Journal of Human Hypertension (2010) 24, 447–457 & 2010 Macmillan Publishers Limited All rights reserved 0950-9240/10 www.nature.com/jhh ORIGINAL ARTICLE Hypertension and paroxysmal atrial fibrillation: a novel predictive role of high sensitivity C-reactive protein in cardioversion and long-term recurrence I Rizos1, AG Rigopoulos1, AS Kalogeropoulos1, S Tsiodras2, S Dragomanovits1, EA Sakadakis1, E Faviou1 and DT Kremastinos1 1 Second Department of Cardiology, University of Athens Medical School, Attikon University Hospital, Athens, Greece and 2Fourth Academic Department of Internal Medicine and Infectious Diseases, University of Athens Medical School, Attikon University Hospital, Athens, Greece The role of inflammation in maintenance of paroxysmal atrial fibrillation (PAF) in patients with hypertension and no other heart disease has not been fully elucidated yet. We investigated the association of various inflammatory markers with cardioversion and recurrence of PAF in patients with hypertension. We studied 75 patients (44 male, mean age 67.9±9.9 years) with PAF (duration from onset of symptomso24 h) secondary to hypertension. None had heart failure or any other ongoing inflammatory process. All patients received anticoagulation and intravenous amiodarone for cardioversion. High sensitivity C-reactive protein (hsCRP), interleukin-6 (IL-6) and tumour necrosis factor (TNF)-a were measured on admission and 48 h later. By 48 h from admission 61/75 patients (81.3%) regained sinus rhythm (cardioverted), whereas 14/75(18.7%) remained in AF (non-cardioverted). hsCRP, IL-6 and TNF-a serum levels on admission were similar between groups. hsCRP at 48 h was the most significant factor correlated with cardioversion outcome (OR: 0.06, 95% CI: 0.01–0.47, P ¼ 0.008). During a 1-year follow-up, AF recurred in 28/61(45.9%) patients. The strongest factor associated with AF recurrence was hsCRP at 48 h X2.27 mg l 1 (hazard ratio: 6.2, 95% CI: 2.2–17.6, P ¼ 0.001). hsCRP at 48 h after admission correlates with cardioversion outcome and may predict long-term AF recurrence. Journal of Human Hypertension (2010) 24, 447–457; doi:10.1038/jhh.2009.89; published online 14 January 2010 Keywords: atrial fibrillation; C-reactive protein; cardioversion; interleukins; recurrence Introduction Atrial fibrillation (AF) is the arrhythmia with the highest prevalence in the clinical setting.1 The pathogenesis of AF involves interactions between atrial structural remodelling as this is affected by underlying heart disease, for example hypertension, coronary artery disease, valvular heart disease and cardiomyopathies, and miscellaneous potential triggers, of which inflammation has gained increased interest recently.2,3 This interest has grown as high sensitivity C-reactive protein (hsCRP), a non-specific inflammatory marker, has been related with the incidence4,5 and persistence of AF.6,7 However, several factors confound the role of inflammation in generation and perpetuation of AF. Correspondence: Dr I Rizos, Second Department of Cardiology, University of Athens Medical School, Attikon University Hospital, 19 Kentauron Street, Filothei 15237, Athens, Greece. E-mail: ioannis.c.rizos@otenet.gr Received 2 June 2009; revised 7 October 2009; accepted 26 October 2009; published online 14 January 2010 As an example, failure to show a correlation between baseline hsCRP levels and maintenance of sinus rhythm after AF cardioversion was attributed to the mismatch between the compared AF and control groups, regarding the cardiovascular disease substrate (hypertension, left ventricular dysfunction, ischemic heart disease) and other risk factors, such as age, smoking and diabetes.8 Moreover, hsCRP levels in lone AF did not differ from those of normal controls, whereas they were significantly higher in hypertensive patients with AF,9 indicating that the hsCRP elevation may simply be a contemporaneous event reflecting underlying cardiovascular or other disease processes rather than be associated with AF itself. In addition, inflammation is a complicated— dynamic—process and the measurement of a single marker at an arbitrary point in time may not adequately portray its relation to AF maintenance. Scarce data about the inflammatory process shortly after the onset of AF have been reported so far.10,11 Inflammation in atrial fibrillation due to hypertension I Rizos et al 448 In this study, we sought to investigate whether the outcome of cardioversion as well as the likelihood of AF recurrence, in a select group of patients with essential hypertension12 and paroxysmal AF (PAF), relate to baseline levels of important inflammatory markers such as hsCRP, interleukin-6 (IL-6) and tumour necrosis factor (TNF)-a and with their levels 48 h after admission. Additionally, we examined if there was an association between the baseline inflammatory load and the level of atrial structural remodelling. Methods Study population This was an observational case–control study of 133 consecutive patients admitted to our department with symptomatic PAF (duration from onset of symptoms o24 h) secondary to hypertension from January 2005 until January 2006. However, 58 subjects were excluded because of the presence of additional comorbidities that complicated the origin of PAF, namely, coronary artery disease, heart failure (left and/or right ventricular dilatation and/or systolic dysfunction), acute pulmonary edema, acute coronary syndromes, lone AF, rheumatic valve disease, sick sinus syndrome, autoimmune diseases, chronic obstructive pulmonary disease, thyrotoxicosis, recent trauma or surgery, known neoplasm, recent or ongoing exposure to non-steroidal antiinflammatory drugs or corticosteroids, renal failure. All remaining patients were under intensive antihypertensive treatment for at least a year with well-controlled arterial blood pressure levels. The diagnosis of AF was accomplished by at least one 12-lead electrocardiogram. Onset of AF was defined precisely by considering the onset of symptoms such as palpitations, shortness of breath and chest discomfort. Study protocol The study protocol was approved by the hospital’s ethics committee and conforms to the principles of the Declaration of Helsinki. Written informed consent was