This is a retrospective cohort study assessing CMR parameters in non-ischemic patients with HTN, LVH, and HF undergoing histological evaluation due to uncertain diagnosis, despite extensive cardiac testing that included genetic testing and advanced cardiac imaging. Patients undergoing evaluation between September 22, 2008 and March 15, 2022, in two tertiary care centers (The Ohio State University Wexner Medical Center, Columbus, OH & University of Pennsylvania, Philadelphia, PA) were included in the analysis. The study was approved by the Institutional Review Boards who waived informed consent in the respective institutions. Inclusion criteria included age ≥ 18 years old, hypertension, LVH (defined as wall thickness ≥ 12 mm on CMR), and HF secondary to HC, ATTR CA, AL CA, or HCM [3, 4]. Prior myocardial infarction, myocarditis, septal ablation or myectomy were excluded. Patients with typical CMR findings of HC, HCM or CA who do not routinely undergo histological assessment were not included in this study. The final diagnosis was based on myocardial histology on EMB or transplant. Figure S1 demonstrates the decision-making process in that patient population.

Patients were classified into three groups based on the histological evaluation – HC, ATTR CA, and AL CA [16]. Additional subtyping based on histology was performed to differentiate AL from ATTR CA [15]. HC was defined as LVH associated with diastolic or systolic left ventricular (LV) dysfunction in patients with persistent systemic HTN, in the absence of other causes for increased afterload and a negative EMB result [4]. None of the histological studies performed in patients with HTN, LVH, and HF of uncertain etiology within the study period were positive for any other etiology such as HCM or Fabry disease. To ensure complete differential diagnosis, CMR findings of patients with established HCM diagnosis (defined as presence of LV wall thickness ≥ 15 mm not explained by loading conditions) confirmed on histological evaluation (of explanted heart or myocardium removed during septal myectomy) were analyzed for comparison with the main three study groups (HC, ATTR CA, AL CA). The electronic medical record was reviewed for demographic and clinical data as well as clinically performed cardiac testing including electrocardiography (ECG), Holter ECG stress testing, echocardiography, coronary computed tomography angiography, single photon emission computed tomography, positron emission tomography, technetium-99 m-pyrophosphate scintigraphy (PYP scan), left heart catheterization, endomyocardial biopsy (EMB), and genetic testing.

Clinical CMR images were acquired on three 1.5T scanners (MAGNETOM Avanto/Avanto fit/Sola, Siemens Healthineers, Erlangen, Germany) using standardized protocols including cine imaging, native and post-contrast T1 mapping, and LGE imaging [17].

Breath-held segmented bSSFP cine images were acquired in patients able to hold their breath in standard 2-, 3- and 4-chamber long axis and stack of short axis (SAX) views covering the entire LV. In cases of difficulties with breath-holding, free-breathing real-time cine images were acquired [18, 19].

T1 relaxation times were measured by the MOLLI (Modified Look-Locker Inversion Recovery) technique, using both prototype (WIP 448 and WIP 780B on Avanto systems) and commercially available sequences (MyoMaps on the Sola system) [20]. Native T1 maps were acquired either with a 3(3)3(3)5 (WIP 448) or with a 5(3 beat)3 acquisition scheme (WIP 780B and Myomaps) with the following acquisition parameters: field of view (240–400) mm x (320–450) mm, acquisition matrix size (192–256) x (108–170), reconstructed spatial resolution (1.4–2.1) x (1.4–2.1) mm, slice thickness 8 mm, flip angle 35º, bandwidth 930–1085 Hz/pixel, TI start time 100–111 ms and TI increment of 80 ms.

Different contrast agents including Gadavist (Bayer Healthcare Pharmaceuticals, 18 patients), Magnevist (Bayer HealthCare Pharmaceuticals, 4 patients), Multihance (Bracco Diagnostics, 4 patients), Dotarem (Guerbet LCC, 6 patients), and Prohance (Bracco Diagnostics, 2 patients) were administered clinically to obtain post-contrast T1 and LGE images during the study period. Contrast dose was based on weight at 0.15–0.2 mmol/kg.

Post-contrast T1 maps were acquired either with a 3(3)3(3)5 (WIP 448) or with a 4(1)3(1)2 acquisition scheme (WIP 780B and Myomaps) 14.5 min (IQR: 11–17) after injection of gadolinium-based contrast agent. The acquisition parameters were as follows: field of view (320–450) mm x (250–400) mm, acquisition matrix size (192–256) x (120–162), reconstructed spatial resolution (1.4–2.3) x (1.4–2.3) mm, slice thickness 8 mm, flip angle 35º, bandwidth 930–1085 Hz/pixel, TI start time 100 ms with increment in 80 ms.

