Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Am J Cardiol. Author manuscript; available in PMC 2010 July 1.
Published in final edited form as:
PMCID: PMC2802058

Relation of Multiple Inflammatory Biomarkers to Incident Atrial Fibrillation


Basic and clinical studies suggest that inflammation predisposes to atrial fibrillation (AF). We assessed the association of 12 circulating inflammatory biomarkers [C-reactive protein, fibrinogen, interleukin-6, intercellular adhesion molecule-1, lipoprotein-associated phospholipase A2 (mass and activity), monocyte chemoattractant protein-1, myeloperoxidase, CD40 ligand, osteoprotegerin, P-selectin, tumor necrosis factor receptor II] with incident AF in 2863 Framingham Offspring Study participants (mean age 60.7 years, SD=9.4, 55% women). During follow-up (median 6 years), 148 participants (43% women) developed incident AF. In multivariable proportional-hazards models, the inflammatory biomarker panel was associated with incident AF (p=0.03). With stepwise selection (p<0.01 for entry and retention), log-transformed osteoprotegerin was associated with incident AF (hazard ratio [HR] per standard deviation 1.30, 95% confidence interval [CI] 1.08–1.56, p=0.006). Adjusting for interim myocardial infarction or heart failure attenuated the association between osteoprotegerin and incident AF (HR 1.18, 95% CI 0.98–1.43, p=0.09). In conclusion, circulating osteoprotegerin concentration was significantly associated with incident AF in our community-based sample, possibly mediated by interim cardiovascular events.

Keywords: atrial fibrillation, inflammation, epidemiology, cohort

Both clinical and experimental evidence supports a contribution of inflammation to the development and persistence of AF, although the exact mechanisms are not well understood. A consistent finding in manifest AF is elevated inflammatory activity at the atrial tissue level.1 In most published studies, patients with AF have higher circulating inflammatory marker concentrations compared to referents.25 Investigators also have reported that the highly proinflammatory state induced by cardiopulmonary bypass surgery leads to increased C-reactive protein (CRP) concentrations, which may be etiologically related to incident AF.4 Prospective studies report an association between the nonspecific inflammatory marker CRP and recurrent6, 7 and incident AF.4



For the Framingham Offspring cohort, 5124 individuals were recruited in the early 1970s and were followed routinely.8 The 7th examination cycle (1998–2001) was attended by 3539 participants who were eligible for analysis. For the present investigation, we excluded individuals with any missing inflammatory biomarker measurements (n=483), incomplete or missing follow-up (n=22), prevalent AF (n=132), and missing covariate data (n=39), leaving 2863 attendees for analysis. The study protocols were approved by the Boston University Medical Center Institutional Review Board. Written informed consent was obtained from all participants at each examination. The authors had full access to the data and take responsibility for its integrity. All authors have read and agreed to the manuscript as written.

All participants underwent regular cardiovascular health status assessments at the Framingham Study clinic every 4 to 8 years. Routine biennial telephone-administered health history updates were performed. Disease risk factors were documented in questionnaires and during physician-administered interviews and physical examinations. Valvular heart disease was considered present if the physician auscultated ≥3 out of 6 systolic, or any diastolic murmur. Current smoking, alcohol consumption, and medication use were indicated by self-report. Diabetes was defined as elevated glucose ≥126 mg/dL or use of hypoglycemic drugs. Systolic blood pressure was determined as the average of 2 measurements performed by a physician. The participants’ physician office visits and hospitalization records were routinely collected.

The diagnosis of AF was made if atrial fibrillation or flutter was present on an electrocardiogram derived from hospital or outpatient records or on a Framingham Study clinic tracing. Biennial health history update calls included a routine question as to whether the participant had experienced interim AF. If AF was reported at the main examination or on the health history update, outside records were routinely sought. All incident AF cases underwent a review by 2 of 3 Framingham cardiologists (RS, DL, EJB). If AF was recorded by history only, 2 investigators had to agree that the evidence for the presence of AF was compelling (cardioversion or institution of anticoagulant or antiarrhythmic medication).