obtained by all study participants and eventually, each one underwent evaluation and treatment according to a standardized protocol described in detail below. After history taking and physical examination on admission, patients had a 12-lead electrocardiogram, an anteroposterior and lateral chest X-ray, and routine laboratory evaluation and biochemical testing. Patients received intravenous amiodarone at a dosing regimen of 5 mg kg 1 bolus in 20 min and then 500–900 mg intravenously for 12 h. For prevention of thromboembolism, anticoagulation with enoxaparin 1 mg kg 1 subcutaneously twice daily was initiated on admission and then acenocoumarol was administered according to ACC/AHA guideJournal of Human Hypertension lines.13 Serum levels of hsCRP, IL-6 and TNF-a were measured on admission and at 48 h after the beginning of intravenous administration of amiodarone. A standard two-dimensional echocardiogram was performed before discharge, using Hewlett Packard Sonos 2000 echocardiographic equipment (Andover, MA, USA). Estimation of maximal left atrial volume was performed as follows: from the apical four- and two-chambers view, left atrial borders were manually traced at end-systole. The biplane and single plane modified Simpson’s formula was applied for maximal left atrial volume measurement.14,15 Images were stored in VHS magnetic tape and two independent observers performed the final estimation of atrial volume off-line. The mean value of three consecutive cardiac cycles was derived as the final atrial volume. All patients were constantly monitored for 48 h and the time of cardioversion to sinus rhythm was recorded. Patients who regained sinus rhythm were advised to seek medical care in the case of any symptom recurrence and were followed regularly on a monthly basis after discharge until the completion of 1 year after cardioversion. Any sustained recurrence (electrocardiographically confirmed and lasting longer than 1 min) of AF was recorded. If necessary, the patient’s physician was also contacted and all the essential information for the AF recurrence was obtained. In addition, information concerning any concurrent illness that might interfere with the probability of recurrence was also sought. During follow-up, any anti-arrhythmic treatment was decided on by the patients’ attending cardiologist at that time. Laboratory methods and blood sampling Blood samples for inflammatory markers measurement were obtained by a peripheral vein on admission and 48 h later and were drawn without stasis into serum tubes. Within an hour after collection the samples were centrifuged at 3.200 g for 10 min at a temperature of about 4 1C. The serum was separated into aliquots and was stored in 80 1C until the assay analysis was performed. Serum levels of TNF-a and of IL-6 were determined by commercial standardized solid phase sandwich enzyme-linked immunosorbent assay methods according to the manufacturer’s instructions (BenderMed Systems, Vienna, Austria and Biosource International, Camarillo, CA, USA). These assays detected only human cytokines with a sensitivity of 2.3 pg ml 1 for TNF-a and of 2 pg ml 1 for IL-6. The clinical samples were kept at 2–8 1C and separated rapidly before storing at 80 1C to avoid loss of bioactive cytokine. hsCRP was measured by microparticle-enhanced immunonephelometric assay (Cardiophase* hsCRP, Dade Behring, Siemens Healthcare Diagnostics, Eschborn, Germany) on a Behring BN ProSpec clinical chemistry analyzer, Inflammation in atrial fibrillation due to hypertension I Rizos et al 449 for which the measuring range is 0.16–10.2 mg l 1, the typical analytical sensitivity is 0.175 for the measurements performed using a sample dilution of 1:20, and the total median intra- and inter-assay coefficient of variation is less than 6% for values below 1.0 mg l 1. Statistical analysis All quantitative variables were tested for normal distribution by Kolmogorov–Smirnov test. Data nonnormally distributed are presented as median and inter-quartile range values (IQR; 25th–75th percentile). Normally distributed variables are expressed as mean and standard deviation values. Univariate analysis for factors significantly associated with the cardioversion and recurrence outcome was performed for all patients. Differences in continuous variables were estimated by Mann– Whitney U-test and two-sample t-test for nonnormally distributed and normally distributed variables, respectively. Differences in categorical data were evaluated by w2-test for normally distributed and Fisher’s exact test for non-normally distributed variables. Correlations were estimated using Spearman’s rank correlation method. Values of Po0.05 were considered statistically significant. Stepwise multivariate logistic regression analysis was performed to define factors significantly associated with the cardioversion and recurrence outcome. All factors with a significance level of Pp0.1, during the univariate analysis, were included in the model. Covariates that were clinically known to be important in AF persistence (age, gender, left ventricular ejection fraction, left atrial volume, diabetes and smoking) were included in the analysis as well. Serum levels of hsCRP and IL-6 were each log transformed (log hsCRP and log IL-6, respectively) for the regression analysis to improve linear fitting. The probability value that was used for entering or removing a variable was P ¼ 0.05 and 0.1, respectively. To assess the most favourable cut-off value, for the factor that correlated most significantly with AF relapse and best predicted the end point of AF recurrence, a receiver operating characteristic curve analysis was applied. The optimal cut-off value corresponded to the value with the best sensitivity and specificity according to the maximal positive and minimal negative likelihood ratios (PLR and NLR). Further analysis of independent predictors for AF recurrence including the optimal cut-off value was performed by means of a Cox proportional hazards regression model. The cumulative risk for AF recurrence, according to the cut-off value of the factor that most significantly associated with AF recurrence, was evaluated by Kaplan–Meier analysis and the difference between patients with and without recurrence