Sage Journals: Discover world-class research

Abstract

BACKGROUND:

Cardiac dysfunction accompanies acute ischemic stroke and affects the effective implementation of early rehabilitation interventions. There is a lack of reference hemodynamic data on cardiac function in the subacute phase of ischemic stroke.

OBJECTIVE:

In this study, we aimed to identify appropriate cardiac parameters for exercise training utilizing a pilot study.

METHODS:

We used a transthoracic electrical bioimpedance non-invasive cardiac output measurement (NICOM) device to monitor cardiac function in real time for two groups [i.e., subacute ischemic stroke inpatients group ( $n=$ 10) and healthy control group ( $n=$ 11)] using a cycling exercise experiment. The parameters of both groups were compared to highlight the cardiac dysfunction in the subacute phase in patients with ischemic stroke.

RESULTS:

We considered stroke volume index (SVI) and systemic vascular resistance index (SVRi) as the primary outcomes, and there was significant intragroup difference (stroke group: $P<$ 0.001; control group: $P<$ 0.001, using one-way ANOVA) and significant intergroup difference at each individual time segment ( $P<$ 0.01, using independent $t$ -test). Among the secondary outcomes, i.e., cardiac index (CI), ejection fraction (EF), end-diastolic volume (EDV), and cardiac contraction index (CTI), we found significant intergroup differences in CI, EF, and CTI scores ( $P<$ 0.01, using independent $t$ -test). Significant interaction with respect to time and group were seen only in the SVRi and CI scores ( $P<$ 0.01, using two-way ANOVA). There was no significant inter- or intra-group differences in EDV scores.

CONCLUSION:

SVRI, SVI, and CI values highlight cardiac dysfunction in stroke patients the most. At the same time, these parameters suggest that cardiac dysfunction in stroke patients may be closely related to the increased peripheral vascular resistance caused by infarction and the limitation of myocardial systolic function.

Keywords

Cycling training exercise ischemic stroke inpatients non-invasive cardiac output poststroke cardiac dysfunction

1. Introduction

Concurrent cardiac dysfunction is common in patients with cerebrovascular disease [1]. Typically, ischaemic stroke alters the autonomic and neurohormonal pathways involved in the control of cardiac function, induces neurovascular uncoupling and disrupts cerebral autoregulation, making cerebral blood flow directly dependent on cardiac function [2, 3, 4, 5].

Previous reviews have estimated that myocardial injury, ischemia-like electrocardiographic changes, and arrhythmias are frequently encountered in acute stroke patients, even in the absence of primary heart disease. These findings support a central nervous system (CNS) origin for these electrocardiographic abnormalities [6, 7, 8]. Sposato et al. pointed out that atrial fibrillation can be detected by cardiac monitoring in up to 20% of patients with stroke [9]. In a meta-analysis of over 50,000 participants, Gunnoo et al. reported that the incidence of cardiovascular complications in the acute phase of ischemic stroke ranged from 3% for myocardial infarction to $>$ 50% for asymptomatic coronary stenosis [10, 11]. Independent research studies by Xiong and Prosser found that after ischemic stroke, 24% of patients had autonomic dysfunction [12, 13], 28% had impaired left ventricular ejection fraction, and 13% to 29% had systolic dysfunction [14]. Likewise, impaired cardiac function after ischemic stroke increases the risk of worse neurological outcome and 90-day disability. Wolf et al. reported that the incidence of acute ischemic stroke increased fivefold for patients with atrial fibrillation [15]. Furthermore, van der Bilt et al. found via a meta-analysis that consisted of 25 studies with a total of 2,690 patients that cardiac dysfunction was associated with an increased risk of death, delayed cerebral ischemia, and poor outcome after stroke [16]. It has been reported that cardiac dysfunction can both worsen pre-existing cerebral damage and cause new brain injury [17].

Hence, currently, early detection of cardiac function in patients with ischemic stroke is regarded as essential for exposure rating and appropriate rehabilitation planning. However, most of the literature to date, including current clinical guidelines, have focused only on monitoring cardiac function within 24 hours of a patient suffering an ischemic stroke [18]. There is a paucity of literature on hemodynamic data collected for guiding cardiac rehabilitation training in patients who are in the subacute phase. According to conventional concepts, the most serious complications after ischemic stroke have been identified in the acute phase. On the contrary, recent studies have shown that damage to cardiac function still accounts for a large proportion in ischemic stroke inpatients [19, 20, 21, 22]. Besides, relevant studies have also pointed out that continuous monitoring of cardiac function in the subacute phase can help to prevent secondary stroke and reduce complications caused by cardiovascular events. Yoshimura et al. evaluated patients at rest and found that cardiac dysfunction that manifested during the acute phase after brain injury usually resolves over the following several weeks alongside improvement of neurological function [23]. Nevertheless, whether cardiac dysfunction has also improved when patients were evaluated during exercise is still controversial.

In patients with ischemic stroke, penumbral perfusion is maintained by collateral flow, and specifically maintained by stabilizing normal mean arterial pressure (MAP) levels in acute interventions [24]. This is because MAP is related to cerebral perfusion (CP), as constant cerebral blood flow (CBF) is maintained over a wide range of MAP due to vessel auto-regulation [25]. According to Deegan et al., MAP also correlates with cardiac output (CO), since MAP is the product of CO and systemic vascular resistance (SVR) [26]. In a study by Marik et al., drug administration schedules were adjusted to enhance CO according to the Frank-Starling curve and SVR via fluid and/or catecholamine administration, to increase MAP levels of patients [27]. Earlier studies have demonstrated that the correlation of MAP levels to CO are volume state-dependent. Therefore, in areas with impaired autoregulation, CO may be a better parameter to focus on, as CBF is dependent on CO rather than on MAP.

