- Email: [email protected]
Postoperative Blood Urea Nitrogen Is Associated With Stroke in Cardiac Surgical Patients
ADULT CARDIAC
Postoperative Blood Urea Nitrogen Is Associated With Stroke in Cardiac Surgical Patients Martinson K. Arnan, MD, MPA, Tyken C. Hsieh, MD, Joseph Yeboah, MD, MS, Alain G. Bertoni, MD, MPH, Gregory L. Burke, MD, MS, Zainab Bahrainwala, MPH, Maura A. Grega, RN, MS, William A. Baumgartner, MD, and Rebecca F. Gottesman, MD, PhD Department of Neurology, Wake Forest University Health Sciences, Winston-Salem, North Carolina; Department of Anesthesia, Division of Cardiac Anesthesia, Mills-Peninsula Medical Center, Burlingame, California; Department of Cardiology, Wake Forest University Health Sciences, Winston-Salem, North Carolina; Division of Public Health Sciences, Wake Forest University Health Sciences, Winston-Salem, North Carolina; Department of Neurology, Johns Hopkins University, Baltimore, Maryland; and Division of Cardiac Surgery, Johns Hopkins University, Baltimore, Maryland
Background. This case-control study identified perioperative risk factors associated with postoperative stroke risk after all cardiac surgical procedures. Methods. Among 5498 adults 18 to 90 years old who underwent cardiac surgical procedures from 2005 to 2010, we identified 180 patients who suffered a stroke within 10 days postoperatively. Controls were randomly selected and frequency matched for sex and age-band to cases. Univariate and multivariate logistic regression analyses were performed to ascertain risk factors for postoperative stroke. Results. Emergency surgical procedures (odds ratio [OR], 3.04; 95% confidence interval [CI], 1.80 to 5.10), current smoking (OR, 1.97; 95% CI, 1.29 to 3.00), peripheral vascular disease (OR, 2.80; 95% CI, 1.41 to 5.53), and previous stroke with residual paralysis (OR, 4.27; 95% CI ,1.18 to 15.38) were associated with increased stroke risk. Preoperative blood pressures were higher in patients with cases than in controls (p < 0.0001). Log of immediate
postoperative blood urea nitrogen (BUN) was higher in patients with cases than in controls (p < 0.0001). In adjusted multivariable logistic regression, postoperative BUN was associated with increased odds of stroke (OR, 2.37 per 25% increase in BUN, p < 0.0001). Postoperative stroke risk was also predicted by emergency surgical procedures (OR, 2.70, p [ 0.014), current smoking (OR, 2.82, p [ 0.002), and preoperative diastolic blood pressure (DBP) (OR, 1.77 for every 10-point increase in DBP, p < 0.0001). Receiver operator characteristic curves indicated that postoperative BUN (area under the curve, 0.855) largely explained the increased postoperative stroke risk. Conclusions. In these analyses, we identified BUN as a marker of heightened postoperative stroke risk after cardiac surgical procedures. Postoperative risk markers may improve assessment of delayed postoperative strokes.
C
coronary artery disease, and peripheral artery disease) have been shown to increase the risk of postoperative stroke [6, 9–15]. Cardiovascular risk factors and evidence of vascular disease are not believed to cause postoperative strokes directly, but they presumably reflect the presence of widespread vascular disease, or they may indicate increased susceptibility to atheroembolic or thrombotic events in any vascular bed [9, 11, 12]. For example, although carotid artery stenosis is associated with postoperative stroke, these strokes frequently occur in the hemisphere contralateral to the stenosis [10, 16]. Existing postoperative stroke risk models are acknowledged to have limited ability to discriminate between high and low levels of stroke risk [17, 18]. Ongoing efforts at improving current risk models aim to find a balance between comprehensiveness of relevant variables and parsimony while maximizing predictive accuracy. Remarkably, although postoperative stroke may be delayed, occurring days after the surgical procedure and after an initial period of no neurologic deficits [9], risk factors identifiable in the immediate postoperative period
ardiac surgical procedures are associated with a postoperative stroke risk that ranges from 0.8% to 9% [1–5]. Stroke after cardiac surgical procedures has been associated with significant increases in morbidity and mortality [6, 7], which lead to substantial economic consequences [8]. Many risk factors for early and delayed strokes after cardiac surgical procedures have been identified, and they include not only intraoperative exposures (eg, cardiopulmonary bypass time or aortic cross-clamp time) [9] but frequently also patient-level factors that are not related to details of the operation itself [9]. Of these patient-level factors, both cardiovascular risk factors (eg, increasing age, sex, hypertension, diabetes, dyslipidemia, and cigarette smoking) and evidence of vascular disease (ascending aorta atherosclerosis, carotid artery stenosis, Accepted for publication Nov 17, 2014. Address correspondence to Dr Arnan, Department of Neurology, Wake Forest Health Sciences, 1 Medical Center Blvd, Winston-Salem, NC 27157; e-mail: [email protected].
