Author + information
- Received April 23, 2007
- Revision received July 16, 2007
- Accepted July 24, 2007
- Published online January 1, 2008.
- Young M. Kim, BSc⁎,1,
- Hassan Kattach, MRCS†,
- Chandi Ratnatunga, FRCS†,
- Ravi Pillai, FRCS†,
- Keith M. Channon, MD, FRCP⁎,2 and
- Barbara Casadei, MD, DPhil, FRCP⁎,2,⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. Barbara Casadei, University Department of Cardiovascular Medicine, John Radcliffe Hospital, Oxford, OX3 9DU, United Kingdom.
Objectives Our goal was to evaluate the role of myocardial nicotinamide adenine dinucleotide phosphate (NADPH) oxidase activity and plasma markers of oxidative stress in the pathogenesis of post-operative atrial fibrillation (AF).
Background Atrial fibrillation is a common complication of cardiac surgery, leading to increased morbidity and prolonged hospitalization. Experimental evidence suggests that oxidative stress may be involved in the pathogenesis of AF; however, the relevance of this putative mechanism in patients undergoing cardiac surgery is unclear.
Methods We measured basal and NADPH-stimulated superoxide production in right atrial appendage samples from 170 consecutive patients undergoing conventional coronary artery bypass surgery. Plasma markers of lipid and protein oxidation (thiorbabituric acid-reactive substances, 8-isoprostane, and protein carbonyls) were also measured in blood samples drawn from a central line before surgery and after reperfusion.
Results Patients who developed AF after surgery (42%) were older and had a significantly increased atrial NADPH oxidase activity than patients who remained in sinus rhythm (SR) (in relative light units/s/μg protein: 4.78 ± 1.44 vs. 3.53 ± 1.04 in SR patients, p < 0.0001). Plasma markers of lipid and protein oxidation increased significantly after reperfusion; however, neither pre-operative nor post-operative measurements differed between patients who developed AF and those who remained in SR after surgery. Multivariate analysis identified atrial NADPH oxidase activity as the strongest independent predictor of post-operative AF (odds ratio 2.41; 95% confidence interval 1.71 to 3.40, p < 0.0001).
Conclusions Atrial NADPH oxidase activity is independently associated with an increased risk of post-operative AF, suggesting that this oxidase system may be a key mediator of atrial oxidative stress leading to the development of AF after cardiac surgery.
Atrial fibrillation (AF) is a frequent complication of most types of cardiac surgery, occurring in 35% to 50% of patients within the first 2 to 5 post-operative days (1,2). Post-operative AF can induce hemodynamic deterioration, increase the risk of stroke, and is invariably associated with prolonged hospitalization and a significant additional cost to patient care (1,2). Little is known of the mechanisms underlying post-operative AF, although emerging evidence indicates that inflammation and oxidative stress associated with cardiac surgery and cardiopulmonary bypass (CPB) may play an important role (3–6). It has recently been reported that myocardial oxidative injury leads to impaired atrial contraction, altered myofibrillar energetics (7), and a reduction in the atrial effective refractory period (AERP) (8), which, in turn, increases the vulnerability to AF (9). Importantly, short-term treatment with high doses of antioxidant and anti-inflammatory agents attenuates AERP shortening and the vulnerability to AF in animal models of rapid atrial pacing (8,10,11), suggesting that atrial electrophysiological remodeling leading to AF may be prevented by inhibiting the myocardial formation of reactive oxygen species (ROS).
An increasing body of experimental evidence indicates that formation of the ROS, superoxide, by cardiovascular nicotinamide adenine dinucleotide phosphate (NADPH) oxidases plays a critical role in the development of hypertension, myocardial hypertrophy, and heart failure (12). We have recently reported that human atrial myocytes have the capacity to produce ROS via a phagocytic-type NADPH oxidase, and that NADPH-stimulated superoxide production is increased in patients with sustained or paroxysmal AF (13). Nicotinamide adenine dinucleotide phosphate oxidase activity is potently stimulated by cytokines (12), suggesting that this oxidase system may be an important link between the systemic inflammatory response to cardiac surgery and myocardial oxidative stress.
Taken together, these data suggest that atrial NADPH oxidase activity may be a more sensitive predictor of ROS-mediated myocardial electrophysiological changes leading to post-operative AF than plasma markers of oxidative stress. To test this hypothesis, we prospectively evaluated the relationship between ROS production in the right atrial myocardium, plasma markers of protein and lipid oxidation, and the occurrence of post-operative AF in 170 consecutive patients who underwent first-time elective coronary artery bypass grafting (CABG) surgery.
