Diagnostic utility of spectral computed tomography in acute myocardial damage without obstructive coronary disease: the CADAMI-SPECTRAL study
C Herrera, A Sanchez, J M Monteagudo, D Bracho, J Maillo, L Gonzalez, J Rodriguez, R A Lopez, A Martin, L Nieto, L Rincon, C Fernandez-Golfin, I Cruz-Gonzalez, P L Sanchez, C Perez del VillarAbstract
Introduction
Patients presenting with acute chest pain, dynamic cardiac biomarkers elevation and non-obstructive coronary arteries require additional testing to establish the underlying diagnosis. Cardiac magnetic resonance (CMR) imaging is the reference standard but has limited availability and contraindications. Novel spectral computed tomography (CT) offers non-invasive assessment of coronary anatomy and pericoronary inflammatory status, as well as quantitative, advanced myocardial tissue characterization within a single examination.
Purpose
To assess whether myocardial perfusion imaging (MPI) and late iodine enhancement (LIE) assessment by spectral CT can detect acute myocardial damage and differentiate among underlying aetiologies.
Methods
in this multicentre study, patients with acute chest pain and troponin cardiac rise referred for CMR were enrolled. Exclusion criteria included ≥ 50% coronary stenosis, prior diagnosis of non-ischemic cardiomyopathy or ischemic heart disease, or contraindications to contrast CMR or CT. Patients were also scheduled to undergo a spectral CT scan during admission. A control cohort consisted of patients without structural heart disease or epicardial coronary artery disease (CAD) who underwent spectral CT for unrelated clinical indications. Automated myocardial segmentation (17-segment AHA model) using TotalSegmentator enabled extraction of median iodine density values from first-pass and LIE acquisitions, global extracellular volume (ECV), apex-to-base gradients. Pericoronary adipose tissue (PCAT) attenuation was also analysed using a semi-automated software. A machine-learning model (XGBoost) integrating quantitative spectral CT parameters was trained for multiclass diagnosis using 10-fold cross-validation, with CMR as ground truth.
Results
among 154 consecutive patients admitted with acute myocardial damage, CMR diagnosis were myocarditis (n=66), MINOCA (n=33), stress cardiomyopathy (n=37), and undetermined aetiology (n=18). The control group included 169 individuals. Patients with stress cardiomyopathy (Table 1) had higher values of first-pass iodine density (2.34±0.44 mg/mL, p<0.001) and global ECV (38.2±3.8%, p<0.001), negative first perfusion apex-to-base gradient (-0.06 ± 0.29, p<0.05) and a apex-to-base ECV gradient (3.7±5.1, p<0.001). PCAT did not differ between groups (p=0.176). The XGBoost model achieved significant multiclass discrimination among the aetiologies, with highest performance for stress cardiomyopathy versus myocarditis (area under the curve: 0.88; Figure 2).
Conclusions
Quantitative spectral CT enables automated myocardial tissue characterization with meaningful diagnostic performance in this clinical scenario. This comprehensive single-test imaging strategy may expand access to tissue characterization beyond CMR and represents a comprehensive alternative when CMR is unavailable or in patients with low pretest probability of coronary disease.Figure 2.