Fasting blood samples obtained from participants were processed promptly, centrifuged and frozen at −80°C. The twelve circulating inflammatory markers comprised: plasma CD40 ligand, CRP, fibrinogen, lipoprotein-associated phospholipase A2 activity and mass, osteoprotegerin, P-selectin, and tumor necrosis factor receptor-II, and serum intercellular adhesion molecule-1, interleukin-6, monocyte chemoattractant protein-1, and myeloperoxidase. Details of specimen type, measurement kit and measurement properties have been published before.9 All mean intra-assay coefficients of variation were less than 7%.

Biomarkers were natural logarithmically transformed for analyses and standardized (mean of 0, standard deviation 1). We used multivariable Cox proportional hazards regression models to relate inflammatory biomarkers to incident AF.10 We adjusted for AF risk factors including age, sex, body mass index, systolic blood pressure, current smoking, diabetes, alcohol consumption category (less than 1 drink per week; 1–7 drinks/week in women or 1–14 drinks/week in men; and >7 drinks/week in women or >14 drinks/week in men), hypertension treatment, left ventricular hypertrophy on the ECG, heart valve disease (heart murmur), history of myocardial infarction and heart failure.

To test the overall inflammation hypothesis, we added all 12 biomarkers to the covariates and used a likelihood ratio test with 12 degrees of freedom. Our primary model forced in clinical covariates, and used a stepwise selection procedure to select biomarkers associated with AF with 2-sided p<0.01 for entry and retention. Similar to prior Framingham multimarker papers, we used a more conservative threshold for significance to account for multiple testing.9 Regression coefficients are presented per 1 SD increase in natural logarithm transformed biomarkers. For the marker selected by the stepwise method, we a) estimated Kaplan Meier survival by tertile; b) conducted a log rank test for trend with tertiles scored as integers; c) estimated unadjusted cumulative incidence of AF by tertile to visually examine the association over time. A C-statistic was calculated for the baseline model and for the model additionally including the retained biomarker.11 The proportional hazards assumption was examined using a Kolmogorov-type supremum test based on cumulative sums of martingale-based residuals over follow-up times and covariate values.12

In addition, bivariate correlation coefficients for the (log) inflammatory biomarkers were calculated. Finally, we present the data for each biomarker analyzed separately.

We assessed potential effect modification of the selected biomarker-AF association by age and sex. To further explore whether the association of inflammatory biomarkers with incident AF was mediated by interim events we conducted analyses adjusting for time-dependent myocardial infarction or heart failure. A secondary model was performed excluding prevalent myocardial infarction and heart failure at baseline. For the single marker analyses we examined what effect size we could detect at 80% power. Analyses were conducted with SAS version 8.1 (Cary, North Carolina, statistical software version.


During follow-up through 2007 (median 6.2 years, maximum 8.0 years), 148 incident AF cases (43% women) were identified. Table 1 shows baseline characteristics of the study sample. In brief, the cohort was middle-aged to elderly and about half of the participants were women. At baseline, heart valve disease, myocardial infarction, and heart failure were uncommon and observed in less than 5% of participants. Of individuals with cardiovascular disease at baseline 38 (11.8%) developed AF, whereas of those without baseline cardiovascular disease 110 (4.3%) developed AF in follow-up.

Table 1
Baseline Characteristics of the Study Sample

Adjusting for established risk factors, the inflammatory biomarkers added as a group were associated with incident AF (p=0.03). In stepwise selection, only osteoprotegerin was significantly associated with AF and had a 30% increased risk of incident AF (hazard ratio [HR] per SD increment in log-marker 1.30, 95% confidence interval [CI] 1.08–1.56, p=0.006). No other inflammatory marker met the p<0.01 entry criterion. Cumulative incidence of AF was highest for individuals in the third tertile of osteoprotegerin (Figure 1). The C-statistic did not change appreciably between the multivariable model with (C-statistic=0.810, 95% CI 0.777–0.844) or without osteoprotegerin (0.808, 95% CI 0.775–0.841) .

Figure 1
Cumulative Incidence of AF by Osteoprotegerin Tertile

For the variable heart murmur, we saw a nominal departure from the proportional hazards assumption (p=0.028). However, when adjusting for the variable (heart murmur yes/no), the estimates for all covariates were similar to those observed without adjustment.