Ó 2015 by The Society of Thoracic Surgeons Published by Elsevier
(Ann Thorac Surg 2015;99:1314–20) Ó 2015 by The Society of Thoracic Surgeons
0003-4975/$36.00 http://dx.doi.org/10.1016/j.athoracsur.2014.11.034
Ann Thorac Surg 2015;99:1314–20
ARNAN ET AL ELEVATED BLOOD UREA NITROGEN IS ASSOCIATED WITH STROKE
disease, vascular risk markers, preoperative vital signs, and preoperative and postoperative chemistry laboratory results, were extracted from electronic health records by study team members.
Statistical Analysis Material and Methods Patients Between January 1, 2005 and December 31, 2010, 5498 patients underwent cardiac surgical procedures at the Johns Hopkins Medical Institution in Baltimore, Maryland. The cardiac surgical procedures included the following: coronary artery bypass graft (CABG), including first-time and repeat procedures; valve operations; combined CABG/valve surgical procedures; combined CABG/other surgical procedures; aortic operations; and other surgical procedures (the Appendix at the end of this article contains a complete list of the procedures). Patients who developed strokelike symptoms postoperatively were identified by the primary team and subsequently examined by the neurology consultation team. Stroke was defined as the development of a new neurologic deficit that could not be attributed to other neurologic conditions such as dementia, seizures, or metabolic derangement. All strokes were diagnosed by a vascular neurologist, and in the majority of patients neuroimaging confirmed the diagnosis. Postoperative strokes diagnosed up to 10 days postoperatively were included. Postoperative strokes were classified as “early” if neurologic deficit was present immediately after emergence from anesthesia and “delayed” if the stroke was diagnosed after an initial postoperative period without a neurologic deficit, consistent with earlier definitions of early versus delayed postcardiac surgical stroke [9]. In these analyses, stroke cases were selected from the Johns Hopkins Cardiac Surgery Stroke Database. In total, there were 204 strokes related to cardiac surgical procedures over the study period (2005 to 2010). However, 24 stroke cases were excluded from analyses because these patients had clinically demonstrable neurologic impairment preoperatively, or they developed stroke symptoms beyond postoperative day 10. Controls were selected at random from 5294 cardiac surgical patients who were not diagnosed with a postoperative stroke by using frequency matching based on sex and 20-year age-band to identify appropriate controls, at a 1:1 ratio. The study was approved by the institutional review board of the Johns Hopkins Medical Institutions.
For univariate analyses, c2 comparisons were made for categorical variables and t tests were used for continuous variables. The t tests were determined conservatively by initially testing for equality of variance. Any variable achieving a p value of 0.1 or less in the univariate analysis and factors considered by consensus to be biologically plausible risk factors were included in the multiple logistic regression analysis. We excluded the redundant variable when we identified collinearity in the multivariate logistic regression. The final multivariable model was developed by backward elimination. All statistical analyses were completed using SAS version 9.2 (SAS Institute, Inc, Cary, NC).
Results In this study, 180 postoperative strokes met the case definition. The rate of stroke by type of procedure is presented in Fig 1. Cases and controls are compared in Table 1. Demographics, comorbidities, and preoperative serum tests were not different between cases and controls, except for hematocrit level and frequency of emergency surgical procedures. The mean age for patients with cases and controls was 63 years ( 14). Most surgical candidates were white, and no statistically significant racial/ethnic difference was observed between the groups. History of hypertension, dyslipidemia, and diabetes was similar between the groups. However, preoperative vital signs revealed significantly higher blood pressures among patients with cases. Baseline history of atrial fibrillation or newly detected atrial fibrillation in the postoperative period was similar among cases and controls. Although a history of chronic obstructive pulmonary disease was not different between patients with cases and controls, there were significantly more current smokers among cases (54.4% versus 37.8%).
Data Collection The Cardiac Surgery Database and the Cardiac Surgery Stroke Database have been prospectively managed and include all patients undergoing cardiac surgical procedures at Johns Hopkins and all patients identified with a stroke by the foregoing criteria, respectively. Relevant risk factors, including a history of cardiovascular events, self-reported smoking status, chronic cardiovascular
Fig 1. Stroke rate by procedure type (2005 to 2010). (CABG ¼ coronary artery bypass graft.)
