and 3 years of CMR experience respectively).
Maximum (at the phase of mitral valve opening), minimum (at the phase of mitral valve closing), and before-contraction (at the phase just before the active contraction of LA) LA contours were automatically detected and manually corrected on standard 4-chamber and 2-chamber views. The LA appendage and pulmonary veins were excluded (Fig.2a and 2b). The LAV was calculated using the biplane area-length method: LAV = 0.85 × A1 × A2 / L. A1 / A2 = the LA areas in the 4-chamber / 2-chamber views; L = the average of the maximal LA length (perpendicular to the mitral annulus plane) in the 4-chamber / 2-chamber views. LAV were indexed to the body surface area (BSA) when the statistical analysis was performed. The measurement of the ASHmax was taken perpendicular to the atrial septum during LA end-diastolic phase (Fig.2c). The LAEF were calculated by formulas supplied in Fig.2d. The LV cavity size and function were automatically sketched and computed by artificial intelligent algorithm in CVI42. If contours’ error occurred, manual correction was carried out.
Observer variability analysis
The intra-observer variability assessment was conducted by a single observer (J.Y.L.). Twenty randomly selected scans recomputed LAV and ASHmax. The second analysis was performed 2 weeks apart from the first assessment. For inter-observer variability assessment, the same 20 scans were analyzed by a second blinded observer (W.Z.). Intra-class correlation coefficients (ICC) were ranked as follows: excellent > 0.90, good 0.75 - 0.90, moderate 0.50 - 0.75 and poor < 0.50 .
The statistical tests were performed using IBM SPSS Statistics for Windows, version 25.0 (IBM Corp, Armonk, NY, USA). We used the Pearson Chi-Square test or the Fisher’s exact test for categorical variables and the one-way analysis of variance or the Kruskal-Wallis test for continuous variables, with post-hoc pairwise comparisons using the Bonferroni correction. Intra-observer and inter-observer agreement were assessed using ICC. Univariate and multivariate logistic regression analyses were performed and receiver operator characteristics (ROC) curves were generated to identify predictors of AL.