Left Atrial Strain and the Risk of Atrial Arrhythmias From Extended Ambulatory Cardiac Monitoring: MESA

Abstract

Background

Abnormalities in left atrial (LA) function often occur before LA structural changes and clinically identified atrial fibrillation (AF). Little is known about the relationship between LA strain and the risk of subclinical atrial arrhythmias detected from extended ambulatory cardiac monitoring.

Methods and Results

A total of 1441 participants of MESA (Multi‐Ethnic Study of Atherosclerosis) completed speckle‐tracking echocardiography and cardiac monitoring during 2016 to 2018 (mean age, 73 years); participants in AF during echocardiography or during the entire cardiac monitoring period were excluded. Absolute values of LA reservoir, booster pump, and conduit strains were measured. We evaluated associations of LA strain with monitor‐detected AF, premature atrial contractions, and supraventricular tachycardia. Primary analyses adjusted for demographic variables, blood pressure, diabetes, smoking, and clinical cardiovascular disease. Cardiac monitoring (median, 14 days) detected AF in 3%. Each SD (4.0%) lower (worse) LA booster pump strain was associated with 84% higher risk of monitor‐detected AF (95% CI, 30%–162%), 39% higher premature atrial contraction frequency (95% CI, 27%–53%), and 19% higher supraventricular tachycardia frequency (95% CI, 10%–29%). Additional adjustment for NT‐proBNP (N‐terminal pro‐B‐type natriuretic peptide), LA volume index, tissue Doppler a′ peak velocity, left ventricular ejection fraction, and global longitudinal strain had little impact on associations. Findings were similar for LA reservoir strain and null for LA conduit strain.

Conclusions

In a multiethnic community‐based cohort, impaired LA strain was an important correlate of subclinical atrial arrhythmias, even after adjustment for conventional measures of LA structure and function.

Nonstandard Abbreviations and Acronyms

LAVI

left atrial volume index

MESA

Multi‐Ethnic Study of Atherosclerosis

SVT

supraventricular tachycardia

Clinical Perspective.

What Is New?

• We found that low left atrial reservoir and booster pump strains were associated with a substantially higher risk of monitor‐detected atrial fibrillation and greater frequency of premature atrial contractions and supraventricular tachycardia.

• These findings add to the literature on impaired left atrial function as a risk factor for clinically recognized atrial fibrillation by extending these findings to subclinical atrial fibrillation and other atrial arrhythmias detected from extended ambulatory cardiac monitoring, a sensitive and unbiased method for identifying arrhythmias.

What Are the Clinical Implications?

• Assessment of left atrial function may help to identify patients at increased risk for these outcomes, which may be of relevance for atrial fibrillation screening and prevention strategies.

Subclinical atrial fibrillation (AF), defined as an episode of AF that escapes clinical detection, is common and associated with clinically recognized AF and ischemic stroke. ¹ , ² However, the temporal relationships and pathophysiologic mechanisms linking subclinical AF to its complications are incompletely understood. ³ , ⁴ An emerging literature has identified electrocardiographic correlates of abnormal left atrial (LA) structure and function, referred to as atrial myopathy, as important risk factors for adverse cardiovascular outcomes. ⁵ For instance, frequent premature atrial contractions (PACs) and runs of supraventricular tachycardia (SVT) are associated with an elevated risk of ischemic stroke, even after adjusting for intercurrent AF. ⁶

Strain imaging techniques that measure the magnitude and rate of myocardial deformation can detect subtle impairments in myocardial function. ⁷ In particular, speckle‐tracking echocardiography is an increasingly used imaging technique that can accurately measure LA strain during the 3 atrial mechanical functions: reservoir, booster pump, and conduit. ⁸ Impaired LA strain often precedes structural changes such as LA enlargement and has been associated with clinically recognized AF and other cardiovascular events in previous studies. ⁹ , ¹⁰ , ¹¹ However, these previous studies relied primarily on routine clinical surveillance for AF detection, which fails to capture a substantial portion of intermittent AF and results in ascertainment bias that differs by race. ¹² Investigating echocardiographic measures of LA strain and subclinical arrhythmias in a multiethnic population setting may yield important insights into the mechanistic relationships between cardiac function, arrhythmogenesis, and cardiovascular outcomes.

Among participants of a multiethnic, community‐based cohort study who completed comprehensive transthoracic echocardiography and extended ambulatory cardiac monitoring, we conducted cross‐sectional analyses evaluating associations of several LA strain measures with monitor‐detected atrial arrhythmias. We hypothesized that impaired LA strain would be associated with a greater burden of atrial arrhythmias.

