We collected consecutive data of patients treated from January 2006 to May 2019 in a tertiary hospital, who were diagnosed with ischemic stroke or transient ischemic attack (TIA) and underwent transthoracic echocardiography (TTE) and TEE simultaneously with the aim of identifying the focus of embolic stroke or TIA Ischemic stroke or TIA was diagnosed based on the presence of neurologic deficits and findings obtained using brain computed tomography or magnetic resonance imaging. Neurologic deficits were identified by an expert in the neurology department at the time of diagnosis in the hospital. Among the patients with ischemic stroke or TIA, only those who met the following criteria performed TEE as follows: (1) relatively young age as under 65, (2) age $\ge$65 years who had recurrent events of ischemic stroke or TIA, (3) multiple embolic strokes, (4) suspicious of embolic cause without the evidence of other definite etiology by the clinician’s decision, or (5) any history of cancer. Therefore, among total of 4206 patients with ischemic stroke or TIA in this period, after excluding the patients who did not meet the criteria specified above, denied to undergo TEE, or with poor systemic conditions unable to tolerate TEE. 442 patients underwent TEE examination. We excluded patients who had previous acute coronary syndrome or coronary artery disease at the left main coronary artery or at least two coronary arteries that were revealed as having 50% or more narrowing on coronary angiography, had valvular heart disease more than moderate degree on TTE or TEE, had heart failure with reduced ejection fraction less than 50%, had not been able to have diastolic parameters measured as E/e’ on TTE, and had been diagnosed with hypertrophic cardiomyopathy, restrictive cardiomyopathy, or constrictive cardiomyopathy. After excluding the patients according to the criteria specified above, a total of 295 patients were finally enrolled in this study. We retrospectively reviewed the patients’ medical records, clinical characteristics, medications, echocardiographic imaging data, and clinical outcomes.

After discharge from the hospital, the patients were scheduled to visit the outpatient clinic regularly for at least 3 months. The primary endpoint was a composite of hospitalization due to heart failure, acute myocardial infarction or acute coronary syndrome, ischemic stroke or TIA, and death from any cause. Heart failure hospitalization was defined based on symptoms of dyspnea of at least New York Heart Association Class III, as well as the need for diuretics or vasodilators, and evidence of pulmonary edema or pleural effusion on chest radiograph. If a patient had more than one clinical outcome, the first event was included in the primary outcome analysis.

Standard 2-dimensional and Doppler measurements were conducted based on recommended guidelines [11]. LV ejection fraction, left atrial (LA) volume index (mL/m${}^2$), and LV mass index (g/m${}^2$) were measured with the apical two- and four-chamber variants using the biplane Simpson’s method. Diastolic parameters such as peak E velocity of early diastolic mitral inflow and early velocity in the diastolic mitral annulus of the septum (e’) were measured using pulse-wave and tissue Doppler. The LA volume index and E/e’, which might be a good parameter for estimating LV filling pressure, were used as indexes of diastolic dysfunction [12]. LV diastolic elastance was calculated using the ratio of E/e’ over the stroke volume and used as a diastolic functional index [13]. Among the echocardiographic parameters, presence of dense mitral annular calcification (MAC) was known to be related with complex aortic atherosclerosis [14]. To determine the relationship between those markers, we also reviewed whether there was dense MAC which was defined as the maximal thickness of the calcification at the base of the mitral leaflets over 5 mm in echocardiography.

On TEE examination, the location and grade of the aortic atheroma were measured using the guidelines provided by the American Society of Echocardiography [15]. Grade 1, intimal thickness 2 mm; 2, intimal thickening of 2–3 mm; 3, atheroma $>$3–5 mm (no mobile/ulcerated components); 4, atheroma $>$5 mm (no mobile/ulcerated components), and 5 as grade 2, 3, or 4 atheroma plus mobile or ulcerated components. The separation of the location of the ascending aorta (AA), aortic arch, and proximal descending thoracic aorta (pDTA) was performed by defining the aortic arch as the portion from the end of the ascending aorta to the origin of the left subclavian artery. We used the TEE image at the level of pDTA to calculate the aortic stiffness index ($\beta$). From this image, the maximum diameter of the aortic lumen during the ejection period (D${}_{max}$) and minimum diameter of the aortic lumen during the pre-ejection period (D${}_{min}$) were measured. Next, the aortic stiffness index was calculated as follows: aortic stiffness index ($\beta$) = ln (systolic blood pressure / diastolic blood pressure) / [(D${}_{max}$ – D${}_{min}$) / D${}_{min}$], where ln indicates the natural logarithm [6]. Aortic distensibility was computed using the following formula: aortic distensibility = (D${}_{max}$ – D${}_{min}$) / (D${}_{min}$$\times$ pulse pressure) [16]. Effective arterial elastance was estimated by calculating the ratio of end-systolic pressure to stroke volume [17]. Total arterial compliance was computed as the stroke volume/pulse pressure.

