Reduced Ejection Fraction and Augmented Right Atrium are Independent Risk Factors of Deep Regional CBF in Cerebral Small Vessel Disease

Background: Systemic cardiac hypoperfusion is a well-acknowledged contributor to ischemic leukoencephalopathy. However, it has remained elusive how atherosclerosis-mediated cardiac remodelling modies cerebral perfusion homeostasis as well as neuroimaging burden in cerebral small vessel disease (CSVD) development. Methods: This retrospective study identied 103 CSVD patients (CSVD burden low 0~1, n=61 and CSVD burden high 2~4, n=42) from Sep. 2017 to Dec. 2019 who underwent transthoracic echocardiography(n=81), structural magnetic resonance imaging and arterial spin labelling. Total CSVD burden was graded according to the ordinal “small vessel disease” rating score (0-4). We investigated the univariate and multivariate linear regression of mean deep regional cerebral blood ow (CBF) as well as logistic regression analysis of CSVD burden high . Results: Augmented right atrial diameter (B coecient, -0.289; 95% CI, -0.578 to -0.001; P=0.049) and decreased left ventricular ejection fraction (B coecient, 32.555; 95% CI, 7.399 to 57.711; P=0.012) were echocardiographic parameters linked with declined deep regional CBF in CSVD patients. Binary logistic regression analysis demonstrated decreased mean deep CBF (OR 0.894; 95% CI 0.811-0.985; P=0.024) was independently associated with higher CSVD burden after adjusted for clinical confounders. Multivariate receiver operating characteristics curve combing clinical risk factors, mean deep CBF and echocardiographic parameters showed predictive signicance for CSVD burden high diagnosis (area under curve=84.25%, 95% CI 74.86% - 93.65%, P<0.0001). Conclusion: The interrelationship of “cardiac -deep regional CBF-neuroimaging burden” reinforces the importance and prognostic signicance of echocardiographic and cerebral hemodynamic assessment in CSVD early-warning. hemoglobin A1c; LDL-C, low density lipoprotein-cholesterol; HDL-C, high density lipoprotein-cholesterol; ApoA1, apolipoprotein A1; ApoB100, apolipoprotein B100; Lp(a), lipoprotein (a); eGFR, estimated glomerular ltration rate


Background
Cerebral small vessel disease (CSVD) is a spectrum of cerebrovascular diseases attributed to arteriolosclerosis, genetic inheritance, infection, autoimmune in ammation, venous collagenosis and other secondary aetiologies such as radiation (1). Arteriolosclerotic CSVD (aCSVD) is the most prevalent category in the elderly and contributes to high worldwide disease burden from stroke(2) and vascular dementia (3). Advanced age and hypertension are the most evidenced epidemiological risk factors(4) and individualized therapy based on modi able cardiovascular factors is the most widely accepted therapeutic and preventive strategy in clinical practice. The pathogenesis of aCSVD remains poorly illustrated and chronic cerebral hypoperfusion secondary to arteriolosclerosis is one of the attributable mechanism (1). Most studies concerned with the relationship between cerebral perfusion and CSVD are based on white matter hyperintensity (WMH) development (5,6). Though persistent cerebral perfusion has been considered as an attributable mechanism of WMH, a longitudinal quantitative neuroimaging study reveals that relative cerebral blood ow (CBF) change has no signi cant correlation with WMH progression (7). Notably, aforementioned studies mainly focus on global CBF alteration but not deep regional CBF which represents the circulation perfusion of parenchymal small vessels.
It is acknowledged that chronic systemic hypoperfusion can contribute to WMH and recent clinical investigations have revealed heart involved mechanisms including cerebral hypoperfusion (8) or sympathetic overactivity in CSVD development (9)(10)(11). Thereinto, left ventricular ejection fraction (LVEF), a well-established cardiac output indicator, reduces considerably in CSVD patients with dementia (12).
However, no considerable effect of LVEF on white matter microstructure damage and deterioration of CBF are observed in healthy middle-aged cohort (13). Actually, WMH is predisposed to lower CBF perfusion in patients with reduced cardiac output but not in patients with normal cardiac output (14). Collectively, how atherosclerosis mediated cardiac remodelling modi es deep regional CBF and subsequent neuroimaging burden in aCSVD patients with normal cardiac function remains unknown. We aimed to preliminarily investigate the interrelationship among echocardiographic parameters, deep regional CBF and CSVD neuroimaging markers in aCSVD patients absent from decompensated heart failure.