METHODS

The data, analytic methods, and study materials will not be made available to other researchers for purposes of reproducing the results or replicating the procedure because of participant privacy issues. Investigators interested in analyzing data from MESA (Multi‐Ethnic Study of Atherosclerosis) may contact the Coordinating Center at the University of Washington or the National Heart, Lung, and Blood Institute Biologic Specimen and Data Repository Information Coordinating Center.

Study Population

MESA is an ongoing cohort study that recruited 6814 men and women aged 45 to 84 years without clinically recognized cardiovascular disease in 2000 to 2002 from 6 sites (Baltimore, MD; Chicago, IL; Forsyth County, NC; Los Angeles County, CA; New York, NY; and St. Paul, MN). ¹³ Study participants completed a baseline examination and 5 follow‐up examinations. In addition, study personnel contacted participants every 9 to 12 months during follow‐up to assess clinical status and to identify new hospitalizations and diagnoses. The study was approved by the institutional review board of all participating sites, and all study participants provided informed consent.

Of the 3303 MESA participants who returned for exam 6 (2016–2018), those who completed contemporaneous comprehensive transthoracic echocardiography and extended ambulatory cardiac monitoring were eligible for analysis (n=1528). We excluded 1 participant without follow‐up for diagnosis of clinically recognized AF after exam 1, 21 participants with <24 hours of cardiac monitoring, and 8 participants with a paced rhythm during cardiac monitoring. We also excluded 57 participants who were in AF during echocardiography or who had 100% burden of AF during cardiac monitoring (Figure 1) because LA booster pump function cannot be assessed during AF and because these participants were likely to have long‐standing AF.

Figure 1. Flow diagram of MESA participants included in this analysis.

AF indicates atrial fibrillation; MESA, Multi‐Ethnic Study of Atherosclerosis; and TTE, transthoracic echocardiography.

Echocardiography Protocol

Participants underwent comprehensive 2‐dimensional, Doppler, M‐mode, and speckle‐tracking echocardiography. ¹⁴ Cardiac structure and function were quantified in accordance with society guideline recommendations. ¹⁵ Left ventricular (LV) ejection fraction and LA volume were measured using Simpson's biplane method of disks from the apical 4‐chamber and apical 2‐chamber views, and LV mass was calculated by the Devereux formula. Peak myocardial velocities during early diastole (e′ velocity) and atrial contraction (a′ velocity)—measurements of LV early diastolic relaxation and atrial contractility, respectively—were quantified at the septal and lateral mitral annulus using pulse‐waved tissue Doppler imaging.

Speckle‐tracking echocardiography was performed by an experienced sonographer blinded to other clinical data, and all strain curves were verified by 2 cardiologists with expertise in echocardiography. Biplane measurements of LA reservoir, booster pump, and conduit strains from the apical 4‐chamber and apical 2‐chamber views were obtained as previously reported. ¹⁴ P‐wave electrocardiographic gating was used for patients in sinus rhythm during echocardiography, and all LA strain measurements were subsequently converted to R‐wave electrocardiographic gating using a standardized formula. ¹⁶ LV global longitudinal strain and LV early diastolic strain rates were calculated as the average of measurements from the apical 4‐chamber, apical 3‐chamber, and apical 2‐chamber views. Measurements were repeated in the apical 4‐chamber view after study participants completed a passive leg raise using a wedge‐shaped pillow as a method of intravascular volume challenge. All values were reported as absolute percentages, with lower strain values corresponding to worse LA function. Echocardiographic analyses were performed using General Electric EchoPAC software version 201 (GE Healthcare, Waukesha, WI).

Ambulatory Cardiac Monitoring

Consenting participants completed extended ambulatory cardiac monitoring with the Zio Patch XT (iRhythm Technologies, Inc., San Francisco, CA), a US Food and Drug Administration–approved single‐channel adhesive ECG monitor that can store up to 14 days of cardiac rhythm data. ¹⁷ Participants mailed the Zio Patch XT to the manufacturer at the conclusion of their monitoring period. Ambulatory ECG data were processed and analyzed by the manufacturer and independently verified by the Epidemiological Cardiology Research Center (Wake Forest University, Winston‐Salem, NC). A subset of participants (n=547) completed a second 14‐day ambulatory monitoring period with a separate Zio Patch XT, and data from both patches were combined for these individuals.

