A schematic overview of the study design is shown in Fig. 1.
Study population
The Atherosclerosis Risk in Communities (ARIC) study is a population-based, prospective cohort study of cardiovascular risk factors in four US communities (Forsyth County, NC; Jackson, MS; suburbs of Minneapolis, MN; and Washington County, MD) [11]. The participants in the baseline period (1987 to 1989) included 15,792 of both genders aged 45 to 64 years. Study participants underwent follow-up visits in 1990-92, 1993-95, 1996-98, 2011-13, 2016-17, and 2018-19. Additionally, ARIC participants received annual follow-up calls since baseline, and survivors had a response rate of ≥ 90%. An extensive questionnaire was collected during each follow-up visit, followed by clinical examination and blood sample testing. Institutional review boards approved the ARIC study, and all participants provided informed consent.
Bạn đang xem: Serum electrolyte concentrations and risk of atrial fibrillation: an observational and mendelian randomization study
The present analysis included a total of 15,792 participants at baseline. Participants with prevalent AF or missing follow-up data for AF (n = 598) were excluded. Furthermore, those with incomplete serum electrolyte data (n = 342) were also excluded, resulting in a final sample size of 14,852.
Serum electrolytes measurement and covariates assessment
At the ARIC central laboratory, serum potassium was measured using an ion-selective electrode (Roche C501 Chemistry Analyzer), while serum magnesium was measured using colorimetric methods on the Roche Cobas 6000 Chemistry Analyzer (Roche Diagnostics; Indianapolis, Indiana). Serum calcium and phosphate were measured in frozen serum samples, using strategies based on o-cresolphthalein complexone and ammonium molybdate respectively. Calcium measurements were adjusted for albumin levels using the following equation: total corrected calcium = measured total calcium (mg/dL) + 0.8 [4.0 – serum albumin (g/dL)].
Xem thêm : Legacy Garlic Shrimp & Grits 26oz
The C-reactive protein was measured at visit 2, and the other covariates were assessed at visit 1. The participants reported the information on age, gender, race, smoking, alcohol intake, education level, history of cardiovascular disease, and use of medications. Body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared. Blood pressure was measured using a random-zero sphygmomanometer after five minutes of rest in the sitting position, and was defined as the average of the second and third measurements taken. The HF and coronary heart disease (CHD) definitions have been previously published [12, 13]. Diabetes mellitus was defined as fasting glucose ≥ 126 mg/dL, non-fasting glucose ≥ 200 mg/dL, treatment for diabetes mellitus, or self-reported physician diagnosis of diabetes. As executed in previously published ARIC studies, each exercise was converted into metabolic equivalent as per the Compendium of Physical Activities [14, 15]. High-density lipoprotein cholesterol (HDL-c) was measured using enzymatic measures, and low-density lipoprotein cholesterol (LDL-c) was calculated based on Friedewald formula [16]. The serum creatinine was measured using a modified kinetic Jaffe method. The left ventricular hypertrophy was defined as left ventricular mass index ≥ 51 g/m2.7 and the QT interval from the digital 12-lead electrocardiogram (ECG) was determined by the NOVACODE program [17, 18].
AF ascertainment
Ascertainment of AF has been described previously and conducted using three methods, i.e., ECG, hospital discharge codes, and death certificates [19, 20]. At each ARIC study visit, a 12-lead ECG was performed using a MAC PC cardiograph (Marquette Electronics Inc, Milwaukee, WI) and transmitted to the ARIC ECG Reading Center for coding, interpretation and storage. The ECG recordings were computer coded and checked by a trained cardiologist at a single reading center to confirm AF diagnosis. Incident AF was identified from hospitalizations or death certificates using ICD-9-CM (International Classification of Diseases, Ninth Revision, Clinical Modification) codes 427.31 (AF) or 427.32 (atrial flutter). An AF discharge code during a hospitalization with open cardiac surgery was excluded in the ARIC Study [19].
Mendelian randomization (MR)
Causal inference using MR relies on the instrumental variable assumptions, which require that the genetic variant must be strongly associated with the exposure, should be independent of any measured and unmeasured confounders, and must influence the outcome through the exposure only and not through any direct or alternative pathways (Fig. 1B). Genetic analysis was done using publicly available, summary-level genome-wide association study (GWAS) data. The study by Nielsen et al. was the largest GWAS for AF (OpenGWAS ID: ebi-a-GCST006414) from six contributing studies of European ancestry to date, including 60,620 cases of AF and 970,216 control subjects [21]. AF was defined based on ICD-9 code 427.3 and ICD-10 code I48. Four separate two-sample MR analysis were performed to test the potential causal associations between serum potassium (OpenGWAS ID: ukb-b-17,881), magnesium (OpenGWAS ID: ukb-b-7372), calcium (OpenGWAS ID: ukb-b-8951), and phosphate (OpenGWAS ID: ukb-d-30810_raw) with the AF risk, estimating the association results in two non-overlapping populations. GWAS data for electrolytes were derived from UK Biobank, a large, prospective cohort study that enrolled > 500,000 people across the United Kingdom from 2006 to 2010 and has received long-term follow-up [22].