ADULT CARDIAC
have been underused in risk assessment models. In this study, we used an age and sex frequency-matched casecontrol study to determine markers of early and delayed postoperative stroke among immediate postoperative and patient-level risk factors.
1315
1316
ARNAN ET AL ELEVATED BLOOD UREA NITROGEN IS ASSOCIATED WITH STROKE
Ann Thorac Surg 2015;99:1314–20
Table 1. Characteristics of Cases and Frequency-Matched Controls by Age-Band and Sexa
ADULT CARDIAC
Variable Age Race Black White Other Unknown Hypertension Dyslipidemia Diabetes Peripheral vascular disease Chronic atrial fibrillation Incident atrial fibrillation Chronic obstructive pulmonary disease Current smoker Emergency (nonelective) Triple vessel disease Acute myocardial infarction Previous stroke Previous stroke (residual paralysis) Low ejection fraction (35%) Preoperative hematocrit Preoperative systolic blood pressure Preoperative Diastolic Blood Pressure Preoperative mean arterial pressure Log Preoperative BUN Inverse preoperative creatinine Log postoperative BUN Inverse postoperative creatinine
Cases N ¼ 180 Stroke n (%)
Controls N ¼ 180 No Stroke n (%)
63 (13)
Odds Ratio (95% Confidence Interval)
63 (14)
41 127 12 0 130 93 53 32 19 43 26 98 61 37 51 20 12
(22.8) (70.6) (6.7) (0.0) (72.2) (52.0) (29.4) (17.9) (10.6) (23.9) (14.6) (54.4) (33.9) (21.0) (28.3) (11.2) (6.7)
27 135 15 3 125 91 46 13 31 33 21 68 26 31 37 14 2
(15.0) (75.0) (8.3) (1.7) (69.4) (50.6) (25.6) (7.2) (17.2) (18.3) (11.7) (37.8) (14.4) (17.2) (20.6) (7.8) (1.7)
38 35.8 133 73 93 3.0 0.98 3.6 0.76
(22.9) (6.3) (25) (14) (16) (0.5) (0.4) (0.5) (0.6)
35 37.9 124 66 86 3.0 1.0 2.9 1.07
(21.0) (5.4) (26) (14) (16) (0.4) (0.3) (0.4) (0.4)
[20.0] [1.0] [36.6] [1.3]
p Value 0.725 0.091
2.80 (1.41-5.53)
1.97 (1.29-3.00) 3.04 (1.80-5.10)
4.27 (1.18-15.38)
0.562 0.791 0.409 0.002b 0.067 0.197 0.410 0.002b <0.0001b 0.362 0.086 0.272 0.013b 0.670 0.0012b 0.003b <0.0001b <0.0001b 0.364 0.159 <0.0001b <0.0001b
[20.0] [1.0] [18.2] [0.93]
Univariate analysis by c2 test and t test (determined conservatively by first testing for equality of variance). Values are presented as means standard deviation or counts (%). Log and inverse function transformations of BUN and creatinine (respectively) were done because violation of normality assumption was noted on diagnostic testing that included Q-Q plots. Log and inverse transformations of BUN and creatinine normalized the data b appropriately. Values in square brackets are nontransformed BUN and creatinine in mg/dL. Significant at the 95% confidence level.
a
BUN ¼ blood urea nitrogen.
Additionally, the frequency of peripheral vascular disease was more than two times higher among cases (17.9 versus 7.2%). A history of cardiovascular disease (defined as myocardial infarction or stroke) was not different between patients with cases and controls. Although a history of stroke was similar between cases and controls, in study participants a history of stroke with residual paralysis was significantly higher among cases (6.7%) than controls (1.7%). Preoperative hematocrit level was significantly lower among patients with cases. Although preoperative blood urea nitrogen (BUN) and creatinine were similar between cases and controls, postoperatively these values were higher in cases than controls. Aortic procedures had the highest postoperative stroke risk of 6.1%, whereas isolated CABG had a stroke rate of approximately 2.5% over the study period. The frequency of aortic procedures was double among patients with cases (Table 2).
In multivariable logistic regression, four factors were found to be strongly associated with increased odds of postoperative stroke (Table 3). Postoperative BUN (odds ratio [OR], 2.37 for every 25% rise in BUN, p < 0.0001), Table 2. Procedure Type and Risk of Stroke Variable Procedure CABG Valve CABG/Valve CABG/Other Aortaa Other a
Cases N ¼ 180 Stroke n (%)
Controls N¼ 180 No Stroke n (%)
p Value 0.010
40 35 20 12 44 29
(22.2) (19.4) (11.1) (6.7) (24.4) (16.1)
59 49 19 7 22 24
(32.8) (27.2) (10.6) (3.9) (12.2) (13.3)
Procedure type associated with highest post-operative stroke risk.