In recent decades, measurement of CO using non-invasive or minimally invasive devices has gained popularity due to its advantages of safety, convenience, and high repeatability [28, 29, 30]. Transthoracic electrical bioimpedance (TEB) is an easy-to-use approach among various non-invasive cardiac output measurements (NICOM) as it provides rapid hemodynamic information in clinical practice [30]. The method of hemodynamic monitoring by TEB was popularized in clinical medicine by the extensive studies by Shoemaker et al. [31], illustrating a calculation of the impedance from the voltage changes by passing a small current through the patient’s thorax. The first derivative $\left({\frac{\mbox{dZ}}{\mbox{dt}}}\right)$ of the impedance waveform is calculated from the time-impedance curve. As the blood flow through the aorta determines most of the change in the impedance, it is possible to compute the mean blood flow in the aorta from the changes in $\left({\frac{\mbox{dZ}}{\mbox{dt}}}\right)$ over time, using the Kubicek formula [32]. This flow is directly proportional to the left ventricular stroke volume (SV). An SV constitutes a key derivation in this method. Using the heart rate and mean arterial blood pressure, which are measured non-invasively and simultaneously along with the SV, several hemodynamic parameters such as CO, systemic vascular resistance, etc. can be calculated.

In this research, we proposed a pilot study to compare the cardiac function in subacute inpatients with ischemic stroke (first onset $<$ 6 months) and healthy controls using the NICOM approach during exercise. We hypothesize that patients continue to sustain cardiac dysfunction in the subacute phase when exercising. We also sought to explore suitable guiding hemodynamic parameters to improve the protocol for early training in ischemic stroke.

2. Methods

This pilot controlled trial was conducted at the Rehabilitation Medicine Department of the Longhua People’s Hospital in Shenzhen City, China. The experimenter was blinded to data analysis, and data analysts did not participate in conducting the experiment. The study was conducted with approval from the Longhua Human Medical Ethics Committee (for Research) [2021(112)]. The study was registered in the clinical registration database under number ChiCTR2100047079.

2.1 Recruitment of study respondents

After obtaining approval for the study from the Institutional Review Board, we screened 50 ischemic stroke inpatients admitted between February and April 2021 to the Longhua People’s Hospital. The inclusion criteria for stroke inpatients were: (1) aged between 30 and 60 years old; (2) diagnosed with a unilateral brain lesion due to ischemic stroke, first onset and within 6 months of onset, (3) had no other neurological deficits or secondary onset; (4) had satisfactory cognition (as measured by the Mini-Mental State Examination [MMSE $>$ 21]) to understand the content or purpose of the study, and to follow simple instructions during the assessment [33]; (5) had sufficient muscle strength (as measured by the Manual Muscle Test [MMT $\geqslant$ 3]) for the lower limb cycling exercise; (6) had not participated in similar cycling exercise any time prior to the current experiment. We initially conducted a pre-test of cycling for 3-min with a 12–16 km/h speed. After the pre-test, we used the Borg fatigue index score to evaluate participants. People who obtained scores of 13–15 were recruited in this study. Furthermore, participants who were pregnant or who had been diagnosed with dysphasia, cardiogenic infarction (either in the heart or brain), or osteoarticular injury (in lower limbs) were also excluded from the research. Subsequently, we recruited healthy adults to form a control group with the following inclusion criteria: (1) aged between 30 and 60 years old; (2) no history of any neurological deficit or cognitive disease; (3) no lower limb motor dysfunction due to any kind of osteoarticular or peripheral neuromuscular disease. Similar to the stroke group, pregnant participants were excluded from the research. There were no gender-specific requirements for the recruitment of participants. All participants were screened by a collaborating clinician to determine eligibility according to the inclusion criteria 3–7 days before the experiment. The experimenter gave an orientation about the content and purpose of the study to all potential participants, and the recruited participants gave their written consent for participation. A final of 10 inpatients with subacute ischemic stroke (i.e., the experimental group, EG) and 11 healthy controls (i.e., the control group, CG) were recruited for the study.

2.2 Experimental setup and protocol

We used a TEB-NICOM device (Enduro, Physioflow@ Manatec, France) to measure the cardiac hemodynamic parameters in both groups in the study. An illustration of the system is shown in Fig. 1A. This commercial monitoring system provides continuous non-invasive cardiac output monitoring. The device recorded ECG and measured chest bioimpedance, which changes with the respiratory and cardiac cycles. Bioimpedance changes associated with ECG were used to calculate stroke volume and other CO parameters automatically every 10 seconds with the NICOM system. The data were instantly displayed on the PC screen and saved for offline processing.

Figure 1.

(A) An illustration of the TEB-NICOM system adopted in this study. (B) A demonstration of the electrode configuration to conduct the non-invasive cardiac output measurement with the device shown above. (The figure is amended from original image of Atlas of Human Anatomy, Xichi Ltd, China).

Before starting the measurement, the participant was instructed to lie on a bed in the supine position. After skin preparation, 6 electrodes were placed on the neck and chest. The configuration of the electrodes is shown in Fig. 1B. The white electrode was placed behind the left side of the neck under the earlobe. The blue electrode was placed in the angle between the left neck clavicle and the sternocleidomastoid muscle. The dark orange electrode was placed in the second intercostal space at the left edge of the sternum. The light orange electrode was placed in the fifth intercostal space at the front of the left axilla. The green electrode was placed in the left paraspinal region flush with the height of xiphoid, and the black electrode was placed directly 1 cm below the green. The six electrodes were connected to a wireless signal transmission box. Subsequently, the experimenter operated the device in the following sequence: launch the detection software; connect the signal transmission box with the NICOM device via Bluetooth; enter individual demographic information; and conduct system calibration.

After the system was calibrated, the participants were informed that the experiment was starting. They were required to first lie in the bed calmly for 3 minutes, then sit up and sit quietly for 3 minutes, and finally they moved to the training platform for the seated cycling exercise for 15 minutes, as shown in Fig. 2. The specific speed and resistance parameter settings for each individual were determined according to the 3-minute cycling pre-test to ensure that all participants received medium-intensity exercise. The resistance and speed remain unchanged during the entire 15 minutes of exercising. Real-time NICOM monitoring was done throughout the whole process. After the exercise ended, the Borg fatigue index score was measured again. All participants got scores of 13–15.

Figure 2.

