This population-based cohort study evaluates the presence of inflammation markers in individuals with atrial fibrillation over a 20-year follow-up.
Key Points
Question
Are markers of systemic and atrial inflammation associated with incident atrial fibrillation in the general population?
Findings
In this population-based cohort study, level of soluble vascular cell adhesion molecule 1 emerged as the only inflammation marker significantly associated with incident atrial fibrillation in a population-based cohort study. The key association was of a dose-response type with considerable strength and was successfully replicated in a second independent cohort.
Meanings
Atrial inflammation as reflected by soluble vascular cell adhesion molecule 1 may be more relevant in atrial fibrillation than systemic inflammation.
Abstract
Importance
Accumulating evidence links inflammation and atrial fibrillation (AF).
Objective
To assess whether markers of systemic and atrial inflammation are associated with incident AF in the general population.
Design, Setting, and Participants
The Bruneck Study is a prospective, population-based cohort study with a 20-year follow-up (n = 909). The population included a random sample of the general community aged 40 to 79 years. Levels of 13 inflammation markers were measured at baseline in 1990. Findings were replicated in a case-control sample nested within the prospective Salzburg Atherosclerosis Prevention Program in Subjects at High Individual Risk (SAPHIR) study (n = 1770). Data analysis was performed from February to May 2016.
Exposures
Levels of 13 inflammation markers.
Main Outcomes and Measures
Incident AF over a 20-year follow-up period in the Bruneck Study.
Results
Of the 909 participants included in the Bruneck Study, mean [SD] age was 58.8 (11.4) years and 448 (49.3%) were women. Among the 880 participants free of prevalent AF (n = 29) at baseline, 117 developed AF during the 20-year follow-up period (incidence rate, 8.2; 95% CI, 6.8-9.6 per 1000 person-years). The levels of soluble vascular cell adhesion molecule 1 (VCAM-1) and osteoprotegerin were significantly associated with incident AF (hazard ratio [HR], 1.49; 95% CI, 1.26-1.78; and 1.46; 95% CI, 1.25-1.69, respectively; P < .001 with Bonferroni correction for both), but osteoprotegerin lost significance after age and sex adjustment (HR, 1.05; 95% CI, 0.87-1.27; P > .99 with Bonferroni correction). Matrix metalloproteinase 9, metalloproteinase inhibitor 1, monocyte chemoattractant protein-1, P-selectin, fibrinogen, receptor activator of nuclear factor-κB ligand, high-sensitivity C-reactive protein, adiponectin, leptin, soluble intercellular adhesion molecule 1, and E-selectin all fell short of significance (after Bonferroni correction in unadjusted and age- and sex-adjusted analyses). The HR for a 1-SD higher soluble VCAM-1 level was 1.34 (95% CI, 1.11-1.62; Bonferroni-corrected P = .03) in a multivariable model. The association was of a dose-response type, at least as strong as that obtained for N-terminal pro-B-type natriuretic peptide (multivariable HR for a 1-SD higher N-terminal pro-B-type natriuretic peptide level, 1.15; 95% CI, 1.04-1.26), internally consistent in various subgroups, and successfully replicated in the SAPHIR Study (age- and sex-adjusted, and multivariable odds ratios for a 1-SD higher soluble VCAM-1 level, 1.91; 95% CI, 1.24-2.96, P = .003; and 2.59; 95% CI, 1.45-4.60; P = .001).
Conclusions and Relevance
Levels of soluble VCAM-1, but not other inflammation markers, are significantly associated with new-onset AF in the general community. Future studies should address whether soluble VCAM-1 is capable of improving AF risk classification beyond the information provided by standard risk scores.
Introduction
Atrial fibrillation (AF) is the most common cardiac arrhythmia, affecting approximately 8.8 million adults in the European Union. Atrial fibrillation is a major contributor to thromboembolic stroke and is associated with significant morbidity and mortality. Aside from well-established clinical risk factors, such as age, hypertension, and heart failure, inflammation has been suggested to contribute to the development of AF, but this hypothesis remains controversial. Evidence has been derived from histologic studies of atrial tissue in patients with AF demonstrating inflammatory infiltrates as well as from a number of epidemiologic studies reporting an association between circulatory inflammation markers, such as C-reactive protein (CRP), osteoprotegerin (OPG), and soluble intercellular adhesion molecule 1 (ICAM-1), with incident AF. A potential role of vascular cell adhesion molecule 1 (VCAM-1) in the development and maintenance of AF has been suggested. As a mediator of leukocyte trafficking, VCAM-1 is relevant to various cardiac diseases, including congestive heart failure as well as ischemic and rheumatic heart disease. An intriguing human atrial tissue culture model and an in vivo pig model demonstrated that local VCAM-1 expression is upregulated during rapid atrial pacing and potentially contributes to an inflammatory and prothrombotic environment within atrial tissue, which was termed “endocardial remodeling.” Whether inflammation related to AF is primarily a systemic or localized phenomenon is still debated.
In this study, we aimed to investigate the association of 13 baseline inflammation markers with incident AF in the long-term, prospective, population-based Bruneck Study, carefully characterize this association, and replicate key findings in another large cohort from the same geographic region.
Methods
Study Population
The Bruneck Study is a prospective, population-based survey on cardiovascular disease, which started in 1990. An age- and sex-stratified random sample of 1000 individuals was drawn from all residents aged 40 to 79 years (n = 4739) living in Bruneck, northern Italy. The response rate was 93.6%, with blood samples for assessment of soluble VCAM-1 (sVCAM-1) available in 909 individuals. Reevaluations were scheduled every 5 years (1995, 2000, 2005, and 2010). Follow-up was 100% complete for clinical end points. The study protocol was reviewed and approved by the University of Verona and Bolzano (Südtiroler Sanitätsbetrieb), and all participants provided written informed consent before entering the study. There was no financial compensation. All characteristics of the participants were assessed by means of standard procedures described previously and detailed in the eMethods in the Supplement.
