Study population
This study was incorporated into the Rotterdam Study (RS). [21]The Rotterdam Study is a prospective population-based cohort study aimed at evaluating the development and progression of chronic disease risk factors in middle-aged and older adults. Between 1990 and 1993, all her residents over the age of 55 in Ommoort, a district of the city of Rotterdam in the Netherlands, were invited to the study. A total of 7983 (78% of all invitees) agreed to participate (RS-I). In 2000, the cohort was expanded with 3011 new participants who became 55 years of age or older or who transitioned into the study area (RS-II). Participants will attend follow-up examinations every 3-4 years.Outcome data on morbidity and mortality were continuously collected from general practitioners in the study area through links with digital files [21].
The Rotterdam study has been approved by the Erasmus MC Medical Ethics Committee (registration number MEC 02.1015) and the Dutch Ministry of Health, Welfare and Sport (Population Screening Act WBO, license number 1071272-159521-PG). The Rotterdam study was registered with the Dutch National Trials Register (NTR; www.trialregister.nl) and the WHO International Clinical Trials Registry Platform (ICTRP; www.who.int/ictrp/network/primary/en/). Catalog number NTR6831. All participants provided written informed consent to participate in the study and to obtain their information from their attending physician.
Body composition was assessed by DXA during the 4th RS-I visit (RS-I-4) and the 2nd RS-II visit (RS-II-2) between 2002 and 2006. it was done. [22]In total, 3724 participants were scanned by DXA. We excluded participants with prevalent AF at baseline (N. = 256). Therefore, 3468 individuals were included in the analysis of DXA-assessed adiposity and incident AF. From 2003 to 2006, all participants visiting the research center were asked to participate in a computed tomography (CT) study aimed at visualizing vascular calcification in multiple vascular beds. .In total, 2524 participants had their CT scanned [23]2230 participants who underwent CT scans of sufficient quality (N.-Exclusions = 294), 91 participants with prevalent atrial fibrillation at baseline were excluded. In total, 2139 participants were included in the CT-assessed adiposity analysis. Further analysis of body composition scores and fat distribution associated with AF risk included participants with both DXA and CT measurements available (n = 1297). (Figure 1)
Study population flow chart
Assessment of body composition
Body composition was assessed by DXA using ProdigyTMs Whole-body fan-beam densitometer (GE Lunar Corp, Madison, WI, USA), where scans were analyzed with enCORE software V13.6 (from GE Lunar), using pre-determined regions of interest according to the manufacturer’s protocol . Total lean mass is the sum of trunk lean mass and extremity lean mass (sum of lean tissue in the arms and legs), and total fat mass includes android fat mass (localized around the waist), gynoid Total fat mass (localized).chest, hips, and thighs), and unspecified fat mass [22, 24]As reported, the reproducibility of the DXA-based measurements was excellent, with Pearson correlation coefficients of 0.991 for total fat mass and 0.994 for total lean mass. [25]The android and gynoid fat mass ratios were also calculated. In addition to the net mass of each fat depot, the percentages of fat mass, android fat mass, and gynoid fat mass were also calculated by dividing each by total body weight.
Assessment of liver and epicardial fat
16 slices (n = 785) or 64 slices (n = 1739) ECG-gated CT scans of the unenhanced heart were performed using a multi-detector CT scanner (Somatom Sensation 16 or 64, Siemens, Forchheim, Germany). Detailed information about the imaging parameters of the scan is provided elsewhere. [23].
Liver fat was assessed using a previously described standardized procedure [23]Briefly, Philips iSite Enterprise software (Royal Philips Electronics NV 2006) was used to place three circular regions of interest on the liver. [26]These regions are delineated throughout the imaged liver tissue (including both left and right liver lobes), carefully selected to include only liver tissue, and devastating, such as large vessels, cysts, and focal lesions. organizations are not included. We then calculated the mean Hounsfield Units (HU) value from these three measurements as a marker of total liver fat. This is a reliable proxy for mean HU values across the liver. [26]Adopted a fully automatic system [27] Quantify the amount of epicardial fat in milliliters. In short, the method consisted of segmentation of the whole heart and calculation of epicardial fat mass. We used a multi-atlas-based approach for segmentation. In this approach, eight manually segmented contrast-enhanced cardiac scans (atlas) were registered (spatially aligned) with her CT scans for every participant. This segmentation was then used to determine the amount of epicardial fat.