A subject was instructed to perform three steps of actions in sequence. (1) lie in bed peacefully for 3 minutes; (2) sit up and sit quietly for 3 minutes; (3) cycling exercise under the seat for 15 minutes.

2.3 Processing of data

The data that was computed online was exported from the NICOM device. Initially, we cut the raw data flow into three segments [i.e., recumbent resting state (T1), sitting resting state (T2), and cycling state (T3)] according to the sequential process of the experiment with manually tagged markers. Then, we performed the data preprocessing within each individual segment, and we deleted 30-second-long data segments both at the beginning and end of the segment to avoid errors caused by switching between tasks. After that we eliminated data points exceeding 3 times the standard deviation beyond the mean value in each segment to reduce ambient noise. Additionally, to observe the gradual CO variances throughout the duration of the cycling exercise, we further divided the T3 segment into six sub-segments (i.e., T3A, T3B, T3C, T3D, T3E, and T3F), which contained data of 150 seconds. In the subsequent analysis, we not only observed the temporal changes in the entire line of data, but also averaged the data points of each segment to explore the differences in CO parameters of the two groups in each time period.

2.4 Statistical analysis

All statistical analyses were performed using IBM SPSS 22. We selected the following main CO parameters as the primary CO parameters in our study: SVI, SVRi, cardiac contraction index (CTI), CI, EF, and EDV. Two-way analysis of variance (ANOVA) was used to evaluate the differences with respect to the independent factors of the group (i.e., the ischemic stroke participants and the healthy controls) and segment period (i.e., T1, T2, T3A, T3B, T3C, T3D, T3E, and T3F). Then, we performed one-way ANOVA to investigate the intragroup difference of either the EG or CG group at different segment periods using the Bonferroni post hoc test. Finally, the intergroup difference of the two groups at each segment period was analyzed using independent $t$ -test. The primary outcomes of the study were the SVI and SVRi scores, and the other scores were the secondary outcomes. This was because SVI reflects the ability of myocardial fiber contraction and plays a role in evaluating prognosis, while SVRi is a quantitative index reflecting cardiac afterload and the patency of cerebral vascular microcirculation. The levels of statistical significance were indicated at 0.05, 0.01, and 0.001 in this study.

3. Results

The cycling exercise and NICOM evaluation were completed by all the recruited participants. No participant felt unwell or withdrew during the entire experiment. The two groups were examined using Fisher exact tests for their demographic data ( $P>$ 0.05), as shown in Table 1. The two groups were not different in terms of age, gender, and handedness. Table 2 presents the results of intergroup differences (including probabilities and effect sizes) with respect to time and group. Table 3 summarizes the probabilities and effect sizes of the intragroup evaluation with respect to each time segment, respectively, for the two groups. Table 4 summarizes the six major CO parameters measured in this study, namely, the means and 95% confidence interval of each parameter together with the independent $t$ -test probabilities.

Table 1
Demographic data of the participants in both groups

Characteristics	Groups		$P$ value
	EG ( $n=$ 10)	CG ( $n=$ 11)
Age (yrs)	53.7 $\pm$ 8.9	44.6 $\pm$ 5.93	0.614
Time since stroke (days)	49.3 $\pm$ 7.2	N/A	N/A
Gender (male/female)	7/3	7/4	0.986
Stroke side (left/right)	6/4	N/A	N/A
Handedness (left/right)	1/9	0/11	0.899

No significance was found between the two groups. EG, Experimental group; CG, Control group.

Table 2

The results of intergroup differences (including probabilities and effect sizes) with respect to time and group by two-way ANOVA

Two-way ANOVA	Time		Group		Time * Group
	Sig	Partial $\eta$ ${}^{2}$	Sig	Partial $\eta$ ${}^{2}$	Sig	Partial $\eta$ ${}^{2}$
SVi	0.002	0.135	0.000	0.187	0.850	0.022
SVRi	0.000	0.570	0.000	0.334	0.001	0.154
CI	0.000	0.554	0.000	0.304	0.000	0.211
CTI	0.019	0.103	0.000	0.253	0.861	0.021
EF	0.439	0.044	0.000	0.196	0.997	0.006
EDV	0.309	0.052	0.344	0.006	0.979	0.010

Table 3

The results of intragroup evaluation by one-way ANOVA, with respect to the timing segments (i.e. T1, T2, T3A, T3B, T3C, T3D, T3E and T3F) for the experimental group (EG) and the control group (CG) respectively

One-way ANOVA	EG		CG
	Sig	Partial $\eta$ ${}^{2}$	Sig	Partial $\eta$ ${}^{2}$
SVi	0.731	0.058	0.001	0.261
SVRi	0.000	0.345	0.000	0.738
CI	0.000	0.330	0.000	0.686
CTI	0.249	0.114	0.137	0.125
EF	0.978	0.021	0.978	0.021
EDV	0.959	0.027	0.342	0.091

Figure 3.

The variation of all CO parameters of both groups are shown in time sequence.

Figure 3 shows the variation of all CO parameters of both groups in the time series. In the figure, the red solid line represents the mean value variations of ischemic stroke patients, while the blue dotted line represents the mean changes in the healthy control group. From the six subgraphs, it is evident that there are differences in some of the CO parameters between the ischemic stroke group and the healthy control group, and there are also similar characteristics. There were significant differences between the two groups in the primary outcome results (i.e., SVI and SVRI), affirming our hypothesis.

3.1 Primary outcomes

Our two-way ANOVA analysis revealed that there were significant differences in SVRi scores with respect to the time ( $P<$ 0.001, partial $\eta^{2}=$ 0.570), group ( $P<$ 0.001, partial $\eta^{2}=$ 0.334), and the interaction between two aspects ( $P<$ 0.001, partial $\eta^{2}=$ 0.211), as listed in Table 2. Moreover, as the experiment progressed, both groups exhibited significant decrement of SVRi from resting state as they performed the cycling (Stroke group: $P<$ 0.001, partial $\eta^{2}=$ 0.345; Control group: $P<$ 0.001, partial $\eta^{2}=$ 0.738, one-way ANOVA, Table 3). The changes mainly occurred in the transition phase of T2, T3A, and T3B segments. Furthermore, independent t-test analysis showed significant intergroup differences in the T3 exercising time segments (i.e., T3A, $P=$ 0.025; T3B, $P=$ 0.002; T3C, $P=$ 0.001; T3D, $P<$ 0.001; T3E, $P<$ 0.001; and T3F, $P<$ 0.001). There was no significant intergroup difference during both resting segments when the participants were lying down or sitting, as shown in Fig. 3A.