CABG ¼ coronary artery bypass graft.
Ann Thorac Surg 2015;99:1314–20
1317
ARNAN ET AL ELEVATED BLOOD UREA NITROGEN IS ASSOCIATED WITH STROKE
Table 3. Multivariable Logistic Regression on Strokea Model 1
Postoperative BUN Smoker Emergency (nonelective surgery) Diastolic Blood pressure
Odds Ratio
p Value
2.04
<0.0001
Model 3
Model 4
Odds Ratio
p Value
Odds Ratio
p Value
Odds Ratio
p Value
2.18 2.48 2.54 1.76
<0.0001 0.003 0.009 <0.0001
2.29 2.42 2.71 1.76
<0.0001 0.004 0.007 <0.0001
2.37 2.82 2.70 1.77
<0.0001 0.002 0.014 <0.0001
a
Model 1 is an unadjusted univariate logistic regression of post-operative BUN with stroke as outcome. Model 2 was developed through backward elimination. The inclusion criterion into Model 2 was p <0.1 in univariate analysis. Model 3 adjusts for postoperative creatinine level. Model 4 adjusts for postoperative creatinine level, low ejection fraction, acute myocardial infarction as presenting diagnosis, and preoperative hematocrit level. Postoperative blood urea nitrogen (BUN) represents the increased odds of stroke for a 25% rise in BUN. Diastolic blood pressure represents the increased odds of stroke for a 10 mm Hg rise. Model 4 was unchanged even after further adjusting for age, procedure type, and history of stroke (results not shown). BUN ¼ blood urea nitrogen.
emergency surgery (OR, 2.70, p ¼ 0.014), current smoking (OR, 2.82, p ¼ 0.002), and preoperative diastolic blood pressure (DBP) (OR, 1.77 for every 10-point increase in DBP, p < 0.0001) remained significantly associated with heightened stroke risk. The model remained significant even after adjusting for biologically plausible confounders. Receiver operator characteristic (ROC) curves indicated that postoperative BUN (area under the curve [AUC], 0.855, Fig 2) largely explained the postoperative stroke risk. Postoperative BUN values greater than 25 mg/dL predicted stroke with sensitivity of 80% and specificity of 75% (see Fig 2). The ROC curve for DBP, emergency surgical procedures, and current smoking as predictors of postoperative stroke resulted in an AUC of 0.700 (Fig 3). The multiple regression model that included all four factors only marginally improved the model (AUC, 0.891, Fig 4).
Comment In these analyses we identified perioperative risk factors that are associated with increased odds of postoperative stroke: low preoperative hematocrit, history of stroke, peripheral vascular disease, preoperatively elevated blood pressure, smoking status, and elevated postoperative BUN. It has been proposed that the primary mechanisms of perioperative brain injury after cardiac surgical procedures are emboli and hypoperfusion in the setting of ischemia/reperfusion injury [19]. Given the
physical manipulation of the ascending aorta in some cardiac operations, many investigators have argued that perioperative strokes are likely a direct result of embolization of material from large vessels [20]. Although in our study procedures on the aorta had the highest postoperative stroke risk of all surgical types (a finding supporting the hypothesis that direct large vessel atheroembolism may contribute to postoperative stroke), we also observed that postoperative strokes may be delayed up to 10 days after the procedure [9]. Although epiaortic ultrasound examinations were not performed routinely at Johns Hopkins during the period under study, the delay between surgical procedure and postoperative stoke raises the possibility that direct atheroembolism from large vessels is not the only mechanism for perioperative strokes; other mechanisms of injury likely play a significant role as well. Patients with evidence of atherosclerotic disease in one vascular territory are likely to have widespread disease in other territories including the ascending aorta and perhaps the intracranial vessels. Any of these areas of high atherosclerotic burden is vulnerable to embolization or thrombosis. This mechanism of disease may explain why aspirin and statins have been found to be protective in high-risk patients undergoing major vascular operations [21–25]. In our univariate analysis, we found that patients with a diagnosis of peripheral vascular disease were significantly more likely to have a stroke as a Fig 2. Logistic regression of probability of stroke versus log of postoperative blood urea nitrogen (BUN). (A) Predicted probability of new stroke. (B) Receiver operator characteristic curve for model.