One hundred seventy consecutive patients undergoing first-time elective CABG surgery were recruited in this study. Exclusion criteria were a preceding history of AF or other atrial arrhythmias, concomitant valvular disease, and the use of anti-arrhythmic drugs other than beta-blockers. The demographic and clinical characteristics of the patients are shown in Table 1. Patients were asked to withdraw treatment with aspirin 3 days before surgery. Similar surgical and anesthetic protocols were followed in all patients. All CABG procedures were performed using the on-pump technique. Post-operatively, heart rate and rhythm were continuously monitored for the first 48 h and at 4-h intervals thereafter for 5 days. Post-operative AF was defined as the characteristic arrhythmia lasting for at least 30 s. The study was approved by the local research ethics committee, and all patients gave written informed consent.
Measurements of superoxide production in right atrial appendage (RAA) homogenates
A sample of the RAA was obtained in all patients at the time of venous cannulation, before the initiation of CPB. Superoxide production was measured in homogenized RAA tissue by lucigenin-enhanced chemiluminescence using a single tube luminometer (Berthold FB12, Berthold, Pforzheim, Germany) modified to maintain the sample temperature at 37oC, as described and validated previously (13). Briefly, basal chemiluminescence from RAA homogenates was measured after subtraction of the background chemiluminescence of the lucigenin (5 μmol/l)-containing buffer. Further measurements were taken after application of NADPH (100 μmol/l). A single investigator blinded to the patients’ post-operative outcome performed all measurements and analyses. Three basal and NADPH-stimulated measurements of superoxide production were obtained and averaged for each patient.
Because it is well-established that high concentrations of lucigenin ≥20 μmol/l) are subject to redox cycling, we (13) have previously validated basal and NADPH-stimulated superoxide measurements in human atrial samples obtained by using 5 μmol/l lucigenin with those obtained by using the luminol analogue L-012 (in the presence of ebselen) and the superoxide dismutase-inhibitable ferricytochrome c reduction, respectively. In both cases measurements were closely correlated with those obtained in parallel by lucigenin-enhanced chemiluminescence (basal measurements: r = 0.88, p < 0.005, n = 8; NADPH-stimulated: r = 0.99, p < 0.0001, n = 11). Specificity of this measurement for superoxide was confirmed by data showing near abolition of the chemiluminescence signal in the presence of specific scavengers such as superoxide dismutase or tiron both in right atrial homogenates and in isolated atrial myocytes (13).
Plasma markers of oxidative stress
Fasting blood samples were drawn from a central line immediately after the induction of anesthesia and 10 min after the administration of protamine in 124 patients.
Thiorbabituric acid-reactive substances (TBARS), a marker of lipid peroxidation, were measured in plasma using a previously described fluorometric technique (14,15). Coefficients of variation for within and between runs were 4% and 7%, respectively. Each sample was run in triplicate. Hemodilution in post-operative samples was adjusted using a correction factor as described previously (16). Plasma protein carbonyl levels (a marker of protein oxidation) were evaluated using an enzyme-linked immunoadsorbent assay method (17) and 8-isoprostane levels using a commercial enzyme immunoassay kit (Cayman Chemical, Ann Arbor, Michigan). Carbonyl content was calculated from the standard curve (prepared using oxidized and reduced bovine serum albumin) and expressed as nmol carbonyl per mg of immunoglobulin. Adjusted (16) total (free and esterified) 8-isoprostane levels in purified plasma samples taken after CPB and reperfusion are given in pg/ml. All measurements were carried out by a single investigator blinded to the patients’ post-operative outcome.
Differences between patients who developed AF and those who maintained sinus rhythm (SR) in the post-operative period were evaluated with a t test or Wilcoxon rank sum test for continuous variables and a chi-square test for categorical variables. To identify the subset of variables that predicted the occurrence of post-operative AF, we entered our variables in a logistic regression model using block entry of variables (SPSS 13.0 for Windows, SPSS Inc., Cary, North Carolina). Criteria for entry and exclusion from this multivariate analysis were p < 0.05 and p ≥ 0.10, respectively; however, variables had been shown to affect the occurrence of post-operative AF (i.e., use of beta-blockers or inhibitors of the renin-angiotensin system  or the activity of the cardiovascular NADPH oxidase, i.e., diabetes mellitus  and treatment with statins ) were also entered into the model. Only main effects (i.e., no interaction terms) were fitted in this model. The association of atrial NADPH oxidase activity with post-operative AF was also assessed by categorization of the patients’ cohort into NADPH oxidase activity quartiles. Data are presented as mean ± SD or as percent of the total (%).
