Atrial fibrillation (AF) occurs in 20%–65% of patients after cardiac surgery requiring cardiopulmonary bypass (CPB) (1–5). AF may be a marker of postoperative complications, including stroke (6), congestive heart failure (4), and increased mortality (1,3,4,6,7). Clinical variables, such as advanced age, hypertension, male gender, and a remote history of previous AF (3), may predispose to postoperative AF; intraoperative surgical variables, including combined valve replacement/coronary artery bypass grafting (CABG) procedures and prolonged aortic cross-clamp and bypass times (2,8), are similarly associated with a more frequent AF incidence. A study of patients undergoing CABG with CPB found that a genetic polymorphism that predisposed to higher levels of interleukin (IL)-6 was more prevalent in patients with postoperative AF, suggesting that the perioperative induction of inflammation may play a role in AF development (9). There is additional evidence for this latter pathophysiology in the non-CPB AF setting; myocarditis precipitates lone episodes of AF (10), and histologic examination of atrial specimens in patients with chronic isolated AF have pointed to the infiltration of inflammatory cells in this pathology (11). We hypothesized that the acute leukocyte inflammatory response during cardiac surgery/CPB might be associated with postoperative AF. Thus, we performed a pilot study to determine if there was any association between perioperative monocyte and neutrophil activation and postoperative AF in cardiac surgery/CPB patients.
After Human Investigation Committee approval and informed consent, we enrolled 72 adults undergoing elective cardiac surgery requiring CPB at Yale-New Haven Hospital who were part of the Multicenter Study of Perioperative Ischemia Research Group study. All patients were in normal sinus rhythm at the time they underwent surgery. As previously described (12,13), all patients underwent CPB using a standardized membrane oxygenator, roller pump, and cardiotomy suction setup. None of the patients had a recent infection, as documented by the absence of fever and lack of positive tissue or blood cultures, and no patients were receiving antibiotics before surgery. Patients did not receive aprotinin or corticosteroids before, during, or after the surgery. Total perioperative red cell transfusion volume was recorded. Because of the relatively small number of patients and the duration of follow-up restricted to the hospital, we did not assess the duration of postoperative AF or its severity but simply occurrence or nonoccurrence.
Patients were continuously monitored by telemetry for 72 h after surgery. The development of postoperative AF was defined by the occurrence of an irregular rhythm between 100 and 160 bpm, the presence of F-waves on the electrocardiogram, and the confirmatory diagnosis of AF by the attending cardiologist.
All blood samples were drawn from the radial artery catheter. Perioperative blood samples to examine leukocyte activation were drawn into 1% paraformaldehyde fixative: (a) immediately before the start of surgery, (b) immediately before and (c) 10 min after aortic cross-clamp release, (d) on arrival in the intensive care unite (ICU), and (e) in the morning of postoperative Day 1. The overall activation response to surgery/CPB was determined by summing the data before and after cross-clamp release and on ICU arrival to calculate the area under the curve of the immediate perioperative period. The rationale for the baseline leukocyte value being drawn after the anesthesia induction but before the start of surgery was to have a starting value that was relatively independent of the individual response to anesthesia and any other preoperative variables. This baseline value would allow subsequent measures to primarily reflect the response to cardiac surgery/CPB.
Leukocyte activation by flow cytometry was examined, as previously reported (14), using monoclonal antibodies to CD45 and CD11b. CD45 is an antigen present on all leukocytes and was used primarily to exclude red blood cells and platelets during the cytometric analysis; CD45 has no utility as an inflammatory marker. By contrast, CD11b is an adhesion receptor that is upregulated on the surface of neutrophils (PMN) and monocytes after their activation and is an accepted marker of cellular (leukocyte) inflammation (15,16). Monocyte and PMN CD11b were expressed as a percentage of the individual patient’s baseline (17). In vitro agonist studies in our laboratory determined that this specific method for detecting CD11b upregulation was 100% sensitive to a ≥10% increase in CD11b surface density induced by either PMN or monocyte activation; the coefficient of variation (CV) of this assay for both leukocyte types was <4%.