Evaluation of atrial fibrillation
Methods for event determination of epidemic and incident AF have been previously described [13, 22, 28, 29]..Confirmation of AF at baseline and follow-up examinations in our study is based on clinical information from the medical records of all participants in the Rotterdam study. The Rotterdam study collected information on medical history and drug use through multiple sources, including baseline home interviews, physical examinations at study centers, pharmacy prescription records, and the National Medical Registry of all primary and secondary discharge diagnoses. Data is collected continuously. , and screening of GP records. In addition, a resting 10-s 12-lead electrocardiogram (ECG) used with the ACTA Gnosis IV ECG recorder (Esaote; Biomedical, Florence Italy) was obtained from all participants at all visits of the Rotterdam Study to validate AF. will be ECG records were stored in digital format and analyzed with the Modular ECG Analysis System (MEANS). ECG records were stored in digital format and analyzed with the Modular ECG Analysis System (MEANS). AF results will then be independently adjudicated by her two research physicians. In case of disagreement, consult a senior cardiologist. Participants were followed from the date of enrollment in the RS until the date of onset of AF, the date of death, the date of loss to follow-up, or January 1, 2014, whichever occurred first.
Evaluation of cardiovascular risk factors
Methods for assessing cardiovascular risk factors are detailed in Additional File 1: Methods. [22, 28, 30].
statistical analysis
Explanatory variables were presented as mean ± standard deviation (SD) or median (interquartile range – IQR) for continuous variables and numerical values (percentages) for categorical variables. as the HU value (a) had a left-skewed non-normal distribution, so we used exponentially transformed values (B.) [B = A3.5/10,000] [23]In addition, since the lower the HU value, the higher the amount of liver fat, the reciprocal (-HU) of the HU value is used to express the level of liver fat.
Values for each fat depot were standardized for direct comparison. Various measures of adiposity, including fat mass, android fat mass, gynoid fat mass, android to gynoid fat ratio, liver fat, and epicardial fat, were newly developed using Cox proportional hazards regression analysis. I checked if it is related to AF. Analysis by DXA measurements also assessed the respective percentages of total body fat mass, android fat mass, and gynoid fat mass in relation to the development of AF. For the first model, we calculated age- and sex-adjusted hazard ratios (HR) and their 95% confidence intervals (CI). Model 2 includes cardiovascular risk factors such as total and HDL cholesterol, history of hypertension, history of diabetes mellitus (DM), history of coronary heart disease (CHD), history of heart failure (HF), history of left ventricular hypertrophy Further adjusted for factors. Smoking status, total alcohol intake, use of lipid-lowering drugs, and use of cardiac medications. In addition, we additionally adjusted total lean mass for model 3 and total fat mass for model 4, respectively. Finally, we used spline analysis of the Cox model to explore potential nonlinear associations between different adiposity and incident AF.However, there was no sign of significant nonlinearity (all P. Results not shown for nonlinearities > 0.05). Sensitivity analyzes tested correction for effects by gender. Moreover, the analysis was repeated after stratifying the participants by BMI. BMI<25 and BMI≧25kg/m category2We also performed all analyzes among participants without common cardiovascular disease (CHD, HF, and stroke) at baseline.
To investigate the potential cumulative effects of adiposity, we additionally included five adipose depots (fat mass, android fat mass, gynoid fat mass, liver fat, and epicardial fat) to generate a body fat score. did. Each adipose depot was scored from 0 to 2 according to its respective tertile. Sum all component scores to get a total score ranging from 0 to 10, with higher scores indicating higher total body fat. Consequently, a Cox proportional hazards regression analysis using the same multivariate adjusted model as described above was performed to investigate the effect of body fat score (as tertiles: score 0–2, score 3–6, and score tertile as the first reference group in 7–10) in incident AF.
Next, we developed a fat distribution pattern. Distribution patterns were derived using principal component analysis (PCA) of fat depot values (including fat mass, android fat mass, gynoid fat mass, liver fat, and epicardial fat). [31]A varimax rotation was used to obtain the potential principal components. A cutoff of 0.5 was used to characterize potential patterns using factor loadings that reflect standardized correlations between each fat depot and body fat distribution pattern. For each participant, a pattern adherence score was constructed by separately summing the pattern fat accumulation observations weighted by the factor loadings corresponding to each of the two patterns. In addition, we used Cox proportional hazards regression analysis to assess how the identified fat distribution patterns (with the first quartile as the reference quartile) were associated with the occurrence of AF.
All missing values of covariates were imputed under the assumption of being missing at random using multiple imputation [32]For multiple imputation, we used all available data to generate five imputation datasets.Statistical significance was considered two-sided P.-value <0.05. Analyzes were performed using R software (R 4.0.3; R Foundation for Statistical Computing, Vienna, Austria).