Atrial arrhythmias of interest were the presence of monitor‐detected AF, the percentage of time in AF (AF burden), the frequency of PACs per hour, and the frequency of runs of SVT per day. Monitor‐detected AF was defined as the presence of an irregularly irregular rhythm with absent P waves at least 30 seconds in duration or the presence of atrial flutter. ¹² The frequency of PACs per hour was calculated as the sum of the number of isolated PACs plus 2 times the number of PAC doublets plus 3 times the number of PAC triplets divided by the total cardiac monitoring period in hours. A run of SVT was defined as ≥4 consecutive PACs, and the frequency of runs of SVT per day was calculated as the total number of runs of SVT divided by the total cardiac monitoring period in days. ¹⁸

Covariates

Sex and race and ethnicity were measured at exam 1 (2000–2002), and the remaining covariates were measured at exam 6 (2016–2018). Three seated, resting blood pressure measurements were obtained using an automated oscillometric method, and the average of the second and third readings were recorded for analyses. Smoking status was self‐reported on a questionnaire, and medication use was assessed by the inventory method. Prevalent diabetes was defined as a fasting plasma glucose value $\ge$126 mg/dL or use of glucose‐lowering treatment, ¹⁹ and treated hypertension was defined as the use of at least 1 antihypertensive medication. Prevalent myocardial infarction and heart failure were determined from ongoing events adjudication. ¹³ Clinically recognized AF from the baseline examination (2000–2002) through December 2015 (called prevalent AF in this analysis of data from 2016–2018) was ascertained from resting 12‐lead ECGs performed at study examinations, hospital discharge diagnosis codes, and Medicare claims data for those enrolled in fee‐for‐service Medicare, as described previously. ²⁰ Glomerular filtration rate was estimated from serum creatinine levels and the Chronic Kidney Disease Epidemiology Collaboration equation. ²¹ NT‐proBNP (N‐terminal pro‐B‐type natriuretic peptide) levels were obtained through the Olink Cardiovascular III panel (Uppsala, Sweden) and were expressed as normalized protein units on a log₂ scale. ⁹

Statistical Analysis

The presence of monitor‐detected AF is a binary outcome. The frequency of PACs per hour and frequency of runs of SVT per day are continuous outcomes with right‐skewed distributions. Because a small number of participants had no PACs or runs of SVT, a value of 1 was added to the total arrhythmia count before performing a natural log transformation for these outcomes. We evaluated cross‐sectional associations of 3 continuous LA strain measures (LA reservoir, booster pump, and conduit strains) with monitor‐detected AF using logistic regression and with continuous outcomes using linear regression. Each LA strain measure was modeled separately, and all primary analyses adjusted for age, sex, race and ethnicity, height, weight, systolic and diastolic blood pressure, treated hypertension, smoking status, diabetes, prevalent cardiovascular disease, and duration of cardiac monitoring. Adjusted associations from logistic regression were expressed as odds ratios, while associations from linear regression were expressed as the ratio of geometric means, which provides the percentage difference in frequency of the arrhythmia per unit difference in LA strain. All measures of association were reported per 1 SD lower (worse) LA strain. Participants with missing values for height (n=1), systolic blood pressure (n=1), diastolic blood pressure (n=1), smoking status (n=2), diabetes (n=17), or treated hypertension (n=6) as well as those with missing LA reservoir (n=56), booster pump (n=72), or conduit strain (n=72) values were excluded from analyses.

Secondary analyses additionally adjusted for estimated glomerular filtration rate, NT‐proBNP, conventional echocardiography measurements (LA volume index [LAVI], LV mass index, LV ejection fraction, septal e′ peak velocity), and speckle‐tracking measures of LV function (global longitudinal strain, early diastolic strain rate). We also evaluated change in LA strain following passive leg raise, a measure of LA compliance or functional reserve, ²² as the exposure variable in separate analyses. To determine whether LA strain provides unique information about AF risk above and beyond tissue Doppler a′ peak velocity, a widely available measure in clinical echocardiography, we also evaluated tissue Doppler a′ peak velocity as an exposure variable and adjusted this measure in analyses of each LA strain variable. We evaluated for effect modification of significant associations by race and ethnicity and LAVI, and we conducted sensitivity analyses in subgroups of participants with normal LAVI and LV ejection fraction, in those without clinically recognized AF, and in those without hypertrophic cardiomyopathy or moderate or greater left‐sided valvular heart disease (aortic stenosis, aortic regurgitation, mitral stenosis, or mitral regurgitation). All analyses were conducted with R version 4.0.2 (R Foundation for Statistical Computing, Vienna, Austria), and a 2‐sided P‐value <0.05 was considered statistically significant.

