Admission Heart Rate Variability is Associated with Depression and Cognition in Patients with Acute Mild-Moderate Ischemic Stroke

Stroke has been shown to cause cardiac autonomic dysfunction. Depression and cognitive impairment are common complications after acute ischemic stroke (AIS). The relationship between poststroke depression (PSD) and cognitive impairment (PCI) and heart rate variability(HRV) was unclear. The purpose of this study was to investigate whether the decreased HRV was related to PSD and PCI in patients with mild-moderate AIS. Methods Changes in HRV after AIS were assessed using the nonlinear fractal dimension (FD) method, and patients within 72 hours of AIS were included in the study. 476 patients were included in this study. All patients underwent mood tests, cognitive test at 3 months. Cognitive and mood state were assessed using the Montreal Cognitive Assessment (MoCA) and the 15-item Stroke Specific Geriatric Depression Scale (GDS), respectively. PSD was defined if GDS ≥5 and PSCI was defined if MoCA<26. We assessed the relationship between FD and PSD and PSCI at 3 months.

3 improving the prognosis of patients.

Background
Stroke was the most important disabling and fatal disease in China [1]. Stroke could lead to physical disability, cognitive and emotional impairments. Post-stroke depression(PSD) was very common after stroke. Previous studies had shown that the prevalence in the first year after stroke was about 41.8% [2]. PSD might be caused by physical disability [3]. PSD had negative effects on the recovery, mortality, quality of life, physical and cognitive functions of stroke patients [4][5][6].
There were many reports about cognitive impairment after stroke [7,8]. Post-stroke cognitive impairment (PSCI) could lead to increased risk of death, decreased functional recovery, and evolve to dementia [9][10][11]. Over time, there appeared to be a complex interaction between depressive symptoms and cognitive function in stroke patients.
Cognitive impairment and depressive symptoms might overlap each other [13,14], and both might coexist in stroke patients [15,16], cognitive impairment might also be caused by depression [17].
It had been reported that stroke could cause cardiac autonomic dysfunction and heart rate variability (HRV) decreased, which reflecting poor parasympathetic regulation. Decreased HRV was also related to depression [19] and poor cognitive function [20]. Because healthy heart rates were slightly irregular and chaotic, chaos theory could better explain the dynamics of heart rate. In physiological and pathological conditions, nonlinear methods could provide new ideas for HRV research [21][22][23]. FD was a characteristic parameter in chaos theory, which could quantify the complexity of HRV and was one of the most commonly used nonlinear methods. FD method had the advantages of high precision and simple calculation. FD was related to dimensional complexity, and it estimates the selfsimilarity of a time interval in time series [24]. 4 The purpose of this study was to discuss whether the reduction of FD was related to depression and cognitive function after mild-moderate AIS.

Study Population
Patients with mild-moderate acute ischemic stroke(AIS) within 72 hours of onset were selected. All patients met the WHO diagnostic criteria, were confirmed by brain CT or diffusion weighted imaging (DWI) magnetic resonance imaging (MRI). Eligible patients were admitted to our stroke ward. The severity of stroke was assessed by the the National Institutes of Health Stroke (NIHSS) score. All study participants or their legal representatives agreed to participate in the trial and they also signed informed consent.
This study was approved by the local ethics committee.

Inclusion criteria
Only patients who met all of the following criteria were included in the study: (1). Aged from 18-80 years old; (2) Acute ischemic stroke occurred for the first time within 72hours after onset; (3) The lesion was single and related to clinical manifestations; (4). Finnishspeaking; (5). Able to co-operate; (6). NIHSS score≤8; (7). All patients were not treated with intravenous thrombolysis or mechanical thrombolysis. All patients received standard medical treatment, including aspirin and lipid-lowering drugs.

Data Collection and Scale Assessment
Detection of 12-lead ECG and HRV in the next morning 9:00 to 10:00 AM, patients with relaxation supine state, in the quiet environment, the room temperature was about 22℃.
The ECG sampling frequency was 1000hz and required 15 minutes to obtain continuous R-R interval sequences, about 2,000 beats. The R-R interval sequences were passes through a filter to eliminate interfering factors, such as noise, artifacts, and premature beats. All R-R interval sequences were automatically edited at first, and then carefully edited manually. Excluding the interfering part, only the part of > 90% pure sinus beats were included in the analysis. Finally, 512 continuous R-R interval sequences were selected for HRV analysis.
The chaotic characteristics of R-R sequences were represented by FD, which was used to quantify the complexity of the R-R dynamic changes, FD equation was FD= logN(ε)/log(1/ ε), ε was the range, which was used to monitor the R-R interval, and N(ε) was the number of the R-R intervals [29]. FD parameters of each subject were calculated automatically by off-line computer software. Decreased FD was defined as FD≤1.05 according to our previous studies [30,31].
Depression was assessed 3 months later. The patients were assessed by an experienced neurologist who blinded to the clinical study and the patient's clinical information.
The Montreal Cognitive Assessment (MoCA) was used for the assessment of cognitive impairment, with a range of 0 to 30 points. A lower MoCA score indicates more severe cognitive impairment, and a score <26 was considered to be cognitive impairment.

Statistical Analysis
Patients were divided into PSD and PSD groups, PSCI and PSCI groups. Demographic characteristics, vascular risk factors, current smoking, and so on were compared between two groups in univariate analysis, distributions of continuous variables were determined by the Kolmogorov-Smirnov test, while Mann-Whitney two sample test was applied in case of non-normal distributions. We used Pearson χ2 test, Fisher exact 2-sided test, or Student t test for data analyze. Adjusting for all confounders (such as age, baseline NIHSS score, gender, BMI, hypertension, current smoking, diabetes, hyperlipidemia, insular stroke, family history of stroke, etiology, and drug use), multivariate logistic regression was used to analyze the relationship between FD and outcomes (PSD, PSCI). The adjusted odds ratio (OR) and 95% confidence interval (CI) represent the analysis results. SPASS 22.0 software was used for all data analyze. P <0.05 was considered as statistically significant.

