Left atrial (LA) remodeling is a predictor of cardiovascular disease (CVD). We performed measurement of the LA function index (LAFI), a composite measure of LA structure and function, in a community-based cohort and here report the distribution and cross-sectional correlates of LAFI.
In 1,719 Framingham Offspring Study participants (54% women, mean age 66 ± 9 years), we derived LAFI from the LA emptying fraction, left ventricular (LV) outflow tract velocity time integral, and indexed maximal LA volume. We used multivariable linear regression to assess the clinical and echocardiographic correlates of LAFI adjusting for age, sex, anthropometric measurements, and CVD risk factors.
The average LAFI was 35.2 ± 12.1. Overall, LAFI declined with advancing age (β = −0.27, P < .001). LAFI was significantly higher (37.5 ± 11.6) in a subgroup of participants free of CVD and CVD risk factors compared with those with either of these conditions (34.5 ± 12.2). In multivariable models, LAFI was inversely related to antihypertensive use (β = −1.26, P = .038), prevalent atrial fibrillation (β = −4.46, P = .001), heart failure (β = −5.86, P = .008), and coronary artery disease (β = −2.01, P = .046). In models adjusting for echocardiographic variables, LAFI was directly related to LV ejection fraction (β = 14.84, P < .001) and inversely related to LV volume (β = −7.03, P < .001).
LAFI was inversely associated with antihypertensive use and prevalent CVD and was related to established echocardiographic traits of LV remodeling. Our results offer normative ranges for LAFI in a white community-based sample and suggest that LAFI represents a marker of pathological atrial remodeling.
Left atrial function index (LAFI) is a composite measure of atrial structural and functional remodeling.
LAFI is inversely associated with age, hypertension, and cardiovascular disease.
LAFI is positively associated with markers of positive cardiac remodeling.
The predictive value of LAFI for incident CVD should be assessed in future studies
Adverse left atrial (LA) remodeling is associated with increased risk of cardiovascular disease (CVD) and CVD-specific and all-cause mortality. Increased LA volume and abnormalities in phasic function are examples of echocardiographic traits that capture aspects of adverse structural and functional LA remodeling and have prognostic significance, particularly as predictors of incident or recurrent atrial fibrillation, heart failure, and cerebrovascular accident. Although LA volume is the recommended LA echocardiographic trait for clinical practice, its prognostic value diminishes when ventricular systolic and diastolic function are considered concomitantly.
A subtle decline in LA function, as detected by impaired LA phasic function (atrial reservoir phase, passive atrial emptying, and atrial systole), is associated with incident and recurrent CVD, adjusting for ventricular function, but such measures are not routinely collected as part of a standard echocardiographic examination. Left atrial function index (LAFI) is a composite measure of LA structure and function that combines information about atrial reservoir function as well as LA size, body habitus, and left ventricular (LV) function (as measured by stroke volume). LAFI might characterize LA remodeling better than currently used volumetric echocardiographic measures. In select populations with CVD, a lower LAFI is associated with an increased risk of incident heart failure, cerebrovascular events, atrial fibrillation recurrence, and all-cause mortality. In this retrospective investigation, we describe the distribution of LAFI and the cross-sectional clinical and echocardiographic correlates of LAFI in a community-based sample.
Materials and Methods
The design and sampling of the Framingham Offspring study were published previously. Briefly, starting in 1971, the children of the original Framingham Heart Study cohort were enrolled and evaluated approximately every 4-8 years. A total of 2,888 participants underwent transthoracic echocardiography with digital image acquisition during examination cycle 8 (2005-8). Participants who were in atrial fibrillation at the time of their echocardiographic examination, who had significant mitral regurgitation on echocardiogram, and who had inadequate atrial images were excluded. We also created a subgroup free of CVD/CVD risk factors ( n = 415) within the general study sample for analysis. This included nonobese participants, free of prevalent hypertension, diabetes, atrial fibrillation, coronary heart disease, cerebrovascular accident, transient ischemic attack, or heart failure. Participants who had one or more of these conditions were included in the subgroup with CVD/CVD risk factors. Laboratory parameters and echocardiographic measures were not considered for creation of CVD/CVD risk factors subgroups. Development of the general study sample and various subgroups is depicted in Supplemental Figure 1 (available at www.onlinejase.com ).