Continuous variables are expressed as mean $\pm$ standard deviation, and categorical variables are presented as counts and percentages. The chi-squared test and independent t-test were used to compare categorical and continuous variables. One-way analysis of variance (ANOVA) and post-hoc analysis (HSD and Bonferroni correction) were used to compare the severity of aortic atheroma and were classified into grades from 1 to 5. The relationships between the independent variables and the severity of diastolic functional impairment were determined using univariate and multivariate linear regression analyses. Clinical outcomes were assessed using Kaplan–Meier analysis, and the predictors of clinical outcomes were evaluated using Cox regression analysis. Statistical significance was set at p 0.05. Data were analyzed using Statistical Package for the Social Sciences version 25.0 (IBM Corporation, Armonk, NY, USA).

Of the 295 patients (mean age, 58.0 $\pm$ 14.1 years), 209 (70.8%) were male. Among the 295 patients, 283 (95.9%) underwent ischemic stroke and the remainder 12 underwent transient ischemic attack (4.1%). Brain lesions were located on the left side in 111 patients (37.6%), on the right side in 108 (36.6%), bilaterally in 41 (13.9%), and in the cerebellum in 16 (5.4%) patients; there were no significant differences between the sexes. On TTE findings, women had lower LV mass index (84.2 $\pm$ 19.1 g/mL vs. 92.4 $\pm$ 20.2 g/mL, p = 0.001) but higher E/e’ (11.8 $\pm$ 3.9 vs. 10.4 $\pm$ 3.5, p = 0.003) and LV diastolic elastance (0.20 $\pm$ 0.10 vs. 0.16 $\pm$ 0.07, p 0.001). Patients with dense MAC were 26 (8.8%) of all study population and women had more patients with dense MAC (16.3% vs. 5.7%, p = 0.004). Although, the prevalence of MAC was higher in aortic atheroma (+) group but it did not reach the statistical significance in our study (11.4% vs. 6.7%, p = 0.164) (Supplementary Table 1). The baseline characteristics and echocardiographic findings of the two groups are described in Tables 1,2.

A total of 132 patients (44.7%) had visible aortic atheromatous plaques. Among them, the grade of atheroma had a similar distribution, except for grade 5. Fifty-eight patients (44%) had aortic atheroma only at the aortic arch, 40 patients (30%) at both the aortic arch and pDTA, and 28 patients (21%) had aortic atheroma only at the pDTA. Six patients (5%) had aortic atheroma in the lesion, including the ascending aorta (Fig. 1). There was no difference in the distribution of atheroma grades between both sexes. The female group had higher dimensions of the aortic root when indexed according to body surface area. Women had higher effective arterial elastance (1.99 $\pm$ 0.60 mmHg/mL vs. 1.78 $\pm$ 0.50 mmHg/mL, p = 0.002) and lower total arterial compliance (1.27 $\pm$ 0.68 mL/mmHg vs. 1.41 $\pm$ 0.44 mL/mmHg, p = 0.033) compared with men. After TEE, the arterial stiffness index and distensibility index were not significantly different between both sexes.

Fig. 1.

Grade distribution of aortic atheroma and location of ischemic stroke. More than 50% of the patients are classified as grade 1 (163 patients, 55%). Fifty-eight patients (44%) have aortic atheroma only at the aortic arch, 40 patients (30%) at both the aortic arch and DTA, and 28 patients (21%) only at the DTA. DTA, descending thoracic aorta.

