Electrocardiogram properties and risk of covert brain infarction and other magnetic resonance imaging abnormalities in a stroke‐free population

Abstract Objectives This study aimed to investigate the association between electrocardiogram (ECG) abnormalities and silent vascular brain injury as defined by cerebral magnetic resonance imaging (MRI) in a stroke‐free community‐based population. Methods A total of 5888 participants were studied from the Cardiovascular Health Study (CHS), a prospective cohort of community‐living older adults. Standard 12‐lead ECGs measured prior to MRI scan were used. MRI scans were conducted at years 4–6 and 10–11. The primary outcome was presence of incident covert brain infarcts (CBIs) on the 2nd MRI examination, excluding previous CBIs and stroke occurrence. Secondary outcomes included white matter, ventricular, and sulcal atrophy on the 1st MRI. Logistic and multiple linear regression models were used to assess the relationship between ECG findings and silent vascular brain injury. Results Left axis deviation before MRI scan was related to presence of incident CBIs (odds ratio [OR]: 1.45; 95% CI: 1.01–2.08, p = .047). A long QT interval was associated with severe white matter hyperintensity (OR: 1.36; 95% CI: 1.04–1.77, p = .024). Minor Q and QS waves with ST‐T abnormalities were positively related to sulcal atrophy (β: 0.43, 95% CI: 0.06–0.81, p = .023). Conclusions Our study found that ECG abnormalities were related to presence of CBIs, white matter hyperintensity, and sulcal atrophy on MRI in a stroke‐free relderly population. Specifically, those with left axis deviation had an increased risk of presence of CBIs.


INTRODUCTION
Silent vascular brain injury occurs at an accelerated rate in elderly people and may cause subsequent overt clinical events, including stroke, vascular cognitive impairment, and death (Appleton et al., 2020;Cannistraro et al., 2019;Kalani et al., 2020;Leung et al., 2017). Subclinical vascular brain injury can be detected using cerebral magnetic resonance imaging (MRI), including white matter hyperintensity, silent infarcts, and gray matter atrophic changes (Appleton et al., 2020;Luoto et al., 2000). It has been reported that the prevalence of covert brain infarct (CBI) in people over 50 years old is higher (20%) than that of overt ischemic stroke (2%−14%) in the United States (Leung et al., 2017). Despite its high prevalence, an understanding of the pathogenesis and biomarkers of silent vascular brain injury is still developing (Cannistraro et al., 2019). In addition, a prevention strategy for CBI and leukoaraiosis has not yet been established (Leung et al., 2017).
Investigators have increasingly focused on studying the effects of cardiovascular physiology on brain infarctions (Lee et al., 2018;Leung et al., 2017;Yaghi et al., 2018). A recent study has reported that electrocardiogram (ECG) abnormalities are associated with cerebrovascular diseases (Danese et al., 2019;Sawano et al., 2020), while the relationship between ECG parameters and silent vascular brain injury has not been assessed. Understanding the relationship between ECG and silent vascular brain injury may contribute to the improvement of subsequent stroke prevention strategies and provide evidence supporting the notion that cardiac structural or functional abnormalities may predispose patients to silent brain vascular injury (Yaghi et al., 2018). Therefore, the present study aimed to investigate the relationship between ECG profile components and silent vascular brain injury, as defined on MRI. We hypothesized that ECG abnormalities prior to the MRI scans would be associated with a greater risk of incident CBI on the follow-up MRI, presence of severe white matter hyperintensity and brain atrophy.

Study population
This study was conducted as part of the Cardiovascular Health Study (CHS), a multisite, population-based longitudinal study designed to study cardio-cerebrovascular diseases in adults aged ≥65 years (Boyle et al., 2021). A total of 5888 participants were recruited from communities beginning in 1989 and followed up for more than 20 years.
The institutional review boards at the University of Washington and each study site approved the study and written informed consent was obtained from all participants. The study conforms with World Medical Association Declaration of Helsinki. Details of the CHS have been previously described (Kuller et al., 2016;Yaghi et al., 2018). In the present study, participants with incomplete baseline information and covariates, missing ECG values, and those with a history of stroke, and transient ischemic attack were excluded. Exclusion criteria of the primary analysis are as follows: missing infarct data at the 1st or 2nd visit, infarcts detected on MRI at the 1st visit, and stroke occurrence during follow-up. In the secondary analysis, participants with available white matter data at the 1st visit were included in the white matter worsening analysis, and those with complete ventricle and sulcal atrophy data on MRI were included in the brain atrophy analysis.