ADULT CARDIAC
Variable
Model 2
ADULT CARDIAC
1318
ARNAN ET AL ELEVATED BLOOD UREA NITROGEN IS ASSOCIATED WITH STROKE
Ann Thorac Surg 2015;99:1314–20
Fig 3. Logistic regression of probability of stroke versus preoperative diastolic blood pressure (DBP), current smoker, and emergency surgical status. (A) Predicted probability of new stroke. (B) Receiver operator characteristic curve for model.
surgical complication. These findings suggest that patients who have a diagnosed vascular disease, regardless of the arterial bed involved, are at a high risk of developing other vascular events such as stroke after cardiac surgical procedures. The most interesting finding in our study was that acutely elevated postoperative BUN emerged as by far the most powerful predictor of postcardiac surgical stroke (OR, 2.37 for every 25% rise in BUN, with ROC AUC of 0.855), greatly exceeding the predictive power of the other more commonly recognized risk factors such as smoking, preoperative hypertension, and emergency surgical procedures. Although there was evidence of a higher risk of renal injury in patients with stroke as reflected by significantly higher creatinine levels in cases than in controls, BUN in multivariable regression analyses was more strongly associated with postoperative stroke than was creatinine levels. The impact of BUN persisted even after adjusting for creatinine levels (see Table 3). We do not believe that the relationship between elevated BUN and stroke risk is causal. Still, this novel finding has important implications. Elevated BUN is generally considered to reflect intravascular volume depletion or inadequate renal perfusion (prerenal azotemia). Possibly, patients with elevated postoperative BUN suffered parallel embolic hits to the brain as well as the kidneys. However, the observation that postoperative BUN was elevated among patients with cases but not controls could suggest that the patients identified postoperatively with strokes had suffered from relatively
Fig 4. Logistic regression of probability of stroke versus log of postoperative blood urea nitrogen (BUN), current smoker, emergency surgical status, and preoperative diastolic blood pressure (DBP). (A) Predicted probability of new stroke. (B) Receiver operator characteristic curve for model.
inadequate renal perfusion during the surgical procedures, even though the same cardiopulmonary bypass perfusion protocol (target mean arterial pressure and cardiac index) was used for both cases and controls. This susceptibility to azotemia among cases may suggest that these patients may have had subclinical renal vascular disease. We suspect that patients with subclinical renal vascular disease may have comorbid overt or subclinical vascular disease in other arterial beds, such as the aorta, the carotid arteries, and possibly intracranial arteries, thus leading to an increased risk of stroke. Elevated BUN in the immediate postoperative period may reflect some underlying vulnerability to vascularly mediated disease. We acknowledge that elevated BUN postoperatively may simply reflect renal injury sustained intraoperatively. However, further analysis of our data (not shown in the Results section) reveals a more complicated picture. We performed a nonparametric correlation procedure (Spearman’s correlation) for BUN and some intraoperative variables. Our analyses revealed that among our patients with stroke, postoperative BUN is only weakly correlated with procedure type (r ¼ 0.11, p ¼ 0.03) and cardiopulmonary bypass time (r ¼ 0.16, p ¼ 0.03) and not correlated with lowest intraoperative mean arterial pressure (r ¼ 0.11, p ¼ 0.13). Additionally, a multiple regression analysis of BUN as a dependent variable, with cardiopulmonary bypass time and lowest intraoperative mean arterial pressure as independent variables, showed that the model was not statistically significant (F ¼ 2.7, p > 0.05). Circulatory arrest time was
Ann Thorac Surg 2015;99:1314–20
ARNAN ET AL ELEVATED BLOOD UREA NITROGEN IS ASSOCIATED WITH STROKE
controls, and this approach precluded our analysis from identifying sex or age as significant risk factors for stroke. Still, we suspect that although female sex and increasing age have been shown to be associated with strokes, the association may be mediated through advanced and widespread vascular disease. In conclusion, we have shown in our analyses that evidence of vascular disease is the likely common mechanism that predisposes cardiac surgical patients to postoperative stroke. We believe that patients with direct evidence of vascular disease such as peripheral arterial disease, with evidence of end-organ damage from advanced vascular disease such as previous stroke, or with subclinical vascular disease (as may be revealed by elevated postoperative BUN) are at highest risk of postoperative stroke regardless of the type of cardiac surgical procedure. Ultimately, learning to identify patients who may be susceptible to adverse outcomes in the setting of routine intraoperative blood pressure goals may help to ameliorate postoperative risk. We wrote and edited this article, and we take full responsibility for its content. Dr Arnan receives research support from a Diversity Supplement to contract N01-HC-95165 from the National Heart, Lung, and Blood Institute. We are grateful to Henry A. Feldman, PhD for his important contributions to this article.