Characteristics of the study subjects
A total of 476 patients were included in the study, including 48.95%(233) males and 51.05%(243) females, with an average age of 65.88±10.14 years (39-80 years), there were 313 patients with hypertension, 144 with diabetes, 235 with hyperlipidemia, and 127 with smoking. The mean NIHSS score was 6.55(±1.93). During the 3-month follow-up, no patients died and no patients No patients were lost to follow-up.

Univariable Models for Predictors of PSCI
PSCI occurred in 158 patients (33.19%) at 3 months. At baseline, age and gender distribution, BMI, the prevalence of hypertension, diabetes mellitus, and blood lipid profile,and soon were similar between the two groups (table 2). Compared with no PSCI group, the prevalence of FD≤1.0 in PSCI was higher1.05(25.68%vs 15.41%, P<0.014).
However, there was no significant difference in FD value between the two groups.

Multivariable Models on the Association between FD≤1.05 and PSD, PSCI
There was an association between FD≤1. 05

Discussion
Cardiac dysautonomia was a common complication of stroke, HRV analysis was a common tool for studying cardiac autonomic control. In previous studies, linear statistical methods (time-domain and frequency-domain methods) were usually used to analyze HRV, which detected the overall amplitude of RR interval fluctuation around its mean value [32].
However, it provided very limited HRV information because the nonlinear mechanism seems to be involved in the origin of heart rate dynamics [33].
FD was one of the most common nonlinear parameters in chaotic characteristics, and could quantify the complexity of HRV [24]. FD algorithm estimated the self-similarity of a time interval in time series, which was related to the complexity of time series. In this study, FD was used to evaluate the status of autonomic function of AIS.

HRV had been the focus of research on biomarkers for depression and cognition after AIS
in the last couple of decades [34], only the traditional time linear method was considere [35,36]. In this study, we investigated the relationship between FD, depression and cognition after AIS Our results showed significant differences between groups of patients with PSD and no PSD. Lower FD values and the higher prevalence of FD≤1.05 in PSD group compared to no PSD group. This result might be due to the reduced ability of the parasympathetic nervous system to regulate heart rate through vagal activity [37,38], and the reduced short-term flexibility of ANS to adapt to environmental and task changes [39-41].
Interestingly, we also founded that the incidence of decreased FD≤1.05 was higher in the PSCI groups, but FD value in early poststroke phase did not differ in the two groups. It was possible that because of the subject we studied, stroke patients are mild-moderate acute ischemic stroke, and the low incidence of PSCI in the mild-moderate stroke. This study 9 showed that cognitive impairment was more likely in patients with FD≤1.05. The results of this study support the relationship between low vagal tone and poor cognitive function.
The prediction of cognitive function by the HR parameters measured could be explained by the well-known effect of post-stroke depression on cognitive function in early phase.
However, it had recently been reported that reduced HRV at rest was associated with reduced whole-brain perfusion, which itself was associated with an increased risk of poststroke dementia. Therefore, the effect of decreased HRV on chronic cognitive function may be mediated by cerebral perfusion insufficiency [42,43].
After adjusting for fully confounders, the results showed that there were significant association of FD≤1.05 with risks of 3-month PSD and PSCI after AIS. To conclude, This study found that FD measurement was helpful in predicting PSD and PSCI in the early stage of stroke, especially in the case of severe aphasia or cognitive impairment, in which questionnaire could not be used.
However, due to some limitations, these results must be carefully interpreted and could not be generalized to all stroke patients. First, the patients were mild-moderate stroke, and severe stroke was excluded. Second, the absence of FD measurements at the 3month, which did not clarify the causal relationship between cardiac autonomic function, depression, and cognitive impairment after stroke. Third, although we adjusted the NIHSS score, the NIHSS score was related to the infarct volume, we did not measure the infarct volume in this study. Fourth, previous studies had shown that bilateral insular lesions might have different effects on cardiac autonomic function. We did not analyze the possible effects of bilateral insular stroke on FD, PSD and PSCI. In addition, larger multicenter clinical studies should be needed to confirm our results, which may provide a better understanding of the pathophysiological mechanisms between poststroke autonomic function, cognitive impairment, and mood disorders. However, despite these limitations, the advantage of our study lies in its large sample size, the adjustment of various confounders in its analysis, and the use of standardized methods to measure FD.

Conclusions
In summary, our findings suggested that decreased FD in AIS was associated with an increased risk of PSD and PSCI at 3 months. Low HRV may be a sign of higher susceptibility to PSD and PSCI. FD has potential predictive value for PSD and PSCI after ischemic stroke.

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the 11 corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.

Funding
This work was funded by the Health and Family Planning Commission of Chengdu (2015009), which is not involved in the database management (collection, analysis, interpretation of data) and has no access to patient information. The funding body did not participate in designing the study or writing the manuscript. The study protocol has undergone peer-review process by the funding body.

Authors' contributions
LYH was responsible for data collection and analysis, as well as the writing of the first draft and subsequent drafts of the paper. RX and JW were responsible for the design and interpretation of the study. LLZ was responsible for data analysis. WD was responsible for the conceptual interpretation of the study, and HY was responsible for the design of FD