ORIGINAL ARTICLE

After decades of female underrepresentation in most clinical studies, there is an increasing interest in the evaluation of the distinctive cardiovascular pheno­types of men and women. (1)

Aside from a different risk factor profile, clinical presentation, and higher susceptibility to myocardial injury associated with psychological stress, women have less coronary calcification, higher rates of micro­vascular disease, and a more hypercoagulable profile, underscoring the role of nonobstructive disease among them. (2-4) Furthermore, independent of smaller (ad­justed for body surface) epicardial coronary arteries, women have higher resting myocardial perfusion and myocardial blood volume compared with men. (5) Indeed, gender differences endure even among the elderly, with males displaying consistently higher coronary calcification and total coronary plaque bur­den despite showing, interestingly, a similar extent of extra-coronary calcification. (6,7)

To date, characterization of myocardial tissue us­ing computed tomography (CT) requires iodinated contrast. Nonetheless, the electron density (ED) of the atoms comprising the tissues, normalized to pure wa­ter, can be directly estimated using dual-layer spectral computed tomography (CT), without modifications of acquisition protocols. (8) Previous studies have sug­gested an incremented value of (non-contrast) ED imaging over iodinated contrast assessment for tissue characterization, although to the best of our knowl­edge, data regarding myocardial ED has not been ex­plored. (9-11)

Therefore, we hypothesized that ED imaging might be able to detect differences indistinguishable for conventional non-contrast CT. To confirm this, we evaluated sex-related differences in myocardial ED and whether, if present, these are related to cardio­vascular risk factors or coronary calcification.

Patients were participants of a prospective single center ob­servational study that explored the ability of whole-blood transcriptome screening test assisted by deep learning to detect coronary calcium, whose main analysis (transcrip­tome) will be reported independently. The study population comprised 200 asymptomatic volunteers (men between 40 and 75 years old and women between 50 and 75 years old), or patients referred for a low-dose chest CT scan for smok­ing. Main exclusion criteria comprised a history of heart, re­nal, or liver failure, previous myocardial infarction, previous coronary revascularization or peripheral vascular disease, active pulmonary disease, immunosuppressant treatment or malignancy under current treatment, and COVID-19 in­fection in the past three months. To avoid age differences, males younger than 50 were further excluded for the present investigation.

All images were obtained with a dual-layer detector CT scan­ner (IQon Spectral CT, Philips Healthcare, Best, The Neth­erlands) using the following parameters: collimation 64 x 0.625 mm; tube voltage 120 kV; current 70-140 mA based on patient size; rotation time 270 ms; slice thickness 2.0 mm. Dose modulation (3D Modulation) and hybrid iterative reconstruction (iDose 5) were used in all cases. This scan­ner enables extraction of spectral data using dual-layered scintillation detectors, being the inner layer an yttrium-based garnet low density scintillator for detection of lower energies, and the outer layer a gadolinium oxysulphide high density scintillator for detection of higher energies. ED es­timation using this scanner is based on a two-base model comprising the combination of Compton scatter-like (SC) and photoelectric-like effects, with a dominating SC com­ponent. Images were assessed using multiparametric side-by-side view of conventional CT (120 kVp) and ED images and were analyzed offline by a cardiac imaging expert who was blinded to all demographic and clinical data. All images were assessed using average multiplanar reconstructions of short axis views (assisted by adjacent landmarks includ­ing the anterior and posterior interventricular groove, and coronary veins and arteries), adjusting window´s width and level at the best discretion of the observer. Using the American Heart Association 16-segment model, regions of interest (ROIs) were placed at the sixteen myocardial seg­ments (Figure 1) starting with those involving the lateral, anterior, and inferior walls, and finally the septal wall. Care was taken to avoid areas with uniform band-like artifacts. Subsequently, the mean Hounsfield units (HU) and mean percent ED relative to water (%EDW) were calculated. It is noteworthy that ROIs were automatically co-localized and identical in size and location in both displays (conventional and ED, Figure 1).

In addition, we evaluated the relationship between myo­cardial ED and the extent of coronary artery calcifications. For this purpose, we evaluated the presence of any calcifi­cation and scored each patient according to the number of vessels involved, and to the numbers of segments involved (CACSIS). (12)

The protocol of the study, in accordance with the declaration of Helsinki and later amendments, was approved by the eth­ics review board (CEI FLENI registration code 4230), and all patients provided their written informed consent.