We used single-nucleotide polymorphisms (SNPs) associated with four serum electrolytes from UK Biobank GWAS at genome-wide significance (P < 5 × 10−6 for potassium, magnesium and calcium; P < 5 × 10−8 for phosphate) as instruments and clumped at linkage disequilibrium R2 < 0.001 (Supplemental Tables 1 to 4). Complete information for data sources is detailed in Supplemental Table 5.
Statistical analysis
Since some variables had incomplete baseline data, multiple imputation was used to impute missing data by chained Eq. [23], since it reduces the selection bias possibility and is preferable to discard observations with missing values [24]. Continuous variables are presented as means ± SD and categorical variables as percentages. The baseline characteristics of the two groups were compared using unpaired t-tests for continuous variables and Chi-square tests for categorical variables. The serum electrolyte levels were categorized into seven groups with cutoffs at the 5th, 20th, 40th, 60th, 80th, and 95th percentiles of the selected four serum electrolyte levels, and used the middle category as the reference group (i.e., 40th to 60th percentile, corresponding to the middle quintile). P values for trend were calculated across the quintile categories using the quintile term. Cox proportional hazards regression modeled the relationship between electrolytes and incident AF events in the ARIC study. Different models were examined to investigate the effects of various confounders on the association between serum electrolytes and AF, i.e., model 1 was adjusted for age, gender, and race, and model 2 was additionally adjusted for the variables in model 1 plus hypertension, diabetes mellitus, smoke, ethanol intake, BMI, left ventricular hypertrophy at ECG, antiarrhythmic drugs, plasma creatinine, metabolic equivalent, LDL-c, C-reactive protein and angiotensin-converting enzyme inhibitors. The results were presented as hazard ratio (HR) and 95% confidence interval (CI). Kaplan-Meier estimates were constructed to show the cumulative AF incidence risk by serum electrolyte quartiles, and differences among quartiles were compared using the log-rank test. Additionally, restricted cubic splines were used to examine the presence of a dose-response association between selected four serum electrolytes and AF. Three knots were chosen for the analysis according to Akaike’s information criterion to provide a smooth and flexible description of the dose-response relationship.
For two-sample MR, the inverse variance-weighted (IVW) method was used to estimate causal effect [25]. IVW was considered the most reliable method if there was no directional pleiotropy (P for MR-Egger intercept > 0.05) [26]. Standard sensitivity analysis, i.e., MR-Egger, weighted median, and weighted mode, was carried out to detect whether there was a violation of key assumptions underlying MR [27, 28]. An estimated intercept term of the MR-Egger regression deviating from zero indicated directional pleiotropy.
Besides, the asymmetry of the funnel plot might indicate a violation of the main MR assumptions. Leave-one-out sensitivity analysis was performed to detect if a single SNP drove an association. F-statistics were calculated to quantify the strength of the genetic instrumental variables. Effect sizes are expressed as odds ratio (OR) alongside 95% confidence intervals (CI). Cochran’s Q test was applied to assess heterogeneity between genetic variants estimates. Variance for electrolyte levels was calculated using the formula of (mathrm R^2=left(mathrmbetatimessqrt{2timesmathrm{MAF}left(1-mathrm{MAF}right)}right)^2) assuming no genetic interactions, as published before [29]. F-statistics for all genetic instruments used in this study were greater than 10, indicating a low likelihood of weak instrumental variable bias.
All analysis were conducted in Stata version 16.0 (StataCorp LP, College Station, TX) and R version 4.2.1 (R Foundation for Statistical Computing). MR analysis was performed using the R-based package “TwoSampleMR” [30]. The P value for GWAS with genome-wide significance was set as less than 5 × 10−6 for potassium, magnesium, calcium and less than 5 × 10−8 for phosphate. A two-sided P value < 0.05 was considered statistically significant for all other analysis.
Nguồn: https://buycookiesonline.eu
Danh mục: Info
This post was last modified on December 8, 2024 3:47 pm