RESULTS

Among the 1441 participants eligible for this analysis, the mean (SD) age was 73 (8) years, and 52% were women (Table 1). Twenty‐five percent of participants were Black, while 21% and 14% were Hispanic and Chinese American, respectively. Four percent had a prior myocardial infarction or heart failure, and 7% had clinically recognized AF. The mean (SD) LAVI was 28 (8) mL/m² while the mean (SD) LV ejection fraction was 62% (5%). The mean (SD) LA reservoir, booster pump, and conduit strains were 27.4% (5.7%), 15.0% (4.0%), and 12.4% (4.0%), respectively, while the mean (SD) change in LA reservoir, booster pump, and conduit strains following passive leg raise were 3.2% (4.5%), −0.4% (4.5%), and 3.6% (4.2%), respectively (Figure S1). Participants with lower LA reservoir strain were older; had a higher prevalence of diabetes, treated hypertension, and clinically recognized AF; and had a higher LAVI and lower LV ejection fraction.

Table 1.

Participant Characteristics by Quartile of Left Atrial Reservoir Strain

	Total Sample N=1441	Quartile 1 (8.3–23.4) N=347	Quartile 2 (23.4–27.1) N=346	Quartile 3 (27.1–31.0) N=346	Quartile 4 (31.0–50.1) N=346
Demographic/clinical characteristics
Age, y, mean (SD)	73 (8)	76 (8)	73 (8)	72 (8)	71 (8)
Height, cm, mean (SD)	165 (10)	163 (10)	166 (10)	166 (10)	166 (10)
Weight, lbs, mean (SD)	171 (38)	173 (38)	175 (40)	170 (38)	166 (36)
Systolic blood pressure, mm Hg, mean (SD)	127 (20)	133 (24)	127 (20)	126 (18)	123 (18)
Diastolic blood pressure, mm Hg, mean (SD)	69 (10)	70 (12)	69 (10)	69 (9)	68 (9)
Female, n (%)	746 (52)	184 (53)	168 (49)	175 (51)	185 (53)
Race and ethnicity, n (%)
White	578 (40)	117 (34)	133 (38)	142 (41)	155 (45)
Chinese American	205 (14)	44 (13)	48 (14)	56 (16)	53 (15)
Black	355 (25)	94 (27)	83 (24)	87 (25)	86 (25)
Hispanic	303 (21)	92 (27)	82 (24)	61 (18)	52 (15)
Smoking status, n (%)
Never	672 (47)	157 (45)	159 (46)	168 (49)	162 (47)
Former	681 (47)	168 (48)	166 (48)	157 (45)	166 (47)
Current	86 (6)	20 (6)	21 (6)	21 (6)	18 (5)
Prevalent diabetes, n (%)	325 (23)	97 (28)	81 (23)	62 (18)	68 (20)
Antihypertensive medication use, n (%)	867 (60)	249 (72)	60 (60)	187 (54)	187 (54)
Previous myocardial infarction or heart failure, n (%)	54 (4)	27 (8)	10 (3)	5 (1)	7 (2)
Prevalent clinically recognized AF, n (%)	97 (7)	46 (13)	18 (5)	16 (5)	12 (3)
Laboratory measurements
eGFR (CKD‐EPI), mL/min per 1.73 m2, mean (SD)	75 (18)	72 (18)	75 (17)	76 (18)	77 (17)
Echocardiography measurements
Left atrial volume index, mL/m2, mean (SD)	28 (8)	32 (10)	28 (7)	26 (7)	25 (5)
Left ventricular mass index, g/m2, mean (SD)	84 (21)	90 (24)	85 (21)	83 (22)	79 (17)
Left ventricular ejection fraction, %, mean (SD)	62 (5)	61 (7)	62 (5)	63 (5)	63 (4)
Septal a′ peak velocity, cm/s, mean (SD)	10.6 (2.2)	9.4 (2.2)	10.4 (2.0)	11.1 (2.0)	11.5 (1.8)
Septal e′ peak velocity, cm/s, mean (SD)	7.3 (2.0)	6.3 (1.8)	7.0 (1.8)	7.5 (1.9)	8.2 (2.0)
Left ventricular global longitudinal strain, %, mean (SD)	19.9 (2.8)	18.6 (3.2)	19.7 (2.5)	20.4 (2.5)	20.8 (2.5)
Left ventricular early diastolic strain rate, 1/s, mean (SD)	1.1 (0.3)	1.0 (0.3)	1.0 (0.3)	1.1 (0.3)	1.2 (0.3)
Left atrial strain measurements
Reservoir strain, %, mean (SD)	27.4 (5.7)
Booster pump strain, %, mean (SD)	15.0 (4.0)
Conduit strain, %, mean (SD)	12.4 (4.0)