of AF was assessed with the log-rank test. Additionally, PLR and NLR were assessed and the exact post-test probability of AF relapse was calculated by Bayes’ Theorem method.16 The pre-test probability corresponded to the likelihood of AF recurrence in 1 year after successful cardioversion as this is observed commonly in clinical practice. Likelihood ratios have several advantages in interpreting the diagnostic value of diagnostic tests. They are independent of disease prevalence and they do not vary in different populations and settings. PLR (sensitivity/(1 specificity)) corrects the true positive rate of a test by the false positive rate, whereas NLR ((1 sensitivity)/specificity) compares the probability of a negative test in diseased subjects, in comparison to the probability of a negative test in those without the disease. Power calculations Formal sample size calculations were performed by analysing data from earlier work by our group.17 Earlier reports showed that the success rate of cardioversion after amiodarone administration in patients with PAF is almost 80% in the first 24 h (1:4 case–control ratio)18 and the recurrence rate of PAF after successful cardioversion with amiodarone is about 50% (1:1 case–control ratio).19 Hence, it was estimated, that to detect a 60% difference in hsCRP levels between patients with a positive and negative cardioversion outcome, a power of 80%, and a two-tailed a of 0.05, 13 non-cardioverted and 50 cardioverted will be required for each arm of the study (1:4 case–control ratio). Likewise, according to the aforementioned criteria, 27 patients with AF recurrence and 27 without AF recurrence will have to be involved, in each arm of the study (1:1 case–control ratio). Results AF cardioversion Seventy-five hypertensive patients with PAF of o24 h duration were studied. Sixty-one patients (81.3%) successfully cardioverted to sinus rhythm (cardioverted), whereas 14 (18.7%) remained in AF 48 h after admission (non-cardioverted). All noncardioverted patients remained in AF during the first week after admission and were excluded from further study. The clinical profiles of the patients are listed in Table 1. Patients who failed to restore sinus rhythm had significantly higher left atrial volume and longer duration of AF from onset of symptoms until admission than those who cardioverted successfully. In addition, the former used thiazide diuretics (in combination with an ACE inhibitor or an angiotensin receptor blocker) in a higher proportion than cardioverted patients. On the contrary, cardioverted and non-cardioverted individuals had no significant difference in both heart rate and blood pressure levels on admission (Table 1). The levels of the examined inflammatory markers on admission were not significantly different beJournal of Human Hypertension Inflammation in atrial fibrillation due to hypertension I Rizos et al 450 Table 1 Clinical characteristics according to cardioversion outcome All patients n ¼ 75 Cardioverted n ¼ 61 Non-cardioverted n ¼ 14 Pa Age, mean±s.d. (years) 67.9±9.9 67.57±9.66 68.71±11.54 Male/female 44/31 34/27 10/4 Smoking, n (%) 16.0 (21.3) 14.0 (23.0) 2.0 (14.3) Diabetes, n (%) 7.0 (10.4) 7.0 (11.5) 0.0 (0.0) AF episodes in the last year, median (IQR) 2.00 (1.0–3.00) 2.00 (1.00–3.00) 1.50 (1.00–3.25) AF duration from onset of symptoms till admission 5.0 (2.0–12.0) 4.0 (2.0–8.0) 11.0 (4.0–15.5) (h), median (IQR) Systolic blood pressure on admission (mm Hg), median (IQR) 135.0 (120.0–150.0) 130.0 (120.0–150.0) 137.5 (127.5–160.0) Diastolic blood pressure on admission (mm Hg), median (IQR) 80.0 (75.0–82.5) 80 (75.0–81.0) 80 (77.0–83.0) Heart rate on admission (beats per minute), mean±s.d. 120.97±23.8 123.4±23.6 111.71±23.0 0.17 0.48 0.11 Previous drug treatment on admission ACE inhibitors, n (%) Angiotensin receptor blockers, n (%) Statins, n (%) Aspirin, n (%) b-adrenergic receptor blockers, n (%) Calcium channel antagonists, n (%) Diuretics, n (%) 0.33 1.00 0.44 1.00 0.54 0.20 0.014 22.0 14.0 12.0 24.0 26.0 21.0 26.0 (29.3) (18.7) (16.0) (32.0) (34.7) (28.0) (34.7) 16.0 12.0 11.0 20.0 20.0 15.0 17.0 (26.2) (19.7) (18.0) (32.8) (32.8) (24.6) (27.9) 6.0 2.0 1.0 4.0 6.0 6.0 9.0 (42.9) (14.3) (7.1) (28.6) (42.9) (42.9) (64.3) 0.73 0.28 0.72 0.34 0.53 0.016 Echocardiographic parameters Left ventricular ejection fraction (%), median (IQR) Left atrial volume (ml), median (IQR) 60.0 (55.0–63.0) 60.0 (56.5–64.5) 50.00 (45.00–70.84) 48.50 (42.25–63.00) 57.0 (50.0–60.0) 0.07 70.00 (49.50–90.25) 0.012 Inflammatory markers Admission hsCRP (mg l 1), median (IQR) Admission IL-6 (pg ml 1), median (IQR) Admission TNF-a (pg ml 1), median (IQR) hsCRP at 48 h (mg l 1), median (IQR) IL-6 (pg ml 1) at 48 h, median (IQR) TNF-a (pg ml 1) at 48 h, median (IQR) 2.10 4.35 13.00 3.24 5.44 12.90 3.01 3.86 13.91 16.29 15.09 13.73 (1.30–5.80) (3.20–5.60) (7.04–17.00) (1.41–9.95) (3.32–12.80) (8.82–20.35) 1.97 4.30 13.80 2.27 4.62 12.38 (1.22–5.56) (3.40–5.60) (7.13–16.98) (1.32–6.53) (3.20–8.01) (8.82–20.65) (1.23–6.58) (1.97–6.09) (9.24–17.60) (4.76–25.63) (5.11–39.24) (10.00–21.65) 0.84 0.35 0.78 0.001 0.011 0.65 Abbreviations: AF, atrial fibrillation; hsCRP, high sensitivity C-reactive protein; IL-6, interleukin-6; IQR, inter-quartile range; TNF, tumour necrosis factor. Bold values stand for statistically significant differences. a P for the comparison between cardioverted and not cardioverted patients. Figure 1 Serum levels of hsCRP, IL-6, and TNF-a in non-cardioverted and cardioverted patients on admission and at 48 h. The black and white bars show median values. The P-value is shown only for statistically significant comparisons. tween non-cardioverted and cardioverted patients. However, patients who failed to cardiovert exhibited significantly higher levels of hsCRP and IL-6 at 48 h after admission in comparison to the corresponding values of cardioverted patients (Figure 1). Values of hsCRP and IL-6 in most non-cardioverted patients Journal of Human Hypertension increased at 48 h in comparison to the corresponding values on admission, whereas they remained relatively stable in most cardioverted patients (Figure 2). In all studied patients, we did not observe any correlation between left atrial size and the examined