Abstract
INTRODUCTION
Heart rate (HR) fragmentation indices quantify breakdown of HR regulation and are associated with atrial fibrillation and cognitive impairment. Their association with brain magnetic resonance imaging (MRI) markers of small vessel disease is unexplored.
METHODS
In 606 stroke‐free participants of the Multi‐Ethnic Study of Atherosclerosis (mean age 67), HR fragmentation indices including percentage of inflection points (PIP) were derived from sleep study recordings. We examined PIP in relation to white matter hyperintensity (WMH) volume, total white matter fractional anisotropy (FA), and microbleeds from 3‐Tesla brain MRI completed 7 years later.
RESULTS
In adjusted analyses, higher PIP was associated with greater WMH volume (14% per standard deviation [SD], 95% confidence interval [CI]: 2, 27%, P = 0.02) and lower WM FA (–0.09 SD per SD, 95% CI: –0.16, –0.01, P = 0.03).
DISCUSSION
HR fragmentation was associated with small vessel disease. HR fragmentation can be measured automatically from ambulatory electrocardiogram devices and may be useful as a biomarker of vascular brain injury.
Keywords: atrial myopathy, brain microbleeds, brain magnetic resonance imaging, brain small vessel disease, heart rate fragmentation, heart rate variability, multi‐ethnic study of atherosclerosis, white matter fractional anisotropy, white matter hyperintensity
1. BACKGROUND
Atrial fibrillation is associated with substantially elevated risks of stroke and with cognitive decline even in the absence of stroke. 1 , 2 However, emerging evidence suggests that underlying structural, contractile, autonomic, or electrophysiological abnormalities of the left atrium, called atrial myopathy, 3 may be causally related to stroke, brain ischemic changes, cognitive decline, and dementia, 4 , 5 and that clinical presentation with atrial fibrillation may represent a late stage in the pathophysiologic process. Sensitive measures of atrial myopathy are needed to allow risk stratification and to facilitate investigation of early interventions to prevent brain abnormalities. To date most efforts have focused on left atrial structural or contractile abnormalities, but additional markers of atrial myopathy may be useful.
Using data from long‐term electrocardiogram (ECG) recordings, we recently developed heart rate (HR) fragmentation metrics that reflect breakdown of the neuroautonomic–electrophysiologic network controlling the sinoatrial node and its regulation of HR. 6 , 7 These HR fragmentation metrics are independently predictive of both atrial fibrillation 8 and cognitive impairment and decline. 9 We hypothesize that HR fragmentation metrics reflect atrial myopathy and may be useful indicators of the risk of central nervous system complications of atrial myopathy. In the setting of the Multi‐Ethnic Study of Atherosclerosis (MESA), we investigated the association of HR fragmentation metrics with three brain magnetic resonance imaging (MRI) markers of small vessel disease: white matter hyperintensity (WMH) volume, total white matter (WM) fractional anisotropy (FA), and brain microbleed presence.
2. METHODS
2.1. Study population
MESA is an observational cohort study of subclinical cardiovascular disease that includes 6814 men and women 45 to 84 years of age and free of clinically recognized cardiovascular disease at study enrollment in 2000 to 2002, recruited at six field centers (Baltimore, Maryland; Chicago, Illinois; Forsyth County, North Carolina; Los Angeles County, California; New York, New York; and St. Paul, Minnesota). 10 Five follow‐up study visits have been completed. Between 2010 and 2013, consenting participants completed home polysomnography in the MESA Sleep Ancillary Study 11 a median (interquartile range [IQR]) of 10 (7, 15) months after the Exam 5 (2010–2012) visit, with overnight ECG recording from a single lead or two leads, processed as previously described. 9 Between 2018 and 2019, a median (IQR) of 6.9 (6.4, 7.3) years after the sleep study, participants in the MESA Atrial Fibrillation Ancillary Study completed a brain MRI. 12 Each field center obtained institutional review board approval and all participants provided written informed consent.
2.2. HR fragmentation
Using the overnight ECG tracings from the polysomnogram (median [IQR] duration 7.7 [6.9, 8.4] hours), we computed HR fragmentation indices. The HR fragmentation index of primary interest prespecified in this analysis was the percentage of inflection points, abbreviated PIP, calculated from the normal‐to‐normal (NN) interval time series on the long‐term ECG recordings. 6 The NN interval refers to the interval between successive QRS complexes on the ECG for non‐ectopic (normal) beats. An inflection point is a change in HR acceleration sign, that is, a transition from HR acceleration to deceleration and vice versa, or from HR acceleration/deceleration to no change and vice versa. 6 Using automated methods, the change in HR acceleration was evaluated across the entire overnight sleep ECG recording, and the summary metric, PIP, was calculated. Thus, PIP quantifies the percentage of adjacent NN intervals in which there is a change in acceleration. Because the ECG signals were sampled at 256 Hz, the resolution of the interbeat interval time series is 1/256 or approximately 4 ms. Therefore, only NN intervals whose difference was > 4 ms or < −4 ms were considered different from each other. Larger values of PIP indicate more HR fragmentation and, we hypothesize, more disruption of normal sinus node regulation of HR, possibly reflecting atrial myopathy. In secondary analyses, we evaluated several related HR fragmentation metrics, all of which are highly correlated: 8 PNNSS (percentage of NN intervals in short segments), PNNLS (percentage of NN intervals in long segments), and ALS (average length of accelerative/decelerative segments), described in Table S1 in supporting information. In a previous publication, 8 we included a figure (Figure A1) with a detailed description of the computation of the HR fragmentation metrics.