The patients with any atheroma (higher than grade 1) had higher aortic stiffness index (14.5 $\pm$ 8.3 kPa vs. 8.8 $\pm$ 6.3 kPa, p 0.001), E/e’ (11.9 $\pm$ 3.6 vs. 10.0 $\pm$ 3.5, p 0.001), LV diastolic elastance (0.19 $\pm$ 0.08 1/mL vs. 0.16 $\pm$ 0.08 1/mL, p 0.001) and LA volume index (32.7 $\pm$ 12.2 mL/m${}^2$ vs. 28.3 $\pm$ 13.1 mL/m${}^2$, p = 0.003) than patients without any atheroma (grade 1). The aortic atheroma grade was significantly related with aortic stiffness index (r = 0.301, p 0.001), E/e’ (r = 0.248, p 0.001), and LV diastolic elastance (r = 0.246, p 0.001), and these correlations were concordant when the group was divided according to sex. Their relationships were more significant in women. The atheroma grade (r = 0.144, p = 0.038) and aortic distensibility (r = –0.136, p = 0.049) in men and aortic stiffness index (r = 0.302, p = 0.005) and distensibility (r = –0.274, p = 0.011) in women were significantly correlated with LA volume index (Fig. 2 and Supplementary Table 2). After further adjustment for age, hypertension, diabetes, estimated glomerular filtration rate, LA volume index, and LV mass index, which were significantly related to diastolic function, the significance of atheroma grade was attenuated. However, when the analyses were performed for each sex, the atheroma grade was significantly and independently related to E/e’ ($\beta$ = 0.181, p = 0.032) in the female group, but not in the male group ($\beta$ = 0.000, p = 0.998). No aortic parameters were independently correlated with LA volume index after adjustment for other confounding factors (Table 3).

Fig. 2.

Relationship between the grade of atheroma and aortic stiffness index, E/e’, and LV diastolic elastance (A–C), and LA volume index (D–F) using univariate simple linear regression. The grade of atheroma is significantly related to aortic stiffness index (A), E/e’ (B), and LV diastolic elastance (C). Among several aortic parameters, the atheroma grade (D) and aortic distensibility (E) are significantly relevant to LA volume index, but aortic stiffness index are not (F). LA, left atrial; LV, left ventricular.

During a mean follow-up of 2.9 $\pm$ 2.6 years, a total of 41 patients (13.9%) experienced clinical outcomes. There was no difference in the incidence of clinical outcomes between both sexes. When the entire cohort was divided according to the grade of aortic atheroma as 1, 2, 3, and at least 4, there was no difference in clinical outcomes (log rank p = 0.976). The subgroups divided based on sex differences also showed no difference in clinical outcomes (men, log rank p = 0.965, women, log rank p = 0.951) (Fig. 3). In multivariate Cox regression analysis of the entire study population, there were no clinical variables independently associated with composite outcomes. In the subgroup of female patients, LA volume index was the only independent prognostic effect with a poor composite outcome (hazard ratio [HR]: 1.07, 95% confidence interval [CI]: 1.02–1.12, p = 0.008); however, the other clinical variables were not (Table 4).

Fig. 3.

Kaplan–Meier analysis for composite outcomes according to the grade of aortic atheroma as 1, 2, 3, and at least 4. There is no difference in clinical outcome (log rank p = 0.976). The subgroups based on sex differences also show no difference in clinical outcomes (men, log rank p = 0.965, women, log rank p = 0.951).

Our study demonstrated that the grade of aortic atheroma was significantly associated with the index of aortic stiffness, E/e’, and LA volume index, which reflects LV diastolic dysfunction. Although it is well known that aortic atheroma is associated with aortic stiffness and the degree of aortic stiffness is related to LV diastolic dysfunction [3, 7], no correlations have been reported in previous studies that directly investigated the correlation between the grade of aortic atheroma and LV diastolic dysfunction. Till date, in patients with ischemic stroke, the presence and degree of aortic atheroma has been demonstrated to be a possible source of embolic events. However, our study revealed that the grade of aortic atheroma was significantly associated with diastolic dysfunction. Atherosclerotic plaque is not only a potential source of embolism in patients with ischemic stroke, but can also be considered as an afterload index that reflects arteriosclerosis. Therefore, treatment that reduces arterial stiffness through prohibition of the renin-angiotensin-aldosterone system as well as statin and antiplatelet treatment should be considered.