Preoperative serum from baseline time point (a) was frozen for quantitative C-reactive protein (CRP) using the Beckman IMMAGE (high sensitivity) CRP immunochemistry reagent (Beckman-Coulter, Fullerton, CA); the CRP assay is sensitive to levels as low as 0.1 mg/L and to CRP changes of ≥0.2 mg/L. Perioperative plasma samples from time points (a), (c), and (e) were frozen for measurement of the released neutrophil inflammatory product myeloperoxidase (MPO) using the Bioxytech® MPO-EIA of OXIS International (Portland, OR). MPO is a granule component of PMN and is released from the cells after inflammatory stimuli that lead to degranulation. The MPO assay is sensitive to levels as low as 25 ng/mL and to MPO changes of ≥40 ng/mL. Whole blood samples in EDTA were also drawn at each time point and examined for leukocyte count and differential using an automated cell counter (STKS; Coulter Electronics, Hialeah, FL). This cell counter is sensitive to leukocyte counts as low as 500/uL and to changes of ≥250/uL.
Perioperative samples at the above time points (a), (d), (e) were drawn for plasma troponin I, which was measured on the Axsym® (Abbott, Abbott Park, IL) EIA system (12), according to the manufacturer’s instructions. This troponin assay is sensitive to levels as low as 0.4 ng/mL and to troponin I changes of ≥0.3 ng/mL. The CVs for all commercial assays were <6%.
The sample size calculation for the study was based on the known upregulation of monocyte and PMN CD11b during cardiac surgery/CPB; these have been shown to have peak increases of 100%–300% of their baseline values and standard deviations between 25% and 75% of the baseline value (17). For a significance level of 0.05, we calculated that a minimum sample size of 20 per group would have 80% power to detect a 50% difference (true group mean difference of 50% of the baseline value) in monocyte and PMN CD11b. If we further assumed a 20% dropout rate and an AF incidence after surgery of 33%, we determined that an enrollment of 72 subjects would be required to yield 24 patients in the AF group.
Although, as noted above, this pilot study was designed only to examine the association of cellular inflammation with AF, it was thought required to also examine previously reported AF risk factors. Thus, the two patient groups were compared for previously cited AF risk factors (Table 1), as well as the specific assays in this study. Using SigmaStat® software (SPSS, Chicago, IL), all variables were first tested to determine whether they exhibited a normal distribution; continuous variables were then examined between AF and non-AF groups using Student’s t-test or the nonparametric Mann-Whitney rank sum test.
All 72 patients completed the study sampling at all time points and underwent 72 h of inpatient monitoring for AF. No patients developed AF clinically after 72 h. Patients who developed AF (n = 26) were compared with patients who remained in sinus rhythm (n = 46) for preoperative and intraoperative characteristics that have been shown to affect risk for AF (Table 1). As in multiple studies (1–3), increased age, longer aortic cross-clamp time, and combined CABG/valve procedures were significantly more common in AF patients than in patients whose rhythm was unchanged after CPB. Sex distribution, incidence of hypertension, and remote history of AF did not differ between the groups. Although the literature suggests that transfusion may affect the inflammatory response to cardiac surgery/CPB (18), the perioperative red cell transfusion volume did not differ between groups (Table 1).
As previously reported (14), cardiac surgery/CPB was associated with significant monocyte activation, as measured by upregulation of monocyte CD11b during the perioperative period. All CD11b values are presented as the percentage of each patient’s baseline value. Monocyte CD11b reached a peak immediately after cross-clamp release and remained increased during reperfusion of the arrested heart. When monocyte CD11b over the perioperative period was summed, patients who developed postoperative AF demonstrated a significantly more frequent increase in monocyte CD11b expression compared with patients whose rhythm did not change after surgery (Fig. 1; P = 0.01).