Ascertainment of AF
Atrial fibrillation was assessed according to the Minnesota coding system (Minnesota code 8.3) by 2 cardiologists (G.E. and M.O.) from electrocardiograms (ECGs). Participants underwent a 12-lead ECG at baseline and during the 4 follow-up examinations and were asked whether they had a history of AF. In addition, Bruneck Hospital provided all ECGs and 24-hour Holter monitoring recordings performed on the participants between 1990 and 2010 (outside the study protocol), and we also had information on ECGs recorded at the general practitioners’ offices. Overall, a mean of 334 standard ECGs per 1000 person-years were available for review. In addition, 39 participants had 1 or more 24-hour Holter monitoring recordings. Moreover, we searched medical records and discharge letters from the Bruneck Hospital back to 1980 and obtained additional records from the general practitioners if the participant reported AF. Of 117 incident AF cases, 28 (23.9%) participants had AF newly detected at scheduled follow-up study visits and 89 (76.1%) cases were determined through clinical records. The type of AF was classified as paroxysmal (self-terminating) or persistent (lasting >7 days or requiring cardioversion). The date of first diagnosis of AF was documented. AF occurring transiently in the context of acute coronary syndrome or cardiac surgery was not included.
Ascertainment of Ischemic AF-Related Stroke
Ischemic stroke and transient ischemic attack were classified according to the criteria of the National Survey of Stroke. Ischemic AF-related stroke was defined as follows: (1) known (prediagnosed) AF or de novo AF ascertained as part of the diagnostic workup of the stroke event and (2) absence of another, more obvious, cause of stroke (eg, cranial vessel dissection or ipsilateral high-grade carotid stenosis with arterio-arterial stroke pattern). Ischemic stroke was determined by interview during study visits, the thorough clinical examinations performed as part of the study protocol, and careful review of databases from Bruneck Hospital. Self-report of ischemic stroke was confirmed by medical records.
Laboratory Methods
Venous blood samples were drawn after an overnight fast and 12 hours of abstinence from smoking. All participants had levels of sVCAM-1 measured in duplicate using a commercially available enzyme-linked immunosorbent assay kit (Bender MedSystems). Intra-assay and interassay coefficients of variation were 4.8% and 11.2%, respectively. Measurements were performed in samples stored at −70°C for 11 years without any cycle of thawing and refreezing. Other baseline inflammation markers were measured by standard procedures: matrix metalloproteinase 9, metalloproteinase inhibitor 1, monocyte chemoattractant protein-1, P-selectin, fibrinogen, OPG, receptor activator of nuclear factor-κB ligand, high-sensitivity CRP, adiponectin, leptin, ICAM-1, and E-selectin. In addition, 75 markers of inflammation, immune activation, or endothelial dysfunction markers were recently determined in samples collected during the 2000 follow-up by proximity extension assay (eMethods in the Supplement). N-terminal pro-B-type natriuretic peptide (NT-proBNP) was measured with a validated, commercially available immunoassay (Elecsys ProBNP; Roche Diagnostics) using established methodology.
Replication Cohort
The Salzburg Atherosclerosis Prevention Program in Subjects at High Individual Risk (SAPHIR) study is a prospective cohort study conducted in 1770 healthy, unrelated individuals (663 women and 1107 men aged 39-67 years) who were recruited by health-screening programs in large companies in and around the city of Salzburg, Austria. The cohort was first examined in 1999-2002, reexamined in 2003-2008, and followed up until September 2013. Methodologic details are described in the eMethods in the Supplement.
Statistical Analysis
Continuous variables are presented as means (SDs) and dichotomous variables are reported as percentages. Person-years of follow-up for each participant were accrued from the 1990 baseline until diagnosis of AF, death, or October 1, 2010, whichever came first. In an exploratory approach, the crude as well as age- and sex-adjusted association of 13 baseline inflammation markers with incident AF was examined using Cox proportional hazard models. Owing to skewed distributions, inflammation markers were loge transformed for analysis. The proportional hazard assumption was tested and confirmed using Schoenfeld residuals for all variables in all models. Potential nonlinear associations with incident AF were examined by fitting penalized splines with 3 df.
In this first exploratory step, only sVCAM-1 emerged as significantly associated with incident AF. This association was further elaborated in multivariable analysis adjusted for age, sex, body mass index, smoking, alcohol consumption, hypertension, NT-proBNP, and high-sensitivity cardiac troponin T (model 1), plus diabetes, glomerular filtration rate, high-sensitivity CRP, and atherosclerosis (model 2). Sensitivity analyses were additionally adjusted for heart failure, total thyroxine level, and socioeconomic status, or PR interval and P axis, or angiotensin-converting-enzyme inhibitor and diuretic therapy, or excluded the first 5 years of follow-up. We defined socioeconomic status on a 3-category scale (low, medium, or high) on the basis of information about occupational status and educational level of the person with the highest income in the household. High socioeconomic status was assumed if the participant had 12 or more years of education or an occupation with an average monthly income of $2000 or more (baseline salary before tax). Low socioeconomic status was defined by 8 years or less of education or an average monthly income of $1000 or lower.
Moreover, the association between sVCAM-1 and new-onset AF was assessed in various subgroups. To further substantiate the concept that local atrial inflammation reflected by sVCAM-1 is more relevant than systemic inflammation, 75 additional markers of inflammation, immune activation, or endothelial dysfunction were measured by proximity extension assay (samples collected in 2000 with a follow-up from 2000 to 2010) and examined in the context of incident AF. To compare the strength of association of sVCAM-1 with incident AF with that of a broadly accepted and replicated risk predictor of AF, multivariable hazard ratio (HR) associated with a 1-SD higher NT-proBNP level was calculated (adjusted for the covariates as model 2).