AF indicates atrial fibrillation; CKD‐EPI, Chronic Kidney Disease Epidemiology Collaboration formula; and eGFR, estimated glomerular filtration rate.

Monitor‐Detected AF

The median (interquartile range) duration of cardiac monitoring was 14.0 (13.6–26.4) days, and AF was identified in 48 (3.3%) (Table S1). Demographic and clinical characteristics among participants with and without monitor‐detected AF are displayed in Table S2. Among participants with monitor‐detected AF, the median (interquartile range) AF burden was 1.6% (0.2%–10.3%). In primary analyses, each SD lower LA reservoir strain was associated with 74% higher risk of monitor‐detected AF (95% CI, 23%–148%) and each SD lower LA booster pump strain was associated with 84% higher risk (95% CI, 30%–162%), while LA conduit strain was not associated (Table 2). Risk estimates for both LA reservoir and booster pump strains were not meaningfully impacted by additional adjustment for estimated glomerular filtration rate, NT‐proBNP, conventional echocardiographic measures, or speckle‐tracking measures of LV function. Changes in LA strain measures following passive leg raise were not associated with AF risk (Table S3). Associations based on LA strain quartiles are displayed in Figure S2. Participants in the first (lowest) quartile compared with the fourth (highest) quartile of LA reservoir strain had a 3.7‐fold greater risk (95% CI, 1.5–10.8) of monitor‐detected AF, and participants in the first and second quartiles compared with the fourth quartile of LA booster pump strain had 4.1‐ (95% CI, 1.4–14.6) and 4.1‐fold greater risks (95% CI, 1.4–14.5), respectively. Excluding 97 participants with prior clinically recognized AF (8 with monitor‐detected AF, 89 without) and 44 participants with hypertrophic cardiomyopathy or left‐sided valvular heart disease (4 with monitor‐detected AF, 40 without) resulted in associations that were significant and nearly identical to findings from the primary analysis (Tables S4 and S5). There was little statistical evidence for effect modification of associations by race and ethnicity or LAVI (Table S6). In the subgroup with normal LAVI, LA booster pump strain was more strongly associated with monitor‐detected AF than LA reservoir strain, and among participants with normal LV ejection fraction, LA reservoir and booster pump strains were associated with comparable risks. Tissue Doppler a′ peak velocity (odds ratio, 1.51 [95% CI, 1.11–2.05]) was less strongly associated with monitor‐detected AF than LA reservoir and booster pump strains. In addition, adjustment for tissue Doppler a′ peak velocity slightly attenuated the association of LA reservoir strain with monitor‐detected AF (odds ratio, 1.54 [95% CI, 1.04–2.31]), but the association for LA booster pump strain was unchanged (odds ratio, 1.84 [95% CI, 1.21–2.85]).

Table 2.

Associations of LA Strain With Monitor‐Detected AF

Continuous exposures	Sample size for primary analysis, N	AF Events	Model 1	Model 2	Model 3	Model 4
			OR (95% CI)	OR (95% CI)	OR (95% CI)	OR (95% CI)
Reservoir strain	1358	44	1.74 (1.23–2.48)	1.62 (1.13–2.36)	1.62 (1.10–2.43)	1.65 (1.08–2.56)
Booster pump strain	1342	41	1.84 (1.30–2.62)	1.80 (1.25–2.62)	1.77 (1.22–2.61)	1.87 (1.25–2.85)
Conduit strain	1342	41	1.03 (0.73–1.47)	0.99 (0.70, 1.42)	0.95 (0.65–1.41)	0.95 (0.65–1.43)

The SDs for LA reservoir, booster pump, and conduit strains were 5.7%, 4.0%, and 4.0%, respectively. Model 1: Adjusted for age, sex, race and ethnicity, height, weight, systolic blood pressure, diastolic blood pressure, hypertension medication use, current smoking status, prevalent diabetes, previous myocardial infarction or heart failure, and ECG monitoring time. Model 2: Model 1 plus adjustment for estimated glomerular filtration rate and N‐terminal pro‐B‐type natriuretic peptide level. Model 3: Model 1 plus adjustment for LA volume index, LV mass index, LV ejection fraction, and septal e′ velocity. Model 4: Model 3 plus adjustment for LV global longitudinal strain and LV early diastolic strain rate. AF indicates atrial fibrillation; LA, left atrial; LV, left ventricular; and OR, odds ratio per 1 SD lower left atrial strain.