References 1. McKhann GM, Grega MA, Borowicz LM Jr, et al. Stroke and encephalopathy after cardiac surgery: an update. Stroke 2006;37:562–71. 2. Martin WR, Hashimoto SA. Stroke in coronary bypass surgery. Can J Neurol Sci 1982;9:21–6. 3. Coffey CE, Massey EW, Roberts KB, et al. Natural history of cerebral complications of coronary artery bypass graft surgery. Neurology 1983;33:1416–21. 4. Breuer AC, Furlan AJ, Hanson MR, et al. Central nervous system complications of coronary artery bypass graft surgery: prospective analysis of 421 patients. Stroke 1983;14:682–7. 5. Junod FL, Harlan BJ, Payne J, et al. Preoperative risk assessment in cardiac surgery: comparison of predicted and observed results. Ann Thorac Surg 1987;43:59–64. 6. Bucerius J, Gummert JF, Borger MA, et al. Stroke after cardiac surgery: a risk factor analysis of 16,184 consecutive adult patients. Ann Thorac Surg 2003;75:472–8. 7. Dacey LJ, Likosky DS, Leavitt BJ, et al. Perioperative stroke and long-term survival after coronary bypass graft surgery. Ann Thorac Surg 2005;79:532–6; discussion 537. 8. Roach GW, Kanchuger M, Mangano CM, et al. Adverse cerebral outcomes after coronary bypass surgery: Multicenter Study of Perioperative Ischemia Research Group and the Ischemia Research and Education Foundation Investigators. N Engl J Med 1996;335:1857–63. 9. Hogue CW Jr, Murphy SF, Schechtman KB, et al. Risk factors for early or delayed stroke after cardiac surgery. Circulation 1999;100:642–7. 10. McKhann GM, Goldsborough MA, Borowicz LM Jr, et al. Predictors of stroke risk in coronary artery bypass patients. Ann Thorac Surg 1997;63:516–21. 11. Gardner TJ, Horneffer PJ, Manolio TA, et al. Stroke following coronary artery bypass grafting: a ten-year study. Ann Thorac Surg 1985;40:574–81. 12. Tuman KJ, McCarthy RJ, March RJ, et al. Morbidity and duration of ICU stay after cardiac surgery: a model for preoperative risk assessment. Chest 1992;102:36–44.
ADULT CARDIAC
not addressed because only a minority of patients underwent circulatory arrest (n ¼ 26; 14.4%) over the study period. We interpret these analyses to mean that the postoperative BUN level cannot be fully explained by intraoperative factors. This lends some support to our conclusion that the rise in BUN among patients with postoperative stroke could reflect renal vascular vulnerability, which may be associated with stroke susceptibility. Additionally, although our population of patients with stroke includes early and delayed strokes, we detected elevated BUN in both groups. The early postoperative strokes make up only 13.5% (n ¼ 24) of our total population of patients with stroke. Among our patients with stroke, there was no significant difference in immediate postoperative mean BUN levels between early and delayed strokes (29.9 mg/dL versus 36.6 mg/dL, p > 0.05). Thus, we suspect that elevated BUN may be associated with both early and delayed strokes. We also noted that emergency surgical status was associated with stroke. The OR for stroke for emergency surgical procedures is 2.70, consistent with the findings in the Northern New England Cardiovascular Disease Study Group’s prediction model for strokes after CABG [26]. It is not immediately clear why emergency surgical procedures are associated with a higher risk of stroke. The association is unlikely to be caused by a difference in operative technique during emergency surgical procedures. More likely, emergency operations are conducted on patients with acutely active disease who could be in a highly inflammatory state. We propose that in the setting of widespread vascular disease, an acute vascular syndrome or an infectious process requiring emergency surgical procedures can increase the predisposition to a postoperative stroke. Alternatively, those patients requiring emergency operations are more hemodynamically unstable at the time of the procedure, and this instability could also contribute to an increased risk of postoperative stroke. We acknowledge that our study has several limitations. First, it is possible that there are unmeasured confounders. However, given that our findings are consistent with previously