Finally, a case-control sample nested within the SAPHIR study was used for replication purposes (eMethods in the Supplement). All reported probability values were 2-sided. In Table 1 and Table 2 and eTable 1 in the Supplement, Bonferroni-corrected P values were calculated to account for multiple testing. Results were considered statistically significant at a P < .05 level. All calculations were performed using SPSS, version 20.0 (IBM Corp); R, version 3.1.0 (R Foundation for Statistical Computing); and Stata/MP, version 12.1 (StataCorp). Data analysis was conducted from February to May 2016.
Table 1. Associations of 13 Baseline Markers of Inflammation With Incident AF in the 880-Participant Bruneck Studya.
| Proteinb | Unadjusted | Adjusted for Age and Sex | ||||
|---|---|---|---|---|---|---|
| HR (95% CI) | P Value | P Valuec | HR (95% CI) | P Value | P Valuec | |
| VCAM-1 | 1.49 (1.26-1.78) | <.001 | <.001 | 1.35 (1.12-1.63) | .001 | .02 |
| OPG | 1.46 (1.25-1.69) | <.001 | <.001 | 1.05 (0.87-1.27) | .63 | >.99 |
| ADIPO | 1.33 (1.09-1.62) | .005 | .07 | 1.17 (0.94-1.45) | .16 | >.99 |
| FIB | 1.22 (1.00-1.50) | .05 | .61 | 0.91 (0.76-1.10) | .33 | >.99 |
| High-sensitivity CRP | 1.19 (0.99-1.43) | .06 | .78 | 0.94 (0.77-1.15) | .56 | >.99 |
| TIMP-1 | 1.11 (0.93-1.32) | .26 | >.99 | 1.01 (0.84-1.21) | .93 | >.99 |
| MCP-1 | 0.92 (0.78-1.10) | .37 | >.99 | 0.88 (0.74-1.06) | .18 | >.99 |
| LYAM3 | 0.94 (0.79-1.11) | .45 | >.99 | 1.10 (0.92-1.31) | .30 | >.99 |
| ICAM-1 | 1.05 (0.88-1.26) | .58 | >.99 | 1.05 (0.86-1.29) | .61 | >.99 |
| MMP-9 | 1.06 (0.87-1.28) | .58 | >.99 | 1.14 (0.95-1.38) | .17 | >.99 |
| LYAM2 | 0.95 (0.80-1.14) | .59 | >.99 | 1.01 (0.84-1.22) | .91 | >.99 |
| LEP | 0.98 (0.81-1.18) | .82 | >.99 | 0.93 (0.72-1.19) | .57 | >.99 |
| RANKL | 0.99 (0.83-1.20) | .96 | >.99 | 1.08 (0.90-1.29) | .42 | >.99 |
Abbreviations: ADIPO, adiponectin; AF, atrial fibrillation; CRP, C-reactive protein; FIB, fibrinogen; HR, hazard ratio; ICAM-1, intercellular adhesion molecule 1; LEP, leptin; LYAM2, E-selectin; LYAM3, P-selectin; MCP-1, monocyte chemoattractant protein-1; MMP-9, matrix metallopeptidase 9; OPG, osteoprotegerin; RANKL, receptor activator of nuclear factor-κB ligand; TIMP-1, metalloproteinase inhibitor 1; VCAM-1, vascular cell adhesion molecule 1.
Markers were assessed in 1990 with 20 years of follow-up (1990-2010). Numbers of events and total participants were 117 and 880, respectively. Markers were loge-transformed for analysis and HRs are given for a 1-SD higher concentration in each marker.
Protein order is arranged by descending significance in unadjusted models.
Bonferroni-corrected P value.
Table 2. Association of sVCAM-1 With Incident AF (1990-2010) in the 880-Participant Bruneck Study.
| Model | HR for a 1-SD Higher sVCAM-1 Level | ||
|---|---|---|---|
| HR (95% CI) | P Value | P Valuea | |
| Unadjusted | 1.50 (1.26-1.80) | <.001 | <.001 |
| Adjusted for age and sex | 1.36 (1.13-1.64) | .001 | .02 |
| Multivariable model 1b | 1.34 (1.10-1.62) | .003 | .04 |
| Multivariable model 2c | 1.34 (1.11-1.62) | .002 | .03 |
| Multivariable model 2, first 5 y of follow-up excluded | 1.37 (1.13-1.67) | .002 | .02 |
| Multivariable model 3d | 1.31 (1.08-1.59) | .007 | .10 |
| Multivariable model 3e | 1.35 (1.12-1.64) | .002 | .02 |
| Multivariable model 3f | 1.35 (1.11-1.63) | .002 | .03 |
Abbreviations: AF, atrial fibrillation; HR, hazard ratio; sVCAM-1, soluble vascular cell adhesion molecule 1.
Bonferroni-corrected P value.
Adjusted for age (years), sex, smoking (0 vs 1), alcohol consumption (grams per day), hypertension (0 vs 1), loge transformed N-terminal pro-B-type natriuretic peptide (picograms per milliliter), and high-sensitivity cardiac troponin T level (undetectable vs tertiles of detectable).
Additionally adjusted for diabetes (0 vs 1), glomerular filtration rate (milliliters per minute), loge transformed high-sensitivity C-reactive protein (milligrams per liter), and atherosclerosis (0 vs 1).
Multivariable model 2, plus heart failure (New York Heart Association classes 1-4 vs 0), total thyroxine level (micrograms per deciliter), and socioeconomic status (high, medium, and low).