Inflammation in atrial fibrillation due to hypertension I Rizos et al 451 Figure 2 Serum hsCRP levels on admission and at 48 h in cardioverted and non-cardioverted patients. The interpolation lines show the trend of each case for the hsCRP serum levels between the two points in time (admission and 48 h). inflammatory markers: hsCRP on admission (r ¼ 0.01, P ¼ 0.9); hsCRP at 48 h (r ¼ 0.1, P ¼ 0.4); IL-6 on admission (r ¼ 0.1, P ¼ 0.4); IL-6 at 48 h (r ¼ 0.2, P ¼ 0.09); TNF-a on admission (r ¼ 0.01, P ¼ 0.9); TNF-a at 48 h (r ¼ 0.1, P ¼ 0.3). On the other hand, hsCRP and IL-6 values at 48 h significantly correlated with the total duration of AF (from onset until cardioversion) (r ¼ 0.399, P ¼ 0.0004 and r ¼ 0.345, P ¼ 0.002, respectively). This correlation remained stable even after adjustment for left atrial volume (r ¼ 0.343, P ¼ 0.003 and r ¼ 0.345, P ¼ 0.003, respectively). There was no significant association between arterial systolic and diastolic blood pressure levels on admission and the examined inflammatory markers, with the only exception being TNF-a serum levels at 48 h, which were positively correlated with baseline systolic (r ¼ 0.234, P ¼ 0.044) and diastolic blood pressure (r ¼ 0.271, P ¼ 0.019). Furthermore, patients with higher heart rate on admission had also higher levels of baseline IL-6 (r ¼ 0.329, P ¼ 0.006). The most significant factor independently associated with cardioversion outcome after adjustment for age, gender, AF duration from onset of symptoms until admission, left ventricular ejection fraction, smoking, diabetes mellitus, thiazide diuretics and left atrial volume was log hsCRP at 48 h (OR: 0.06, 95% CI: 0.01–0.47, P ¼ 0.008). In addition, log IL-6 at 48 h significantly correlated with the cardioversion outcome (OR: 0.16, 95% CI: 0.03–0.92, P ¼ 0.04). AF recurrence Throughout the 1-year follow-up period, 28 (45.9%) of the cardioverted patients had at least one episode of sustained AF recurrence. None of them had any of the conditions included in the exclusion criteria of the study at the time of recurrence. Patients who had a recurrence also reported more AF episodes during the year before admission. In addition, the total time from onset of symptoms until cardioversion was slightly longer in patients with AF recurrence, but this difference was not statistically significant. There was no significant difference between patients with AF recurrence and those who remained in sinus rhythm during the 1-year follow-up regarding systolic and diastolic blood pressure levels and heart rate values on admission (Table 2). Levels of hsCRP on admission and at 48 h as well as IL-6 levels on admission were higher in patients with AF recurrence, whereas no difference was observed in IL-6 at 48 h and in TNF-a serum levels on admission and at 48 h between the two groups (Figure 3). The number of relapses during the 1-year followup correlated positively with hsCRP levels at 48 h (r ¼ 0.425, P ¼ 0.001), hsCRP on admission (r ¼ 0.326, P ¼ 0.01) and the number of AF episodes during the last year before hospital admission (r ¼ 0.353, P ¼ 0.005). Of all the examined inflammatory markers, only hsCRP at 48 h had a strong negative correlation with the time that patients remained free of AF recurrence (r ¼ 0.374, P ¼ 0.003). Additionally, the only two cardioverted patients that had a marked increase in hsCRP at 48 h (Figure 2) also had early recurrence of AF in the first and second month of follow-up, respectively. Binary logistic regression analysis showed that after adjustment for age, gender, left atrial volume, left ventricular ejection fraction, diabetes, smoking and the number of AF episodes during the last year, log hsCRP at 48 h was the most significant factor associated with AF recurrence during the 1-year follow-up (OR: 5.96, CI: 1.65–21.52, P ¼ 0.006). Log hsCRP on admission was also associated with AF relapse after adjustment for the same confounders (OR: 4.58, CI: 1.23–16.99, P ¼ 0.02). The optimal cut-off value of hsCRP at 48 h was determined by the receiver operating characteristic curve analysis (Figure 4) at 2.27 mg l 1 with a sensitivity of 72% and a specificity of 68%. Cox proportional hazards regression model after adjusting for hsCRP on admission, IL-6 on admission, age, Journal of Human Hypertension Inflammation in atrial fibrillation due to hypertension I Rizos et al 452 Table 2 Clinical characteristics of cardioverted patients (n ¼ 61) according to AF recurrence during the 1-year follow-up No recurrence Recurrence Pa Number of patients (%) Age, mean±s.d. (years) Male/female Smoking, n (%) Diabetes, n (%) AF episodes in the last year, median (IQR) AF duration from onset of symptoms until admission (h), median (IQR) Time elapsed from admission until cardioversion (h), median (IQR) Total time from onset of symptoms until cardioversion (h), median (IQR) Systolic blood pressure on admission (mm Hg), median (IQR) Diastolic blood pressure on admission (mm Hg), median (IQR) Heart rate on admission (beats per minute), mean±s.d. 33.0 (54.1) 66.00±9.02 22/11 9.0 (27.3) 3.0 (9.1) 1.0 (1.0–2.0) 3.0 (2.0–6.0) 10.0 (6.0–13.5) 14.0 (10.5–24.5) 135.0 (120.0–150.0) 80.0 (75.0–80.0) 120.89±28.51 28.0 (45.9) 69.48±10.23 12/16 5.0 (17.9) 4.0 (14.3) 3.0 (2.0–4.0) 4.0 (2.0–10.0) 12.0 (5.0–24.0) 21.0 (10.75–31.25) 120 (114.8–148.8) 80.0(75.0–84.38) 126.04±17.32 — 0.09 0.06 0.38 0.46 0.006 0.18 0.35 0.132 0.68 0.37 0.43 Echocardiographic parameters Left ventricular ejection fraction (%), median (IQR) Left atrial volume (ml), median (IQR) 60.0 (57.0–63.5) 49.00 (40.65–62.00) 60.0 (55.0–65.0) 48.00 (45.00–67.00) 0.70 0.54 Inflammatory markers Admission hsCRP (mg l 1), median (IQR) Admission IL-6 (pg ml 1), median (IQR) Admission TNF-a (pg ml 1), median (IQR hsCRP at 48 h (mg l 1), median (IQR) IL-6 (pg ml 1) at 48 h, median (IQR) TNF-a (pg ml 1) at 48 h, median (IQR) 1.71 4.21 13.00 1.72 4.69 10.70 4.13 4.80 12.60 3.50 4.44 14.70 0.02 0.04 0.81 0.004 0.70 0.33 (1.06–2.60) (2.71–5.04) (7.52–15.92) (1.07–3.64) (3.86–7.96) (8.63–16.45) (1.33–13.90) (3.78–10.40) (5.29–24.00) (2.03–9.95) (3.18–8.15) (9.07–20.83) Abbreviations: AF, atrial fibrillation; hsCRP, high sensitivity C-reactive protein; IL-6, interleukin-6; IQR, inter-quartile range; TNF, tumour necrosis