Silent myocardial infarction fatty scars detected by coronary calcium score CT scan in diabetic patients without history of coronary heart disease

In the present study, we retrospectively evaluated CAC-CT scans and other clinical characteristics of a sub-group of patients of the DISCO cohort.

The DISCO cohort [13], a retrospective descriptive monocentric cohort, was designed in the Diabetology Department of the Cardiovascular Hospital Louis Pradel, in Lyon (France). Seven hundred thirty-two diabetic patients  > 40 years old without history of CHD who had a CAC assessment between 01-Jan-2015 and 31-Dec-2016 were systematically included. CAC-CT was performed to identify very high-risk patients that would undergo further examinations for the detection of silent myocardial ischemia (as recommended by guidelines at the time of the study) [14]. Data regarding several biographic, biological and lifestyle parameters were collected in electronic medical records, with a high completeness (98%). This data collection was performed after agreement of the local ethics committee (Hospices Civils de Lyon, N°19-111). The database was declared to the national data protection committee (Commission nationale de l'informatique et des libertés, N°19-234) and all the patients received an information notice about the study in order to collect their possible refusal of participation, in compliance with the legislation in place at the time of the study (French bioethics law Jardé).

Out of the DISCO cohort (n = 732), patients were included in the present sub-study based on the Agatston score. Only those with a CACS = 0 AU and those with a CACS ≥ 300 AU were further evaluated. These two groups correspond to a very low CV risk level and to a high CV risk, respectively [15] (Fig. 1). Hence, these two groups are two clinically distinct populations implying different management (primary vs secondary prevention).

CT scans were performed according to routine recommended clinical protocols on a commercial CT scanner (Brilliance 64, Philips). Acquisition and reconstruction parameters are summarized in Supplemental Material-Table 1.

Agatston score was calculated during clinical routine by the radiologist in charge of the shift with a dedicated commercially available software (HeartBeat-CS, IntelliSpace Portal, Philips). Centiles were calculated based on the Multi-Ethnic Study of Atherosclerosis (MESA) database.

CT scans of included patients were retrospectively analyzed, blinded to CAC score results, in consensus by two experienced observers in cardiovascular imaging.

Exams were reviewed on the PACS for the presence of intramyocardial fat. If the scan was deemed positive or doubtful, the images were exported to a commercial multimodality viewer (IntelliSpace Portal, Philips) for further analysis.

Intramyocardial fat was defined as the presence of a hypodense area inside the myocardium of the left ventricle with an average density ≤ 10 HU [9, 10]. The localization of each IMFS was defined according to the 17 segments AHA classification. In addition, its distribution inside the myocardial wall was noted as subendocardial, transmural (> 50% of the myocardial thickness) or located in the papillary muscle. Due to the difficulty in distinguishing the myocardium of the interventricular septum and the apical segments from the ventricular cavity, some IMFSs in this restricted area could not be categorized with certainty regarding their localization in the myocardial wall. Therefore, they were labelled as “septal” or “apical”. Fat containing areas of the right ventricular wall, left ventricle trabeculae and those with a clear epicardial localization were not taken into account for the purpose of this study.

For each IMFS, the surface, density and standard deviation of the density were calculated on the multiplanar reconstruction where the deposit appeared the largest. For each patient with multiple IMFS, the cumulative and average surface of all lesions were calculated.

For patients having IMFS, images of scintigraphy performed within 1 year of the CAC-CT were retrospectively evaluated blinded to the surface and the topography of the IMFS. Analysis was performed by two independent experienced observers; doubts were settled by a third experienced observer.

Statistical analysis was performed with SPSS (IBM, version 21) and R (R project, version 4.0.4).

The categorical variables are presented as frequencies and percentages. The χ² test or Fisher’s test was employed to analyze differences in categorical variables. For continuous variables, normality was assessed with Shapiro-Wilk’s tests and QQ-plots. Normally distributed values were expressed as average ± standard deviation. Values that were not normally distributed were expressed as median and interquartile range. Interquartile range was indicated as the difference between Q3 and Q1. T-Student’s and Mann-Whitney’s tests were employed accordingly to the distribution to assess differences of continuous and ordinal variables, respectively.