We also evaluated two conventional short‐term HR variability metrics measured on the same Exam 5 sleep study ECGs (2010–2012): SDNN (standard deviation of NN intervals) and rMSSD (root mean square of successive differences in NN intervals), also described in Table S1. The average NN interval across the sleep study ECG (AVNN), was also measured, from which the average HR of non‐ectopic beats can be calculated as 60,000/AVNN.
2.3. Brain MRI
The brain MRI was conducted as previously described, 12 including T1‐weighted, T2‐weighted, T2 fluid attenuated inversion recovery (FLAIR), axial 2D echo‐planar diffusion tensor imaging, and a susceptibility‐weighted imaging (SWI) quantitative susceptibility mapping (QSM) sequence. 5 The brain MRI parameters of interest for this analysis were markers of small vessel disease of the brain associated with increased risk of stroke and dementia: 13 total WMH volume, WM integrity as measured by total WM FA, and the presence of one or more microbleeds. Tissue segmentation to gray and white matter was performed using T1‐weighted images and a multiatlas segmentation method. 14 WMH volume was calculated from FLAIR images by a deep‐learning method. 15 WM FA, a scalar ranging from 0 to 1, is a measure of WM integrity calculated from diffusion tensor imaging and reflects the degree to which water diffusion is limited to a single dimension. Values of WM FA that are lower than normal reflect worse WM structural integrity. 16 FA in total WM was extracted in native space after registering T1 images and segmentations to FA maps. Microbleeds were initially identified by a deep learning–based segmentation method that used T2‐weighted images and QSM data to identify and differentiate microbleeds from iron deposits. 17 Identified putative lesions were then reviewed by a radiologist who made the final classification, as previously described. 5
2.4. Participant characteristics
Participant characteristics including self‐reported sex, race and ethnicity, and educational attainment were obtained at study entry. At the 2010 to 2012 (Exam 5) study visit, systolic and diastolic blood pressure, weight, and height were measured, and total cholesterol was measured in fasting blood specimens. Low‐density lipoprotein cholesterol was calculated using the Friedewald equation. Participants reported their age, cigarette smoking status, history of hypertension, history of diabetes, and use of medications. The glomerular filtration rate was estimated using the Chronic Kidney Disease Epidemiology Collaboration creatinine equation. 18 The apnea–hypopnea index, based on all apneas and hypopneas with 4% desaturations per hour of sleep, was determined from the overnight sleep study. 11 At telephone contacts every 9 to 12 months throughout MESA follow‐up, participants were asked to identify new hospitalizations and diagnoses. Medical records were obtained and myocardial infarction, heart failure, 19 and stroke 20 events during follow‐up to the date of the sleep study were adjudicated; atrial fibrillation was identified from diagnosis codes as previously reported. 21
2.5. Statistical analysis
WMH volume was log transformed due to its skewed distribution. We used linear regression with robust standard errors to estimate the ratio of geometric means, which provides the percent difference in WMH volume per 1‐standard deviation (SD) increment in PIP. WM FA was expressed as a z score in SD units, and we used linear regression with robust standard errors to estimate the difference in SD units in WM FA associated with each 1‐SD increment in PIP. For the presence of microbleeds, we used Poisson regression with robust standard errors to estimate the prevalence ratio associated with each 1‐SD increment in PIP.
RESEARCH IN CONTEXT
Systematic review: We searched PubMed for studies of heart rate (HR) fragmentation or HR variability in relation to brain magnetic resonance imaging markers of small vessel disease. Existing studies of HR variability did not analyze small vessel disease markers and there were no analyses of HR fragmentation.
Interpretation: In the Multi‐Ethnic Study of Atherosclerosis, we derived newly developed HR fragmentation indices from overnight sleep study electrocardiograms. In multivariable analyses, we found that greater fragmentation was associated with greater white matter hyperintensity volume and lower white matter fractional anisotropy.
Future directions: Additional investigation is needed to confirm that HR fragmentation indices can define populations at risk for subclinical vascular brain injury. If confirmed, measurement of fragmentation using non‐invasive electrocardiogram monitoring with ambulatory devices may inform decisions about therapy to reduce vascular risk factors.