published risk stratification schemes we believe our analysis adds value to what is already known in the field of perioperative stroke assessment. Second, neuroimaging was not obtained postoperatively for asymptomatic patients. Therefore, there is a possibility that strokes were underdiagnosed in this study. The institutional postoperative stroke rate at Johns Hopkins is consistent with the rates reported by other institutions. Thus, we are unlikely to be significantly underdiagnosing clinically significant strokes. Third, our findings are specific to one institution and a finite set of surgeons with a certain level of experience. Consequently, the findings from our institution may not be generalizable to other clinical settings. However, the thrust of our findings is that evidence of vascular disease (either overt or subclinical) is associated with postoperative strokes and likely has general applicability. Finally, we used frequency matching by gender and age-band to select
1319
ADULT CARDIAC
1320
ARNAN ET AL ELEVATED BLOOD UREA NITROGEN IS ASSOCIATED WITH STROKE
13. Grover FL, Hammermeister KE, Burchfiel C. Initial report of the Veterans Administration Preoperative Risk Assessment Study for Cardiac Surgery. Ann Thorac Surg 1990;50:12–26; discussion 27–8. 14. Djaiani G, Fedorko L, Borger M, et al. Mild to moderate atheromatous disease of the thoracic aorta and new ischemic brain lesions after conventional coronary artery bypass graft surgery. Stroke 2004;35:e356–8. 15. Eagle KA, Guyton RA, Davidoff R, et al. ACC/AHA 2004 guideline update for coronary artery bypass graft surgery: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Committee to Update the 1999 Guidelines for Coronary Artery Bypass Graft Surgery). Circulation 2004;110; p. e340–437. 16. Li Y, Walicki D, Mathiesen C, et al. Strokes after cardiac surgery and relationship to carotid stenosis. Arch Neurol 2009;66:1091–6. 17. Head SJ, Osnabrugge RL, Howell NJ, et al. A systematic review of risk prediction in adult cardiac surgery: considerations for future model development. Eur J Cardiothorac Surg 2013;43:e121–9. 18. Hornero F, Martin E, Paredes F, et al. Stroke after coronary artery bypass grafting: preoperative predictive accuracies of CHADS2 and CHA2DS2VASc stroke risk stratification schemes. J Thorac Cardiovasc Surg 2012;144:1428–35. 19. Shaw PJ, Bates D, Cartlidge NE, et al. An analysis of factors predisposing to neurological injury in patients undergoing coronary bypass operations. Q J Med 1989;72:633–46. 20. Borger MA, Ivanov J, Weisel RD, et al. Stroke during coronary bypass surgery: principal role of cerebral macroemboli. Eur J Cardiothorac Surg 2001;19:627–32. 21. Fleisher LA, Beckman JA, Brown KA, et al. 2009 ACCF/AHA focused update on perioperative beta blockade incorporated into the ACC/AHA 2007 guidelines on perioperative cardiovascular evaluation and care for noncardiac surgery: a Report of the American College of Cardiology Foundation/ American Heart Association Task Force on Practice Guidelines. Circulation 2009;120:e169–276. 22. Kertai MD, Boersma E, Westerhout CM, et al. A combination of statins and beta-blockers is independently associated with a reduction in the incidence of perioperative mortality and nonfatal myocardial infarction in patients undergoing abdominal aortic aneurysm surgery. Eur J Vasc Endovasc Surg 2004;28:343–52. 23. Feringa HH, van Waning VH, Bax JJ, et al. Cardioprotective medication is associated with improved survival in patients with peripheral arterial disease. J Am Coll Cardiol 2006;47: 1182–7. 24. Feringa HH, Bax JJ, Karagiannis SE, et al. Elderly patients undergoing major vascular surgery: risk factors and medication associated with risk reduction. Arch Gerontol Geriatr 2009;48:116–20. 25. Lau WC, Froehlich JB, Jewell ES, et al. Impact of adding aspirin to beta-blocker and statin in high-risk patients
Ann Thorac Surg 2015;99:1314–20
undergoing major vascular surgery. Ann Vasc Surg 2013;27: 537–45. 26. Charlesworth DC, Likosky DS, Marrin CA, et al. Development and validation of a prediction model for strokes after coronary artery bypass grafting. Ann Thorac Surg 2003;76:436–43.