We performed receiver-operating characteristic (ROC) analysis to examine the predictive capacity of the monocyte CD11b model for postoperative development of AF using either the sum of perioperative CD11b or the single monocyte CD11b value after cross-clamp release because the latter may be more easily determined in real time. The area under each ROC curve was calculated to indicate the variability of the prediction model. ROC identified a 500% value for the sum of CD11b and a 150% value for the single CD11b value after cross-clamp release as the optimal deflection points. The sensitivity and specificity for summed CD11b were 58% and 76%, respectively; the respective values for post–cross-clamp CD11b were 65% and 70%. The positive predictive values for summed and single CD11b were 58% and 55%, respectively, and the negative predictive values were 76% and 78%, respectively. The respective areas under the 2 curves were 68% and 70%.
All seven patients who underwent combined CABG/valve procedure developed postoperative AF (8). Thus, to determine if monocyte activation leading to AF was associated with CABG-only patients, we excluded valve surgery patients from the monocyte CD11b analysis. The sum of perioperative monocyte CD11b expression remained significantly larger in CABG-only AF patients (n = 19) compared with CABG-only non-AF patients (n = 46; P = 0.037).
The absolute number of circulating monocytes was stable before cross-clamp release and then increased after cross-clamp removal. Patients who developed postoperative AF (n = 26) demonstrated a significantly larger increase in the number of circulating monocytes than non-AF patients (n = 46; Fig. 2; P = 0.007). This association between the monocyte increase and subsequent AF remained statistically significant when valve surgery patients (n = 7) were excluded from the analysis, as above (P = 0.039).
Like monocytes, PMN were activated during the perioperative period, with CD11b values peaking just before cross-clamp release (Fig. 3). Unlike monocytes however, total perioperative PMN CD11b was not significantly higher in AF patients (n = 26) compared to the non-AF group (n = 46; Fig. 3; P= 0.057), nor was PMN CD11b significantly different when valve surgery patients (n = 7) were excluded, as above (P = 0.072). We performed ROC analysis for peak PMN CD11b. The optimal deflection point was at 200% for PMN CD11b, yielding a sensitivity of 69% and specificity of 54%; the negative and positive predictive values for this cutoff were 76% and 46%, respectively. Area under the curve was 72%. ROC for the sum of perioperative PMN CD11b showed similar results (data not shown). Similar to monocytes, the absolute number of circulating PMN increased more than two-fold after cross-clamp release, and this increase was significantly greater in all AF patients (Fig. 4; P = 0.005) and in CABG-only AF patients (P = 0.047). When we examined plasma levels of the released PMN product MPO, we found that MPO increased significantly during CPB, but both total and peak MPO levels were not significantly different between non-AF and AF patient groups (P ≥ 0.44 for both).
Preoperative CRP was measured at baseline (after the induction but before surgery) to gauge the presurgical inflammatory state of patients; we then compared these values in the two patient groups. Patients who developed postoperative AF had baseline CRP levels (mean ± sd, 2.2 ± 3.2 mg/L) that were not significantly different from patients with preserved postoperative sinus rhythm (1.8 ± 2.1 mg/L; P= 0.6). When we examined preoperative CRP values by ROC analysis, the result was a straight-line curve with no optimal deflection point (area under the curve, 47%); thus, as further indicated by the mean ± sd values shown here, there was no optimal CRP cutoff value to distinguish between AF and non-AF groups.
Peak troponin I levels during the perioperative period were similar between the AF (21 ± 12 ng/mL) and non-AF (33 ± 39 ng/mL) groups (P= 0.47). Thus, this global measure of myocardial damage/ischemia was not associated with postoperative AF.
This pilot study has examined the association between postoperative AF and specific cellular markers of perioperative inflammation in cardiac surgery/CPB patients. Compared with patients with preserved sinus rhythm, patients who developed postoperative AF had greater perioperative monocyte activation, indicated by upregulation of CD11b, but PMN activation did not differ between AF and non-AF groups. We also found a larger increase in the circulating absolute numbers of both monocytes and PMN after aortic cross-clamp release in AF patients. It is important to note that monocyte CD11b, as well as monocyte and PMN counts, remained significantly associated with postoperative AF, even when excluding valve surgery patients, a subset of CPB patients at high risk for postoperative AF.