A minimally adjusted model (Model 1) was constructed with adjustment for age (as a linear term), sex, race and ethnicity, and MESA field center; a more fully adjusted model (Model 2) included AVNN, body mass index, systolic and diastolic blood pressure, diabetes, use of antihypertensive medication, current smoking, and a history of a cardiac event (myocardial infarction, heart failure, or atrial fibrillation) before the measurement of HR fragmentation. Covariates included in the more fully adjusted model were those previously associated with brain MRI markers of small vessel disease. 12 All models with WMH volume as the dependent variable were additionally adjusted for total intracranial volume. In sensitivity analyses, we examined whether additional adjustment for estimated glomerular filtration rate or apnea–hypopnea index altered the associations of interest. We explored the possibility of non‐linear associations of PIP with the brain MRI measures using splines. We examined differences in associations by age, race and ethnic group, history of prior cardiac events, and the severity of sleep apnea as defined by apnea–hypopnea index ≥15 versus < 15. The secondary analyses of alternate HR fragmentation metrics and conventional HR variability metrics used the same methods as the primary analysis. The log base 2 transform was applied to rMSSD due to its skewed distribution.
3. RESULTS
Of the 4716 participants who attended the 2010 to 2012 study visit (Exam 5), 1960 completed polysomnography and had HR fragmentation indices calculated (Figure 1). Of these 1960 participants, 613 participated in the brain MRI ancillary study. Seven participants with a history of clinically detected stroke before the sleep study were excluded from the analysis and several participants had inadequate diffusion or QSM imaging, leaving 606, 604, and 597 participants with available data for WMH volume, FA, and microbleeds, respectively (Figure 1). The mean (SD) age was 67 (8) years at the time of the sleep study and 74 (8) years at the time of the brain MRI, 54% were women, and the participants self‐identified as Black (25%), Chinese (16%), Hispanic (19%), and White (40%).
FIGURE 1.
Study inclusion diagram. MRI, magnetic resonance imaging; MESA, Multi‐Ethnic Study of Atherosclerosis; WMH, white matter hyperintensity
Among participants who attended the 2010 to 2012 (Exam 5) study visit, compared to MESA participants not included in these analyses, those included were on average younger, had completed more education, and had a lower prevalence of treated hypertension and diabetes mellitus (Table S2 in supporting information). A larger proportion of those included were Chinese, and a smaller proportion had a history of a cardiac event before the sleep study.
The HR fragmentation metric PIP ranged from 41% to 84%, with a mean (SD) of 57% (7%). Higher PIP, indicating greater HR fragmentation, was strongly associated with advanced age (Figure 2); each 10‐year increment in age was associated with a mean difference of 3% in PIP. Higher PIP was also associated with female sex, higher systolic blood pressure, use of antihypertensive medication, and diabetes (Table 1). The distributions and frequency of the brain MRI measures are shown in Table 2. Scatterplots of log (WMH volume) with PIP and total WM FA with PIP are shown in Figures 3A and 3B; locally weighted scatterplot smoothing (LOWESS) curves show that the associations appear to be approximately linear. Figure 3C shows overlaid histograms of PIP in those with and without microbleeds.
FIGURE 2.
Adjusted mean percentage of inflection points (PIP) on long‐term electrocardiographic recordings for three age groups of Multi‐Ethnic Study of Atherosclerosis participants (vertical bars indicate 95% confidence intervals; adjusted for age, race/ethnicity, and field center)
TABLE 1.
MESA participant characteristics in 2010 to 2012 (Exam 5) overall and by quartiles of PIP.
| Overall | Quartiles of percentage of inflection points | ||||
|---|---|---|---|---|---|
| n = 606 |
41%–52% n = 151 |
53%–56% n = 152 |
57%–62% n = 151 |
63%–84% n = 152 |
|
| Age, years, n (%) | |||||
| 55–64 | 293 (48) | 100 (66) | 91 (60) | 65 (43) | 37 (24) |
| 65–74 | 183 (30) | 38 (25) | 38 (25) | 54 (36) | 53 (35) |
| 75–92 | 130 (21) | 13 (9) | 23 (15) | 32 (21) | 62 (41) |
| Women, n (%) | 328 (54) | 66 (44) | 90 (59) | 86 (57) | 86 (57) |
| Race or ethnicity, n (%) | |||||
| Black | 151 (25) | 34 (23) | 39 (26) | 38 (25) | 40 (26) |
| Chinese | 94 (16) | 30 (20) | 21 (14) | 22 (15) | 21 (14) |
| Hispanic | 117 (19) | 30 (20) | 33 (22) | 27 (18) | 27 (18) |
| White | 244 (40) | 57 (38) | 59 (39) | 64 (42) | 64 (42) |
| High school education or less, n (%) | 149 (25) | 32 (21) | 39 (26) | 39 (26) | 39 (26) |
| BMI, kg/m2, mean (SD) | 28 (5) | 27.8 (4.5) | 28 (4.8) | 28.8 (5.7) | 27.2 (4.9) |
| Current smoking, n (%) | 34 (6) | 7 (5) | 9 (6) | 10 (7) | 8 (5) |
| LDL cholesterol, mg/dL, mean (SD) | 109 (32) | 113 (31) | 118 (36) | 106 (30) | 100 (28) |
| SBP, mmHg, mean (SD) | 121 (19) | 118 (18) | 121 (20) | 122 (18) | 124 (20) |
| DBP, mmHg, mean (SD) | 69 (10) | 70 (9) | 70 (11) | 69 (9) | 69 (10) |
| Hypertension medication, n (%) | 273 (45) | 50 (33) | 64 (42) | 75 (50) | 84 (55) |
| Diabetes mellitus, n (%) | 82 (14) | 18 (12) | 15 (10) | 19 (13) | 30 (20) |
| Cardiac disease, * n (%) | 21 (3) | 1 (1) | 5 (3) | 8 (5) | 7 (5) |
| AVNN, ms, mean (SD) | 958 (130) | 939 (118) | 951 (120) | 974 (120) | 967 (155) |
Abbreviations: AVNN, average normal‐to‐normal interval; BMI, body mass index; DBP, diastolic blood pressure; LDL, low‐density lipoprotein; MESA, Multi‐Ethnic Study of Atherosclerosis; PIP, percentage of inflection points; SBP, systolic blood pressure; SD, standard deviation.