Study population
One hundred and three patients (inclusion work ow shown in Fig. 1) admitted to neurology department in Third A liated Hospital of Sun Yat-sen University from Sep. 2017 to Dec. 2019 matched the inclusion criteria below: 1) age ≥ 40y; 2) at least one of the following atherosclerotic risk factors: smoking (≥ 10 cigarettes/day for at least 10y); excessive alcohol consumption (≥ 15 drinks per week for men and ≥ 8 drinks per week for women); body mass index (BMI) 25; hypertension; diabetes mellitus; impaired glucose tolerance(IGT) or impaired fasting glucose(IFG); coronary heart disease; hyperlipemia; hyperhomocysteinemia; symptomatic stroke history; 3) magnetic resonance imaging (MRI) con rmed recent subcortical small infarct (RSSI) or complaining of CSVD common symptoms including cognitive decline, gait disturbance or bradykinesia, emotional stress(anxiety or depression) and mixture of the above symptoms. 4) MRI neuroimaging met the STandards for ReportIng Vascular changes on nEuroimaging (15); 5) No visible moderate-severe intracranial atherosclerotic stenosis in MR angiography. Vascular aetiologies including traumatic, infectious, neoplastic, autoimmune, toxic, metabolic causes and large vessels occlusion, cardiac embolism were all excluded. All included patients completed standard brain MRI including ASL and 81 patients undergone comprehensive transthoracic echocardiography (TTE). The demographic information, medical history, laboratory biomedical test and auxiliary examinations were reviewed from the electronic medical record system. Carotid artery ultrasonography was performed for measuring intima media thickness and excluding individuals with atherosclerotic stenosis 50%. This clinical investigation was approved by the medical ethics committee of the Third A liated Hospital of Sun Yat-Sen University and was performed in accordance with principles illustrated in Declaration of Helsinki. All participants had been given the informed consent.

Acquisition Of Echocardiographic Data
Two-dimensional doppler echocardiography was conducted mainly to screen potential cardiac risk factors or cardiogenic emboli but there were only 81 echocardiographic records could be retrieved in hospital information system. We collected echocardiographic parameters including aortic root diameter ( MRI data was analysed by experienced neuroradiologists (Guo N, Zhang K) blinded to clinical information in Functool software on GE AW workstation. We focused on deep regional CBF given that penetrating small vessels lack of collateral circulation were more vulnerable to cerebral hypoperfusion. For CBF analysis, a rectangular region of interest (ROI) with 1cm × 4 cm area were delineated in the deep regions of both hemispheres symmetrically for ve consecutive levels including centrum semiovale, roof of lateral ventricles, body of third ventricle, upper level of third ventricle and lower level of third ventricle.
Average CBF of both hemispheric deep regions on the same level was recorded as the deep regional CBF of each level. Finally, the average CBF of ve levels was de ned as the mean deep regional CBF. Based on the de nitions and imaging principal summarized in STandards for ReportIng Vascular changes on nEuroimaging (15), total CSVD neuroimaging burden was assessed according to an ordinal CSVD score(0-4)(16). One point was scored for matching each of the following four categories: number of lacunae ≧ 1; number of CMBs ≧ 1; moderate to severe basal ganglia perivascular space (BG-PVS) (17); periventricular WMH Fazekas scale 3(extending into the deep white matter) or deep WMH Fazekas scale 2 ~ 3(early con uent or con uent) (5). Additionally, cerebral atrophy was scored according to global cortex atrophy rating scale (GCA) from 0-3 (0 = absent, 1 = mild, 2 = moderate, 3 = severe) (18). Finally, we dichotomized total 103 patients as CSVD Burden low (CSVD burden scale 0 ~ 1) and CSVD Burden high (CSVD burden scale 2 ~ 4).