Discrete variables are reported as counts (%) and contin­uous variables are presented as mean ± standard devia­tion or median (interquartile range, IQR) for non-uniform distributions. Comparisons between groups of continuous variables were performed using a t test for independent variables, one-way analysis of variance (ANOVA), or the Mann-Whitney U test; whereas comparisons of categorical variables were performed using Fisher exact test, and the chi square test (multiple groups). To identify independent predictors of an incremented %EDW, we built a multiple logistic regression model using the 75th percentile %EDW as dependent variable. Myocardial segments were re-evalu­ated by an independent cardiologist in a randomly selected subset of 20 cases (320 segments), and the agreement be­tween observers was analyzed using intraclass correlation coefficients (ICC; two-way random effects model absolute agreement, and average measurement) with 95% confi­dence intervals. All statistical analyses were performed us­ing SPSS software, version 22.0 (Chicago, Illinois, USA). A two-sided p value of less than 0.05 indicated statistical significance.

Two hundred patients were enrolled in the main study between June and October 2021, of which 29 men younger than 50 years old were excluded to at­tain a balanced age comparison, leading to a total of 171 patients (77 men and 94 women) in this analy­sis. Mean age was 60.2±6.8 years, without differenc­es between groups (men 60.4±6.9 years, vs. women 60.0±6.8 years, p=0.714). We did not identify sig­nificant differences between groups regarding rates of diabetes (p=0.282), smoking (p=0.334), and hy­percholesterolemia (p=0.076), whereas men showed a higher prevalence of hypertension (53% vs. 34%, p=0.012) and obesity (48% vs. 31%, p=0.034). Com­pared to women, men had larger extent of coronary calcification (coronary artery calcification segment in­volvement score, CACSIS: men 2.34±2.9, vs. women 1.38±2.3, p= 0.016).

Myocardial attenuation levels (HU) were not related to sex or coronary risk factors (Table 1), whereas %EDW was significantly higher among males (104.5±0.2 %EDW, vs. 104.3±0.2 %EDW, p<0.0001). Despite the narrow differences, there was no overlap between the error bars based on 95% confidence intervals of the population mean (Figure 2).

Myocardial %EDW was also higher among pa­ tients with diabetes (104.5±0.2 %EDW vs. 104.4±0.2 %EDW, p=0.007), hypertension (104.5±0.2 %EDW, vs. 104.4±0.2 %EDW, p=0.001), and obesity (104.5±0.2 %EDW, vs. 104.4±0.2 %EDW, p=0.004). Hypercholes­terolemia (p=0.722), smoking history (p=0.503), and treatment with statins (p=0.184) were not related to the myocardial ED. After discrimination according to body mass index (BMI) categories (BMI <25 kg/m2; 25-29.9 kg/m m2 2; and ≥30 kg/m2), myocardial attenu­ation levels were not related to sex among patients with normal weight (p=0.242), overweight (p=0.913), or obesity (p=0.445); whereas myocardial ED was significantly higher among males irrespective of the BMI (normal weight: 104.5±0.2 %EDW vs. 104.3±0.2 %EDW, p=0.046; overweight: 104.5±0.2 %EDW vs. 104.4±0.2 %EDW, p=0.002; and obesity: 104.6±0.3 %EDW vs. 104.4±0.2 %EDW, p=0.006). The extent of coronary artery calcification was unrelated to neither the myocardial attenuation (p=0.554), nor the myo­cardial ED (p=0.382). When discriminated in tertiles (Table 2), we identified a significant relationship be­tween the myocardial %EDW and sex (p<0.0001), dia­betes (p=0.038), and hypertension (p=0.049).

Regarding reproducibility, the interobserver reli­ability was modest when assessing the myocardial attenuation levels (ICC 0.58, 95%IC -0.07 - 0.83) and good by means of %EDW (ICC 0.86, 95%CI 0.64-0.94).