Regarding the SVi scores, we observed significant differences with respect to time ( $P=$ 0.002, partial $\eta^{2}=$ 0.135) and group ( $P<$ 0.001, partial $\eta^{2}=$ 0.187), whereas no significant difference was observed with respect to the term of interaction (i.e., time*group). Only the control group demonstrated significant intragroup differences during the experiment process ( $P=$ 0.001, partial $\eta^{2}=$ 0.261, Table 3), as measured by one-way ANOVA. Yet, the trend of the whole SVI curve of the stroke group was relatively stable, as shown in Fig. 3B. The trend of SVI curve of the control group had relatively decreased in the transition phase between T1and T2, and then increased at the beginning of the T3A phase. Significant intergroup differences could be observed in the following time segments, respectively: T1, $P=$ 0.016; T3B, $P=$ 0.042; T3D, $P=$ 0.026; T3E, $P=$ 0.016; and T3F, $P=$ 0.024, in our independent $t$ -test analysis.

3.2 Secondary outcomes

In addition to the primary outcomes, we noticed that other parameters also revealed the similarities and differences in cardiac function between the two groups of participants during cycling, as shown in Fig. 3.

Figure 3C indicates the variation in CI scores across the whole experiment process for both the stroke and control groups. Significant differences were observed with respect to the term of time ( $P<$ 0.01, partial $\eta^{2}=$ 0.554) and group ( $P<$ 0.01, partial $\eta^{2}=$ 0.304), as well as the interaction between both aspects ( $P<$ 0.001, partial $\eta^{2}=$ 0.211) as measured by two-way ANOVA. Significant intragroup differences were marked for both groups in the subgraph, i.e., for the stroke group, $P<$ 0.001, partial $\eta^{2}=$ 0.330; and for the control group, $P<$ 0.001, partial $\eta^{2}=$ 0.686, as per one-way ANOVA analysis with the Bonferroni post-hoc test. The CI scores of the stroke group increased steadily with time as the experiment progressed. Meanwhile, the CI scores of the control group switched with the change of position from lying to sitting, where the CI index decreased first, and then increased rapidly as the respondents began cycling. The significant difference between the groups was as follows from the sitting resting state to the end of the experiment, i.e., T2, $P=$ 0.006; T3A, $P=$ 0.031; T3B, $P=$ 0.004; T3C, $P=$ 0.002; T3D, $P<$ 0.001; T3E, $P<$ 0.001; T3F, $P=$ 0.002, as per the independent $t$ -test analysis.

Figures 3D and E present the means of CTI and EF scores, respectively. It is evident from the subgraphs that although the relevant parameters of the two groups were different in different time segments of the

Table 4
A summarization of the six major CO parameters measured in this study, including the means and 95% confidence interval of each parameter together with the independent $t$ -test probabilities to compare the intergroup difference at each time segment (i.e. T1, T2, T3A, T3B, T3C, T3D, T3E and T3F)