PACs and SVT

The median (interquartile range) frequency of PACs during the cardiac monitoring period was 4.2 (1.4–20.0) per hour. In primary analyses, each SD lower LA reservoir strain was associated with 42% higher PAC frequency (95% CI, 30%–56%), and each SD lower LA booster pump strain was associated with 39% higher PAC frequency (95% CI, 27%–53%) (Table 3). LA conduit strain was only weakly associated with PAC frequency (geometric mean ratio, 1.12 [95% CI, 1.02–1.24]). Restricted cubic splines demonstrated little evidence of nonlinearity of these relationships (Figure S3), and adjustment for additional potential confounding variables had little impact on risk estimates. In secondary analyses of change in LA strain following passive leg raise, only booster pump strain was associated with PAC frequency. Quartile analyses for LA reservoir and booster pump strains demonstrated graded, dose‐dependent relationships with the frequency of PACs per hour (Figure 2). Compared with the fourth (highest) quartile of LA booster pump strain, participants in the first, second, and third quartiles had 149% (95% CI, 94%–220%), 95% (95% CI, 52%–150%), and 46% (95% CI, 14%–86%) higher PAC frequency, respectively; findings were similar for quartiles of LA reservoir strain. The findings were nearly identical after excluding participants with clinically recognized AF (Table S7), and there was little evidence of interaction by LAVI or race and ethnicity.

Table 3.

Associations of LA Strain With the Frequency of PACs and the Frequency of Runs of SVT

Continuous exposures	Sample size for primary analysis, N	Model 1	Model 2	Model 3	Model 4
		GMR (95% CI)	GMR (95% CI)	GMR (95% CI)	GMR (95% CI)
Frequency of PACs
Reservoir strain	1358	1.42 (1.30–1.56)	1.30 (1.19–1.43)	1.36 (1.23–1.51)	1.39 (1.24–1.55)
Booster pump strain	1342	1.39 (1.27–1.53)	1.29 (1.18–1.42)	1.31 (1.18–1.44)	1.33 (1.19–1.48)
Conduit strain	1342	1.12 (1.02–1.24)	1.07 (0.98–1.18)	1.11 (1.00–1.23)	1.13 (1.01–1.26)
Frequency of runs of SVT
Reservoir strain	1358	1.19 (1.10–1.29)	1.10 (1.02–1.19)	1.15 (1.05–1.25)	1.14 (1.04–1.25)
Booster pump strain	1342	1.19 (1.10–1.29)	1.11 (1.03–1.21)	1.16 (1.07–1.26)	1.15 (1.05–1.26)
Conduit strain	1342	1.04 (0.96–1.12)	1.00 (0.93–1.08)	1.00 (0.92–1.09)	1.00 (0.91–1.10)

The SDs for LA reservoir, booster pump, and conduit strains were 5.7%, 4.0%, and 4.0%, respectively. Model 1: Adjusted for age, sex, race and ethnicity, height, weight, systolic blood pressure, diastolic blood pressure, hypertension medication use, current smoking status, prevalent diabetes, previous myocardial infarction or heart failure, and ECG monitoring time. Model 2: Model 1 plus adjustment for estimated glomerular filtration rate and N‐terminal pro‐B‐type natriuretic peptide level. Model 3: Model 1 plus adjustment for LA volume index, LV mass index, LV ejection fraction, and septal e′ velocity. Model 4: Model 3 plus adjustment for LV global longitudinal strain and LV early diastolic strain rate. GMR indicates geometric mean ratio of frequency of PACs or frequency of runs of SVT per 1 SD lower LA strain; LA, left atrial; LV, left ventricular; PAC, premature atrial contraction; and SVT, supraventricular tachycardia.

Figure 2. Associations of LA strain quartiles with (A) the frequency of PACs and (B) the frequency of runs of SVT.