Multivariable model 2, plus PR interval (milliseconds), and P axis (≤30° vs >30°).
Multivariable model 2, plus angiotensin-converting-enzyme inhibitor therapy (0 vs 1) and diuretic therapy (0 vs 1).
Results
Of the 909 participants in the Bruneck Study (mean [SD] age, 58.8 [11.4] years; 448 [49.3%] women), 29 (3.2%) had prevalent AF. After exclusion of prevalent AF cases, 880 individuals formed the study population for the longitudinal analysis. In the 20-year follow-up period, 117 (13.3%) participants developed AF (incidence rate, 8.2; 95% CI, 6.8-9.6 per 1000 person-years); 44 (5.0%) had paroxysmal AF during the entire follow-up, and 73 (8.3%) progressed to persistent AF.
Table 1 reports the association between baseline inflammation markers and incident AF. In univariable analyses, sVCAM-1 and OPG were significantly associated with incident AF when corrected for multiple testing. After adjustment for age and sex, this association remained significant for sVCAM-1 only. Median (interquartile range [IQR]) baseline sVCAM-1 level was 615.6 ng/mL (501.0-805.1 mg/mL). Table 3 depicts baseline characteristics by low, middle, and high sVCAM-1 level groups (tertile groups). Significant and mostly small quantitative differences were obtained for age, glucose, total thyroxine, high-sensitivity CRP, NT-proBNP, high-sensitivity cardiac troponin T, and high-density and low-density lipoprotein cholesterol levels, PR interval and P axis in the baseline ECG; as well as for the frequency of diabetes across sVCAM-1 categories. The incidence of AF was 4.4 (range, 2.7-6.3) per 1000 person-years in the low compared with 9.5 (range, 7.0-12.4) in the middle sVCAM-1 group and 11.4 (range, 8.5-14.5) in the high sVCAM-1 group. In Cox proportional hazards models adjusted for age and sex, sVCAM-1 was significantly associated with incident AF (HR for a 1-SD higher sVCAM-1, 1.36; 95% CI, 1.13-1.64; Bonferroni-corrected P = .02). Results remained similar with additional adjustment for other risk predictors of AF, angiotensin-converting enzyme inhibitor and diuretic therapy, or ECG variables, and after exclusion of the first 5 years of follow-up (Table 2). The use of penalized splines indicated an adequate fit of a linear dose-response association (eFigure 1 in the Supplement). The association between sVCAM-1 and new-onset AF was robust in various subgroups, including individuals with and without heart failure, hypertension, and atherosclerosis (Figure). The strength of association was at least as high as that obtained for NT-proBNP (multivariable HR for a 1-SD higher NT-proBNP level, 1.15; 95% CI, 1.04-1.26). As anticipated, all other baseline markers of inflammation were not associated with AF, and this extended to 75 markers of inflammation, immune activation, or endothelial dysfunction measured from samples of the 2000 evaluation of the Bruneck Study (eTable 1 in the Supplement).
Table 3. Baseline Characteristics According to sVCAM-1 Level Groups in the 909-Participant Bruneck Study.
| Characteristic | sVCAM-1 Levela | P Valueb | ||
|---|---|---|---|---|
| Low (n = 303) |
Middle (n = 303) |
High (n = 303) |
||
| Demographic variable | ||||
| Age, mean (SD), y | 54.9 (10.8) | 60.2 (11.2) | 61.3 (11.3) | <.001 |
| Women, No. (%) | 156 (51.5) | 138 (45.5) | 154 (50.8) | .27 |
| Socioeconomic status, No. (%)c | ||||
| Low | 162 (53.5) | 187 (61.7) | 216 (71.3) | .07 |
| Medium | 76 (25.1) | 66 (21.8) | 45 (14.9) | |
| High | 65 (21.5) | 50 (16.5) | 42 (13.9) | |
| Anthropometric variables, mean (SD) | ||||
| Weight, kg | 68.7 (12.1) | 68.2 (11.4) | 68.6 (13.1) | .46 |
| Height, m | 1.7 (0.1) | 1.7 (0.1) | 1.7 (0.1) | .65 |
| BMI | 24.9 (3.5) | 24.8 (3.6) | 25.2 (4.3) | .46 |
| Carotid ultrasonography, No. (%) | ||||
| Atherosclerosis | 102 (33.7) | 139 (45.9) | 141 (46.5) | .61 |
| ECG variablesd | ||||
| PR interval, mean (SD), milliseconds | 152.9 (18.1) | 157.4 (22.3) | 160.5 (24.2) | .03 |
| cQT interval, mean (SD), milliseconds | 407.7 (23.3) | 408.0 (23.9) | 411.3 (25.1) | .54 |
| P axis ≤30°, No. (%) | 46 (15.3) | 59 (19.9) | 73 (24.7) | .02 |
| P-wave terminal force, mean (SD), µV × ms | 2650.9 (2037.2) | 2539.2 (1786.3) | 2611.0 (1928.1) | .41 |
| Lifestyle and vascular risk variables | ||||
| Smoking, No. (%) | 90 (29.7) | 67 (22.1) | 65 (21.5) | .16 |
| Alcohol consumption, mean (SD), g/d | 31.6 (43.7) | 32.1 (40.6) | 29.4 (38.4) | .40 |
| Diabetes, No. (%) | 9 (3.0) | 20 (6.6) | 35 (11.6) | .01 |
| Glucose, mean (SD), mg/dL | 97.1 (13.9) | 101.9 (21.7) | 103.9 (23.9) | .01 |
| Hypertension, No. (%) | 174 (57.4) | 201 (66.3) | 193 (63.7) | .39 |
| Systolic BP, mean (SD), mm Hg | 143.4 (20.5) | 146.2 (22.8) | 146.9 (21.0) | .57 |
| Diastolic BP, mean (SD), mm Hg | 88.5 (9.6) | 89.3 (10.4) | 88.8 (10.1) | .50 |
| HDL-C, mean (SD), mg/dL | 57.9 (13.8) | 57.0 (15.2) | 54.5 (13.3) | .003 |