In models excluding prevalent myocardial infarction or heart failure at baseline, osteoprotegerin was similarly associated with AF (HR 1.27, 95% CI 1.05–1.55, p=0.02) compared to the final model including and adjusting for prevalent disease. The final model did not show a statistically significant interaction between osteoprotegerin and age or sex for the risk of incident AF. If occurrence of interim myocardial infarction and heart failure (n=87) was introduced as a time-dependent covariate, the panel of circulating biomarkers (p=0.14), and osteoprotegerin (HR 1.18, 95% CI 0.98–1.43, p=0.09) were not statistically significantly associated with AF.

Pairwise correlation analysis revealed minor to moderate correlations between inflammatory biomarkers. Modest bivariate correlations were observed between osteoprotegerin and CRP (r=0.15), fibrinogen (r=0.14), intercellular adhesion molecule-1 (r=0.18), interleukin-6 (0.20), monocyte chemoattractant protein-1 (r=0.12), and tumor necrosis factor receptor-II (r=0.31) (all p<0.0001). No significant correlations were observed for osteoprotegerin with the other inflammatory markers. Maximum correlation coefficients were seen for CRP and fibrinogen (r=0.52) and CRP and interleukin-6 (r=0.47).

Because different investigators have examined varying inflammatory biomarkers in relation to AF, in secondary analyses we examined the multivariable-adjusted relation of the biomarkers analyzed individually in relation to incident AF (Table 2). We had 80% power to detect a hazard ratio of 1.27 per SD at p<0.05 and 1.37 at p<0.005 level for individual biomarkers. Lipoprotein-associated phospholipase A2 mass showed a borderline significant association with incident AF (p=0.02, Table 2). We did not observe significant associations between incident AF and CRP (HR 1.05; 95% CI 0.87–1.26; p=0.61) or interleukin-6 (HR 1.08; 95% CI 0.91–1.29; p=0.36).13

Table 2
Results from Multivariable-Adjusted Proportional Hazards Regression Models: Each Circulating Inflammatory Biomarker Modeled Separately


We examined the association of 12 circulating inflammatory biomarkers with incident AF in a contemporary community-based middle-aged to elderly cohort. The biomarker osteoprotegerin was significantly associated with incident AF. The relation between osteoprotegerin and incident AF was attenuated, and no longer achieved statistical significance in models accounting for interim myocardial infarction and heart failure, suggesting that interim cardiovascular events may partially mediate the osteoprotegerin–AF association.

The pathophysiology of the inflammation-AF relation is an area of increasing scientific scrutiny. There is sound evidence from animal14 and human15 data that during ongoing atrial dysrhythmia higher levels of oxidative stress and inflammation can be measured locally and systemically, and atrial pro-inflammatory and pro-thrombotic gene expression is altered.1 Atrial oxidative damage experienced during AF alters myofibrillar energetics and may lead to myocyte necrosis. Apoptosis and necrosis induce low-grade inflammation, paving the way for structural alterations. The histological substrate of these processes are inflammatory and fibrotic changes of the atrial myocardium.16 Inflammation thus may perpetuate adverse electrical and structural remodeling and promote disease chronicity. A more novel hypothesis is that enhanced inflammatory activity, as seen after cardiac surgery, may precede the onset of AF. This notion is supported by recent epidemiological findings4 and our current results, which establish a temporal relation between inflammation and long-term AF incidence.

There are several possible interpretations for why we did not observe the previously reported4 association between CRP concentrations and incident AF. Aviles and colleagues examined participants in the Cardiovascular Health Study and observed an adjusted HR for AF per 1-standard deviation in log CRP of 1.24 (95% CI 1.11 to 1.40; P<0.001).4 Cardiovascular Health Study investigators examined a single inflammatory biomarker, had a significantly older participant mean age (73±5 years), a higher prevalence of cardiovascular disease, and many more incident AF events (n=897) than our study. It should be noted that our upper 95% confidence interval boundary was 1.26, which included the Cardiovascular Health Study’s point estimate (1.24). We acknowledge that with 148 events our statistical power was modest for main effects and was underpowered for statistical interactions. Hence, we cannot exclude an interaction between inflammation and age (or prevalent cardiovascular disease) and risk of AF, accounting for the differences between the 2 studies.