The study protocol was approved by the Boston University Medical Center Institutional Review Board, the University of California, San Francisco, School of Medicine Review Board, and the University of Massachusetts Medical School Review Board, and all participants provided written informed consent.
Left Atrial Volumetric Assessment
We performed offline analysis of echocardiographic images from 1,795 participants with LA imaging of sufficient quality to enable LA volumetric measurement from apical two and four-chamber views. Prior studies have demonstrated that the presence of atrial fibrillation rhythm at the time of echocardiogram affects LA contractile function and significantly decreases LAFI. We therefore excluded the participants who were in atrial fibrillation at the time of echocardiogram ( n = 40). Since atrial fibrillation is strongly associated with LA remodeling, the participants with history of atrial fibrillation who were in sinus rhythm during the echocardiogram were included in the current study to assess the ability of LAFI to capture LA remodeling associated with atrial fibrillation. Participants with moderate or higher degrees of mitral regurgitation on the echocardiogram ( n = 36, quantified using color Doppler by the maximum systolic proximal mitral regurgitation jet height) were also excluded, leading to the final sample size of 1,719. The baseline characteristics of included and excluded participants are presented in Supplemental Table 1 (available at www.onlinejase.com ). Excluded participants had significantly higher body mass index (29 kg/m 2 vs 28 kg/m 2 for included participants; P = .001), use of antihypertensive medications (57% vs 52% in included participants; P = .002), prevalence of atrial fibrillation (10% vs 5% in included participants; P < .001), and heart failure (4% vs 2% in the included participants; P = .003). Other clinical and demographic variables were similarly distributed between the included and excluded participants.
Two sonographers performed LA volume measurement. Serial quality control iterations were performed to maximize the inter- and intraobserver correlation. During each of these iterations, the maximal and minimal LA volumes were measured for 20 randomly selected participants by both sonographers. Sonographers were trained between serial iterations, and the interobserver coefficients of variation between the sonographers measured during the final quality control iteration for maximum (LAmax) and minimum LA volume (LAmin) measurement were 2.6% and 3.8%, respectively. Intraobserver coefficients of variation, derived similarly by remeasuring the LAmax and LAmin for 20 randomly selected participants, for LAmax and LAmin measurements were 3.4% and 4.4%, respectively. We could not calculate the inter- and intraobserver variability for LAFI because none of the observers made all the measurements required for calculation of LAFI in our study (LA volumes and LVOT-VTI). It is probable that the variability in LAFI calculation might be compounded due to variability in the individual components.
After completion of the final iteration of quality control, the sonographers reviewed the quality of echocardiographic images for all the Framingham Offspring examination 8 participants and converted the saved images into a digital format. Left atrial images were deemed “inadequate” if any of the following were present: (1) endocardial borders were not well visualized, (2) posterior wall of LA was not visualized (i.e., LA foreshortening was present), or (3) recorded cardiac cycles contained a premature beat. LA volumes deemed inadequate by one sonographer were then reviewed by the other sonographer or an investigator (D.D.M.) to confirm. Outlier values (mean ± 3 * SD) were investigated and r-measured.
The LAmax and LAmin were obtained using the recommended Simpson’s biplane summation of disks method on a Digisonics DigiView System Software (ver. 220.127.116.11, Digisonics, Houston, TX). The original images were optimized for the volumetric LV and valvular assessment. The LAmax and LAmin used in this analysis were determined by averaging LAmax and LAmin measurements from the apical two- and four-chamber views. The LA emptying fraction (LAEF) was calculated as ([LAmax − LAmin]/LAmax) × 100. The LAmax index was calculated by dividing LAmax by the body surface area. The LAFI was calculated using a previously validated formula ( Figure 1 ) :
LAFI = LAEF ∗ LVOT – VTI LAmax index .