Independent $T$ -test	Group	SVi		SVRi		CI		CTI		EF		EDV
		Mean	Sig	Mean	Sig	Mean	Sig	Mean	Sig	Mean	Sig	Mean	Sig
T1	EG	41.4 (33.6 $\sim$ 49.2)	0.016	2308.8 (2021.3 $\sim$ 2596.3)	0.153	3.4 (3.0 $\sim$ 3.9)	0.886	142.2 (104.2 $\sim$ 180.1)	0.004	59.5 (47.6 $\sim$ 71.4)	0.061	130.2 (108.6 $\sim$ 151.7)	0.621
	CG	49.6 (43.2 $\sim$ 56.0)		2108.3 (1782.8 $\sim$ 2433.7)		3.4 (3.1 $\sim$ 3.7)		221.4 (152.7 $\sim$ 290.0)		68.2 (60.3 $\sim$ 76.1)		134.8 (114.6 $\sim$ 154.9)
T2	EG	40.9 (33.9 $\sim$ 48.0)	0.482	2220.2 (1917.9 $\sim$ 2522.4)	0.475	3.6 (3.2 $\sim$ 4.1)	0.006	134.8 (98.4 $\sim$ 171.2)	0.068	58.2 (48.9 $\sim$ 67.5)	0.131	131.6 (107.6 $\sim$ 155.5)	0.453
	CG	43.3 (35.4 $\sim$ 51.2)		2327.9 (1960.7 $\sim$ 2695.1)		3.2 (3.0 $\sim$ 3.4)		184.2 (111.3 $\sim$ 257.1)		64.5 (55.5 $\sim$ 73.6)		124.1 (103.1 $\sim$ 145.1)
T3A	EG	45.6 (35.1 $\sim$ 56.2)	0.120	1795.2 (1486.8 $\sim$ 2103.6)	0.025	4.5 (3.8 $\sim$ 5.1)	0.031	165.9 (115.6 $\sim$ 216.2)	0.015	60.8 (50.3 $\sim$ 71.4)	0.050	139.4 (116.7 $\sim$ 162.1)	0.966
	CG	52.4 (44.0 $\sim$ 60.8)		1489.8 (1224.0 $\sim$ 1755.7)		5.2 (4.4 $\sim$ 6.0)		260.0 (159.0 $\sim$ 360.9)		70.0 (60.4 $\sim$ 79.6)		139.0 (114.6 $\sim$ 163.3)
T3B	EG	46.6 (34.9 $\sim$ 58.4)	0.042	1695.2 (1387.6 $\sim$ 2002.9)	0.002	4.7 (3.9 $\sim$ 5.6)	0.004	181.6 (122.5 $\sim$ 240.6)	0.013	62.4 (51.6 $\sim$ 73.2)	0.036	138.7 (112.8 $\sim$ 164.5)	0.636
	CG	56.3 (47.8 $\sim$ 64.8)		1239.6 (980.8 $\sim$ 1498.5)		6.3 (5.1 $\sim$ 7.5)		300.8 (175.7 $\sim$ 425.8)		72.8 (62.4 $\sim$ 83.2)		144.0 (119.4 $\sim$ 168.6)
T3C	EG	47.3 (35.5 $\sim$ 59.2)	0.061	1743.3 (1367.2 $\sim$ 2119.4)	0.001	4.8 (3.9 $\sim$ 5.8)	0.002	184.8 (128.3 $\sim$ 241.2)	0.031	62.4 (52.8 $\sim$ 72.1)	0.051	140.7 (112.1 $\sim$ 169.4)	0.670
	CG	56.2 (47.8 $\sim$ 64.7)		1207.3 (959.6 $\sim$ 1455.0)		6.5 (5.4 $\sim$ 7.6)		296.7 (154.6 $\sim$ 438.8)		72.2 (60.6 $\sim$ 83.8)		146.0 (118.9 $\sim$ 176.4)
T3D	EG	46.6 (35.4 $\sim$ 57.7)	0.026	1803.1 (1425.6 $\sim$ 2180.5)	0.000	4.8 (3.8 $\sim$ 5.8)	0.000	180.0 (123.9 $\sim$ 236.2)	0.019	62.1 (52.7 $\sim$ 71.5)	0.035	139.2 (111.6 $\sim$ 166.8)	0.517
	CG	57.0 (48.4 $\sim$ 65.7)		1157.3 (920.5 $\sim$ 1394.0)		6.7 (5.6 $\sim$ 7.9)		304.6 (160.5 $\sim$ 448.6)		72.6 (61.0 $\sim$ 84.2)		147.4 (118.3 $\sim$ 176.4)
T3E	EG	46.5 (36.3 $\sim$ 56.6)	0.016	1761.9 (1431.0 $\sim$ 2091.9)	0.000	4.9 (3.9 $\sim$ 5.8)	0.000	175.3 (119.1 $\sim$ 231.4)	0.011	61.5 (52.4 $\sim$ 70.7)	0.023	140.4 (113.7 $\sim$ 167.0)	0.531
	CG	57.3 (48.6 $\sim$ 66.1)		1115.0 (863.2 $\sim$ 1366.8)		7.0 (5.7 $\sim$ 8.3)		306.4 (170.5 $\sim$ 442.3)		72.7 (61.4 $\sim$ 84.1)		148.4 (117.7 $\sim$ 179.1)
T3F	EG	46.8 (36.8 $\sim$ 56.8)	0.024	1711.5 (1389.2 $\sim$ 2033.9)	0.000	5.0 (3.9 $\sim$ 6.1)	0.002	178.3 (118.3 $\sim$ 238.3)	0.018	61.7 (52.6 $\sim$ 70.7)	0.026	140.8 (117.2 $\sim$ 164.4)	0.579
	CG	56.8 (48.2 $\sim$ 65.4)		1117.5 (852.1 $\sim$ 1382.9)		7.1 (5.6 $\sim$ 8.7)		299.0 (163.2 $\sim$ 434.8)		72.4 (61.4 $\sim$ 83.4)		147.5 (117.4 $\sim$ 177.6)

experiment, the trend of the mean value of the two groups was quite similar. This is consistent with the results of two-way ANOVA, which shows significant differences with respect to time ( $P=$ 0.019, partial $\eta^{2}=$ 0.103 for CTI) and to group ( $P<$ 0.001, partial $\eta^{2}=$ 0.253 for CTI; $P<$ 0.001, partial $\eta^{2}=$ 0.196 for EF), with no significant results with respect to the interaction between the two aspects. Meanwhile, the trend of change in the CTI and EF curve shown in Fig. 3D–E did not show significant statistical intragroup difference as calculated with one-way ANOVA ( $P>$ 0.05). Nevertheless, significant intergroup differences exhibited in T1 ( $P=$ 0.004), T3A ( $P=$ 0.015), T3B ( $P=$ 0.013), T3C ( $P=$ 0.031), T3D ( $P=$ 0.019), T3E ( $P=$ 0.011), and T3F ( $P=$ 0.018) in CTI scores, and they appeared in T3B ( $P=$ 0.036), T3D ( $P=$ 0.035), T3E ( $P=$ 0.023), and T3F ( $P=$ 0.026) in EF scores, as measured by independent $t$ -test.

As shown in Fig. 3F, there were no significant intergroup differences or intragroup differences in EDV scores. The similarity of both curves was relatively high in the subgraph when compared to the curves in subgraphs of other hemodynamic data.

4. Discussion

In this study, we compared the CO parameters of subacute ischemic stroke inpatients with those of healthy adults during a bicycle exercise test. The results confirmed our previous conjecture that although some stroke patients did not show cardiac dysfunction at rest, and their CO parameters were similar to those of healthy people, the characteristics of cardiac dysfunction could be prominently seen in the process of exercise initiation and exercise maintenance. Table 4 shows the CO parameters and the independent T-test results in detail at each time segment (i.e., T1, T2, T3A, T3B, T3C, T3D, and T3F). The radar map in Fig. 4 provides a holistic perspective to show the changes across time of all the six CO parameters of both groups.

Figure 4.

A demonstration of the timing changes of the parameters of the two groups of subjects in each stage of the experiment via the radar map.