The y axis is the geometric mean ratio of (A) the frequency of PACs and (B) the frequency of runs of SVT, and the x axis is the measure of LA strain. For each measure, the referent group is the highest quartile of LA strain (quartile 4). LA indicates left atrial; PAC, premature atrial contractions; and SVT, supraventricular tachycardia.

The median (interquartile range) frequency of runs of SVT during the cardiac monitoring period was 0.4 (0.1–1.2) per day. In primary analyses, each SD lower LA reservoir strain was associated with 19% higher (95% CI, 10%–29%) SVT frequency, and each SD lower LA booster pump strain was also associated with 19% higher (95% CI, 10%–29%) SVT frequency, while LA conduit strain was not associated. Restricted cubic splines demonstrated little evidence of nonlinearity of these relationships, and risk estimates did not meaningfully change with additional adjustment for potential confounding variables. Changes in LA strain measures following passive leg raise were not associated with SVT frequency. Quartile analyses for LA reservoir and booster pump strains also demonstrated graded, dose‐dependent relationships with the SVT frequency, and the magnitude of associations was similar after excluding participants with clinically recognized AF. There was little evidence of effect modification by race and ethnicity or LAVI.

DISCUSSION

In this community‐based, multiethnic population, we found that low LA reservoir and booster pump strains were associated with a substantially higher risk of monitor‐detected AF and a greater frequency of PACs and SVT. The strength of these associations persisted after adjustment for NT‐proBNP, conventional echocardiographic measures of LA structure and function, and LV structure and function, and in subgroups with normal LAVI, normal LV ejection fraction, no prior clinically recognized AF, and no hypertrophic cardiomyopathy or valvular heart disease. For PACs and SVT, we also observed evidence of linear, dose‐dependent relationships. These findings add to the literature on impaired LA function as a risk factor for clinically recognized AF by extending these findings to subclinical AF and other atrial arrhythmias detected from extended ambulatory cardiac monitoring, a sensitive and unbiased method for identifying arrhythmias.

Previous studies evaluating associations of LA function with incident AF in the general population have identified abnormal LA reservoir function as an important risk factor. Early efforts to quantify reservoir function assessed changes in LA volume during the cardiac cycle, and in the Framingham Offspring Study, lower echocardiographic LA function index, a derived measure incorporating the volumetric assessment of LA reservoir function, was associated with higher risk of incident clinically recognized AF after a median follow‐up of 8 years. ²³ These findings were corroborated using highly sensitive strain imaging techniques to measure reservoir function, and multiple studies of community‐dwelling individuals demonstrated that low baseline echocardiographic LA reservoir strain was associated with incident clinically recognized AF. ⁹ , ²⁴ , ²⁵

An emerging literature suggests that abnormal LA booster pump function may be a stronger risk factor than LA reservoir function for incident clinically recognized AF in the general population. In a recent analysis in the Copenhagen City Heart Study, each 1% lower LA booster pump strain was associated with an 8% higher risk of incident clinically recognized AF during 5 years of follow‐up, while a 1% lower LA reservoir strain was associated with only a 5% higher risk of incident AF. ²⁶ Similarly, among 203 elderly participants with stroke risk factors who received a long‐term implantable loop recorder, lower baseline LA booster pump strain by cardiac magnetic resonance was more strongly associated with incident AF after 40 months of continuous monitoring than LA reservoir strain. ²⁷ Our study, which used sensitive measurements of LA function from echocardiography and arrhythmia detection from ambulatory cardiac monitoring, corroborates these findings and highlights the importance of LA booster pump function as an independent risk factor for monitored‐detected AF in a generalizable population of community‐dwelling individuals. LA booster pump function reflects the ability of the LA to contract, whereas LA conduit function represents the ability of the LA to empty into the LV in early diastole; thus, it is not surprising that worse LA booster pump strain (reflective of LA contractile dysfunction) is more closely associated with AF compared with LA conduit strain, which is likely more reflective of the health of the LV in early diastole compared with the LA.

Few studies have evaluated associations of LA function with supraventricular ectopy. In a recent analysis in MESA that included most of the participants studied in the present study, Heckbert et al ¹⁸ examined the longitudinal associations of LA function measured by cardiac magnetic resonance with the frequency of PACs and runs of SVT ≈6 years later. After adjustment for AF risk factors and LV structure, LA reservoir strain was not significantly associated with PAC frequency or runs of SVT. In the Atherosclerosis Risk in Communities Study, in contrast, higher LA booster pump and reservoir strains from echocardiography were associated with lower frequencies of PACs and atrial tachycardia ≈3 to 6 years later. ²⁸ In our study, which contained contemporaneous measures of echocardiographic LA strain and supraventricular ectopy, we observed linear and dose‐dependent associations of LA reservoir and booster pump strains with PAC frequency and runs of SVT per day.