Statistical analysis
Data were reported as mean ± standard deviation (SD) for normally distributed variables, median (interquartile range, IQR) for skewedly distributed quantitative variables respectively and numbers (percentages) for qualitative variables. Univariate analysis of clinical factors between patients with CSVD Burden low and CSVD Burden high was assessed by Student's t tests or Mann-Whitney U tests depending on variables distribution. Pearson χ 2 test was used for categorical comparison. Multivariate linear regression analysis was performed to determine the contribution of clinical and echocardiographic parameters to mean deep CBF. Backward stepwise (likelihood ratio) binary logistic regression model was constructed to determine factors associated with higher CSVD neuroimaging burden. Statistical analysis was performed in SPSS 25.0 (IBM, Armonk, New York) and P 0.05 was considered of statistical signi cance.

Clinical characteristics
In total, 103 patients including 64 males (62.1%, mean age 63.00 ± 11.57 year) and 39 females (37.9%, mean age 63.56 ± 12.27 year) were included into the present study. Among this study population, primary causes for admission were listed brie y as follow: "lacunar stroke" (33, 32.0%), "gait disturbance or bradykinesia" (23, 22.3%), "cognition decline" (14, 13.6%), "anxiety or depression" (14, 13.6%), mixed symptoms of above (19,18.4%). The clinical characteristics including demographic information, and laboratory biomedical factors between CSVD Burden low and CSVD Burden high groups were summarized in Table 1. Patients with CSVD Burden high had higher prevalence of hypertension, diabetes mellitus and symptomatic stroke accompanied with higher fasting plasma glucose but lower eGFR compared to patients with CSVD Burden low .  Univariate linear regression and correlation analysis of deep regional CBF The average of ve levels of deep regional CBF was de ned as the "mean deep regional CBF".

Multivariate linear regression analysis of mean deep CBF and contribution of mean deep CBF to CSVD burden high
In multivariate linear regression analysis as showed in Table 3, in addition to hypertension (B coe cient, 3.141; 95% CI, 0.201 to 6.080; P = 0.037), augmented RAD (B coe cient, -0.289; 95% CI, -0.578 to -0.001; P = 0.049) and decreased LVEF (B coe cient, 32.555; 95% CI, 7.399 to 57.711; P = 0.012) were structural and functional echocardiographic parameters linked with declined mean deep regional CBF in CSVD patients respectively. As shown in Fig. 2c, multivariate linear regression model demonstrated predictive signi cance of clinical risk factors and echocardiographic parameters for evaluating mean deep CBF (R 2 = 0.328, P = 0.002). Variance in ation factors (VIF) of all included variables were less than 1.85, indicating no collinearity among included variables in the regression model.

Analysis Of Attributable Factors To Higher Csvd Burden
In binary logistic regression analysis as showed in Fig. 3b (Fig. 3c).