Osteoprotegerin, a tumor necrosis factor receptor family member, which was not strongly correlated with any of the other measured inflammatory markers, has not been related to incident AF previously. Apart from its role in bone metabolism, osteoprotegerin has been associated with arterial calcification, endothelial function, and atherosclerosis17 in animal models. Whereas osteoprotegerin itself exerts protective effects on vascular remodeling and calcification, elevated circulating concentrations seem to reflect a counterregulatory overproduction of the protein which thus becomes associated with adverse outcome.

Attributable to the influence on vascular wall composition, osteoprotegerin has been related to pulse wave velocity, an indicator of vascular stiffness.18 Aortic stiffness mirrored by pulse pressure has been reported to predict incident AF.19 In addition, osteoprotegerin may play a pathophysiological role in left ventricular remodeling and heart failure,2022 conditions, for which gene expression of osteoprotegerin is consistently increased23 and which are both strongly related to future AF.

In our secondary analyses, including MI and CHF as time-dependent variables, the tests for the complete set of biomarkers and osteoprotegerin alone no longer reached statistical significance. The weakening of the apparent relation between inflammatory markers and AF is consistent with the concept that interim cardiovascular disease either is a confounder, or is part of the pathophysiological pathway connecting inflammation and the manifestation of AF.20 If our findings are confirmed in other studies, further research is required to elucidate whether the risk association of osteoprotegerin with AF is attributable to mechanisms such as cardiovascular remodeling, and interim cardiovascular disease events, or whether there is a direct relation between osteoprotegerin and AF risk.

Our data were derived from a well-characterized community-based sample with extensive routine ascertainment of potential clinical confounders, strict quality control of inflammatory biomarker measurements and rigorous clinical ascertainment of the outcome. Due to the relatively small number of incident AF cases, we had modest power to detect relations of inflammatory markers to incident AF, particularly if the effect size was small. To reduce the burden of multiple testing, we à priori decided to run multivariable-adjusted models for the primary analyses. Since inflammatory biomarkers may lie in the pathophysiological pathway from risk factors (e.g. body mass index) to incident AF, our models may be over-adjusted and mask relevant associations.

A more comprehensive picture of the actions of osteoprotegerin might have been derived if measurements of the receptor activator of nuclear factor-kappa B ligand (RANKL) and its receptor (RANK) had been available to better reflect the osteoprotegerin-RANK-RANKL axis in relation to AF.24 We provided data only on single occasion inflammatory biomarker measurements, which did not account for potential biomarker variability over time, although for some markers such variability is known to be relatively low.25 Participants with AF, especially asymptomatic ones, may have been missed, which might bias the study results towards the null. Finally, our data need confirmation in independent studies.

If inflammation can be established as a relevant AF risk factor as was shown for osteoprotegerin and incident heart failure and myocardial infarction,20 it would be a potentially modifiable one. Lifestyle modifications like exercise, weight loss and diet can positively diminish pro-inflammatory states. In addition, drugs such as statins and renin angiotensin aldosterone blocking drugs have been noted to lower circulating osteoprotegerin concentrations,26 and to reduce the risk of incident AF.27, 28 The extent to which statins and renin-angiotensin aldosterone blocking drugs mechanistically reduce AF via modulation of inflammation or prevention of the development of the risk factors for AF (such as hypertension, diabetes, myocardial infarction and heart failure) remains to be determined.


Supported by NIH/NHLBI contract N01-HC-25195, 6R01-NS 17950, and NIH grants HL064753. HL076784, AG028321 (EJB), 1 RO1HL71039 (RSV); NIH Research career award K24 HL04334 (RSV); Deutsche Forschungsgemeinschaft (German Research Foundation) Research Fellowship SCHN 1149/1–1 (RS); 1K23HL083102, Doris Duke Charitable Foundation Clinical Scientist Development Award (SK). Lipoprotein-associated phospholipase A2 activity and mass measurements were provided to the Heart Study by GlaxoSmithKline at no cost.


Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.


The authors report no conflicts of interest.

Reference List

1. Nakamura Y, Nakamura K, Fukushima-Kusano K, Ohta K, Matsubara H, Hamuro T, Yutani C, Ohe T. Tissue factor expression in atrial endothelia associated with nonvalvular atrial fibrillation: possible involvement in intracardiac thrombogenesis. Thromb Res. 2003;111:137–142. [PubMed]
2. Roldan V, Marin F, Blann AD, Garcia A, Marco P, Sogorb F, Lip GY. Interleukin-6, endothelial activation and thrombogenesis in chronic atrial fibrillation. Eur Heart J. 2003;24:1373–1380. [PubMed]
3. Chung MK, Martin DO, Sprecher D, Wazni O, Kanderian A, Carnes CA, Bauer JA, Tchou PJ, Niebauer MJ, Natale A, Van Wagoner DR. C-reactive protein elevation in patients with atrial arrhythmias: inflammatory mechanisms and persistence of atrial fibrillation. Circulation. 2001;104:2886–2891. [PubMed]
4. Aviles RJ, Martin DO, pperson-Hansen C, Houghtaling PL, Rautaharju P, Kronmal RA, Tracy RP, Van Wagoner DR, Psaty BM, Lauer MS, Chung MK. Inflammation as a risk factor for atrial fibrillation. Circulation. 2003;108:3006–3010. [PubMed]
5. Mukamal KJ, Tolstrup JS, Friberg J, Gronbaek M, Jensen G. Fibrinogen and albumin levels and risk of atrial fibrillation in men and women (the Copenhagen City Heart Study) Am J Cardiol. 2006;98:75–81. [PubMed]
6. Conway DS, Buggins P, Hughes E, Lip GY. Prognostic significance of raised plasma levels of interleukin-6 and C-reactive protein in atrial fibrillation. Am Heart J. 2004;148:462–466. [PubMed]
7. Liu T, Li G, Li L, Korantzopoulos P. Association between C-reactive protein and recurrence of atrial fibrillation after successful electrical cardioversion: a meta-analysis. J Am Coll Cardiol. 2007;49:1642–1648. [PubMed]
8. Dawber T, Meadors G, Moore F., Jr Epidemiological approaches to heart disease: the Framingham Study. Am J Public Health Nations Health. 1951;41:279–281. [PubMed]
9. Schnabel R, Larson MG, Dupuis J, Lunetta KL, Lipinska I, Meigs JB, Yin X, Rong J, Vita JA, Newton-Cheh C, Levy D, Keaney JF, Jr, Vasan RS, Mitchell GF, Benjamin EJ. Relations of inflammatory biomarkers and common genetic variants with arterial stiffness and wave reflection. Hypertension. 2008;51:1651–1657. [PMC free article] [PubMed]
10. Cox DR, Oakes D. Analysis of Survival Data. London, UK: Chapman and Hall; 1984. Analysis of Survival Data; p. 201.
11. Pencina MJ, D’Agostino RB. Overall C as a measure of discrimination in survival analysis: model specific population value and confidence interval estimation. Stat Med. 2004;23:2109–2123. [PubMed]
12. Lin D, Wei LJ, Ying Z. Checking the Cox model with cumulative sums of martingale-based residuals. Biometrika. 1993;80:557–572.
13. Marcus GM, Whooley MA, Glidden DV, Pawlikowska L, Zaroff JG, Olgin JE. Interleukin-6 and atrial fibrillation in patients with coronary artery disease: data from the Heart and Soul Study. Am Heart J. 2008;155:303–309. [PMC free article] [PubMed]
14. Wijffels MC, Kirchhof CJ, Dorland R, Allessie MA. Atrial fibrillation begets atrial fibrillation. A study in awake chronically instrumented goats. Circulation. 1995;92:1954–1968. [PubMed]