Although another marker of PMN activation, plasma MPO, increased during the perioperative period, MPO levels were not significantly associated with AF. However, a caveat for this interpretation is that plasma markers vary considerably because of perioperative fluid management, making plasma measurements less likely to prove significant. For this latter reason, we did not elect to measure circulating levels of atrial natriuretic peptide, a specific marker of atrial changes after cardiac surgery (19). A second caveat for our lack of association between PMN activation and postoperative AF is that increased CD11b expression on PMN may cause them to rapidly marginate from the circulation or egress into the myocardium (20). Finally, our prestudy power calculation targeted a 50% true group mean difference for both PMN and monocytes, but the greater upregulation of CD11b in PMN may have obscured a significant association that might have been clearer with larger study numbers. However, the significance of previous AF risk factors (age and CPB time) in this study suggests that our cohort of subjects was not skewed. Older age alone has been associated with a decreased lymphocytic response to surgery/CPB(21), but the monocyte and neutrophil inflammatory stimulus during surgery/CPB is not affected by age (14).
The variability of our CD11b prediction model was significant, as indicated by the area under the ROC curves and their positive and negative predictive values; thus, these findings require confirmation in a larger number of patients. There are other limitations to our study. Although we measured the total perioperative red cell transfusion volume, we did not record the number and type of other transfused blood products. Monitoring for AF was limited to 72 hours after surgery; thus, it is possible that patients may have developed clinically silent AF after that point.
Interestingly, the preoperative inflammatory status of the patients, as indicated by high-sensitivity serum CRP levels, did not differ between AF and non-AF patients. CRP, as a marker of systemic inflammation, has been shown in large population studies to predict cardiovascular events and stroke, both of which may be sequelae of AF, and increased CRP has been demonstrated in a general AF population. Specifically, those patients who developed AF within 24 hours before sampling had higher CRP values than those in sinus rhythm (22). The authors speculated that AF may persist because of atrial structural changes that are promoted by inflammation, a situation that may also describe AF postcardiac surgery/CPB. However, our data suggest that the primary inflammatory insult leading to AF occurs during the operative/CPB procedure. Indeed, the inflammatory (IL-6) response to CPB has been positively correlated with the duration of CPB (23). Because this latter study was rigorously conducted in the absence of blood-blood product transfusion, their findings also directly support our data showing that the perioperative inflammatory response was independent of transfusion, e.g., no difference in blood volume use between AF and non-AF groups. Although postsurgery/CPB AF has been hypothesized to be a consequence of inadequate cardioplegic protection of the atria (24), we did not find an association between postoperative AF and generalized cardiac damage/ischemia, as assayed by peak perioperative levels of troponin I.
Primary inflammatory cardiac disorders including myocarditis, pericarditis, and some cardiomyopathies have been shown to produce AF via mechanisms that involve atrial infiltration by inflammatory cells (25). Even in the absence of a clear pathology for AF, histologic examination of atrial biopsies suggests an underlying inflammatory process in most patients with lone AF (11). Atrial damage has been associated with postoperative AF (10), but the role of perioperative inflammation associated with cardiac surgery/CPB has only recently been considered in the pathology of AF. Yared et al. (26) found that preoperative dexamethasone decreased the incidence of AF after cardiac surgery/CPB, but the effect of corticosteroids on leukocyte counts/activation peri-CPB was not studied.
The concept that inflammation resulting from cardiac surgery/CPB might contribute to the pathophysiology of AF was further suggested by Gaudino et al. (9), who found that post-CABG patients manifesting AF had significantly higher IL-6 levels. Moreover, a polymorphism in the promoter region of the IL-6 gene, which is associated with higher plasma IL-6 levels, was also an independent predictor of postoperative AF. Investigators have confirmed that cardiac surgery/CPB increases the circulating levels of the inflammatory IL-6 cytokine (23); whether IL-6 specifically targets atrial tissue or whether this is a general marker of the perioperative inflammatory state that contributes to development of AF is unknown.