Discussion
In the present study, we observed a linear relationship between RAD, LVEF and mean deep regional CBF whose decline contributed to higher CSVD neuroimaging burden, suggesting that echocardiographic and cerebral hemodynamic parameters should be taken into consideration for neuroimaging burden assessment in CSVD patients absent from decompensated heart failure.
The value of LVEF, a well-established predictor for left ventricular systolic function was positively associated with deep CBF after adjustment of conventional cerebrovascular risk factors including age, hypertension, symptomatic stroke and eGFR in CSVD patients. Low LVEF prolongs arterial input function and contributes to larger hypoperfusion volumes as well as poorer poststroke outcome (19). In addition, LVFS another index of left ventricular systolic function was positively associated with deep regional CBF.
We supposed that left ventricular systolic function modi ed CBF perfusion of deep regional territories which were more susceptible to ischemia due to insu cient collateral circulation. Contrarily, no signi cant correlation was found between E/A ratio and deep regional CBF, indicating left ventricular systolic function but not diastolic function tipped the balance in cerebral perfusion.
Notably, RAD was negatively correlated with mean deep regional CBF. The association between right atrium enlargement and CBF perfusion has seldom been elucidated previously. It has been con rmed that right atrium enlargement is usually in parallel with elevated right atrial pressure (20). Moreover, elevated right atrial pressure indicates increased cerebrovascular resistance and cerebrovenous congestion (21). It is supposed that dilated right atrium may re ect elevated right atrial pressure and central venous pressure indirectly which impedes e ux of glymphatic drainage from PVS. Inadequate glymphatic clearance may ultimately contribute to accumulation of toxic metabolic by-products and subsequent neurovascular unit (NVU) immunoin ammation (22). Right atrial hemodynamics involved mechanism in aCSVD development is supported by the evidence that mean right atrial pressure is independently associated with higher WMH volume in chronic valvular heart disease (23). Collectively, cardiac structural and functional remodelling participates in aCSVD development via complicated "heart-brain axis" as showed in Supplementary Fig. 1.
It has been long believed that chronic permanent hypoperfusion is a hallmark in the pathophysiology of aCSVD (24,25). Cerebral hypoperfusion is linked with BBB leakage in CSVD patients with WMH or normal-appearing white matter (6). BBB leakage followed by NVU in ammation and energy metabolic disturbance cascade thereby is a predominant pathogenesis of aCSVD (1,26). Consistently, recent systemic review and meta-analysis con rm global CBF reduction in CSVD patients especially for those with WMH (27). Particularly, the present study focused on deep medullary territories CBF supply instead of global CBF in aCSVD population absent from chronic heart failure. The small penetrating vessels are the main CBF supply of deep parenchyma whose ischemia are predisposed to the development of WMH, CMBs and PVS. In accordance with the correlation between deep CBF in broader-zone regions (centrum semiovale level and roof of lateral ventricles level) and the burden of WMH, deep CMBs, mean deep regional CBF were independently linked with higher CSVD burden.
Though it seems no direct correlation between LVEF or RAD with CSVD neuroimaging markers, ROC curve introduced with echocardiographic parameters revealed higher predictive signi cance for CSVD burden high diagnosis. The reasons concerned with less strong direct relationship between echocardiographic parameters and CSVD neuroimaging marker burden are summarized below. Firstly, patients with cardiac embolism and moderate-severe intracranial atherosclerotic stenosis were excluded and all participants maintained normal LVEF in the present study. It is supposed that the cerebral perfusion and CSVD neuroimaging markers are more vulnerable to cardiac hemodynamics alteration in chronic heart failure but not in normal heart function state. Actually, patients with decompensated LVEF would bene t more remarkably from increased cardiac output than those with compensated LVEF (28). Secondarily, though increased LVEF may increase the CBF supply in aCSVD patients, persistently enhanced cardiac output and arterial pulsatility may contribute to cerebrovascular remodelling and elevation of cerebrovascular resistance concomitantly (29). Subsequently, long-term effect of cerebral hemodynamic stress may counteract the modest protective effect from subtle elevated LVEF in aCSVD patients absent from decompensated heart failure. It is worth noting that the regulation of cardiac output on cerebral hemodynamics is a complicated integrated effect but not direct "point to point" effect (30). Collectively, further studies strati ed heart function grade are warranted to explore the cardiac hemodynamic effects on CSVD burden.
Supplementary Fig. 1 Schematic of "heart-brain axis" mechanism Firstly, chronic arteriolosclerosis contributes to long-term microcirculation ischemia. Furthermore, decreased LVEF suggestive of cardiac output reduction results in small vessel CBF supply decline. Secondarily, enlargement of right atrium suggestive of increases cerebral venous return resistance follows by decreased interstitial uid return. Finally, the insu cient para-arterial in ux and para-venous e ux contributes to inadequate glymphatic clearance and PVS in ammation which are demonstrated as CSVD neuroimaging markers, such as WMH and CMBs in radiology.
There are some limitations in our study. Firstly, the hemodynamic data from cardiac catheterization was not available. Compared to transthoracic echocardiography, cardiac catheterization data is more convincing to elucidate the link between cardiac and cerebral hemodynamics. However, it is reasonable and reliable for CSVD patients to assess cardiac structure and function via echocardiogram instead of cardiac catheterization given the invasive operational risk and medical cost. Secondarily, it is a crosssectional study lack of follow-up data to validate the causality relation between deep regional CBF alteration and CSVD neuroimaging burden.

Conclusion
In conclusion, LVEF and RAD are functional and structural echocardiographic parameters modifying deep regional CBF whose decline is an indicator of poor small vessel circulation and higher CSVD burden.
Feasible echocardiographic and deep regional CBF assessment will provide prognostic signi cance for CSVD burden early-warning.

Declarations
Ethics approval and consent to participates This clinical study was approved by the medical ethics committee of the Third A liated Hospital of Sun Yat-Sen University and was performed in accordance with the principles illustrated in Declaration of Helsinki. All participants had been given the informed consent.

Consent for publication
No applicable.

Availability of data and materials
The datasets used and analysed during the current study are available from the corresponding author on reasonable request.
Competing interest