| LDL-C, mean (SD), mg/dL | 141.1 (34.5) | 138.6 (41.6) | 135.2 (38.4) | .02 |
| High-sensitivity CRP, mean (SD), mg/L | 1.4 (0.8-2.5)c | 1.4 (1.0-2.4)c | 1.7 (1.0-3.8)c | .02 |
| Total thyroxine level, mean (SD), µg/dL | 6.8 (1.6) | 6.7 (1.3) | 7.3 (1.9) | <.001 |
| NT-proBNP, median (IQR), pg/mL | 60 (36-104) | 88 (52-160) | 104 (60-196) | <.001 |
| Detectable hs-cTnT, No. (%) | 193 (63.7) | 209 (69.0) | 231 (76.2) | .03 |
| eGFR, mean (SD), mL/min | 86.1 (13.7) | 81.5 (13.7) | 79.8 (15.3) | .15 |
| Cardiovascular disease, No. (%) | ||||
| History of MI | 2 (0.7) | 9 (3.0) | 6 (2.0) | .27 |
| Heart failure (NYHA class ≥1) | 41 (13.5) | 75 (24.8) | 95 (31.4) | .23 |
Abbreviations: BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); BP, blood pressure; cQT, corrected QT interval; CRP, C-reactive protein; ECG, electrocardiogram; eGFR, estimated glomerular filtration rate; HDL-C, high-density lipoprotein cholesterol; hs-cTnT, high-sensitivity cardiac troponin T; IQR, interquartile range; LDL-C, low-density lipoprotein cholesterol; MI, myocardial infarction; NT-proBNP, N-terminal pro-B-type natriuretic peptide; NYHA, New York Heart Association; sVCAM-1, soluble vascular cell adhesion molecule 1.
SI conversion factors: To convert glucose to millimoles per liter, multiply by 0.0555; HDL-C and LDL-C to millimoles per liter, multiply by 0.0259; CRP to nanomoles per liter, multiply by 9.524; and thyroxine to nanomoles per liter, multiply by 12.871.
Data are presented according to sVCAM-1 level groups at baseline. A low level was 543 ng/mL or less; middle, greater than 543 and less than 722 ng/mL; and high, 722 ng/mL or greater.
P value adjusted for age and sex.
We defined socioeconomic status on a 3-category scale (low, medium, or high) on the basis of information about occupational status and educational level of the person with the highest income in the household. High socioeconomic status was assumed if the participant had 12 or more years of education or an occupation with an average monthly income $2000 or more (baseline salary before tax). Low socioeconomic status was defined by 8 years or less of education or an average monthly income of $1000 or lower.
Electrocardiogram factors were measured in participants without AF or a pacemaker (n = 868).
Figure. Association Between Soluble Vascular Cell Adhesion Molecule 1 and Incident Atrial Fibrillation in Various Subgroups in the Bruneck Study.
Hazard ratios (HRs) were derived from Cox proportional hazards models adjusted for age and sex and calculated for a 1-SD higher level of loge-transformed baseline soluble vascular cell adhesion molecule 1. MI indicates myocardial infarction.
In the SAPHIR study, 5 of 1770 participants (0.3%) had a diagnosis of AF at baseline. In a nested case-control analysis considering 60 individuals with new-onset AF (14 years’ median follow-up) and 114 controls, the sVCAM-1 level was significantly associated with future AF (conditional logistic regression: age- and sex-adjusted and multivariable odds ratio (OR) for a 1-SD higher sVCAM-1 level, 1.91; 95% CI, 1.24-2.96; and 2.59; 95% CI, 1.45-4.60; P = .003 and P = .001, respectively). Baseline characteristics of participants in the SAPHIR study are summarized in eTable 2 in the Supplement.
Our study yielded preliminary evidence for an association between the level of sVCAM-1 and ischemic AF-associated stroke in the Bruneck Study (n = 24; multivariable HR for a 1-SD higher sVCAM-1 level, 1.54; 95% CI, 1.01-2.35; P = .048), but not ischemic stroke of other etiology (n = 77; multivariable HR for a 1-SD higher sVCAM-1 level, 0.85; 95% CI, 0.66-1.09; P = .20) (Table 4). Participants with incident ischemic AF-related stroke were significantly older and had higher median NT-proBNP levels than did those with non-AF ischemic-related stroke (mean [SD] age 70.3 [1.3] vs 62.6 [1.3] years, P = .003; median [IQR] NT-proBNP 286 [122-767] vs 98 [54-185] pg/dL, P < .001). No other significant differences in clinical characteristics were found between these 2 groups. The incidence rate was 1.3 (95% CI, 0.9-2.0) per 1000 person-years for AF-related ischemic stroke and 4.4 (95% CI, 3.5-5.5) per 1000 person-years for non–AF-related ischemic stroke.
Table 4. Association Between sVCAM-1 and Ischemic Stroke in the Bruneck Studya.
| Characteristic | Events, No. | Incidence Rate per 1000 Person-years (95% CI) | HR (95% CI) | P Value |
|---|---|---|---|---|
| All ischemic strokes | 101 | 5.8 (4.7-7.0) | 0.97 (0.78-1.20) | .75 |
| AF-related ischemic stroke | 24 | 1.3 (0.9-2.0) | 1.54 (1.01-2.35) | .048 |
| Non–AF-related ischemic stroke | 77 | 4.4 (3.5-5.5) | 0.85 (0.66-1.09) | .20 |
Abbreviations: AF, atrial fibrillation; HR, hazard ratio; sVCAM-1, soluble vascular cell adhesion molecule 1.