ECG examination
In the primary analysis, ECG abnormalities prior to the 1st MRI scan were used. ECGs were coded according to the Minnesota Code (MC) (Auer et al., 2012). Electrocardiographic abnormalities were defined according to previous publications and MC codes (Auer et al., 2012;Bussink et al., 2013;Denes et al., 2007;Silva et al., 2021

MRI evaluation
The first brain MRI scans for CHS participants were conducted in years 4-6 (1991−1994) and the second MRI scans were conducted in years 10-11 (1997−1999). Two independent neuroradiologists reviewed the scans using a standard rule to identify the number, size, and location of brain infarcts, as well as the white matter grade scored from 0 (no changes) to 9 (most pronounced changes) (Kalani et al., 2020;Yaghi et al., 2018). Brain atrophy was also measured, considering the sizes of the ventricles and sulci ranging from 0 (smallest) to 9 (largest), as detailed previously Luoto et al., 2000).
The primary outcome was presence of incident CBIs on the 2nd MRI examination, excluding those with an infarct detected on the 1st MRI and stroke occurrence before the 2nd MRI exam. CBI on MRI was defined as an area of abnormal signal ≥3 mm in diameter within one vascular distribution with no mass effect. Those with positive MRI findings must be asymptomatic. Secondary outcomes included white matter, ventricular, and sulcal atrophy on the 1st MRI. Severe white matter hyperintensity was defined as grades 3−9. Severe brain atrophy was defined as grades 3−9 atrophy of the ventricle or sulci (Rosano et al., 2005).

Covariates
The latest information on the covariates for the primary and secondary analyses was collected before the initial MRI. These variables included age, sex, race, body mass index, systolic blood pressure, antihypertensive drug therapy, smoking status, diabetes mellitus, congestive heart failure, myocardial infarction, and atrial fibrillation, as previously defined .

Statistical analysis
Continuous variables are presented as means (standard deviations), and categorical variables are reported as frequencies (percentages).
Differences in baseline characteristics between the groups were tested using the Student's t-test for continuous variables and the chi-square test for categorical variables.
In the primary analysis, a logistic regression model with a prospective cohort study design was used to assess the relationship between all measured ECG abnormalities 4 years before the 1st MRI scan and presence of incident CBIs on the 2nd MRI scan.
Secondary analyses were conducted in a cross-sectional design.
We further evaluated the relationship between measured ECG abnormalities at baseline, severe white matter hyperintensity, and severe brain atrophy on the 1st MRI scan with logistic regression models. A multiple linear regression model was used to assess the association among ECG abnormalities, degree of white matter hyperintensity and brain atrophy grade. Sensitivity analyses were conducted to examine the association of ECG abnormalities before 1st MRI scan with presence of incident CBI on the 2nd MRI scan, excluding participants who had hypertension and atrial fibrillation participants. Because numerous studies provided evidence that hypertension was associated with CBI as well as ischemic stroke and hypertension was the most important risk factor for covert vascular brain injury (Cannistraro et al., 2019;Kalani et al., 2020;Leung et al., 2017;Wardlaw et al., 2013).
Atrial fibrillation is the most common cardiac factor related with stroke occurrence. Previous studies ever thought that atrial fibrillation may influence the association between atrial disease and CBI (Kamel et al., 2015;Yaghi et al., 2018). A prior study ever performed sensitivity analysis that excluded participants with atrial fibrillation when evaluating the relationship between left atrial abnormality and vascular brain injury (Kamel et al., 2015).
Furthermore, exploratory analysis was performed to evaluate the association between ECG abnormalities before 1st MRI scan and clinically defined nonlacunar ischemic stroke occurrence before the 2nd MRI scan with a logistic regression model. Only the first occurrence of nonlacunar ischemic stroke was included. The first-time nonlacunar ischemic stroke occurred after 1st MRI scan and before the 2nd MRI scan. Adjustment covariates were specified a priori and included age, sex, race, body mass index, systolic blood pressure, antihypertensive drug therapy, smoking status, diabetes mellitus, congestive heart failure, myocardial infarction, atrial fibrillation, left atrial dimension, and all ECG-measured parameters in all analyses. Statistical significance was set at p < .05. Statistical analyses were conducted using the SPSS ver. 24.0 and R ver. 3.5.3.