15. Neuman RB, Bloom HL, Shukrullah I, Darrow LA, Kleinbaum D, Jones DP, Dudley SC., Jr Oxidative stress markers are associated with persistent atrial fibrillation. Clin Chem. 2007;53:1652–1657. [PMC free article] [PubMed]
16. Frustaci A, Chimenti C, Bellocci F, Morgante E, Russo MA, Maseri A. Histological substrate of atrial biopsies in patients with lone atrial fibrillation. Circulation. 1997;96:1180–1184. [PubMed]
17. Bucay N, Sarosi I, Dunstan CR, Morony S, Tarpley J, Capparelli C, Scully S, Tan HL, Xu W, Lacey DL, Boyle WJ, Simonet WS. osteoprotegerin-deficient mice develop early onset osteoporosis and arterial calcification. Genes Dev. 1998;12:1260–1268. [PubMed]
18. Kim SM, Lee J, Ryu OH, Lee KW, Kim HY, Seo JA, Kim SG, Kim NH, Baik SH, Choi DS, Choi KM. Serum osteoprotegerin levels are associated with inflammation and pulse wave velocity. Clin Endocrinol (Oxf) 2005;63:594–598. [PubMed]
19. Mitchell GF, Vasan RS, Keyes MJ, Parise H, Wang TJ, Larson MG, D’Agostino RB, Sr, Kannel WB, Levy D, Benjamin EJ. Pulse pressure and risk of new-onset atrial fibrillation. JAMA. 2007;297:709–715. [PubMed]
20. Omland T, Ueland T, Jansson AM, Persson A, Karlsson T, Smith C, Herlitz J, Aukrust P, Hartford M, Caidahl K. Circulating osteoprotegerin levels and long-term prognosis in patients with acute coronary syndromes. J Am Coll Cardiol. 2008;51:627–633. [PubMed]
21. Avignon A, Sultan A, Piot C, Mariano-Goulart D, Thuan Dit Dieudonne JF, Cristol JP, Dupuy AM. Osteoprotegerin: a novel independent marker for silent myocardial ischemia in asymptomatic diabetic patients. Diabetes Care. 2007;30:2934–2939. [PubMed]
22. Crisafulli A, Micari A, Altavilla D, Saporito F, Sardella A, Passaniti M, Raffa S, D’anneo G, Luca F, Mioni C, Arrigo F, Squadrito F. Serum levels of osteoprotegerin and RANKL in patients with ST elevation acute myocardial infarction. Clin Sci (Lond) 2005;109:389–395. [PubMed]
23. Omland T, Drazner MH, Ueland T, Abedin M, Murphy SA, Aukrust P, de Lemos JA. Plasma osteoprotegerin levels in the general population: relation to indices of left ventricular structure and function. Hypertension. 2007;49:1392–1398. [PubMed]
24. Hofbauer LC, Khosla S, Dunstan CR, Lacey DL, Boyle WJ, Riggs BL. The roles of osteoprotegerin and osteoprotegerin ligand in the paracrine regulation of bone resorption. J Bone Miner Res. 2000;15:2–12. [PubMed]
25. Van Hoydonck PG, Schouten EG, Temme EH. Reproducibility of blood markers of oxidative status and endothelial function in healthy individuals. Clin Chem. 2003;49:963–965. [PubMed]
26. Nellemann B, Gormsen LC, Dollerup J, Schmitz O, Mogensen CE, Rasmussen LM, Nielsen S. Simvastatin reduces plasma osteoprotegerin in type 2 diabetic patients with microalbuminuria. Diabetes Care. 2007;30:3122–3124. [PubMed]
27. Fauchier L, Pierre B, de Labriolle A, Grimard C, Zannad N, Babuty D. Antiarrhythmic effect of statin therapy and atrial fibrillation a meta-analysis of randomized controlled trials. J Am Coll Cardiol. 2008;51:828–835. [PubMed]
28. Healey JS, Baranchuk A, Crystal E, Morillo CA, Garfinkle M, Yusuf S, Connolly SJ. Prevention of atrial fibrillation with angiotensin-converting enzyme inhibitors and angiotensin receptor blockers: a meta-analysis. J Am Coll Cardiol. 2005;45:1832–1839. [PubMed]