factor. Bold values stand for statistically significant differences. a P for the comparison between patients with and without AF recurrence. Figure 3 Serum levels of hsCRP, IL-6 and TNF-a on admission and at 48 h in patients with and patients without recurrence of AF during follow-up. The black and white bars show median values. The P-value is shown only for statistically significant comparisons. gender, PAF episodes in the last year, left atrial volume, ejection fraction, diabetes, smoking and left atrial volume showed that this cut-off value of hsCRP at 48 h was the most powerful marker associated with AF recurrence (hazard ratio: 6.2, CI: 2.2–17.6, P ¼ 0.001). Kaplan–Meier survival analysis revealed that 74% of patients with hsCRP at 48 h o2.27 mg l 1 maintained sinus rhythm at the end of 1-year follow-up, whereas the corresponding percentage for patients with hsCRP at 48 h X2.27 mg l 1 value was only 33.3%, that is 2.2 times lower (Figure 5). The PLR and NLR for this cut-off value were 2.14 and 0.42, respectively, and by Journal of Human Hypertension considering that pre-test recurrence probability for 1 year after cardioversion is up to 75%,20 the posttest probability for PAF relapse in 1 year after cardioversion for patients with hsCRP at 48 h X2.27 mg l 1 is about 88%. Discussion The relationship between inflammation and maintenance of AF in a population restricted to subjects with hypertension is, for the first time in the literature, presented in this study. Several studies Inflammation in atrial fibrillation due to hypertension I Rizos et al 453 Figure 4 Receiver operating characteristic (ROC) curve for hsCRP serum levels at 48 h as predictors to AF recurrence within 1 year after admission. Figure 5 Kaplan–Meier survival estimates for the cumulative risk of AF recurrence during follow-up in two groups of patients divided by the optimal cut-off value for hsCRP at 48 h. have shown that higher baseline levels of hsCRP are associated with failure of cardioversion and with higher AF relapse rates.6,7,10,19,21–23 Others, however, do not prove that hsCRP is a potent determinant of either cardioversion outcome or AF recurrence.8,24,25 The divergence in these results could be attributed, at least in part, to the significant variability of the precipitating cardiovascular substrate (including heart failure, valvular heart disease, cardiomyopathies) and AF type (paroxysmal, persistent or permanent). In fact, it appears that AF duration and the type of underlying heart disease are significant confounding factors that may alter the factual relation of inflammatory markers with AF cardioversion and recurrence likelihood. Acknowledging the above, we attempted to assemble a more homogeneous population, by enroling only patients with hypertension and preserved cardiac function. In addition, our study is the first to show that baseline hsCRP failed to predict cardioversion outcome in patients with PAF and essential hypertension. Rather, increased levels of hsCRP and IL-6 measured 48 h after admission strongly correlated with failure of cardioversion. This suggests that people sustaining an inflammatory process for a more prolonged period have increased likelihood of unsuccessful cardioversion. On the other hand, levels of hsCRP—both on admission and at 48 h—and IL-6 baseline levels were significantly higher in cardioverted patients who had PAF recurrence during 1-year follow-up. Moreover, in the multivariate analysis, hsCRP at 48 h after admission was the most powerful factor associated with PAF relapse. Our results suggest that hsCRP levels shortly after cardioversion with amiodarone could be useful in identifying patients with higher propensity for early AF recurrence. Recently, other similar studies that included patients with PAF have shown that pre-cardioversion CRP is an important predictor for cardioversion outcome and maintenance of sinus rhythm after successful cardioversion.10,19 The usage of a non-ultrasensitive assay to measure CRP serum levels in these studies may have led to an overstatement of the importance of these associations. To further corroborate this, a recent meta-analysis investigating the role of CRP in restoration of sinus rhythm and AF recurrence revealed a great heterogeneity in the analysed studies and regarded differences in the used CRP assay as an important source for this heterogeneity.26 CRP is a circulating acute-phase reactant considered to be the prototypic downstream marker of inflammation. CRP is synthesized by the liver primarily in response to IL-6 and reaches its highest levels 48–72 h after the initiation of the inflammatory process.27 Regarding the relation of CRP with AF, it is unclear whether acute CRP elevations may be related to the AF incidence or maintenance. The former appears to be true in cardiac surgeryinduced AF, in which the peak increase of hsCRP concentration is noticed 48–72 h after cardiac surgery and coincides with the peak AF occurrence.28 The additional second measurement of the inflammatory markers, 48 h after the initial assessment on admission, was used in our study in an attempt to elucidate the dynamics of such complex inflammatory associations. Provided that our patients had AF duration of up to 24 h on admission, it can be ascertained that the second measurement at 48 h after admission was actually taken in all patients inside the interval of 48–72 h after the onset of AF. Thus, hsCRP measured at 48 h after admission for PAF may offer a more explicit expression of the dynamic progress of the suspected initial inflammatory event compared with its corresponding value on admission. We have already shown that IL-2 represents a potent trigger of this process, as it depicts its highest concentration closer Journal of Human Hypertension Inflammation in atrial fibrillation due to hypertension I Rizos et al 454 to AF onset and mainly in AF subjects with increased hsCRP values at 48 h.17 Interestingly, hsCRP exhibits a rising tendency in the overwhelming majority of the non-cardioverted patients in the first 48 h, whereas it remains almost stable in the other group (Figure 2). The concurrent progression of acute inflammatory response and the propensity of AF to persist were probably further reinforced