In this cross-sectional study, we enrolled 54 patients with Chagas’ heart disease to undergo CMR. Patients with serologic confirmed Chagas’ disease diagnosis were referred to the study from our cardiomyopathy outpatient clinic. Exclusion criteria were history of myocardial infarction or coronary artery disease, more than two risk factors for CAD or diabetes mellitus (for those patients without anatomical confirmation of coronary arteries free of significant stenosis, by invasive coronary angiography or by coronary computed tomography angiography), valvular heart disease, previous viral myocarditis, other phases of cardiomyopathy, creatinine clearance below 30 ml/min/1.72 m
2 and a contraindication to perform CMR. All patients underwent a brief interview prior to the MRI exam, which included information on height, weight, medical history and previous exams. We enrolled three subgroups at different stages of disease progression based on a classification of Chagas’ disease and grouped by the outpatient clinic: 1) a group of 16 patients without evidence of cardiac involvement by ECG, chest radiography and echocardiography called indeterminate phase of Chagas’ disease (IND), 2) a group of 17 patients who had cardiac phase without left ventricular systolic dysfunction (CPND), determined by ejection fraction equal or superior to 55 % by a routine clinical echocardiography analysis and electrocardiographic abnormalities (right bundle branch block with left anterior hemiblock) or 3) a group of 21 patients with the cardiac phase with left ventricular systolic dysfunction determined by ejection fraction inferior to 55 % by echocardiography analysis (CPD). The use of echocardiography for the definition of the left ventricular dysfunction and for the classification of chagasic cardiopathy phase followed the recommendation of the current guidelines for Chagas heart disease [
21]. All CMR scans and image acquisitions were uneventful. All patients signed written informed consent approved by our local ethic committee (Comissão de Ética para Análise de Projetos de Pesquisa (CAPPesq) do Hospital das Clínicas da Faculdade de Medicina da Universidade de São Paulo) under number 0054/11.
CMR methods
CMR was performed in all patients with a Philips 1.5 T scanner (Achieva, Philips, Best, The Netherlands). Images were acquired and coupled to the ECG during breath-hold, at left ventricle long and short axis, in the exact same location in the different image sequences. This allowed a precise comparison between cardiac function and regional myocardial structure. Cine images were acquired using SSFP sequence with TR 3.5 ms, TE 1.5 ms, flip angle 60 °, receiver bandwidth ± 125 kHz, field of view (FOV) of 35 × 35 cm, 256 × 148 matrix, temporal resolution 35 ms, 8.0 mm slice thickness, with 2 mm gap between the slices.
The detection of T2W was performed with a triple inversion-recovery pulse fast spin-echo (FSE) and long TE (always TE > 70 ms), with breath-hold, in short axis view of the left ventricle, which was carefully prescribed in order to guarantee the presence of an adequate skeletal muscle area for reference. In this sequence, we used the volumetric body coil acquisition to guarantee volume homogeneity and avoid signal dropout, as usually occurs with the surface coils. The parameters were: TR 2RR interval, TE 80-120 ms, Echo train length 24 (ETL), TI 140 ms, slice thickness 10 mm, with no gap (keeping the same center slice as for cine and LGE sequences) field of view 34 × 38 cm, matrix 128 × 128. MEGE acquisitions were performed pre and immediately post-gadolinium injection, with the same scanner parameters and calibrations. We applied a spin-echo free breathing sequence, in axial plane tipped inferiorly towards the left to align with the inferior wall of the left ventricle, with no change in acquisition parameters before and immediately after intravenous injection of 0.1 mmol/kg gadolinium based contrast (gadoteric acid, Gd-DOTA, Guerbet Aulnay-Sous-Bois - France), at a 2 ml/sec rate, followed by 10 ml of saline. The sequence began immediately after injection and lasted 3 to 4 min; the images reflect a gadolinium enhancement for an average time of 2 min. The sequence parameters were the following: Field of view should be adjusted to include the left upper arm, echo train length 2 to 4, slice thickness of 8 to 10 mm, 128 y-lines, and four signals averages.