The sVCAM-1 level was treated as a continuous variable. Cox proportional hazard models were used to calculate hazard ratios associated with a 1-SD higher sVCAM-1 level. Models were adjusted for age (years), sex, smoking (0 vs 1), alcohol consumption (grams per day), hypertension (0 vs 1), loge N-terminal pro-B-type natriuretic peptide (picograms per milliliter), high-sensitivity cardiac troponin T level (undetectable vs tertiles of detectable), diabetes (0 vs 1), glomerular filtration rate (milliliters per minute), high-sensitivity C-reactive protein (milligrams per liter), and atherosclerosis (0 vs 1).
Discussion
In an exploratory approach to identify inflammation markers relevant for the occurrence of AF in the general community, sVCAM-1 was the only one significantly associated with long-term risk of AF. The association was of a dose-response type and considerable strength, independent of a large number of clinical and laboratory characteristics, and persisted in various subgroups, including individuals with and without heart failure, hypertension, or atherosclerosis (Figure). This key finding was successfully replicated in the SAPHIR study.
Being a member of the immunoglobulin superfamily, VCAM-1 is a cell surface protein and expressed by several cell types, especially endothelial cells, but also macrophages, smooth muscle, and dendritic cells. Soluble forms of VCAM-1 arise from excess surface expression or increased shedding. In addition, VCAM-1 is a key player in the adhesion and transmigration of leukocytes from circulation into surrounding tissue and thus contributes to tissue inflammation (eFigure 2 in the Supplement). Although only low levels are expressed on resting endothelium and endocardium, marked upregulation of VCAM-1 is elicited by various stress conditions. In a number of human heart diseases, reactive oxygen species and hemodynamic factors have been shown to enhance cardiac VCAM-1 expression, and subsequent cell infiltration and inflammation in heart tissue gives rise to sustained cardiac remodeling, fibrosis, and dysfunction.
Accumulating evidence suggests that atrial remodeling is involved in the pathophysiology of AF, and it is tempting to speculate that VCAM-1 in circulation partly reflects local endocardial activation. A 2-fold increase in endocardial VCAM-1 expression was found in patients with paroxysmal and persistent AF. Furthermore, prominent atrial upregulation of VCAM-1 was observed during rapid atrial pacing in in vitro and in vivo experiments. Studies showed that angiotensin II receptor blockers reduce the occurrence of AF and downregulation of adhesion molecules within the atrium was proposed as a potential underlying mechanism. In epidemiologic studies, levels of sVCAM-1 were associated with the presence of AF and highest in patients with left atrial appendage thrombi. To our knowledge, the only prospective study so far to assess the association of VCAM-1 with new-onset AF was conducted in patients with coronary artery disease undergoing cardiac bypass surgery; the findings demonstrated that high sVCAM-1 level predicts postoperative AF manifestation.
In an exploratory analysis within the Framingham study, OPG (but not 11 other markers of inflammation) was significantly associated with incident AF (multivariable HR, 1.30; 95% CI, 1.08-1.65; P = .006). Our study also yielded a significant association between OPG and AF in univariable analysis; however, the significance was lost after adjustment for age given a particularly strong correlation between age and OPG (r = 0.52). Whereas the Women's Health Study reported a weak but significant association between levels of ICAM-1 and incident AF (multivariable HR, 1.11; 95% CI, 1.02-1.20; P = .02), the Framingham Study and our study found no such association.
Several studies observed an association between CRP and incident AF, but results from the Framingham Study were inconsistent and our study and other recent studies yielded no association with adjustment for standard AF risk factors. Negative findings in the Bruneck Study extend to a large number of markers of systemic inflammation and immune activation (eTable 1 in the Supplement) measured by a custom-made proteomics chip that accords with recent findings in 2 Swedish studies using the same chip. Overall, current epidemiologic evidence supporting a link between systemic inflammation and AF is weak. The present study is supportive of the emerging concept that local endocardial activation and remodeling is more relevant.
Endocardial VCAM-1 expression has been reported to promote thrombogenicity and left atrial appendage clot formation. Epidemiologic evidence on this finding, however, is inconsistent, with 1 study demonstrating an association between sVCAM-1 levels and cardiovascular events in patients with AF and another showing no such association. Both studies relied on hospital-based series of patients with AF. Our study adds new data to this controversy and demonstrated a significant association between sVCAM-1 and AF-associated ischemic stroke but not for ischemic stroke of other cause in the general population (Table 4).
Strengths and Limitations
Strengths of our study are the well-characterized, community-based cohort with a more than 90% participation rate, long-term follow-up, and rigorous assessment of a large number of clinical and laboratory characteristics. In addition, we were able to replicate key findings in an independent population-derived cohort using the same sVCAM-1 assay. Some limitations have to be mentioned as well. Detection of paroxysmal AF remains a diagnostic challenge. In the near future, systematic long-term recording of AF will become a reality in population-based studies given the recent development of easy-to-use, noninvasive cardiac monitors (eg, external loop recorders). Such recording, however, was not feasible in the time period of the present study (1990-2010). Notwithstanding, the Bruneck Study ranks among the prospective population studies with the most thorough ascertainment of AF relying on a mean of 334 ECGs per 1000 person-years. We measured sVCAM-1 and other inflammation markers only once at baseline in the Bruneck Study. Serial measurements would have better accounted for within-person fluctuations in inflammatory marker levels over time. In the SAPHIR study, consideration of sVCAM-1 levels measured both at baseline and during follow-up further enhanced predictive accuracy.