Cardiac surgery requiring CPB produces a variable systemic inflammatory response (27), both overtly and in laboratory measures (21), in the degree of cellular activation and the specific cell types affected. The operative factors thought to play a role in this inflammatory state include the surgical procedure, the obligate blood-biomaterial contact of CPB, the temporary ischemic state of the myocardial and renal beds, exposure to endotoxin, and pharmacologic manipulation with protamine reversal. If we had examined samples after the initiation of surgery but before CPB, we might have been able to assess the association of AF with early leukocyte activation caused by surgery alone. Failing that, our assessment of leukocyte activation was founded on the in toto combination of cardiac surgery and CPB.
Based on in vitro and in vivo studies by our laboratory and others (14,28), it is clear that cardiac surgery/CPB causes a leukocyte inflammatory response. However, only monocyte CD11b upregulation perioperatively was significantly associated with postoperative AF in this study. CD11b is the β2-integrin that mediates leukocyte adhesion to vascular endothelial cells and leukocyte migration from the vasculature into tissues (20). Whether circulating CD11b-upregulated monocytes specifically exit into atrial tissues during CPB is unknown but may be histologically examined in future studies. The long circulation half-life for monocytes and their ability to transform into long-lived tissue macrophages may also have a role in the inflammatory response leading to postsurgical AF.
Pre- or perioperative risk assessment has the potential to minimize the number of patients requiring intervention to prevent AF and, thus, reduce toxicity caused by antiarrhythmic therapy (29). Multiple factors during cardiac surgery/CPB have been implicated in monocyte activation, including the complement system (17), cardiotomy suction (30), and the contact activation pathway (27,31). If larger studies can confirm the findings of this pilot investigation, perioperative monocyte activation may not only identify patients at risk for postoperative AF, but also define preventive strategies. Given that the prevalence of AF after CPB has changed little over the past 10 years (5,7), this postoperative complication merits aggressive investigation into its prevention.
1. Mathew J, Parks R, Savino J, et al. Atrial fibrillation following coronary artery bypass graft surgery: predictors, outcomes, and resource utilization. JAMA 1996;276:300–6.
2. Maisel W, Rawn J, Stevenson W. Atrial fibrillation after cardiac surgery. Ann Intern Med 2001;135:1061–73.
3. Hogue C, Hyder M. Atrial fibrillation after cardiac operation: risks, mechanisms, and treatment. Ann Thorac Surg 2000;69:300–6.
4. Almassi G, Schowalter T, Nicolosi A, et al. Atrial fibrillation after cardiac surgery: a major morbid event? Ann Surg 1997;226:501–13.
5. Andrews T, Reimold S, Berlin J, Antman E. Prevention of supraventricular arrhythmias after coronary artery bypass surgery: a meta-analysis of randomized control trials. Circulation 1991;84:III236–44.
6. Stanley T, Mackensen G, Grocott H, et al. The impact of postoperative atrial fibrillation on neurocognitive outcome after coronary artery bypass graft surgery. Anesth Analg 2002;94:290–5.
7. Hravnak M, Hoffman L, Saul M, et al. Resource utilization related to atrial fibrillation after coronary artery bypass grafting. Am J Crit Care 2002;11:228–38.
8. Siebert J, Anisimowicz L, Lango R, et al. Atrial fibrillation after coronary artery bypass grafting: does the type of procedure influence the early postoperative incidence? Eur J Cardiothorac Surg 2001;19:455–9.
9. Gaudino M, Andreotti F, Zamparelli R, et al. The -17G/C interleukin-6 polymorphism influences postoperative interleukin-6 levels and postoperative atrial fibrillation: is atrial fibrillation an inflammatory complication? Circulation 2001;108:II195–9.
10. Falk R. Etiology and complications of atrial fibrillation: insights from pathology studies. Am J Cardiol 1998;82:10N–7.
11. Frustaci A, Chimente C, Fulvio B, et al. Histological substrate of atrial biopsies in patients with lone atrial fibrillation. Circulation 1997;96:1180–4.