AF is the final common end point of pathologic LA remodeling and is a consequence of 4 principal mechanisms: electrical remodeling, structural remodeling, autonomic nervous system changes, and calcium‐handling abnormalities. ²⁹ Structural remodeling is characterized by atrial enlargement and tissue fibrosis, and in patients with paroxysmal and persistent AF, the presence and extent of impaired echocardiographic LA strain correlates with the degree of atrial fibrosis. ³⁰ Impaired echocardiographic LA reservoir strain is also more strongly correlated with the grade of LA myocardial fibrosis than maximal LAVI among patients with severe mitral regurgitation referred for cardiac surgery. ³¹

However, impaired LA function identified by speckle‐tracking echocardiography may be more than simply a biomarker of irreversible LA fibrosis and instead may identify at‐risk individuals who have not yet sustained permanent atrial structural remodeling. In an animal study of hypertensive rats, Shah et al ³² demonstrated that reduced LV strain identified impaired cardiomyocyte calcium handling before manifest cardiac fibrosis, and emerging clinical data further suggest that impaired LA function is reversible and recoverable. ³³ , ³⁴ Because calcium abnormalities are integral to the pathogenesis and progression of AF, impaired LA function could become an upstream therapeutic target to prevent the development of fixed atrial remodeling. ²⁹

Strengths of this study include the deeply phenotyped participants of a multiethnic community‐based cohort study, high‐quality measures of LA strain, and sensitive measures of arrhythmia from extended ambulatory cardiac monitoring. In particular, we used biplane measurements of LA strain that were obtained from a standardized comprehensive transthoracic echocardiography study with dedicated imaging of the LA. We acknowledge that the cross‐sectional study design cannot inform temporality of observed associations between abnormal LA function and subclinical atrial arrhythmias, but we intentionally excluded participants in AF during echocardiography or during the entire cardiac monitoring period to limit measurement error and reduce the potential for reverse causation. Despite the careful measurement of and adjustment for potential confounding variables, residual confounding is possible. The limited number of monitor‐detected AF events limits precise estimation of the associations between abnormal LA function and subclinical AF.

In conclusion, abnormal LA reservoir and booster pump functions were associated with higher risk for monitor‐detected AF, PAC frequency, and SVT frequency in a multiethnic community‐based sample. An accumulating body of evidence suggests that abnormal LA function may be an important risk factor for not only clinically detected AF but also several subclinical atrial arrhythmias. Assessments of LA function may help to identify patients at increased risk for these outcomes, which may be of relevance for AF screening and prevention strategies.

Sources of Funding

This 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, N01‐HC‐95165, N01‐HC‐95166, N01‐HC‐95167, N01‐HC‐95168, and N01‐HC‐95169 from the National Heart, Lung, and Blood Institute; and by grants UL1‐TR‐000040, UL1‐TR‐001079, and UL1‐TR‐001420 from the National Center for Advancing Translational Sciences. Dr Huber was supported by National Heart, Lung, and Blood Institute grant T32HL007828. Drs Bertoni and Shah were supported by National Heart, Lung, and Blood Institute grant R01HL127028. Dr Heckbert was supported by National Heart, Lung, and Blood Institute grant R01HL127659. Dr Floyd was supported by National Heart, Lung, and Blood Institute grant R01HL142599. The research reported here is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Disclosures

Dr Floyd has consulted for Shionogi, Inc. Dr Shah has received research grants from Actelion, AstraZeneca, Corvia, Novartis, and Pfizer; and has received personal fees from Abbott, Actelion, AstraZeneca, Amgen, Aria CV, Axon Therapies, Bayer, Boehringer‐Ingelheim, Boston Scientific, Bristol‐Myers Squibb, Cardiora, Coridea, CVRx, Cyclerion, Cytokinetics, Edwards Lifesciences, Eidos, Eisai, Imara, Impulse Dynamics, Intellia, Ionis, Ironwood, Lilly, Merck, MyoKardia, Novartis, Novo Nordisk, Pfizer, Prothena, Regeneron, Rivus, Sanofi, Shifamed, Tenax, Tenaya, and United Therapeutics. The remaining authors have no disclosures to report.

Supporting information

Tables S1

Figures S1–S3