Seventy-one patients (42%) developed AF after CABG surgery, and, as reported previously (1), the occurrence of AF was most common on the second and third post-operative days. No significant differences between groups were found with respect to treatment, cardiovascular risk factors and co-morbidity, duration of CPB or aortic cross-clamp, and number of grafted vessels (Table 1). However, patients who developed post-operative AF were slightly older and had a higher basal and NADPH-stimulated superoxide production than those who remained in SR (Fig. 1). In-hospital post-operative treatment with beta-blockers was initiated in a similar percentage of patients from each group (i.e., in 63% of SR patients and in 69% of patients who developed AF, p = 0.43); however, these agents tended to be started slightly later in patients who developed AF after surgery (Table 1). There was no difference in the post-operative use of angiotensin-converting enzyme inhibitors or angiotensin II receptor blockers (i.e., these agents were initiated in 67% of patients who remained in SR and 68% of patients who developed AF, p = 0.41) or in the time of initiation of treatment after surgery (Table 1).
Plasma levels of TBARS and protein carbonyls increased significantly after reperfusion (Fig. 2), reflecting the increase in systemic oxidative stress associated with cardiac surgery. However, neither pre-operative nor post-operative plasma levels of TBARS or protein carbonyl differed between patients who developed AF and those who remained in SR after surgery (Fig. 2). Equally, the perioperative increase in these plasma markers of oxidative stress was not significantly associated with the occurrence of post-operative AF (p = 0.31 for TBARS and p = 0.45 for protein carbonyls). Similar findings were obtained when post-operative plasma levels of 8-isoprostane were compared (28.5 ± 2.9 pg/ml in patients who developed AF vs. 28.0 ± 3.8 pg/ml in patients who remained in SR, p = 0.86). There was no correlation between myocardial basal or NADPH-stimulated superoxide production and plasma markers of oxidative stress (not shown).
Atrial NADPH-stimulated superoxide release increased with age (r = 0.20, p = 0.009) but was not significantly affected by pre-operative treatment with statins or inhibitors of the angiotensin-converting enzyme/angiotensin II receptors (p = 0.36 and p = 0.81, respectively), by left ventricular ejection fraction (LVEF) (p = 0.52), or by the diagnosis of diabetes mellitus (in relative light units [RLU]/s/μg: 4.23 ± 1.62 in patients with diabetes vs. 3.99 ± 1.28 in nondiabetic patients, n = 43 and 127, respectively, p = 0.33) or history of congestive heart failure (p = 0.89). In contrast, basal superoxide production from the RAA increased with the severity of left ventricular dysfunction: a) 0.064 ± 0.03 RLU/s/μg in patients with an LVEF >50%; b) 0.072 ± 0.03 RLU/s/μg in patients with an LVEF between 30% to 50%; and c) 0.089 ± 0.05 RLU/s/μg in patients with an LVEF <30%, n = 83, 65, and 22, respectively; p = 0.002 for a vs. c and p = 0.04 for b vs. c) and was significantly higher in patients with a clinical history of congestive heart failure (0.068 ± 0.03 RLU/s/μg vs. 0.086 ± 0.05 RLU/s/μg in patients with a history of congestive heart failure, n = 149 and 21, respectively, p = 0.02).
Multivariate analysis applied in a model-selection procedure identified myocardial NADPH-stimulated superoxide production as the most important independent predictor of post-operative AF in our patients (odds ratio 2.41; 95% confidence interval 1.71 to 3.40, p < 0.0001) (Table 2); age was no longer associated with an increased risk of post-operative AF after the addition of NADPH-stimulated superoxide production to the model. Characterization of our population according to quartiles of atrial NADPH-stimulated superoxide confirmed a strong association between these measurements and the occurrence of post-operative AF (Table 3).