For example, in this study, there was no difference in SVRi between the two groups in the resting state, which reflected that the cardiac afterload function in the subacute stage of ischemic stroke improved since the ischemic stroke when the patient was at rest. According to this phenomenon, researchers speculated that the cardiac function of patients with ischemic stroke can recover within one month after the onset of stroke [34]. However, in this study, we found that the SVRi of ischemic stroke inpatients was significantly higher than that of the healthy control group throughout, from the beginning to the end of cycling exercise. On the one hand, it directly indicates that the peripheral vascular resistance of ischemic stroke patients could hardly be autoregulated under exercising state, and on the other hand, it indirectly suggests that the cerebral microcirculation may not be unobstructed during this period. In the past, the view that the risk of cardiac dysfunction of stroke mainly occurs within 24 hours has been supported by ECG research [35], but given the current need for early rehabilitation training, there is still an urgent lack of data pertaining to ECG monitoring during exercise in patients in the subacute phase. Future research in this aspect will be helpful to improve the prompt warning of risk in early training and assist in the formulation of rehabilitation planning.

Another primary outcome we have focused on is the SVI score. At present, this index can be used to predict the recovery of patients’ cardiac and cerebral function and to warn of secondary stroke risk. A recent study has shown that the decrease of SVI is closely related to time spent below saturation of 90% [36]. Real-time SVI monitoring in early rehabilitation training is helpful to remind therapists to pay attention to patients’ saturation, thus improving the ability for early warning of risk and avoiding injury. In this study, the SVI value of the patients was significantly lower than that of healthy controls in five time segments, i.e., T1, T3B, T3D, T3E, and T3F. This result suggests two possible phenomena: one is that the myocardial systolic function of patients with subacute stroke had not fully recovered, which is based on the gentle slope of the SVI curve throughout the whole exercising process; the other is that patients with subacute stroke are intolerant to more than 10 minutes of moderate exercise during early rehabilitation training. In the experiment, the patients’ saturation in the whole process was not lower than 90% based on continuous monitoring, and the Borg fatigue index of patients was also within the appropriate range after the exercise. Despite this, the SVI of patients still did not increase significantly in the group, suggesting that there could be an obstacle to myocardial systolic function during exercise.

In this study, apart from the relevant parameters of the primary outcome, the results of the CI scores are also noteworthy. Resting CI score has been widely used to evaluate cardiac function in clinical practice. In this study, we found that the resting CI of the healthy participants decreased significantly from the lying position to the sitting position, and then increased rapidly from the sitting position to exercise initiation, while that of the stroke patients were relatively stable throughout the whole process. The phenomenon is consistent with the two-way ANOVA results with respect to the interaction aspect. In the process of exercise initiation and exercise maintenance, the CI value of ischemic stroke patients remained significantly lower than that of healthy controls. We believe that not only the CI value during rest, but also the CI value during exercise has clinical significance for ischemic stroke rehabilitation. When seen along with the SVI value, a possible explanation is that the impairment of myocardial contractility has not been overcome at this stage, hence appropriate cardiopulmonary function training is needed to assist the progress of exercise rehabilitation. In a study by Fuhrer et al., the authors found that there was a correlation between CI value and MAP value, and MAP directly affected the functional recovery of lateral branch perfusion in the penumbra. Therefore, paying attention to CI value in the process of early rehabilitation training will not only help to improve the efficiency of exercise training from the cardiac perspective, but also facilitate the spontaneous recovery of the central nervous system.

The diversity and innovation of medical devices contribute greatly to improving the quality and efficiency of medical services [37]. Artificial intelligence (AI) can be used in medicine to diagnose, treat and predict diseases [38]. The results of Almina et al.’s study showed that the overall performance of AI in cardiovascular application models was above 83%, which could be used as a diagnostic tool in the future [39]. Rachel et al. found that the introduction of machine learning (ML) algorithms into medical device (MD) management strategies benefit healthcare organizations not only in terms of improving the safety and quality of patient diagnosis and treatment, but also in terms of cost optimization and resource management [40].

5. Conclusions

In this study, we confirmed our hypothesis that patients with subacute stroke still have significant cardiac dysfunction. We monitored the cardiac output function of two groups in real time using non-invasive cardiac output. On further comparison, we found that of the six parameters observed, SVRI, SVI and CI values were the most indicative of cardiac dysfunction in stroke patients. At the same time, these parameters suggest that cardiac dysfunction in stroke patients may be closely related to the increased peripheral vascular resistance caused by infarction and the limitation of myocardial systolic function.

Footnotes

Conflict of interest

None to report.

Funding

This study was supported by the Traditional Chinese Medicine Bureau of Guangdong Province (Grant number 20211346) and Shenzhen Science and Technology Program (Grant number: RCBS20210609104401012).

References

Koroshetz

Benner

Vangel

Melinosky

Arsava

Ayata

Zhu

Schwamm

Sorensen

. Neuroanatomic correlates of stroke-related myocardial injury. Neurology. 2006 May 9; 66(9): 1325-9. doi: 10.1212/01.wnl.0000206077.13705.6d. Epub 2006 Mar 8. PMID: 16525122.

Battaglini

Robba

Lopes da Silva

Dos Santos Samary

Leme Silva

Dal Pizzol

Pelosi

Rocco

PRM

. Brain-heart interaction after acute ischemic stroke. Crit Care. 2020 Apr 21; 24(1): 163. doi: 10.1186/s13054-020-02885-8. PMID: 32317013; PMCID: PMC7175494.

Zhang

Liu

Luo

Liao

Tan

Bai

Sun

Chen

. A non-invasive real-time monitoring system for cytotoxic brain edema in post ischemic stroke based on near-field coupling. Technol Health Care. 2021; 29(5): 963-978. doi: 10.3233/THC-202685. PMID: 33427707.

Guan

Collet

Mazowita

Claydon

. Autonomic nervous system and stress to predict secondary ischemic events after transient ischemic attack or minor stroke: Possible implications of heart rate variability. Front Neurol. 2018 Mar 5; 9: 90. doi: 10.3389/fneur.2018.00090. PMID: 29556209; PMCID: PMC5844932.

Tranmer

Keller

Kindt

Archer

. Loss of cerebral regulation during cardiac output variations in focal cerebral ischemia. J Neurosurg. 1992 Aug; 77(2): 253-9. doi: 10.3171/jns.1992.77.2.0253. PMID: 1625014.

Cheshire

Jr Saper

. The insular cortex and cardiac response to stroke. Neurology. 2006 May 9; 66(9): 1296-7. doi: 10.1212/01.wnl.0000219563.87204.7d. PMID: 16682656.