Following the acquisition of spin-echo images, an additional dose of gadolinium-based contrast (0.1 mmol/kg) was injected for LGE acquisitions, using a typical inversion-recovery prepared gradient-echo acquired 10 to 20 min after the contrast, with the following parameters: TR 7.1 ms, TE 3.1 ms, flip angle 20°, cardiac phases 1, views per segment 16 to 32, matrix 256 X 192, slice thickness 8 mm, gap between slices 2 mm and field of view 32 to 38 cm, inversion time 150 to 350 ms, receiver bandwidth 31.25 kHz, number of excitations 1, acquisition every other heart beat. Additionally, we performed another sequence as described above but with an inversion time TI of 600 ms for further characterization of cardiac thrombus.
Data analysis
The tests were analyzed in workstation with the CMR-42 software version 4.0.3 (Calgary Circle Cardiovascular Imaging Inc., Canada). The analyses were performed by an experienced professional blinded to the clinical classification and LGE results, randomly assigned using the Lake Louise criteria for myocarditis by CMR [
22]. End-systolic, end-diastolic LV volumes, and LVEF were measured applying Simpson’s method. Segmental LGE transmurality and myocardial function were scored using standard LV 17-segment model. The T2W were assessed qualitatively by visual analysis and assisted by a 2SD thresholding technique that could be used by the observer to define the segments involved by myocardial edema. Additionally, a quantitative analysis used the average signal intensity ratio of the heart muscle and skeletal muscle, (pectoral or biceps). Ratios above 1.9 indicate positivity for the presence of myocardial edema. Patients were considered positive for myocardial edema if they had at least one of the above criteria. The assessment of MEGE was performed by detecting myocardial signal intensity gain after gadolinium injection and comparing it to the gain occurred in skeletal muscle, all in the same images. A signal increase ratio superior to 4 was considered positive for MEGE. If skeletal muscle had excessive increase in signal, superior to 20 %, myositis was to be considered and, to avoid underestimation in the MEGE ratio, an absolute myocardial signal increase superior to 45 % was used as a positive criterion [
17,
23].
Late gadolinium enhancement patterns were classified as subendocardial, midwall, subepicardial, or transmural. The analysis included the number of LV segments involved by late gadolinium enhancement and an assisted automatic quantitative analysis to estimate LGE mass by a thresholding technique with pixel signal intensity 3 standard deviation above the mean of normal myocardial considered as myocardial fibrosis. In Chagas cardiomyopathy, heterogeneous and patchy myocardial fibrosis is the most frequent pattern, which challenges the use of Full Width at Half Maximum (FWHM) technique, where it is crucial to define a clear homogenous myocardial fibrosis area as a reference. By our knowledge, no definition on ideal threshold for detecting myocardial fibrosis in LGE images of Chagas cardiomyopathy patients is available. The authors performed a pilot analysis and felt that 3 standard deviation (SD) cutoff above normal myocardial signal intensity had the best agreement with visual LGE in a range of Chagas cardiomyopathy severity, which led the authors to choose in this study the 3SD threshold instead of 5SD (used in some studies involving hypertrophic cardiomyopathy). Quantification of late enhancement in the left ventricular myocardium used CMR42 software (Circle CVI, Calgary Canada).
Statistical analysis
Comparisons of normally distributed continuous variables were performed by the Student
t test and one-way analysis of variance with Bonferroni test for multiple comparisons. The Fisher exact test was used for proportions comparisons. The nonparametric test for discrete variables and non-normal continuous variables was Kruskal-Wallis rank test. Normality was determined by Shapiro-Francia W’ test. Simple linear regression was used between the MF mass and LVEF, end-diastolic volume, and end-systolic volume. Stata 8.0 (Stata Corp., College Station, Texas) was used, and
p < 0.05 (two-tailed) considered statistically significant. Given the exploratory nature of the study, formal calculations of sample size were not performed. Based on an article published by our group [
13], which investigated myocardial fibrosis in 51 patients with similar characteristics, and considering the myocardial fibrosis and inflammation as part of the same pathophysiological process, we chose to include at least 50 patients.