History of clinically recognized myocardial infarction, heart failure, or atrial fibrillation.
TABLE 2.
Summary of brain MRI measures in participants with brain MRI sequence data and heart rate fragmentation metrics.
| Brain MRI measure | N in analysis | Mean (SD), median (IQR), or N (%) |
|---|---|---|
| Total intracranial volume, mL, mean (SD) | 606 | 1360 (147) |
| Total white matter hyperintensity volume, mL, median (IQR) | 606 | 2.7 (1.2, 6.9) |
| Log‐transformed white matter hyperintensity volume, mean (SD) | 606 | 1.0 (1.3) |
| White matter fractional anisotropy * , mean (SD) | 604 | 0.39 (0.02) |
| N with microbleeds (%) | 597 | 189 (32) |
Abbreviations: IQR, interquartile range; MRI, magnetic resonance imaging; SD, standard deviation.
Fractional anisotropy is a scalar with values between 0 and 1.
FIGURE 3.
A, Scatterplot of log(white matter hyperintensity volume) and percentage of inflection points (PIP), with a locally weighted scatterplot smoothing (LOWESS) curve. B, Scatterplot of white matter fractional anisotropy and PIP, with a LOWESS curve. C, Histogram of PIP in participants with no microbleeds and in participants with one or more microbleeds
In the multivariable analysis with minimal adjustment (Model 1), each SD increment in PIP was associated with greater WMH volume and lower total WM FA (Table 3). With additional adjustment for cardiovascular risk factors and history of cardiovascular disease in Model 2, the associations of PIP with WMH volume and total WM FA were attenuated but remained statistically significant. Additional adjustment for estimated glomerular filtration rate or apnea–hypopnea index did not alter the associations. PIP was not associated with microbleed prevalence in either model (Table 3). There was no evidence for non‐linearity of the association of PIP with WMH volume and total WM FA, and associations did not differ by age, race and ethnic group, history of prior cardiac events, or the severity of sleep apnea.
TABLE 3.
Multivariable analysis of the association of PIP with three brain MRI measures.
| Model 1 | Model 2 | |||
|---|---|---|---|---|
| Brain MRI measure | Estimate (95% CI) | P‐value | Estimate (95% CI) | P‐value |
| WMH volume, % difference per SD of PIP | 16 (5, 29) | 0.005 | 14 (2, 27) | 0.02 |
| WM fractional anisotropy, difference in SD units per SD of PIP | −0.11 (−0.18, −0.03) | 0.007 | −0.09 (−0.16, −0.01) | 0.03 |
| Microbleeds, prevalence ratio per SD of PIP | 1.10 (0.97, 1.26) | 0.15 | 1.10 (0.96, 1.25) | 0.17 |
Note: Model 1 adjusted for age, sex race and ethnicity, field center. Models for WMH volume also adjusted for total intracranial volume.
Model 2 additionally adjusted for average normal‐to‐normal interval (AVNN), systolic blood pressure, diastolic blood pressure, use of hypertension medication, current smoking, diabetes mellitus, body mass index, and history of clinically recognized myocardial infarction, heart failure, or atrial fibrillation.
Abbreviations: CI, confidence interval; MRI, magnetic resonance imaging; PIP, percentage of inflection points; WMH, white matter hyperintensity.
In secondary analyses (Table S3 in supporting information), the alternative HR fragmentation metric ALS was associated with WMH volume and WM FA in Model 2; all associations with microbleeds were null. Conversely, higher levels of the two conventional HR variability indices, SDNN and rMSSD, were not associated with any of the brain MRI measures in Model 2.
4. DISCUSSION
In this analysis of > 600 MESA participants with measures of HR fragmentation from long‐term ECG recordings and with brain MRI, greater HR fragmentation as measured by PIP was associated with two measures of vascular brain injury: greater WMH volume and lower total WM FA. These associations persisted after adjustment for cardiovascular risk factors. The MRI markers of vascular brain injury we studied are strongly associated with dementia; 22 thus, HR fragmentation is of interest as a potential marker of subclinical vascular brain injury and dementia risk.