Chen

Venkat

Seyfried

Chopp

Yan

Chen

. Brain-heart interaction: Cardiac complications after stroke. Circ Res. 2017 Aug 4; 121(4): 451-468. doi: 10.1161/CIRCRESAHA.117.311170. PMID: 28775014; PMCID: PMC5553569.

Christensen

Fogh Christensen

Boysen

. Abnormalities on ECG and telemetry predict stroke outcome at 3 months. J Neurol Sci. 2005 Jul 15; 23664(1-2): 99-103. doi: 10.1016/j.jns.2005.03.039. PMID: 15935384.

Andrew

Thrift

Cadilhac

. The prevalence, impact and economic implications of atrial fibrillation in stroke: What progress has been made? Neuroepidemiology. 2013; 40(4): 227-39. doi: 10.1159/000343667. Epub 2013 Jan 24. PMID: 23364221.

10.

Gunnoo

Hasan

Khan

Slark

Bentley

Sharma

. Quantifying the risk of heart disease following acute ischaemic stroke: A meta-analysis of over 50,000 participants. BMJ Open. 2016 Jan 20; 6(1): e009535. doi: 10.1136/bmjopen-2015-009535. PMID: 26792217; PMCID: PMC4735313.

11.

Wang

Zhang

Bao

Peng

Yang

Zhuang

Ning

Qiao

Qin

Chen

. Experimental study on differential diagnosis of cerebral hemorrhagic and ischemic stroke based on microwave measurement. Technol Health Care. 2020; 28(S1): 289-301. doi: 10.3233/THC-209029. PMID: 32364161; PMCID: PMC7369055.

12.

Xiong

Tian

Leung

Soo

YOY

Chen

VHL

Mok

VCT

Chu

WCW

Wong

Leung

TWH

. Autonomic dysfunction predicts clinical outcomes after acute ischemic stroke: A prospective observational study. Stroke. 2018 Jan; 49(1): 215-218. doi: 10.1161/STROKEAHA.117.019312. Epub 2017 Dec 4. PMID: 29203690.

13.

Prosser

MacGregor

Lees

Diener

Hacke

Davis

; VISTA Investigators. Predictors of early cardiac morbidity and mortality after ischemic stroke. Stroke. 2007 Aug; 38(8): 2295-302. doi: 10.1161/STROKEAHA.106.471813. Epub 2007 Jun 14. PMID: 17569877.

14.

Wolf

Abbott

Kannel

. Atrial fibrillation as an independent risk factor for stroke: The framingham study. Stroke. 1991 Aug; 22(8): 983-8. doi: 10.1161/01.str.22.8.983. PMID: 1866765.

15.

van der Bilt

Hasan

Vandertop

Wilde

Algra

Visser

Rinkel

. Impact of cardiac complications on outcome after aneurysmal subarachnoid hemorrhage: A meta-analysis. Neurology. 2009 Feb 17; 72(7): 635-42. doi: 10.1212/01.wnl.0000342471.07290.07. PMID: 19221297.

16.

Samuels

. The brain-heart connection. Circulation. 2007 Jul 3; 116(1): 77-84. doi: 10.1161/CIRCULATIONAHA.106.678995. PMID: 17606855.

17.

European Stroke Organisation (ESO) Executive Committee; ESO Writing Committee. Guidelines for management of ischaemic stroke and transient ischaemic attack 2008. Cerebrovasc Dis. 2008; 25(5): 457-507. doi: 10.1159/000131083. Epub 2008 May 6. PMID: 18477843.

18.

Sposato

Cipriano

Saposnik

Ruíz Vargas

Riccio

Hachinski

. Diagnosis of atrial fibrillation after stroke and transient ischaemic attack: A systematic review and meta-analysis. Lancet Neurol. 2015 Apr; 14(4): 377-87. doi: 10.1016/S1474-4422(15)70027-X. Epub 2015 Mar 4. PMID: 25748102.

19.

Christensen

Fogh Christensen

Boysen

. Abnormalities on ECG and telemetry predict stroke outcome at 3 months. J Neurol Sci. 2005 Jul 15; 234(1-2): 99-103. doi: 10.1016/j.jns.2005.03.039. PMID: 15935384.

20.

Yoshimura

Toyoda

Ohara

Nagasawa

Ohtani

Kuwashiro

Naritomi

Minematsu

. Takotsubo cardiomyopathy in acute ischemic stroke. Ann Neurol. 2008 Nov; 64(5): 547-54. doi: 10.1002/ana.21459. PMID: 18688801.

21.

Wang

Liu

Fan

Liu

. Effects of ferulic acid on regulating the neurovascular unit: Implications for ischemic stroke treatment. World J Tradit Chin Med. 2022; 8: 210-7.

22.

Shen

Zhu

Liang

Zhang

. Effect of moxibustion on cardiac remodeling and myocardial function in rats with exercise-induced fatigue. World J Tradit Chin Med. 2021; 7: 254-7.

23.

Fuhrer

Reinhard

Niesen

. Paradigm change? Cardiac output better associates with cerebral perfusion than blood pressure in ischemic stroke. Front Neurol. 2017 Dec 22; 8: 706. doi: 10.3389/fneur.2017.00706. PMID: 29312128; PMCID: PMC5743655.

24.

Reinhard

Rutsch

Lambeck

Wihler

Czosnyka

Weiller

Hetzel

. Dynamic cerebral autoregulation associates with infarct size and outcome after ischemic stroke. Acta Neurol Scand. 2012 Mar; 125(3): 156-62. doi: 10.1111/j.1600-0404.2011.01515.x. Epub 2011 Apr 6. PMID: 21470192.

25.

Deegan

Devine

Geraghty

Jones

Ólaighin

Serrador

. The relationship between cardiac output and dynamic cerebral autoregulation in humans. J Appl Physiol (1985). 2010 Nov; 109(5): 1424-31. doi: 10.1152/japplphysiol.01262.2009. Epub 2010 Aug 5. PMID: 20689094; PMCID: PMC2980368.