Since LAFI is a derived measure dependent upon the measurement of LA volumes and LV outflow tract (LVOT) velocity-time integral (VTI), we could not calculate the interobserver variability for it because no two observers made all the measurements required for calculation of LAFI.
Echocardiographic Covariates and Definitions
M-mode and two-dimensional echocardiographic images were acquired using a standard protocol in parasternal and apical views as previously described. Leading-edge technology was used to make M-mode measurements, and all final measures were derived using the average of measurements over three or more cardiac cycles. M-mode measurements used for the current study included end-diastolic LV septal wall thickness (SWT), posterior wall thickness (PWT), LV end-diastolic diameter (LVEDD), and LV end-systolic diameter (LVESD). Left ventricular mass was calculated using a previously validated formula: 0.8 [1.04 (LVEDD + SWT + PWT) 3 − (LVEDD) 3 ] + 0.6 g. LV end-diastolic (LVEDV) and end-systolic volumes (LVESV) were calculated in apical two-chamber view using Simpson’s method. The LV ejection fraction (LVEF) was defined as ([LVEDV − LVESV]/LVEDV) × 100, and stroke volume was calculated as LVEDV – LVESV. The LVOT diameter was measured in parasternal long-axis view. LVOT pulsed wave Doppler tracings were not available for analyses, and the LVOT-VTI was derived by dividing the stroke volume by LVOT area (3.14 × (LVOT diameter/2) 2 ). Since we excluded the participants with significant mitral regurgitation, the value of LVOT-VTI derived using stroke volume should be similar to the pulsed wave Doppler derived value.
Ascertainment of Cardiovascular Risk Factors and CVD Definitions
At each Framingham Heart Study Offspring examination, participants undergo a detailed medical history and physical examination, including measurement of systolic and diastolic blood pressure by a study physician and weight and height by technicians. Venous sampling is performed after an overnight fasting. Participants also undergo laboratory evaluation for CVD risk factors, including fasting blood glucose, serum creatinine, and blood cholesterol concentrations. Standard enzymatic methods are used for measurement of serum total cholesterol, high-density lipoprotein, and serum creatinine.
Participants with systolic blood pressure ≥140 mmHg and/or diastolic blood pressure ≥90 mmHg and/or who were taking antihypertensive medications were categorized to have hypertension. Diabetes was defined as fasting plasma glucose ≥126 mg/dL and/or treatment with medications for diabetes. Criteria for defining atrial fibrillation, coronary heart disease, cerebrovascular accident, transient ischemic attack, and heart failure in the Framingham Offspring Study have been described elsewhere. Participants were considered to be current smokers if they reported smoking one or more cigarettes on a daily basis during the year preceding their heart study examination. Body mass index was calculated as the weight in kilograms divided by the square of height in meters. Obesity was defined as a body mass index ≥30 kg/m 2 . Estimated glomerular filtration rate was calculated by the Chronic Kidney Disease Epidemiology Collaboration equation using contemporaneously measured serum creatinine.
Baseline participant characteristics are presented as numbers and percentages for categorical variables and as mean ± SD for continuous variables. Baseline participant characteristics were compared between the subgroup free of CVD/CVD risk factors and the subgroup with CVD/CVD risk factors using t -test for continuous variables and χ 2 test for the categorical variables. We performed natural logarithmic transformation for variables observed to have a skewed distribution. Distribution histograms are presented to depict the variation of LAFI with age and by sex. Scatter plots for depicting the association of LAFI with LA volumes are presented. Separate histograms are presented for the subgroup free of CVD/CVD risk factors.