Phase sensitive inversion recovery (PSIR) LGE images were acquired with a free-breathing motion-corrected and averaged single-shot inversion recovery prepared bSSFP method (MOCO LGE with 8 averages) in standard 2-, 3- and 4-chamber long axis and SAX views and/or with a free-breathing single-shot single-average inversion recovery prepared bSSFP method in a SAX stack covering the LV. Representative acquisition parameters were as follows: field of view (240–400) mm x (320–500) mm, acquisition matrix size (102–192) x (192–256), reconstructed spatial resolution (1.4–2.6) x (1.4–2.6) mm, slice thickness 8 mm, flip angle 40–55º, and bandwidth 975–1532 Hz/pixel.

CMR studies were anonymized and analyzed using SuiteHeart (Neosoft, LLC, Pewaukee, Wisconsin). LV volumes and LVEF were measured from SAX images that covered the LV. Native and post-contrast myocardial T1 values were measured within the septum on the mid SAX maps, excluding confluent LGE. Septal T1 was measured, in order to avoid measurement biases in quantification of ECV and myocardial fibrosis [21]. ECV was calculated using the standard formula based on T1 mapping and hematocrit obtained on the day of the CMR study [22]. Elevated ECV was defined as ≥ 29% and was based on institutionally established normative data (same threshold in both tertiary care centers).

The presence, pattern, and extent of LGE on long axis (2-, 3-, and 4-chamber views) and SAX images were assessed by two level 3 trained CMR readers blinded to clinical information. LGE patterns were described as follows: subendocardial, midwall, subepicardial, and transmural LGE. Presence of LGE in papillary muscles (PM) and base-apex gradient was also assessed [7, 8]. Extent of LGE was reported using the full-width half-maximum technique according to the American Heart Association [AHA] 17-segment model [23, 24]. To assess inter-observer variability, two level 3 CMR readers performed the analysis of T1 mapping and LGE using a random set of 10 patients. To assess the intra-observer variability, one of the readers performed the analysis of T1 mapping and LGE twice in the same set of 10 patients at least two weeks apart.

Cardiac histology was based either on EMB or native myocardium examination in patients undergoing heart transplant. EMB consisted of 5 specimens obtained from the RV septal wall. Histologic sections were examined by standard hematoxylin and eosin, trichrome, iron, and Congo red stains. Immunohistochemical stains using antibodies directed against serum amyloid A, transthyretin, and kappa and lambda light chains were conducted in case of positive Congo red stain. If the CA classification could not be made, mass spectrometry was performed. Electron microscopy studies were performed depending on the histological results and pathologist’s assessment.

Categorical variables are summarized using frequency (percentage), and comparison between groups was performed using the chi-square test or exact test. Continuous variables are expressed as mean ± standard deviation (SD) for normal distributions and median + interquartile range for non-normal distributions. The distribution of continuous variables was determined using skewness, kurtosis, visual inspection of the histogram, and QQ plot. T test or Mann–Whitney U test was used to compare differences between the two groups for normally and non-normally distributed variables respectively. Comparisons between three or more groups were made with a one-way analysis of variance (ANOVA). Given unequal sample sizes among groups, the Welch’s test was performed.

With regards to reproducibility assessment, intra- and inter-observer reproducibility for CMR parameters were analyzed using the Bland–Altman method, and intra-class correlation coefficients (ICC, two-way random, absolute agreement, and single measure) for continuous variables, and the Kappa value and proportion in agreement for categorical variables. For continuous variables, agreement was considered excellent when ICC > 0.74, good when ICC = 0.60–0.74, fair when ICC = 0.40–0.59, and poor when ICC < 0.4. For categorical variables, Kappa value was interpreted as follows: values ≤ 0 as no agreement and 0.01–0.20 as none to slight, 0.21–0.40 as fair, 0.41– 0.60 as moderate, 0.61–0.80 as substantial, and 0.81–1.00 as almost perfect agreement. A two-sided P-value of < 0.05 was considered statistically significant. A two-sided P-value of < 0.05 was considered statistically significant. The statistical analyses were performed using IBM SPSS Statistics for Windows, version 22.0 (IBM Corp., Armonk, N.Y., USA) and R software, version 4.0.3 (The R Foundation, Vienna, Austria).