12. Mathew J, Rinder C, Howe J, et al. Platelet PLA2
polymorphism enhances risk of neurocognitive decline after cardiopulmonary bypass. Ann Thorac Surg 2001;71:663–6.
13. Rinder C, Mathew J, Rinder H, et al. The platelet PlA2 polymorphism and platelet activation are associated with increased troponin I release after cardiopulmonary bypass. Anesthesiology. In press.
14. Rinder C, Bonan J, Rinder H, et al. Cardiopulmonary bypass induces leukocyte-platelet adhesion. Blood 1992;79:1201–5.
15. Xu X, Hakansson L. Degranulation of primary and secondary granules in adherent neutrophils. Scand J Immunol 2002;55:178–88.
16. Borregaard N, Cowland J. Granules of the human neutrophilic polymorphonuclear leukocyte. Blood 1997;89:3503–21.
17. Rinder C, Rinder H, Smith B, et al. Blockade of C5a and C5b-9 generation inhibits leukocyte and platelet activation during extracorporeal circulation. J Clin Invest 1995;96:1564–72.
18. Hamada Y, Kohtani T, Nakata T, et al. Blood transfusion under cardiopulmonary bypass is a possible inducer for inflammation. Kyobu Geka 2001;54:835–8.
19. Sakai TL, Whitten CW, O’Flaherty DN, et al. Changes in plasma atrial natriuretic peptide concentration during heart transplantation. J Cardiothorac Vasc Anesth 1992;6:686–91.
20. Smith C, Marlin S, Rothlein R, et al. Cooperative interactions of LFA-1 and Mac-1 with intercellular adhesion molecule-1 in facilitating adherence and transendothelial migration of human neutrophils in vitro
. J Clin Invest 1989;83:2008–17.
21. Rinder CS, Rinder HM, Tracey JB, et al. Lymphocyte and monocyte subset changes during cardiopulmonary bypass: effects of aging and gender. J Lab Clin Med 1997;129:592–602.
22. Chung MK, Martin DO, Sprecher D, et al. C-reactive protein elevation in patients with atrial arrhythmias: inflammatory mechanisms and persistence of atrial fibrillation. Circulation 2001;104:2886–91.
23. Whitten CW, Hill GE, Ivy R, et al. Does the duration of cardiopulmonary bypass or aortic cross-clamp, in the absence of blood and/or blood product administration, influence the IL-6 response to cardiac surgery?. Anesth Analg 1998;86:28–33.
24. Ascione R, Caputo M, Calori G, et al. Predictors of atrial fibrillation after conventional and beating heart coronary surgery: a prospective, randomized study. Circulation 2000;102:1530–5.
25. Nattel S. New ideas about atrial fibrillation 50 years on. Nature 2002;415:219–25.
26. Yared JP, Starr NJ, Torres FK, et al. Effects of single dose, postinduction dexamethasone on recovery after cardiac surgery. Ann Thorac Surg 2000;69:1420–4.
27. Edmunds LJ. Inflammatory response to cardiopulmonary bypass. Ann Thorac Surg 1998;66:S12–6.
28. Kappelmayer J, Bernabei A, Gikakis N, et al. Upregulation of Mac-1 surface expression on neutrophils during simulated extracorporeal circulation. J Lab Clin Med 1993;121:118–26.
29. Amar D. Postoperative atrial fibrillation. Heart Disease 2002;4:117–23.
30. Aldea G, Soltow L, Chandler W, et al. Limitation of thrombin generation, platelet activation, and inflammation by elimination of cardiotomy suction in patients undergoing coronary artery bypass grafting treated with heparin-bonded circuits. J Thorac Cardiovasc Surg 2002;123:742–55.
© 2005 International Anesthesia Research Society
31. Sundaram S, Gikakis N, Hack C, et al. Nafamostat mesilate, a broad spectrum protease inhibitor, modulates platelet, neutrophil, and contact activation in simulated extracorporeal circulation. Thromb Haemost 1996;75:76–82.