This is the first prospective study to compare the potential value of myocardial and systemic markers of oxidative stress in predicting the occurrence AF after CABG surgery. Our findings clearly show that NADPH-stimulated superoxide production in the RAA is independently associated with an increased risk of post-operative AF. Plasma markers of oxidative stress increased significantly after CPB and reperfusion; however, neither pre-operative nor post-reperfusion levels of plasma TBARS, protein carbonyl, or 8-isoprostane discriminated between patients who developed AF and those who maintained SR in the post-operative period. These findings indicate that ROS production in the atrial myocardium may be a more accurate indicator of myocardial oxidative stress and its functional correlates (e.g., reduction in AERP and arrhythmogenesis) (8,10,11) than systemic markers of oxidative stress and suggest that targeting tissue-specific sources of superoxide production may prove a more successful therapeutic strategy than systemic antioxidant interventions.
Oxidative stress and AF
Myocardial oxidative damage is associated with atrial electrophysiological remodeling and higher inducibility of AF in animal models of rapid atrial pacing, both of which are prevented by treatment with high doses of antioxidant or anti-inflammatory agents (8,10,11). These findings are not easily reconciled with the disappointing results of antioxidant vitamins trials in humans. Poor patient selection and monitoring of the antioxidant effects of treatment have been implicated in the therapeutic failure of antioxidant agents in humans (20). However, it is equally plausible that interventions that are at least partially effective in reducing plasma markers of lipid and protein oxidation in humans may not lead to a significant reduction in oxidative injury in the tissue of interest. This hypothesis is supported by increasing evidence indicating that both myocardial and vascular tissue contain constitutively active and highly modulated oxidase systems and that local ROS production plays an important role in the pathogenesis of common myocardial disease states (12). Importantly, controlled clinical trials have recently demonstrated that interventions that prevent activation of NADPH oxidases by vasoactive hormones or inflammatory cytokines (i.e., angiotensin II receptor blockers , statins , or steroids ) are also highly effective in preventing the occurrence of AF in hypertensive subjects (23) and in patients after cardiac surgery (24,25).
Sources of ROS in the human myocardium
Nicotinamide adenine dinucleotide phosphate oxidase–dependent ROS production in the myocardium or vessel wall has been implicated in the pathogenesis of a wide array of cardiovascular diseases (12), including paroxysmal or chronic AF (13). As excess ROS production may lead both to nitric oxide (NO) scavenging (and the production of the powerful pro-oxidant molecule peroxynitrite) and to ‘uncoupling’ of NO synthases (which causes these enzymes to produce superoxide rather than NO ), it remains unclear to which extent the pathophysiological correlates of increased myocardial superoxide production are due to direct oxidation of key proteins substrates or to a reduction in myocardial NO bioavailability.
We have previously reported that a membrane-bound NOX2-containing NADPH oxidase is the main source of ROS production in the human atrial myocardium and isolated atrial myocytes (13). Here we show that atrial NADPH oxidase activity is an independent predictor of the occurrence of post-operative AF. It could be argued that measurements in RAA samples obtained after reperfusion might have reflected perioperative myocardial oxidative injury more accurately. However, NADPH oxidases have not been convincingly implicated with the oxidative stress that accompanies ischemia and reperfusion (27). Indeed, our preliminary data show no significant difference in NADPH oxidase activity in RAA samples obtained before CPB and shortly after reperfusion in the same cohort of patients (unpublished data, Y. M. Kim, May 2005). In contrast, several vasoactive substances (e.g., angiotensin II, endothelin I) and inflammatory mediators (e.g., interleukin-6 and tumor necrosis factor-alpha) that are released in response to cardiac surgery and CPB are now known to be potent activators of cardiovascular NADPH oxidases (12). Thus, the perioperative inflammatory response may have a major role in stimulating atrial NADPH oxidase activity, which, in turn, would increase myocardial oxidative stress leading to electrophysiological changes (i.e., shortening of the AERP ) that facilitate the new onset of AF in the early post-operative period. In other words, atrial NADPH oxidase may serve as an important link between systemic inflammation and tissue oxidative injury. In agreement with this hypothesis, basal superoxide measurements did not differ between patients who maintained SR and those who developed AF after surgery. Whereas superoxide production in response to NADPH reflects the stimulated NADPH oxidase activity, basal superoxide release results from the background activity of several oxidase systems and as such is less likely to bear a relationship with the inflammation-driven post-operative myocardial oxidative stress.
We considered that a lower signal-to-noise ratio may have contributed to our inability to detect significant differences in basal superoxide measurements between SR and post-operative AF patients; however, the use of a single vial luminometer (modified to maintain our samples at a physiological temperature) increases the sensitivity of this technique and largely circumvents the problems in detecting low levels of superoxide that are encountered when using standard plate luminometers. Our observation that basal—but not NADPH-stimulated—atrial superoxide production increased proportionally with the severity of left ventricular dysfunction suggests that these measurements may reflect different pathophysiological mechanisms.