by the positive association of hsCRP at 48 h with the total duration of AF. Furthermore, the two cases in the cardioverted group, in which a marked rise in hsCRP levels after 48 h was noted, also showed quick AF relapse into the first 2 months after successful cardioversion. In support to our observations, a recent study of patients with persistent, nonvalvular AF and a homogeneous disease substrate has shown a gradual decrease in serum hsCRP within 1 month after successful cardioversion, probably conveying the waning of the initial inflammatory activity.23 Our study provides an hsCRP cut-off value of 2.27 mg l 1, assessed at 48–72 h after AF onset, as a practical measure for the clinician treating PAF. Higher values indicated an approximately 6.2 times greater risk for arrhythmia recurrence in the first year after successful cardioversion, whereas 74% of patients, which had lower values, remained in sinus rhythm during the 1-year follow-up. Moreover, after analysis with the likelihood ratio form of the Bayes’ theorem, the post-test probability for AF recurrence in patients with hsCRP at 48 h X2.27 was approximately 88% within 1 year after successful cardioversion. The observed cut-off value is remarkably higher in comparison to the one referred by Watanabe et al.22 (0.6 mg l 1), whereas the recurrence rate is lower in our study (46 versus 76%). It is likely, that the enrolment of patients with more severe and variable cardiovascular disease substrate (that involved heart failure, coronary artery disease, valvular heart disease and cardiomyopathies), might have significantly increased the propensity for AF relapse in the study by Watanabe et al. Furthermore, the probable increased structural and electrophysiological deterioration of the heart under such conditions might require a less intense inflammatory trigger for AF to recur. Finally, the immense variability regarding the duration of AF (range 1–265 days) in that study could also justify the aforementioned differences, as increased AF duration promotes structural and electrophysiological changes that potentially contribute to the persistence of AF.29 The impact of IL-6 and TNF-a on AF also remains questionable. An earlier study of patients with persistent AF who underwent direct current cardioversion failed to show any significant relation between baseline IL-6 and the incidence or cardioversion outcome of AF.21 On the contrary, another study revealed increased IL-6 serum levels in patients with AF.30 Given that IL-6 is considered to be a representative mediator of endothelial inflamJournal of Human Hypertension mation, this result could be explained in part by the effect of underlying comorbidities, such as coronary artery disease, rather than a direct association with AF.30 Another study showed higher levels of IL-6 and TNF-a in patients with PAF compared to healthy volunteers;11 however, the population sample was too small to substantiate any safe conclusions. In our study, pre-cardioversion serum levels of IL-6 were positively correlated with heart rate on admission, indicating that higher states of sympathetic activity may promote endothelial inflammation and thus elevated levels of IL-6. In addition, baseline IL-6 was also significantly associated with long-term AF recurrence in the univariate analysis. Nonetheless, after controlling for other covariates, baseline IL-6 lost the significance of the association with AF relapse, which may be attributed to the small sample size of our study. With regards to TNF-a, we did not find any important associations with the cardioversion outcome and long-term maintenance of sinus rhythm except for a weak correlation of TNF-a at 48 h with baseline systolic and diastolic blood pressure values. These findings may be compatible with a focal inflammatory progression in fibrillating atria that is not capable of inducing significant changes in serum TNF-a. Indeed, TNF-a represents a mediator of systemic inflammation, especially in conditions with extensive immune disorders such as autoimmune diseases, sepsis, heart failure, ischemiareperfusion injury and cardiac allograft rejection.31 All such conditions, however, met the exclusion criteria of our study. Earlier clinical and experimental studies have linked atrial structural and electrical remodelling to the maintenance of AF.32,33 Inflammatory mediators (IL-1b, IL-6 and TNF-a) can directly affect the cardiac structural integrity by decreasing collagen synthesis and procollagen mRNA expression in cardiomyocytes and increasing the breakdown of collagen by enhanced matrix metalloproteinases activity.34–36 It has been recently shown that atrial structural remodelling, assessed by left atrial diameter, correlated positively with hsCRP levels in patients with PAF.23,37,38 Such a relation, however, was not supported by our results. A possible explanation could be the difference in the way we assessed left atrial size, by measuring left atrial volume instead of atrial diameter. Second, even though the positive relation between hsCRP and left atrial diameter was clearly illustrated in the univariate analysis, the above studies failed to reproduce this association in the multivariate model. Heart rate and blood pressure levels on admission had no association with cardioversion or recurrence outcome. However, any such conclusion regarding the aforementioned relations would be precarious, as AF with high ventricular response, observed commonly in PAF and in our study cohort, may significantly alter sympathetic nerve activity and thus affect blood pressure and heart rate levels.39 Inflammation in atrial fibrillation due to hypertension I Rizos et al 455 Study limitations This study shows a dynamic progression of the inflammatory process in PAF patients with hypertension who failed to restore sinus rhythm; however, it uses only two early time points for the assessment of this effect. The alteration in the levels of inflammatory markers could be illustrated more obviously with further measurements in more instances or before the onset of PAF. Unfortunately, the latter could not be assessed because our patients were urgently admitted to the hospital and the time of AF onset could not be predicted. As a result of this, we were