Conclusions
We found that sVCAM-1 levels, but not markers of systemic inflammation, are significantly and independently associated with incident AF. Although causality cannot be firmly established from observational studies, a causal interpretation between VCAM-1 and AF is favored by several characteristics of the association (Hill’s criteria of causation) including its considerable magnitude and dose-response type, temporality, consistency in various subgroups, and replication in a second independent cohort. Furthermore, the association is biologically plausible and coherent with previous animal and clinical investigations. Our data add to the preliminary evidence that VCAM-1 is involved in atrial remodeling and may well assist in improving AF risk prediction and in selecting the right individuals for intensified AF screening (eg, implantable loop recorder) among patients with ischemic stroke of undetermined origin. Further studies are required to elaborate findings toward these clinical applications.
eMethods. Detailed Methodology
eTable 1. Associations Between 75 Markers of Inflammation, Immune Activation, or Endothelial Dysfunction and Incident Atrial Fibrillation in the Bruneck Study
eTable 2. Baseline Characteristics in the SAPHIR Study
eFigure 1. Penalized Spline Fit of the Association Between Soluble Vascular Cell Adhesion Molecule 1 and Incident Atrial Fibrillation in the Bruneck Study
eFigure 2. Mechanisms and Epidemiological Evidence Linking Vascular Cell Adhesion Molecule 1 to Atrial Fibrillation
References
- 1.Krijthe BP, Kunst A, Benjamin EJ, et al. . Projections on the number of individuals with atrial fibrillation in the European Union, from 2000 to 2060. Eur Heart J. 2013;34(35):2746-2751. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Chen LY, Sotoodehnia N, Bůžková P, et al. . Atrial fibrillation and the risk of sudden cardiac death: the Atherosclerosis Risk in Communities Study and Cardiovascular Health Study. JAMA Intern Med. 2013;173(1):29-35. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Wolf PA, Abbott RD, Kannel WB. Atrial fibrillation as an independent risk factor for stroke: the Framingham Study. Stroke. 1991;22(8):983-988. [DOI] [PubMed] [Google Scholar]
- 4.Guo Y, Lip GY, Apostolakis S. Inflammation in atrial fibrillation. J Am Coll Cardiol. 2012;60(22):2263-2270. [DOI] [PubMed] [Google Scholar]
- 5.Aviles RJ, Martin DO, Apperson-Hansen C, et al. . Inflammation as a risk factor for atrial fibrillation. Circulation. 2003;108(24):3006-3010. [DOI] [PubMed] [Google Scholar]
- 6.Conen D, Ridker PM, Everett BM, et al. . A multimarker approach to assess the influence of inflammation on the incidence of atrial fibrillation in women. Eur Heart J. 2010;31(14):1730-1736. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Schnabel RB, Larson MG, Yamamoto JF, et al. . Relation of multiple inflammatory biomarkers to incident atrial fibrillation. Am J Cardiol. 2009;104(1):92-96. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Yamashita T, Sekiguchi A, Iwasaki YK, et al. . Recruitment of immune cells across atrial endocardium in human atrial fibrillation. Circ J. 2010;74(2):262-270. [DOI] [PubMed] [Google Scholar]
- 9.Blankenberg S, Barbaux S, Tiret L. Adhesion molecules and atherosclerosis. Atherosclerosis. 2003;170(2):191-203. [DOI] [PubMed] [Google Scholar]
- 10.Chen MC, Chang HW, Juang SS, et al. . Percutaneous transluminal mitral valvuloplasty reduces circulating vascular cell adhesion molecule-1 in rheumatic mitral stenosis. Chest. 2004;125(4):1213-1217. [DOI] [PubMed] [Google Scholar]
- 11.Malik I, Danesh J, Whincup P, et al. . Soluble adhesion molecules and prediction of coronary heart disease: a prospective study and meta-analysis. Lancet. 2001;358(9286):971-976. [DOI] [PubMed] [Google Scholar]
- 12.Tsutamoto T, Wada A, Maeda K, et al. . Angiotensin II type 1 receptor antagonist decreases plasma levels of tumor necrosis factor alpha, interleukin-6 and soluble adhesion molecules in patients with chronic heart failure. J Am Coll Cardiol. 2000;35(3):714-721. [DOI] [PubMed] [Google Scholar]
- 13.Goette A, Bukowska A, Lendeckel U, et al. . Angiotensin II receptor blockade reduces tachycardia-induced atrial adhesion molecule expression. Circulation. 2008;117(6):732-742. [DOI] [PubMed] [Google Scholar]
- 14.Hammwöhner M, Ittenson A, Dierkes J, et al. . Platelet expression of CD40/CD40 ligand and its relation to inflammatory markers and adhesion molecules in patients with atrial fibrillation. Exp Biol Med (Maywood). 2007;232(4):581-589. [PubMed] [Google Scholar]
- 15.Kiechl S, Lorenz E, Reindl M, et al. . Toll-like receptor 4 polymorphisms and atherogenesis. N Engl J Med. 2002;347(3):185-192. [DOI] [PubMed] [Google Scholar]
- 16.Schett G, Kiechl S, Bonora E, et al. . Vascular cell adhesion molecule 1 as a predictor of severe osteoarthritis of the hip and knee joints. Arthritis Rheum. 2009;60(8):2381-2389. [DOI] [PubMed] [Google Scholar]
- 17.Willeit K, Pechlaner R, Egger G, et al. . Carotid atherosclerosis and incident atrial fibrillation. Arterioscler Thromb Vasc Biol. 2013;33(11):2660-2665. [DOI] [PubMed] [Google Scholar]