At univariate analysis, sex was the only variable as­sociated with a high %EDW (p=0.011), whereas age (p=0.702), diabetes (p=0.154), obesity (p=0.073), smoking (p=0.454), hypertension (p=0.421), and CACSIS (p=0.842), were not. At multiple logistic re­gression analysis including sex, obesity, diabetes, and hypertension in the model, male sex was identified as the only independent predictor of a high %EDW (OR 2.51, 95%CI 1.23-5.34, p=0.016), whereas obesity (OR 1.97, 95%CI 0.91-4.14, p=0.0823), diabetes (OR 1.89, 95%CI 0.76-5.02, p=0.206), and hypertension (OR 1.53, 95%CI 0.74-3.46, p=0.284) remained outside the model.

The main finding of our study was the identification of significant sex-related differences in myocardial ED, with male sex being identified as the only independ­ent predictor of a high %EDW. Furthermore, myocar­dial ED was significantly higher in patients with car­diovascular risk factors, perhaps indicating a common pathophysiological process. Interestingly, attenuation levels estimated using conventional CT were unable to detect such associations.

Women’s heart is not just a smaller kind of men’s. As aforementioned, several studies have identified substantial phenotypical differences between men and women that exceed the most common structural and functional aspects, including the microstructural architecture even at a cellular, metabolic, and electri­cal level. (13-16)

Our findings are in line with the results of a large cohort of healthy biobank participants among whom the cardiovascular magnetic resonance radiomic pro­file was evaluated. In that study, both male sex and cardiovascular risk factors were associated with a dimmer and less texturally complex myocardium. No­tably, in keeping with our findings, Raisi-Estabragh et al. identified diabetes and hypertension as the risk factors with the closest associations with left ventricu­lar features, but not smoking. (17) Of note, we did not find a relationship between the extent of coronary artery calcification and myocardial ED, although our findings did not address the effect of flow-limiting le­sions in myocardial ED.

The dual-layer configuration of single-source dual energy CT scanners such as the one used in the pre­sent study allows simultaneously obtaining spectral data, and a significant reduction of image noise (re­flected in this study by the exceedingly low standard deviation), without affecting the routine workflow of the CT scanner, and makes it compatible for ret­rospective reconstruction. (18,19) For decades, the main purpose of ED imaging was radiation therapy planning, although single-energy CT requires scan­ner-specific calibration curves for dose calculation performed by skilled medical physicists. (20) In turn, dual-energy CT allows direct calculation of the ED on a voxel basis. In a phantom study using the same dual-layer spectral CT scanner as in our study, Hua et al. demonstrated a high accuracy of ED measurements compared with the expected values, with a median de­viation from all tissue inserts ranging from 0.1% to 1.1%. In keeping with our findings, such negligible error might explain to some extent the incremental value of ED over conventional CT to detect subtle differences between tissues that cannot be conveyed without using contrast-enhanced images. Notably, in the study of Hua el at. all soft tissue materials near water ED (adipose, brain, breast, and liver) were well separated, and results were not sensitive to acquisi­tion or reconstruction parameters. (21)

The seemingly synchronic effect of male sex and cardiovascular risk factors might potentially be relat­ed to a shared pathophysiological process such as fi­brosis, or water or fat content, although interpretation of our results in this regard should be cautious since they don’t offer explanations about the underlying pathophysiological mechanisms related to observed differences. (8) In this regard, despite the significant differences between genders and the relationship with risk factors, it should be acknowledged that the clini­cal relevance of our results is uncertain. Nonetheless, if confirmed, our findings suggest that ED might be­come in the future a valuable unsophisticated tool for non-contrast CT myocardial tissue characterization.

It should be stressed that it is unclear whether the improved discrimination compared with conventional imaging was related to the intrinsic value of ED im­aging, or to the almost lack of image noise and mini­mal error associated with such approach. (8) In this regard, a previous study has shown that compared with other spectral parameters such as the effective Z number, ED imaging has the lowest deviation (within 1%) from phantom inserts. (21) It should also be rec­ognized that our results might be influenced by the challenging multiplanar reformatting and ROI place­ment in non-contrast images. Finally, despite the ef­fect of motion artifacts for non-coronary imaging is mild, future studies are warranted in order to confirm our findings using ECG-gated ED imaging. (22,23)

In conclusion, ED imaging was able to identify my­ocardial tissue differences that conventional CT was unable to discriminate, with a higher %EDW in men and in patients with cardiovascular risk factors.

https://creativecommons.org/licenses/by-nc-sa/4.0/