RESULTS
A total of 5888 participants were enrolled in the CHS study. The selection process of the study population is shown in Figure 1. A total of 1130 participants were eligible for the primary analysis, 2781 for the white matter analysis, and 2780 for the brain atrophy analysis. The baseline information of the participants included in the primary and secondary analyses is presented in had severe brain atrophy.
The frequency distribution of ECG abnormalities in the incidence of CBI and different white matter grades is shown in Figure 2. The percentage of covert infarcts in the left axis deviation group was higher than that in the no left axis deviation population. Participants with long QT intervals on ECG accounted for a higher percentage of white matter grades 3−9 than those with normal QT intervals.
The main results are listed in Table 2. Left axis deviation measured at year 4 was associated with presence of incident CBI on the 2nd MRI scan from a fully adjusted model in Table 2. Other ECG indicators were not significantly related to the risk of covert brain infarcts. Sensitivity analyses were performed excluding hypertensive and atrial fibrillation participants (Supplemental Table S1). Data show a strong relationship between left axis deviation and risk of covert brain infarcts after full adjustment (OR, 1.79, 95% CI: 1.04−3.10, p = .037). Minor isolated ST-T abnormalities were related with incident stroke in a fully adjusted model (OR, 1.51, 95% CI: 1.03-2.21, p = .036, Supplemental Table S2).
A long QT interval was associated with severe white matter hyperintensity after full adjustment (Table 3). The relationship between ECG abnormalities and white matter and brain atrophy grades is listed in

DISCUSSION
In this community-dwelling population-based longitudinal study, left axis deviation at baseline was associated with following presence of incident CBIs in a stroke-free elderly population. A long QT interval

F I G U R E 1
The selection process of the study population. at baseline was associated with severe white matter hyperintensity.
A positive relationship was observed between minor Q and QS waves with ST-T abnormalities at baseline and sulcal atrophy grade.
The relationship between cardiac physiological indicators and covert cerebrovascular abnormalities has been widely reported (Cannistraro et al., 2019;Godin et al., 2011;Gottesman et al., 2010;Kamel et al., 2015;Lee et al., 2018;Leung et al., 2017;Wardlaw et al., 2013;Yaghi et al., 2018). In particular, blood pressure has been demonstrated to be related to covert brain infarction and white matter hyperinten-sities on MRI (Godin et al., 2011;Gottesman et al., 2010;Kamel et al., 2015;Leung et al., 2017;Wardlaw et al., 2013). The mean right atrial pressure was likewise associated with white matter hyperintensities (Lee et al., 2018). Another study found that a larger left atrial diameter was related to brain infarcts (Yaghi et al., 2018). Regarding ECG examination, one study reported that isolated nonspecific ST segment and T wave abnormalities were related to a higher risk of ischemic stroke (Sawano et al., 2020). Another cohort study found that a silent myocardial infarction detected on ECG was associated with incident TA B L E 1 Basic characteristics of all participants in our study cohort. ischemic stroke (Merkler et al., 2021). However, evidence regarding the link between ECG parameters and subclinical vascular brain injury on MRI is lacking.

Primary analysis Secondary analysis
The primary finding of the present study was that left axis deviation may be related with high risk of covert brain infarcts in the strokefree elderly population. The relationship between left axis deviation and presence of incident CBIs could be due to several factors. First, the total activation time may be longer in the population with a left axis deviation. Thus, activation of the basal anterolateral region may be delayed (Abu-Alrub et al., 2021). Second, left axis deviation was related to left-sided structural abnormalities, specifically LV diastolic and systolic dysfunction, as well as LV dilation (Lui et al., 2021). These two reasons may cause a decrease in the flow velocity in the left ventricle and cerebral blood flow, thereby contributing to stasis and ischemia of small-vessel territories and brain infarcts (Cannistraro et al., 2019;Di Tullio et al., 1999;Nakanishi et al., 2017;Wardlaw et al., 2013;Yaghi et al., 2018). This result was consistent with a previous study that reported that LV concentric hypertrophy carries a higher independent TA B L E 2 Relationship between ECG abnormalities and covert brain infarcts. .808 †The number of participants with ST elevation was 6, so the confidence interval was too wide to show it. Models adjusted for age, sex, race, body mass index, systolic blood pressure, antihypertensive drug therapy, smoking status, diabetes mellitus, congestive heart failure, myocardial infarction, atrial fibrillation, left atrial dimension, and all ECG-measured parameters.

TA B L E 3
Relationship between ECG abnormalities with severe white matter hyperintensity and brain atrophy.