It is possible that atrial stretch and remodelling may affect the activity of NADPH oxidases, which, in turn, may contribute to the recently reported association between atrial volume and post-operative AF (28). Unfortunately, because the majority of patients who have cardiac surgery in our institution have their pre-operative assessment (e.g., coronary angiogram and echocardiography) elsewhere, we did not have access to accurate echocardiographic measurements of atrial volume; thus, the relationship between atrial size, NADPH oxidase activity, and post-operative AF could not be investigated.
Plasma markers of lipid and protein oxidation increased significantly shortly after reperfusion, in agreement with the notion that cardiac surgery and CPB increase systemic oxidative stress; however, neither marker predicted post-operative AF. More robust indexes of lipid peroxidation, such as the determination of urinary levels of F2-isoprostanes by gas chromatography/mass spectrometry (29), might have provided a more accurate assessment of systemic oxidative stress in our patients; however, performing these complex and time-consuming measurements on a large scale is impractical and unlikely to detect myocardial oxidative stress as accurately as measurements of atrial NADPH oxidase activity. In agreement with this hypothesis, we found no significant correlation between measurements of atrial superoxide production and plasma markers of oxidative stress (not shown).
We showed that atrial NADPH oxidase activity is independently associated with an increased risk of AF in patients undergoing conventional CABG surgery, whereas the post-operative rise in plasma markers of protein and lipid oxidation bears no association with this outcome. Our data suggest that preventing the activation of cardiovascular-specific oxidases may be more effective in averting myocardial oxidative injury and its functional consequences than the systemic administration of antioxidant vitamins. Testing this hypothesis may provide an important key for understanding the role of oxidative stress in human cardiovascular disease.
The authors would like to thank Dr. Philip E. James and Mrs. Joan Parton from the Wales Heart Research Institute, Cardiff University School of Medicine for helpful discussions on measurements of reactive species and oxidative stress and for carrying out the thiorbabituric acid-reactive substances, 8-isoprostane, and protein carbonyl measurements. The authors are grateful to Professor Martin Farrall (Department of Cardiovascular Medicine, University of Oxford) for his invaluable help with the statistical analysis of the data.
- Abbreviations and Acronyms
- atrial effective refractory period
- atrial fibrillation
- coronary artery bypass grafting
- cardiopulmonary bypass
- left ventricular ejection fraction
- nicotinamide adenine dinucleotide phosphate
- nitric oxide
- right atrial appendage
- relative light units
- reactive oxygen species
- sinus rhythm
- thiorbabituric acid-reactive substances
- Received April 23, 2007.
- Revision received July 16, 2007.
- Accepted July 24, 2007.
- American College of Cardiology Foundation
- Aranki S.F.,
- Shaw D.P.,
- Adams D.H.,
- et al.
- Mihm M.J.,
- Yu F.,
- Carnes C.A.,
- et al.
- Carnes C.A.,
- Chung M.K.,
- Nakayama T.,
- et al.
- Allessie M.,
- Ausma J.,
- Schotten U.
- Shiroshita-Takeshita A.,
- Schram G.,
- Lavoie J.,
- Nattel S.
- Shiroshita-Takeshita A.,
- Brundel B.J.J.M.,
- Lavoie J.,
- Nattel S.
- Griendling K.K.,
- Sorescu D.,
- Ushio Fukai M.
- Kim Y.M.,
- Zhang Y.H.,
- Guzik T.J.,
- et al.
- Wasowicz W.,
- Neve J.,
- Peretz A.
- Nightingale A.K.,
- James P.P.,
- Morris-Thurgood J.,
- et al.
- Guzik T.J.,
- Mussa S.,
- Gastaldi D.,
- et al.
- Maack C.,
- Kartes T.,
- Kilter H.,
- et al.
- Griendling K.K.,
- FitzGerald G.A.
- Wassmann S.,
- Laufs U.,
- Muller K.,
- et al.
- Marumo T.,
- Schini-Kerth V.B.,
- Brandes R.P.,
- Busse R.
- Wachtell K.,
- Lehto M.,
- Gerdts E.,
- et al.
- Patti G.,
- Chello M.,
- Candura D.,
- et al.
- Vasquez-Vivar J.,
- Kalyanaraman B.,
- Martasek P.,
- et al.
- Becker L.B.