not able to have measurements of inflammatory markers before the onset of AF in subjects with such previous history. Furthermore, the small number of evaluated patients did not allow us to clearly elucidate the effect of other factors such as gender or age in PAF recurrences, for example female and older patients were more likely to recur. The study, however, was restricted to a homogeneous group of subjects with hypertension and no evidence of other heart disease, thus limiting the confounding effects of factors such as left ventricular dysfunction, known to affect atrial integrity and inflammatory indices. Finally, the majority of the subjects with sinus rhythm restoration were not under a constant treatment protocol after hospital discharge. Thus, we were not able to evaluate any effect of antiarrhythmic and/or other drug treatment on AF recurrence likelihood. Until further studies confirm these data, it is premature to extrapolate our results in patients with other forms of underlying cardiac diseases or other forms of the arrhythmia, like ‘lone’ AF or persistent AF. Finally, the pathophysiologic linkage between the development of AF and the elevation of inflammatory indices, especially hsCRP, as well as the source of these agents, has to be clarified. It is still ambiguous whether the elevation of inflammatory markers has a causal or a causative relation with AF, or, eventually, if it is the result of the underlying cardiac disease or another concurrent inflammatory process. In conclusion, our results suggest that in patients with preserved left ventricular injection fraction, essential hypertension and PAF, baseline hsCRP was not associated with cardioversion outcome. On the contrary, the corresponding levels 48 h after admission were the most influential factor correlated with cardioversion outcome and AF recurrence in 1-year follow-up. Thus, measurement of hsCRP shortly after cardioversion could serve as a potent marker of probable recurrence within the first year in hypertensive patients with PAF. This is indicative of an important role of sustained inflammatory process and validation of these observations in patients with distinct and homogeneous cardiac substrates is necessary. Furthermore, our findings are suggestive of a potential therapeutic benefit, after an aggressive anti-inflammatory treatment, for example with anti-inflammatory agents,19 involving the improvement of AF cardioversion outcome and the reduction of AF recurrence rate in patients with arterial hypertension. The risks and benefits of such an approach should be further studied in carefully designed clinical trials. What is known about this topic K Baseline levels of serum high sensitivity C-reactive protein (hsCRP) have been correlated with incidence and maintenance of AF. K Several reports failed to show such a relation because of the confounding effect of numerous, underlying cardiovascular disorders (coronary artery disease, heart failure, cardiomyopathies, valvular heart disease). K In the operating setting, early incidence of atrial fibrillation (AF) after heart surgery was accompanied by acute increase of postoperative hsCRP serum levels. What this study adds K This study is focused on the likelihood assessment of the cardioversion and recurrence of paroxysmal AF (PAF) in a strictly select group of patients, in which the only trigger of the arrhythmia is hypertension. K Baseline hsCRP levels failed to predict cardioversion outcome of PAF. A sustained inflammatory process, as this is mainly reflected by the acute elevation of hsCRP 48 after admission, appears to be the most important factor in determining the cardioversion outcome of PAF. K hsCRP levels, shortly after successful cardioversion of PAF, appear to be a better predictor of long-term AF recurrence compared with baseline levels, in patients with hypertension. Conflict of interest The authors declare no conflict of interest. References 1 Go AS, Hylek EM, Phillips KA, Chang Y, Henault LE, Selby JV et al. Prevalence of diagnosed atrial fibrillation in adults: national implications for rhythm management and stroke prevention: the AnTicoagulation and Risk Factors in Atrial Fibrillation (ATRIA) Study. JAMA 2001; 285(18): 2370–2375. 2 Engelmann MD, Svendsen JH. Inflammation in the genesis and perpetuation of atrial fibrillation. Eur Heart J 2005; 26(20): 2083–2092. 3 Boos CJ, Anderson RA, Lip GY. Is atrial fibrillation an inflammatory disorder? Eur Heart J 2006; 27(2): 136–149. 4 Chung MK, Martin DO, Sprecher D, Wazni O, Kanderian A, Carnes CA et al. C-reactive protein elevation in patients with atrial arrhythmias: inflammatory mechanisms and persistence of atrial fibrillation. Circulation 2001; 104(24): 2886–2891. 5 Aviles RJ, Martin DO, Apperson-Hansen C, Houghtaling PL, Rautaharju P, Kronmal RA et al. Inflammation as a risk factor for atrial fibrillation. Circulation 2003; 108(24): 3006–3010. 6 Malouf JF, Kanagala R, Al Atawi FO, Rosales AG, Davison DE, Murali NS et al. High sensitivity C-reactive protein: a novel predictor for recurrence Journal of Human Hypertension Inflammation in atrial fibrillation due to hypertension I Rizos et al 456 7 8 9 10 11 12 13 14 15 16 17 18 19 20 of atrial fibrillation after successful cardioversion. J Am Coll Cardiol 2005; 46(7): 1284–1287. Wazni O, Martin DO, Marrouche NF, Shaaraoui M, Chung MK, Almahameed S et al. C reactive protein concentration and recurrence of atrial fibrillation after electrical cardioversion. Heart 2005; 91(10): 1303–1305. Cosgrave J, Foley JB, Bahadur K, Bennett K, Crean P, Walsh MJ. Inflammatory markers are not associated with outcomes following elective external cardioversion. Int J Cardiol 2006; 110(3): 373–377. Ellinor PT, Low A, Patton KK, Shea MA, MacRae CA. C-Reactive protein in lone atrial fibrillation. Am J Cardiol 2006; 97(9): 1346–1350. Dernellis J, Panaretou M. C-reactive protein and paroxysmal atrial fibrillation: evidence of the implication of an inflammatory process in paroxysmal atrial fibrillation. Acta Cardiol 2001; 56(6): 375–380. Sata N, Hamada N, Horinouchi T, Amitani S, Yamashita T, Moriyama Y et al. C-reactive protein and atrial fibrillation. Is inflammation a consequence or a cause of atrial fibrillation? Jpn Heart J 2004; 45(3): 