Regarding the magnitude of the associations, for each 1‐SD (7%) increment in PIP in Model 2, we observed mean differences of +14% in WMH volume (which corresponds to 0.13 SD of log‐transformed WMH volume) and −0.09 SD in WM FA. For context, in the Rotterdam Study over 5 to 6 years of follow‐up, the adjusted hazard ratio for development of adjudicated dementia associated with a 1‐SD increment in log‐transformed WMH volume was 1.57 (95% confidence interval [CI]: 1.03 to 2.38) 23 and for a 1‐SD decrement in global WM FA was 1.54 (95% CI: 1.25 to 1.92). 24
In our secondary analysis of additional HR fragmentation metrics (Table S3), all the point estimates for the associations with WMH volume and FA were in the hypothesized direction, but most of them did not reach statistical significance. This congruence of results is not surprising as all of the HR fragmentation metrics are correlated with one another.
Published studies of conventional HR variability metrics in relation to brain MRI findings have focused on structural and resting state connectivity measures but not MRI markers of small vessel disease. 25 For example, studies in healthy volunteers, most < 60 years old, indicate that higher rMSSD is associated with greater orbitofrontal cortical thickness. 26 , 27 , 28 The present analysis did not find associations of rMSSD or SDNN with brain MRI findings, but we studied MRI markers of small vessel disease rather than cortical thickness, and the MESA participants were older with a substantial prevalence of obesity, hypertension, and diabetes. We are not aware of other studies of HR fragmentation in relation to MRI markers of small vessel disease.
We hypothesize that fibrosis, stress, inflammation, and genetic predisposition affect sinoatrial tissue and cardiac autonomic nervous system function, leading to atrial myopathy and disruption of the network controlling beat‐to‐beat HR dynamics that manifest as HR fragmentation. 8 , 29 Atrial pathology has also been associated with the development of subclinical and clinically recognized atrial fibrillation. 30 Atrial fibrillation–related cerebral emboli and brain hypoperfusion may both contribute to cognitive decline. 31 , 32 , 33 It is not clear whether HR fragmentation is itself causally related to development of atrial fibrillation or to cardiovascular and neurologic sequelae, or whether it is a biomarker of the atrial pathology. In either case, if HR fragmentation proves to be a powerful biomarker of the underlying atrial pathology, it can be useful as a predictor of clinically relevant outcomes, especially important in this burgeoning era of wearable technology. 34
We previously demonstrated in the MESA cohort that increased HR fragmentation was associated with greater risk of atrial fibrillation, 8 lower cognitive test scores both cross‐sectionally and longitudinally, and with greater decline in scores over a 6‐year period. 9 Together, the findings regarding cognition and brain MRI markers suggest that higher values of PIP may be a useful marker of subclinical cognitive decline and vascular brain injury.
Our study does not address the likely complex mechanistic associations among HR fragmentation, atrial myopathy, and brain injury. One specific factor that may link the emergence of HR fragmentation with atrial myopathy is autonomic dysfunction. 8 , 35 Healthy vagal activity has been shown to suppress cytokine activation. 36 , 37 In contrast, decreased vagal function has been linked to pro‐inflammatory processes 29 , 38 and atrial electrical remodeling that leads to atrial fibrillation. 39 , 40 These three processes (autonomic dysfunction, inflammation, and atrial electrical remodeling) appear to form a vicious cycle, each perpetuating the other. Important support for this conjecture comes from studies indicating that autonomic neuromodulation (low‐level vagus nerve stimulation) may reverse proarrhythmic atrial remodeling, 8 , 35 and that vagus nerve stimulation has prominent anti‐inflammatory effects. 41 , 42 Cardiac neuroautonomic impairment (vagal and sympathetic) may also result in arterial blood pressure dysregulation, which in turn may lead to suboptimal cerebral perfusion. 43 , 44 , 45 Dysregulation of blood pressure itself has been linked to development of WMH and lacunar infarctions. 46
Strengths of the present analysis include the large number of participants with both long‐term ECG monitoring and brain MRI, the racial and ethnic diversity of the MESA cohort, the high field strength (3‐Tesla) of the MRI, and the automated methods used for deriving the brain MRI measures of vascular brain injury. However, the participants included in this analysis had fewer co‐morbidities than those not included. The participants we studied may have had less HR fragmentation and less vascular brain injury than others of similar age, limiting our ability to detect significant associations. HR fragmentation and vascular brain injury likely develop slowly over time but were each assessed at only one time point; thus, it is not possible with the present data to determine the temporal sequence of the exposures and outcomes. Our analyses were extensively adjusted for carefully measured clinical characteristics, but to the extent that confounding characteristics were not recognized or accurately measured, residual confounding remains a possibility. Information on the presence of lacunar infarcts is not currently available from the brain MRIs. Our findings will need to be replicated in other populations.