To assess associations of different variables with LAFI, stepwise multivariable linear regression models were used, forcing age and sex into each model. Two separate regression models were created: one model for LAFI in relation to clinical and demographic correlates and a second model for LAFI in relation to echocardiographic correlates. The clinical and demographic regression model consisted of candidate variables selected based on prior data demonstrating their associations with LA remodeling or LAFI. Eligible clinical and demographic variables included age, sex, current smoking status, systolic blood pressure, diastolic blood pressure, antihypertensive medication use, diabetes mellitus, prevalent atrial fibrillation, prevalent heart failure, prevalent cerebrovascular accident and/or transient ischemic attack, prevalent coronary heart disease, estimated glomerular filtration rate less than 60 mL/min/1.73 m 2 , and ratio of serum total cholesterol to high-density lipoprotein cholesterol. The echocardiographic regression model, in addition to age and sex, included LVEF, LVEDV, and LV mass. We also tested for effect modification of LAFI with age and sex for each model.
In secondary analysis, we studied the association of LAFI with clinical and demographic factors and with echocardiographic variables among the subgroup of participants who had normal LA size based on consensus criteria (LAmax index <34 mL/m 2 , n = 1,293; Supplemental Figure 1 , available at www.onlinejase.com ). Two separate regression models were used, as detailed above: the clinical and demographic model and the echocardiographic model.
A P value of <.05 in two-tailed tests was considered statistically significant. All statistical analyses were performed using SAS (v9.4, SAS Institute Inc., Cary, NC) and SPSS (IBM SPSS version 24, Chicago, IL) software.
The baseline demographic, clinical, and laboratory characteristics of the general study sample, subgroup free of CVD/CVD risk factors, and subgroup with CVD/CVD risk factors are presented in Table 1 . Overall, the study participants were middle-aged and older adults, and 54% were women. Two-thirds of the participants had hypertension. Five percent had a history of atrial fibrillation ( Table 1 ) but were not in atrial fibrillation at the time of the echocardiogram. The subgroup free of CVD/CVD risk factors had lower average age, and a significantly higher proportion were women compared to the subgroup with CVD/CVD risk factors. The echocardiographic characteristics of the study sample and various subgroups are presented in Table 2 . The mean LAFI was 35.2 ± 12.1 for the overall sample and was significantly higher among the subgroup free of CVD/CVD risk factors when compared with the subgroup with CVD/CVD risk factors (37.5 ± 11.6 for CVD/CVD risk factor-free group vs 34.5 ± 12.2 for subgroup with CVD/CVD risk factors; P < .001).
|Variable||All participants ( N = 1,719)||Subgroup free of CVD/CVD risk factors ( n = 415)||Subgroup with CVD/CVD risk factors ( n = 1,304)||P value|
|Age (years)||66 ± 9||63 ± 9||67 ± 9||<.001|
|Women||928 (54%)||271 (65%)||657 (50%)||<.001|
|Body mass index (kg/m 2 )||28 ± 5||24 ± 3||29 ± 5||<.001|
|Current smoker||167 (10%)||55 (13%)||112 (9%)||.005|
|Systolic blood pressure (mm Hg)||128 ± 17||118 ± 11||131 ± 17||<.001|
|Diastolic blood pressure (mm Hg)||73 ± 10||71 ± 8||74 ± 11||<.001|
|Antihypertensive medication use||880 (51%)||—||880 (67%)||—|
|Diabetes mellitus||257 (15%)||—||257 (20%)||—|
|Prevalent atrial fibrillation ∗||89 (5%)||—||89 (7%)||—|
|Prevalent heart failure||32 (2%)||—||32 (3%)||—|
|Prevalent CVA or TIA||64 (4%)||—||64 (5%)||—|
|Prevalent coronary artery disease||175 (10%)||—||175 (13%)||—|
|Estimated glomerular filtration rate less than 60 mL/min/1.73 m 2||275 (16%)||29 (7%)||275 (16%)||<.001|
|Total/HDL cholesterol ratio||3.47 ± 1.05||3.29 ± 0.98||3.52 ± 1.06||<.001|