441–445. Lip GY, Felmeden DC, Li-Saw-Hee FL, Beevers DG. Hypertensive heart disease. A complex syndrome or a hypertensive ‘cardiomyopathy’? Eur Heart J 2000; 21(20): 1653–1665. Fuster V, Ryden LE, Asinger RW, Cannom DS, Crijns HJ, Frye RL et al. ACC/AHA/ESC guidelines for the management of patients with atrial fibrillation: executive summary. A report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines and the European Society of Cardiology Committee for Practice Guidelines and Policy Conferences (committee to develop guidelines for the management of patients with atrial fibrillation): developed in collaboration with the North American Society of Pacing and Electrophysiology. J Am Coll Cardiol 2001; 38(4): 1231–1266. Pritchett AM, Jacobsen SJ, Mahoney DW, Rodeheffer RJ, Bailey KR, Redfield MM. Left atrial volume as an index of left atrial size: a population-based study. J Am Coll Cardiol 2003; 41(6): 1036–1043. Wang Y, Gutman JM, Heilbron D, Wahr D, Schiller NB. Atrial volume in a normal adult population by twodimensional echocardiography. Chest 1984; 86(4): 595–601. Sookoian S, Castano G, Burgueno AL, Gianotti TF, Rosselli MS, Pirola CJ. A diagnostic model to differentiate simple steatosis from nonalcoholic steatohepatitis based on the likelihood ratio form of Bayes theorem. Clin Biochem 2009; 42(7–8): 624–629. Rizos I, Tsiodras S, Rigopoulos AG, Dragomanovits S, Kalogeropoulos AS, Papathanasiou S et al. Interleukin2 serum levels variations in recent onset atrial fibrillation are related with cardioversion outcome. Cytokine 2007; 40(3): 157–164. Chevalier P, Durand-Dubief A, Burri H, Cucherat M, Kirkorian G, Touboul P. Amiodarone versus placebo and classic drugs for cardioversion of recent-onset atrial fibrillation: a meta-analysis. J Am Coll Cardiol 2003; 41(2): 255–262. Dernellis J, Panaretou M. Relationship between Creactive protein concentrations during glucocorticoid therapy and recurrent atrial fibrillation. Eur Heart J 2004; 25(13): 1100–1107. Naccarelli GV, Wolbrette DL, Khan M, Bhatta L, Hynes J, Samii S et al. Old and new antiarrhythmic drugs for Journal of Human Hypertension 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 converting and maintaining sinus rhythm in atrial fibrillation: comparative efficacy and results of trials. Am J Cardiol 2003; 91(6A): 15D–26D. Conway DS, Buggins P, Hughes E, Lip GY. Predictive value of indexes of inflammation and hypercoagulability on success of cardioversion of persistent atrial fibrillation. Am J Cardiol 2004; 94(4): 508–510. Watanabe E, Arakawa T, Uchiyama T, Kodama I, Hishida H. High-sensitivity C-reactive protein is predictive of successful cardioversion for atrial fibrillation and maintenance of sinus rhythm after conversion. Int J Cardiol 2006; 108(3): 346–353. Kallergis EM, Manios EG, Kanoupakis EM, Mavrakis HE, Kolyvaki SG, Lyrarakis GM et al. The role of the post-cardioversion time course of hs-CRP levels in clarifying the relationship between inflammation and persistence of atrial fibrillation. Heart 2008; 94(2): 200–204. Korantzopoulos P, Kolettis TM, Kountouris E, Siogas K, Goudevenos JA. Variation of inflammatory indexes after electrical cardioversion of persistent atrial fibrillation. Is there an association with early recurrence rates? Int J Clin Pract 2005; 59(8): 881–885. Aguiar-Souto P, Toquero-Ramos J, Fernandez-Lozano I. Discordant regulation of CRP and NT-proBNP plasma levels after electrical cardioversion of persistent atrial fibrillation. Pacing Clin Electrophysiol 2006; 29(12): 1452. 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(15): 1642–1648. Gabay C, Kushner I. Acute-phase proteins and other systemic responses to inflammation. N Engl J Med 1999; 340(6): 448–454. Bruins P, te Velthuis H, Yazdanbakhsh AP, Jansen PG, van Hardevelt FW, de Beaumont EM et al. Activation of the complement system during and after cardiopulmonary bypass surgery: postsurgery activation involves C-reactive protein and is associated with postoperative arrhythmia. Circulation 1997; 96(10): 3542–3548. Allessie MA, Konings K, Kirchhof CJ, Wijffels M. Electrophysiologic mechanisms of perpetuation of atrial fibrillation. Am J Cardiol 1996; 77(3): 10A–23A. Roldan V, Marin F, Blann AD, Garcia A, Marco P, Sogorb F et al. Interleukin-6, endothelial activation and thrombogenesis in chronic atrial fibrillation. Eur Heart J 2003; 24(14): 1373–1380. Meldrum DR. Tumor necrosis factor in the heart. Am J Physiol 1998; 274(3 Pt 2): R577–R595. Allessie MA, Boyden PA, Camm AJ, Kleber AG, Lab MJ, Legato MJ et al. Pathophysiology and prevention of atrial fibrillation. Circulation 2001; 103(5): 769–777. Wijffels MC, Kirchhof CJ, Dorland R, Allessie MA. Atrial fibrillation begets atrial fibrillation. A study in awake chronically instrumented goats. Circulation 1995; 92(7): 1954–1968. Siwik DA, Chang DL, Colucci WS. Interleukin-1beta and tumor necrosis factor-alpha decrease collagen synthesis and increase matrix metalloproteinase activity in cardiac fibroblasts in vitro. Circ Res 2000; 86(12): 1259–1265. Sarkar S, Vellaichamy E, Young D, Sen S. Influence of cytokines and growth factors in ANG II-mediated Inflammation in atrial fibrillation due to hypertension I Rizos et al 457 collagen upregulation by fibroblasts in rats: role of myocytes. Am J Physiol 2004; 287(1): H107–H117. 36 Li YY, McTiernan CF, Feldman AM. Proinflammatory cytokines regulate tissue inhibitors of metalloproteinases and disintegrin metalloproteinase in cardiac cells. Cardiovasc Res 1999; 42(1): 162–172. 37 Watanabe T, Takeishi Y, Hirono O, Itoh M, Matsui M, Nakamura K et al. C-Reactive protein elevation predicts the occurrence of atrial structural remodeling in patients with paroxysmal atrial fibrillation. Heart Vessels 2005; 20(2): 45–49. 38 Psychari SN, Apostolou TS, Sinos L, Hamodraka E, Liakos G, Kremastinos DT. Relation of elevated C-reactive protein and interleukin-6 levels to left atrial size and duration of episodes in patients with atrial fibrillation. Am J Cardiol 2005; 95(6): 764–767. 39 Wasmund SL, Li JM, Page RL, Joglar JA, Kowal RC, Smith ML et al. Effect of atrial fibrillation and an irregular ventricular response on sympathetic nerve activity in human subjects. Circulation 2003; 107(15): 2011–2015. Journal of Human Hypertension