5. CONCLUSION
Our findings lend support to the hypothesis that cardiac neuroautonomic regulation as reflected in the HR fragmentation index, PIP, may help to define populations at risk for subclinical vascular brain injury. With recent developments in mobile monitoring technology, non‐invasive ambulatory ECG monitoring overnight or for longer periods is now practical, and the measurement of HR fragmentation from these ECG recordings can be automated. Thus, if HR fragmentation is indeed a biomarker of atrial myopathy and its brain complications, information from non‐invasive ECG monitoring may inform clinical care, including decisions about therapy to reduce vascular risk factors.
CONFLICT OF INTEREST STATEMENT
S.R. reports consulting fees from Jazz Pharma, Eli Lilly, and Apnimed Inc., unrelated to this study. The other authors report no conflicts of interest. Author disclosures are available in the supporting information.
CONSENT STATEMENT
All Multi‐Ethnic Study of Atherosclerosis participants provided written informed consent.
Supporting information
Supporting Information
Supporting Information
ACKNOWLEDGMENTS
The authors thank the other investigators, the staff, and the participants of the MESA study for their valuable contributions. A full list of participating MESA investigators and institutions can be found at http://www.mesa‐nhlbi.org. This MESA research was supported by contracts 75N92020D00001, HHSN268201500003I, N01‐HC‐95159, 75N92020D00005, N01‐HC‐95160, 75N92020D00002, N01‐HC‐95161, 75N92020D00003, N01‐HC‐95162, 75N92020D00006, N01‐HC‐95163, 75N92020D00004, N01‐HC‐95164, 75N92020D00007, 75N92019C00011, N01‐HC‐95165, N01‐HC‐95166, N01‐HC‐95167, N01‐HC‐95168, and N01‐HC‐95169 and grants R01 HL127659 and R01 HL144510 from the National Heart, Lung, and Blood Institute; by grant R01 EB030362‐14S1 from the National Institute of Biomedical Imaging and Bioengineering; and by grants UL1‐TR‐000040, UL1‐TR‐001079, and UL1‐TR‐001420 from the National Center for Advancing Translational Sciences (NCATS). Additional support was provided by the National Institute on Aging and grants R01 AG080821 and R01 AG070867. The funding sources had no role in the study design, the analysis and interpretation of data, the writing of the report, or the decision to submit the article for publication.
Heckbert SR, Jensen PN, Erus G, et al. Heart rate fragmentation and brain MRI markers of small vessel disease in MESA. Alzheimer's Dement. 2024;20:1397–1405. 10.1002/alz.13554
REFERENCES
- 1. Virani SS, Alonso A, Benjamin EJ, et al. Heart disease and stroke statistics‐2020 update: a report from the american heart association. Circulation. 2020;141:e139‐e596. doi: 10.1161/CIR.0000000000000757 [DOI] [PubMed] [Google Scholar]
- 2. Kalantarian S, Ruskin JN. Atrial fibrillation and cognitive decline: phenomenon or epiphenomenon? Cardiol Clin. 2016;34:279‐285. doi: 10.1016/j.ccl.2015.12.011 [DOI] [PubMed] [Google Scholar]
- 3. Goette A, Kalman JM, Aguinaga L, et al. EHRA/HRS/APHRS/SOLAECE expert consensus on atrial cardiomyopathies: definition, characterization, and clinical implication. Heart Rhythm. 2017;14:e3‐e40. doi: 10.1016/j.hrthm.2016.05.028 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Wang W, Zhang MJ, Inciardi RM, et al. Association of echocardiographic measures of left atrial function and size with incident dementia. JAMA. 2022;327:1138‐1148. doi: 10.1001/jama.2022.2518 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Austin TR, Jensen PN, Nasrallah IM, et al. Left atrial function and arrhythmias in relation to small vessel disease on brain mri: the multi‐ethnic study of atherosclerosis. J Am Heart Assoc. 2022;11:e026460. doi: 10.1161/JAHA.122.026460 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Costa MD, Davis RB, Goldberger AL. Heart rate fragmentation: a new approach to the analysis of cardiac interbeat interval dynamics. Front Physiol. 2017;8:255. doi: 10.3389/fphys.2017.00255 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Costa MD, Davis RB, Goldberger AL. Heart rate fragmentation: a symbolic dynamical approach. Front Physiol. 2017;8:827. doi: 10.3389/fphys.2017.00827 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Costa MD, Redline S, Soliman EZ, Goldberger AL, Heckbert SR. Fragmented sinoatrial dynamics in the prediction of atrial fibrillation: the Multi‐Ethnic Study of Atherosclerosis. Am J Physiol Heart Circ Physiol. 2021;320:H256‐H271. doi: 10.1152/ajpheart.00421.2020 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Costa MD, Redline S, Hughes TM, Heckbert SR, Goldberger AL. Prediction of cognitive decline using heart rate fragmentation analysis: the multi‐ethnic study of atherosclerosis. Front Aging Neurosci. 2021;13:708130. doi: 10.3389/fnagi.2021.708130 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Bild DE, Bluemke DA, Burke GL, et al. Multi‐ethnic study of atherosclerosis: objectives and design. Am J Epidemiol. 2002;156:871‐881. [DOI] [PubMed] [Google Scholar]