Secondary outcomes
White matter grade 3-9 Brain atrophy 3-9 Models adjusted for age, sex, race, body mass index, systolic blood pressure, antihypertensive drug therapy, smoking status, diabetes mellitus, congestive heart failure, myocardial infarction, atrial fibrillation, left atrial dimension, and all ECG-measured parameters.
risk for silent brain infarcts in a multiethnic stroke-free general population (Nakanishi et al., 2017). Another potential explanation may be that left axis deviation and CBI share vascular risk factors, which are the causes of brain infarcts (Yaghi et al., 2018). After excluding participants with hypertension, the association between left axis deviation and covert brain infarcts was more apparent. It may be inferred that the pathology of silent vascular brain injury caused by left axis deviation is different from that of hypertension (Cannistraro et al., 2019).
There may be several uncertain pathological conditions that can lead to a silent vascular brain injury. Although two other studies reported that a left atrial abnormality was related to prevalent brain infarcts, our results provide further evidence that a cardiac structural or functional abnormality may occur before silent brain infarcts. Moreover, when we analyzed the association between ECG abnormalities and following ischemic stroke occurrence, left axis deviation was not related with risk of stroke. It may indicate that the risk factors of stroke and silent vascular brain injury may be different in the same time period, which may be related to the disease progression.
A long QT interval was found to be associated with severe white matter hyperintensity. A recent study included patients from the out-patient Cognitive Impairment and Dementia Center of a hospital and reported that a prolonged QT interval was prevalent in patients with dementia and in patients with higher leukoaraiosis scale scores (Danese et al., 2019). Although the community population in our study was different from that of the other study, our results confirmed the results of the latter with larger sample size. A previous study suggested that cerebral vascular load is associated with prolonged QTc (Danese et al., 2019). Patients with lesions in autonomic cardiac centers might have a change in the vascular origin, which causes unstable blood pressure with subsequent dysfunction of cerebral perfusion and alteration of the white matter (Dias et al., 2013). Previous studies found that a silent myocardial infarction was associated with the risk of stroke (Merkler et al., 2021;Merkler et al., 2019). However, the relationship between silent myocardial infarction and silent vascular brain injury is unknown. Our study found that minor Q and QS waves with ST-T abnormalities were positively related to the sulcal atrophy grade. Minor Q and QS waves with ST-T abnormalities belong to silent myocardial infarction (Gibson et al., 2018). One possible explanation may be that silent myocardial infarction indicates ischemic changes in vessels secondary to endothelial dysfunction and impaired autoregulation, which may also lead to small-vessel diseases, including sulcal atrophy, detected on MRI (Cannistraro et al., 2019;Wardlaw et al., 2013).
However, some limitations still need to be considered. First, each participant underwent MRI twice during the entire follow-up period.
We could not acquire dynamic changes in brain vascular findings on MRI. Participants who underwent both MRI scans may be healthier than those who did not, limiting generalizability. Second, the populations in the different analyses were not consistent. There were different missing values regarding covert brain infarcts, white matter, and brain atrophy on MRI. The results may have been altered if all participants had completed both MRI scans and were included in both analyses . Third, routine ECG examination was not performed in all participants in the CHS. Fourth, our study was performed in an elderly population aged ≥65 years, and the results may differ in younger populations (Merkler et al., 2021). Last, there may be uncontrolled confounding factors in the study, such as undiagnosed atrial fibrillation and imaging interrater variability over the time scale.

CONCLUSION
In a stroke-free community-dwelling population, our study found that left axis deviation may occur before covert brain infarcts. We likewise found that a long QT interval was associated with severe white matter hyperintensity and that minor Q and QS waves with ST-T abnormalities were positively related to sulcal atrophy. These results may reveal novel risk factors and explain the causes of silent vascular brain injuries.
Further studies are needed to determine the detailed pathology of small-vessel disease lesions on MRI.

AUTHOR CONTRIBUTIONS
FG and JH conceived and designed the research; FG, CC, and FL organized the data; FG and CC performed the data analysis; all authors interpreted the results of statistical analysis; FG, CC, and FL wrote the manuscript; JH revised the manuscript critically. All the authors have reviewed the manuscript.

Thank the NHLBI Biologic Specimen and Data Repository Information
Coordinating Center for providing data source. This study does not necessarily reflect the opinions or views of the CHS or the NHLBI.

CONFLICT OF INTEREST STATEMENT
The authors have no relevant financial or nonfinancial interests to disclose.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available in Biologic Specimen and Data Repositories Information Coordinating Center (BioLINCC) at https://biolincc.nhlbi.nih.gov/studies/chs/.

ETHICS STATEMENT
The institutional review boards at the University of Washington and each study site approved the study and written informed consent was obtained from all participants.