- 18.Prineas RJ, Crow RS, Blackburn H. The Minnesota Code Manual of Electrocardiographic Findings. Boston, MA: John Wright PSB, 1982:203. [Google Scholar]
- 19.Walker AE, Robins M, Weinfeld FD. The National Survey of Stroke: clinical findings. Stroke. 1981;12(2, pt 2)(suppl 1):I13-I44. [PubMed] [Google Scholar]
- 20.Assarsson E, Lundberg M, Holmquist G, et al. . Homogenous 96-plex PEA immunoassay exhibiting high sensitivity, specificity, and excellent scalability. PLoS One. 2014;9(4):e95192. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Januzzi JL, van Kimmenade R, Lainchbury J, et al. . NT-proBNP testing for diagnosis and short-term prognosis in acute destabilized heart failure: an international pooled analysis of 1256 patients: the International Collaborative of NT-proBNP Study. Eur Heart J. 2006;27(3):330-337. [DOI] [PubMed] [Google Scholar]
- 22.Melmer A, Lamina C, Tschoner A, et al. . Body adiposity index and other indexes of body composition in the SAPHIR study: association with cardiovascular risk factors. Obesity (Silver Spring). 2013;21(4):775-781. [DOI] [PubMed] [Google Scholar]
- 23.Eilers PH, Marx BD. Flexible smoothing with B-splines and penalties. Stat Sci. 1996;11(2):89-121. [Google Scholar]
- 24.Hijazi Z, Lindbäck J, Alexander JH, et al. ; ARISTOTLE and STABILITY Investigators . The ABC (age, biomarkers, clinical history) stroke risk score: a biomarker-based risk score for predicting stroke in atrial fibrillation. Eur Heart J. 2016;37(20):1582-1590. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Hope SA, Meredith IT. Cellular adhesion molecules and cardiovascular disease; part I: their expression and role in atherogenesis. Intern Med J. 2003;33(8):380-386. [DOI] [PubMed] [Google Scholar]
- 26.Yamazaki KG, Ihm SH, Thomas RL, Roth D, Villarreal F. Cell adhesion molecule mediation of myocardial inflammatory responses associated with ventricular pacing. Am J Physiol Heart Circ Physiol. 2012;302(7):H1387-H1393. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Healey JS, Baranchuk A, Crystal E, et al. . Prevention of atrial fibrillation with angiotensin-converting enzyme inhibitors and angiotensin receptor blockers: a meta-analysis. J Am Coll Cardiol. 2005;45(11):1832-1839. [DOI] [PubMed] [Google Scholar]
- 28.Zhang L, Ding R, Zhen Y, Wu ZG. Relation of urotensin II levels to lone atrial fibrillation. Am J Cardiol. 2009;104(12):1704-1707. [DOI] [PubMed] [Google Scholar]
- 29.Verdejo H, Roldan J, Garcia L, et al. . Systemic vascular cell adhesion molecule-1 predicts the occurrence of post-operative atrial fibrillation. Int J Cardiol. 2011;150(3):270-276. [DOI] [PubMed] [Google Scholar]
- 30.Marott SC, Nordestgaard BG, Zacho J, et al. . Does elevated C-reactive protein increase atrial fibrillation risk? a Mendelian randomization of 47,000 individuals from the general population. J Am Coll Cardiol. 2010;56(10):789-795. [DOI] [PubMed] [Google Scholar]
- 31.Schnabel RB, Larson MG, Yamamoto JF, et al. . Relations of biomarkers of distinct pathophysiological pathways and atrial fibrillation incidence in the community. Circulation. 2010;121(2):200-207. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Lind L, Sundström J, Stenemo M, Hagström E, Ärnlöv J. Discovery of new biomarkers for atrial fibrillation using a custom-made proteomics chip. Heart. 2016;heartjnl-2016-309764. [DOI] [PubMed] [Google Scholar]
- 33.Svennberg E, Lindahl B, Berglund L, et al. . NT-proBNP is a powerful predictor for incident atrial fibrillation—validation of a multimarker approach. Int J Cardiol. 2016;223:74-81. [DOI] [PubMed] [Google Scholar]
- 34.Breitenstein A, Glanzmann M, Falk V, et al. . Increased prothrombotic profile in the left atrial appendage of atrial fibrillation patients. Int J Cardiol. 2015;185:250-255. [DOI] [PubMed] [Google Scholar]
- 35.Ehrlich JR, Kaluzny M, Baumann S, Lehmann R, Hohnloser SH. Biomarkers of structural remodelling and endothelial dysfunction for prediction of cardiovascular events or death in patients with atrial fibrillation. Clin Res Cardiol. 2011;100(11):1029-1036. [DOI] [PubMed] [Google Scholar]
- 36.Pinto A, Tuttolomondo A, Casuccio A, et al. . Immuno-inflammatory predictors of stroke at follow-up in patients with chronic non-valvular atrial fibrillation (NVAF). Clin Sci (Lond). 2009;116(10):781-789. [DOI] [PubMed] [Google Scholar]
- 37.Hill AB. The environment and disease: association or causation? Proc R Soc Med. 1965;58:295-300. [PMC free article] [PubMed] [Google Scholar]
- 38.Alonso A, Krijthe BP, Aspelund T, et al. . Simple risk model predicts incidence of atrial fibrillation in a racially and geographically diverse population: the CHARGE-AF consortium. J Am Heart Assoc. 2013;2(2):e000102. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eMethods. Detailed Methodology
eTable 1. Associations Between 75 Markers of Inflammation, Immune Activation, or Endothelial Dysfunction and Incident Atrial Fibrillation in the Bruneck Study
eTable 2. Baseline Characteristics in the SAPHIR Study
eFigure 1. Penalized Spline Fit of the Association Between Soluble Vascular Cell Adhesion Molecule 1 and Incident Atrial Fibrillation in the Bruneck Study
eFigure 2. Mechanisms and Epidemiological Evidence Linking Vascular Cell Adhesion Molecule 1 to Atrial Fibrillation