- 11. Chen X, Wang R, Zee P, et al. Racial/ethnic differences in sleep disturbances: the multi‐ethnic study of atherosclerosis (MESA). Sleep. 2015;38:877‐888. doi: 10.5665/sleep.4732 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Austin TR, Nasrallah IM, Erus G, et al. Association of brain volumes and white matter injury with race, ethnicity, and cardiovascular risk factors: the multi‐ethnic study of atherosclerosis. J Am Heart Assoc. 2022;11:e023159. doi: 10.1161/JAHA.121.023159 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Wardlaw JM, Smith C, Dichgans M. Small vessel disease: mechanisms and clinical implications. Lancet Neurol. 2019;18:684‐696. doi: 10.1016/S1474-4422(19)30079-1 [DOI] [PubMed] [Google Scholar]
- 14. Doshi J, Erus G, Ou Y, et al. Alzheimer's Neuroimaging Initiative. MUSE: mUlti‐atlas region segmentation utilizing ensembles of registration algorithms and parameters, and locally optimal atlas selection. Neuroimage. 2016;127:186‐195. doi: 10.1016/j.neuroimage.2015.11.073 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Doshi J, Erus G, Habes M, DavatzikosC. DeepMRSeg: a convolutional deep neural network for anatomy and abnormality segmentation on MR images. arXiv preprint arXiv. 2019;1907:02110. [Google Scholar]
- 16. Haight T, Bryan RN, Erus G, et al. White matter microstructure, white matter lesions, and hypertension: an examination of early surrogate markers of vascular‐related brain change in midlife. Neuroimage Clin. 2018;18:753‐761. doi: 10.1016/j.nicl.2018.02.032 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Rashid T, Abdulkadir A, Nasrallah IM. DEEPMIR: a deep neural network for differential detection of cerebral microbleeds and iron deposits in MRI. Sci Rep. 2021;11:14124. doi: 10.1038/s41598-021-93427-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Levey AS, Stevens LA, Schmid CH, et al. A new equation to estimate glomerular filtration rate. Ann Intern Med. 2009;150:604‐612. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Bluemke DA, Kronmal RA, Lima JA, et al. The relationship of left ventricular mass and geometry to incident cardiovascular events: the MESA (Multi‐Ethnic Study of Atherosclerosis) study. J Am Coll Cardiol. 2008;52:2148‐2155. doi: 10.1016/j.jacc.2008.09.014 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Longstreth WT Jr, Gasca NC, Gottesman RF, Pearce JB, Sacco RL. Adjudication of transient ischemic attack and stroke in the multi‐ethnic study of atherosclerosis. Neuroepidemiology. 2018;50:23‐28. doi: 10.1159/000486174 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Austin TR, Wiggins KL, Blackshear C, et al. Atrial fibrillation in an African‐American cohort: the jackson heart study. Clin Cardiol. 2018;41:1049‐1054. doi: 10.1002/clc.23020 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Habes M, Pomponio R, Shou H, et al. The Brain Chart of Aging: machine‐learning analytics reveals links between brain aging, white matter disease, amyloid burden, and cognition in the iSTAGING consortium of 10,216 harmonized MR scans. Alzheimers Dement. 2021;17:89‐102. doi: 10.1002/alz.12178 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Ikram MA, Vrooman HA, Vernooij MW, et al. Brain tissue volumes in relation to cognitive function and risk of dementia. Neurobiol Aging. 2010;31:378‐386. doi: 10.1016/j.neurobiolaging.2008.04.008 [DOI] [PubMed] [Google Scholar]
- 24. Cremers LGM, Wolters FJ, de Groot M, et al. Structural disconnectivity and the risk of dementia in the general population. Neurology. 2020;95:e1528‐e1537. doi: 10.1212/WNL.0000000000010231 [DOI] [PubMed] [Google Scholar]
- 25. Matusik PS, Zhong C, Matusik PT, Alomar O, Stein PK. Neuroimaging studies of the neural correlates of heart rate variability: a systematic review. J Clin Med. 2023;12. doi: 10.3390/jcm12031016 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Koenig J, Abler B, Agartz I, et al. Cortical thickness and resting‐state cardiac function across the lifespan: a cross‐sectional pooled mega‐analysis. Psychophysiology. 2021;58:e13688. doi: 10.1111/psyp.13688 [DOI] [PubMed] [Google Scholar]