Overall 83 patients (21%) with fatty scars in the left ventricle myocardium were identified, of which 28 were in the group with a CACS = 0 (11.6%) and 55 in the group with a CACS ≥ 300 (37.6%), OR = 4.67 (95% CI = 2.78–7.84; p < 0.001) (Table 1 and Fig. 2). The difference was statistically significant even when all the IMFS in the apical segment 14 were excluded, leaving 17 lesions in the group CACs = 0 (7%) and 37 lesions in the group CACS ≥ 300 (25.5%; p ≤ 0.001). One example of IMFS per CAC group is shown in Figs. 3 and 4.

The number of IMFS, the cumulative surface and the average surface of fat lesions per patient who had an IMFS were similar in patients in the groups CACS = 0 and CACS ≥ 300 (Table 1). The distribution of the lesions in the 17 segments and the localization in the ventricular wall are reported in Table 1 and Fig. 5 respectively. Overall, a similar prevalence of IMFS was found in the sudendocardial and the transmural localization (Table 1).

Data for patients without and with IMFS split in the two CACS groups are presented in Table 2.

In the group with CACS = 0, the only significant differences were a mild increase in diastolic blood pressure in the subgroup with IMFS (75.5 vs 72 mmHg; p = 0.04) and more patients treated with calcium channel blockers in the subgroup without IMFS (17.4% vs 0; p = 0.02).

Differences between patients with IMFS that had a CACS = 0 and a CACS ≥ 300 are shown in the last column of Table 2. Among patients with IMFS, patients with CACS ≥ 300 were 11 years older at the time of the CAC assessment (67.2 ± 9.4 vs 56.43 ± 7.39 years), had a 7 years longer duration of diabetes (25.9 (13.0) vs 19.4 (15.5) years) and a 15 mL/min/1.73 m² reduced glomerular filtration rate (GFR) (83 (25.0) vs 98 (22.3) 15 mL/min/1.73 m²) as compared to patients with CACS = 0 (all p < 0.05). The same subgroup of patients also had more frequently hypertension (75% vs 50%; p = 0.03) and was more likely to include present or past smokers (45% vs 21%; p = 0.006). Furthermore, among patients with IMFS, patients with CACS ≥ 300 had more frequently (almost 9 patients out of 10 vs less than one-third) carotid plaques at ultrasound (p < 0.001).

Data of patients without and with IMFS is presented in Table 3 and additional data about medications in these groups in Supplemental Material—Table 3.

Patients with IMFS were about 4 years older at CAC assessment (63.5 ± 9.3 vs 58.6 ± 10.5 years; p = 0.001), had more coronary calcifications (CACS increased of 496 AU and 86 centiles (p < 0.001)), had a 3 years longer duration of diabetes (21.9 (14.0) vs 18.9 (15.0) years; p = 0.008) and had a 9 mL/min reduced renal function as compared to patients without IMFS (86 (26.0) vs 95 (25.0) mL/min/1.73 m; p = 0.002).

Patients using statins had a 1.75 (95% IC = 1.07–2.88; p = 0.03) higher risk and those showing carotid plaques at ultrasound had a risk increase of 3.03 folds (95% IC = 1.43–6.39; p = 0.004) of having IMFS (Table 3).

Multiple logistic regression (Table 4) showed a negative association between IMFS and levels of glycated haemoglobin (OR = 0.80; 95% IC = 0.66–0.96; p = 0.019) and confirmed a strong positive association with CACS ≥ 300 (OR = 5.12; 95% IC = 2.66–9.85; p < 0.001). Age at CAC was among the predictors attained in the reduced model but was not significant.

Eighty-one patients (20% of the total), of which only one in the CACS = 0 group, underwent a scintigraphy within 1 year from the CAC assessment. Twenty-nine exams were performed in patients with IMFS, out of which 13 were classified as pathological. Among these, 8 were deemed positive for necrosis with or without associated ischemia, 2 doubtful or with small lesions and 3 with ischemia without necrosis. Among these 8 patients, 6 had at least one SMI fat deposit on the CAC-CT and 2 had two. In all the 8 cases, the localization of at least one of the lesions visible on the scintigraphy corresponded to that of at least one of the IMFS.