In our study, the grade of aortic atheroma was independently correlated with LV diastolic dysfunction, which was represented by E/e’ in women ($\beta$ = 0.181, p = 0.032) but not in men. Various studies have demonstrated the association between aortic stiffness and diastolic dysfunction in women [9, 13, 17, 18]. Shim et al. [9] demonstrated this relationship by using the central pulse pressure as an indices of aortic stiffness and revealed that the associations of central pulse pressure with e’ velocity and E/e’ were significant in female (r = 0.38, p = 0.001, r = –0.36, p = 0.001, respectively). In another study, the index of aortic stiffness was used as aortic characteristic impedance (Z${}_c$), and the study demonstrated that higher Z${}_c$ was associated with elevated E/A ratio, grade of diastolic dysfunction, and index of ventricular-arterial coupling (effective arterial elastance/LV end-systolic elastance, Ea/Ees), especially in women [18]. Our study showed the relationship between atheroma grade, which was revealed to be associated with aortic stiffness index and diastolic dysfunction in women. Increased aortic stiffness causes increased LV afterload, impaired LV relaxation, reduced coronary perfusion, and induced subendocardial ischemia and fibrosis, leading to considerable LV diastolic dysfunction, especially in women [19, 20]. This implies that the ventricular-vascular interaction is more prominent in women. Therefore, such an indicator of aortic stiffness is an important determinant to the pathophysiology of HFpEF. It is well known that prevalence of the HFpEF is more dominant in female, compared with male population. This is because the women have more susceptibility to cardiac dysfunction such as lower myocardial compliance, the inefficiency of cardiac metabolism, and vascular uncoupling than men [10]. In female patients, a small-sized LV chamber would be associated with abnormal LV chamber function and those conditions cause elevated LV filling pressure, which makes elevated E/e’ [13]. Considering these underlying issues, solving the ventricular-vascular interaction is a significant treatment target of HFpEF [21]. Furthermore, it can be concluded that the reduction of LV afterload could be an important treatment strategy in female patients with ischemic stroke. This strategy is important because female gender is different from male in terms of the risk factors, subtype and severity of the ischemic stroke, and prognosis. In women, cardio-embolic infarction was more and lacunar infarction was less, and acute ischemic stroke was regarded as a severe comorbidity and had high mortality, compared with men [22].

In our study, a total cardiovascular event occurred in 14% of patients during the mid-term follow-up duration. Among the whole study population, the LA volume index, as an important index of diastolic function, was an independent poor prognosticator (HR: 1.07, 95% CI: 1.02–1.12, p = 0.008). However, the indices related to aortic atheroma or stiffness were not independently related to the LA volume index, regardless of sex differences. In a previous study, an increased LA volume index in HFpEF patients was associated with poor prognosis [23] and the results revealed that patients with a higher LA volume index ($>$40 mL/m${}^2$) and higher maximal velocity of tricuspid regurgitation ($>$3.1 m/s) had a considerably greater risk of composite outcomes (p = 0.014) than those with others. Our study also showed similar results, but the atheroma grade was not directly related to poor prognosis. Further prognostic research is therefore required. In a previous study, more than half of the patients with a silent lacunar stroke without any cognitive dysfunction showed minor alteration of neuropsychological function [24]. These changes were associated with the presence of the silent lacunar infarcts on the earlier status of small vessel disease. Based on these results of this research, in acute stroke patients with aortic atheroma, the future directions of this study will be able to determine whether the silent cerebral ischemia would be associated with further poorer functional outcomes.

There are some considerable limitations in this study. First, this was a retrospective study and there were no available data on regular follow-up echocardiographic exams to distinguish the changes in vascular hemodynamics after the diagnosis of ischemic stroke. Second, there are limitations in the accurate measurement of the grade of aortic atheroma, especially in patients with atheroma located at the proximal aortic arch, which is difficult to access using TEE. Third, the mid-term follow-up duration was relatively short (mean 2.9 years), which might not be sufficient to evaluate the relationship between aortic atheroma and clinical prognosis. Forth, since not all patients diagnosed with ischemic stroke or TIA underwent TEE, there is a limitation in selection bias.