- 27. Yoo HJ, Thayer JF, Greening S, et al. Brain structural concomitants of resting state heart rate variability in the young and old: evidence from two independent samples. Brain Struct Funct. 2018;223:727‐737. doi: 10.1007/s00429-017-1519-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Yoo HJ, Nashiro K, Min J, et al. Heart rate variability (HRV) changes and cortical volume changes in a randomized trial of five weeks of daily HRV biofeedback in younger and older adults. Int J Psychophysiol. 2022;181:50‐63. doi: 10.1016/j.ijpsycho.2022.08.006 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Goldberger JJ, Arora R, Buckley U, Shivkumar K. Autonomic nervous system dysfunction: jACC focus seminar. J Am Coll Cardiol. 2019;73:1189‐1206. doi: 10.1016/j.jacc.2018.12.064 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Shen MJ, Arora R, Jalife J. Atrial Myopathy. JACC Basic Transl Sci. 2019;4:640‐654. doi: 10.1016/j.jacbts.2019.05.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Lavy S, Stern S, Melamed E, Cooper G, Keren A, Levy P. Effect of chronic atrial fibrillation on regional cerebral blood flow. Stroke. 1980;11:35‐38. [DOI] [PubMed] [Google Scholar]
- 32. Petersen P, Kastrup J, Videbaek R, Boysen G. Cerebral blood flow before and after cardioversion of atrial fibrillation. J Cereb Blood Flow Metab: official journal of the International Society of Cerebral Blood Flow and Metabolism. 1989;9:422‐425. doi: 10.1038/jcbfm.1989.62 [DOI] [PubMed] [Google Scholar]
- 33. Terrosu P. Association between heart and dementia… keep an eye on the left atrium. Eur Heart J Suppl. 2022;24:I186‐I189. doi: 10.1093/eurheartjsupp/suac092 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Wilson S, Ardle RM, Tolley C, Slight S. Usability and acceptability of wearable technology in the early detection of dementia. Alzheimers Dement. 2022;18(Suppl 2):e059820. doi: 10.1002/alz.059820 [DOI] [PubMed] [Google Scholar]
- 35. Stavrakis S, Nakagawa H, Po SS, Scherlag BJ, Lazzara R, Jackman WM. The role of the autonomic ganglia in atrial fibrillation. JACC Clin Electrophysiol. 2015;1:1‐13. doi: 10.1016/j.jacep.2015.01.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Bonaz B, Sinniger V, Pellissier S. Anti‐inflammatory properties of the vagus nerve: potential therapeutic implications of vagus nerve stimulation. J Physiol. 2016;594:5781‐5790. doi: 10.1113/JP271539 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Jessen F, Kaduszkiewicz H, Daerr M, et al. Anticholinergic drug use and risk for dementia: target for dementia prevention. Eur Arch Psychiatry Clin Neurosci. 2010;260(Suppl 2):S111‐115. doi: 10.1007/s00406-010-0156-4 [DOI] [PubMed] [Google Scholar]
- 38. Pavlov VA, Tracey KJ. The cholinergic anti‐inflammatory pathway. Brain Behav Immun. 2005;19:493‐499. doi: 10.1016/j.bbi.2005.03.015 [DOI] [PubMed] [Google Scholar]
- 39. Packer M. Characterization, pathogenesis, and clinical implications of inflammation‐related atrial myopathy as an important cause of atrial fibrillation. J Am Heart Assoc. 2020;9:e015343. doi: 10.1161/JAHA.119.015343 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Chung MK, Martin DO, Sprecher D, et al. C‐reactive protein elevation in patients with atrial arrhythmias: inflammatory mechanisms and persistence of atrial fibrillation. Circulation. 2001;104:2886‐2891. [DOI] [PubMed] [Google Scholar]
- 41. Huston JM, Tracey KJ. The pulse of inflammation: heart rate variability, the cholinergic anti‐inflammatory pathway and implications for therapy. J Intern Med. 2011;269:45‐53. doi: 10.1111/j.1365-2796.2010.02321.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42. Stavrakis S, Humphrey MB, Scherlag B, et al. Low‐Level vagus nerve stimulation suppresses post‐operative atrial fibrillation and inflammation: a randomized study. JACC Clin Electrophysiol. 2017;3:929‐938. doi: 10.1016/j.jacep.2017.02.019 [DOI] [PubMed] [Google Scholar]
- 43. Collins O, Dillon S, Finucane C, Lawlor B, Kenny RA. Parasympathetic autonomic dysfunction is common in mild cognitive impairment. Neurobiol Aging. 2012;33:2324‐2333. doi: 10.1016/j.neurobiolaging.2011.11.017 [DOI] [PubMed] [Google Scholar]
- 44. Schaich CL, Malaver D, Chen H, et al. Association of heart rate variability with cognitive performance: the multi‐ethnic study of atherosclerosis. J Am Heart Assoc. 2020;9:e013827. doi: 10.1161/JAHA.119.013827 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45. Wolters FJ, Mattace‐Raso FU, Koudstaal PJ, Hofman A, Ikram MA. Heart brain connection collaborative research g. orthostatic hypotension and the long‐term risk of dementia: a population‐based study. PLoS Med. 2016;13:e1002143. doi: 10.1371/journal.pmed.1002143 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46. Gomez‐Angelats E, de La Sierra A, Sierra C, Parati G, Mancia G, Coca A. Blood pressure variability and silent cerebral damage in essential hypertension. Am J Hypertens. 2004;17:696‐700. doi: 10.1016/j.amjhyper.2004.05.002 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supporting Information
Supporting Information