26.

Marik

Monnet

Teboul

. Hemodynamic parameters to guide fluid therapy. Ann Intensive Care. 2011 Mar 21; 1(1): 1. doi: 10.1186/2110-5820-1-1. PMID: 21906322; PMCID: PMC3159904.

27.

Rali

Buechler

Van Gotten

Waters

Shah

Haglund

Sauer

. Non-invasive cardiac output monitoring in cardiogenic shock: The NICOM Study. J Card Fail. 2020 Feb; 26(2): 160-165. doi: 10.1016/j.cardfail.2019.11.015. Epub 2019 Nov 18. PMID: 31751786.

28.

Joosten

Desebbe

Suehiro

Murphy

Essiet

Alexander

Fischer

Barvais

Van Obbergh

Maucort-Boulch

Cannesson

. Accuracy and precision of non-invasive cardiac output monitoring devices in perioperative medicine: A systematic review and meta-analysis†. Br J Anaesth. 2017 Mar 1; 118(3): 298-310. doi: 10.1093/bja/aew461. PMID: 28203792.

29.

Saugel

Thiele

Hapfelmeier

Cannesson

. Technological assessment and objective evaluation of minimally invasive and noninvasive cardiac output monitoring systems. Anesthesiology. 2020 Oct 1; 133(4): 921-928. doi: 10.1097/ALN.0000000000003483. PMID: 32773696; PMCID: PMC7494612.

30.

Sathyaprabha

Pradhan

Rashmi

Thennarasu

Raju

. Noninvasive cardiac output measurement by transthoracic electrical bioimpedence: Influence of age and gender. J Clin Monit Comput. 2008 Dec; 22(6): 401-8. doi: 10.1007/s10877-008-9148-6. Epub 2008 Nov 13. PMID: 19005768.

31.

Shoemaker

Bishop

Appel

Van de Water

Harrington

Wang

Patil

. Multicenter trial of a new thoracic electrical bioimpedance device for cardiac output estimation. Crit Care Med. 1994 Dec; 22(12): 1907-12. PMID: 7988125.

32.

Kubicek

Karnegis

Patterson

Witsoe

Mattson

. Development and evaluation of an impedance cardiac output system. Aerosp Med. 1966 Dec; 37(12): 1208-12. PMID: 5339656.

33.

Folstein

McHugh

. Mini-mental state. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res. 1975 Nov; 12(3): 189-98. doi: 10.1016/0022-3956(75)90026-6. PMID: 1202204.

34.

Lindsay

Bayley

McDonald

Graham

Warner

Phillips

. Toward a more effective approach to stroke: Canadian best practice recommendations for stroke care. CMAJ. 2008 May 20; 178(11): 1418-25. doi: 10.1503/cmaj.071253. Erratum in: CMAJ. 2008 Jul 1; 179(1): 56. PMID: 18490636; PMCID: PMC2374857.

35.

Adeoye

Ogah

Ovbiagele

Akinyemi

Shidali

Agyekum

Aje

Adebayo

Akinyemi

Kolo

Appiah

Iheonye

Kelechukwu

Ganiyu

Olunuga

Akpa

Olagoke

Sarfo

Wahab

Olowookere

Fakunle

Akpalu

Adebayo

Nkromah

Yaria

Ibinaiye

Ogbole

Olumayowa

Lakoh

Calys-Tagoe

Olowoyo

Innocent

Tiwari

Arnett

Godwin

Ayotunde

Akpalu

Obiora

Joseph

Omisore

Jenkins

Lackland

Owolabi

Isah

Dambatta

Komolafe

Bock-Oruma

Melikam

Imoh

Sunmonu

Gebregziabher

Olabisi

Armstrong

Onyeonoro

Sanya

Agunloye

Ogunjimi

Arulogun

Farombi

Obiabo

Obiako

Owolabi

; SIREN Team as part of the H3Africa Consortium. Prevalence and Prognostic Features of ECG Abnormalities in Acute Stroke: Findings From the SIREN Study Among Africans. Glob Heart. 2017 Jun; 12(2): 99-105. doi: 10.1016/j.gheart.2017.01.002. Epub 2017 Mar 14. PMID: 28302557; PMCID: PMC5582979.

36.

Huhtakangas

Saaresranta

Bode

Bloigu

Huhtakangas

. Stroke volume predicts nocturnal hypoxemia in the acute ischemic stroke after intravenous thrombolysis. J Stroke Cerebrovasc Dis. 2019 Sep; 28(9): 2481-2487. doi: 10.1016/j.jstrokecerebrovasdis.2019.06.017. Epub 2019 Jul 3. PMID: 31277996.

37.

Almir

Lejla

Lemana

. Chapter 74 – Inspection of medical devices. Clinical Engineering Handbook, Second Edition. 2020; 491-497.

38.

Badnjević

Avdihodžić

Gurbeta Pokvić

. Artificial intelligence in medical devices: Past, present and future. Psychiatr Danub. 2021 May; 33(Suppl 3): S336-S341. PMID: 34010259.

39.

Almina

Marijana

Lemana

, et al. Review of artificial intelligence application in cardiology. 2020 9th Mediterranean Conference on Embedded Computing (MECO). 2020; pp. 1-5.

40.

Kovačević

Gurbeta Pokvić

Spahić

, et al. Prediction of medical device performance using machine learning techniques: Infant incubator case study. Health Technol. 2020; 10: 151-155. doi: 10.1007/s12553-019-00386-5.

Measurement of non-invasive cardiac output during cycling exercise in ischemic stroke inpatients: A pilot study

Abstract

BACKGROUND:

OBJECTIVE:

METHODS:

RESULTS:

CONCLUSION:

Keywords

1. Introduction

2. Methods

2.1 Recruitment of study respondents

2.2 Experimental setup and protocol

2.4 Statistical analysis

3. Results

Table 1 Demographic data of the participants in both groups

3.2 Secondary outcomes

Footnotes

Conflict of interest

Funding

References

Table 1
Demographic data of the participants in both groups