Appendix: List of Other Cardiac Surgical Procedures Pacemaker implantation/removal Pericardial window Sternal wire removal Extracorporeal membrane oxygenator implantation/removal Ventricular septal defect repair Anomalous pulmonary venous return repair Apical conduit Ventricular assist device implantation/removal Intraaortic balloon pump implantation/removal Heart transplantation Evacuation of hematoma Automatic implantable cardioverter-defibrillator implantation/removal Heart/lung transplant Patent foramen ovale closure Pericardiectomy Incision and drainage Debridement and washout Atrial septal defect repair Removal of foreign body Subaortic membrane resection Left ventricular myotomy and myectomy Exploratory median sternotomy Cardiac tumor resection Pulmonary embolectomy Inferior vena cava tumor resection Transmyocardial laser revascularization Ligation of coronary artery fistula Atrial mass removal Radiofrequency ablation Pulmonary artery repair Myxoma resection Atrial septectomy Fontan procedure Unroofing anomalous right coronary artery
Postoperative Blood Urea Nitrogen Is Associated With Stroke in Cardiac Surgical Patients Martinson K. Arnan, MD, MPA, Tyken C. Hsieh, MD, Joseph Yeboah, MD, MS, Alain G. Bertoni, MD, MPH, Gregory L. Burke, MD, MS, Zainab Bahrainwala, MPH, Maura A. Grega, RN, MS, William A. Baumgartner, MD, and Rebecca F. Gottesman, MD, PhD Department of Neurology, Wake Forest University Health Sciences, Winston-Salem, North Carolina; Department of Anesthesia, Division of Cardiac Anesthesia, Mills-Peninsula Medical Center, Burlingame, California; Department of Cardiology, Wake Forest University Health Sciences, Winston-Salem, North Carolina; Division of Public Health Sciences, Wake Forest University Health Sciences, Winston-Salem, North Carolina; Department of Neurology, Johns Hopkins University, Baltimore, Maryland; and Division of Cardiac Surgery, Johns Hopkins University, Baltimore, Maryland
Background. This case-control study identified perioperative risk factors associated with postoperative stroke risk after all cardiac surgical procedures. Methods. Among 5498 adults 18 to 90 years old who underwent cardiac surgical procedures from 2005 to 2010, we identified 180 patients who suffered a stroke within 10 days postoperatively. Controls were randomly selected and frequency matched for sex and age-band to cases. Univariate and multivariate logistic regression analyses were performed to ascertain risk factors for postoperative stroke. Results. Emergency surgical procedures (odds ratio [OR], 3.04; 95% confidence interval [CI], 1.80 to 5.10), current smoking (OR, 1.97; 95% CI, 1.29 to 3.00), peripheral vascular disease (OR, 2.80; 95% CI, 1.41 to 5.53), and previous stroke with residual paralysis (OR, 4.27; 95% CI ,1.18 to 15.38) were associated with increased stroke risk. Preoperative blood pressures were higher in patients with cases than in controls (p < 0.0001). Log of immediate
postoperative blood urea nitrogen (BUN) was higher in patients with cases than in controls (p < 0.0001). In adjusted multivariable logistic regression, postoperative BUN was associated with increased odds of stroke (OR, 2.37 per 25% increase in BUN, p < 0.0001). Postoperative stroke risk was also predicted by emergency surgical procedures (OR, 2.70, p [ 0.014), current smoking (OR, 2.82, p [ 0.002), and preoperative diastolic blood pressure (DBP) (OR, 1.77 for every 10-point increase in DBP, p < 0.0001). Receiver operator characteristic curves indicated that postoperative BUN (area under the curve, 0.855) largely explained the increased postoperative stroke risk. Conclusions. In these analyses, we identified BUN as a marker of heightened postoperative stroke risk after cardiac surgical procedures. Postoperative risk markers may improve assessment of delayed postoperative strokes.
C
coronary artery disease, and peripheral artery disease) have been shown to increase the risk of postoperative stroke [6, 9–15]. Cardiovascular risk factors and evidence of vascular disease are not believed to cause postoperative strokes directly, but they presumably reflect the presence of widespread vascular disease, or they may indicate increased susceptibility to atheroembolic or thrombotic events in any vascular bed [9, 11, 12]. For example, although carotid artery stenosis is associated with postoperative stroke, these strokes frequently occur in the hemisphere contralateral to the stenosis [10, 16]. Existing postoperative stroke risk models are acknowledged to have limited ability to discriminate between high and low levels of stroke risk [17, 18]. Ongoing efforts at improving current risk models aim to find a balance between comprehensiveness of relevant variables and parsimony while maximizing predictive accuracy. Remarkably, although postoperative stroke may be delayed, occurring days after the surgical procedure and after an initial period of no neurologic deficits [9], risk factors identifiable in the immediate postoperative period
ardiac surgical procedures are associated with a postoperative stroke risk that ranges from 0.8% to 9% [1–5]. Stroke after cardiac surgical procedures has been associated with significant increases in morbidity and mortality [6, 7], which lead to substantial economic consequences [8]. Many risk factors for early and delayed strokes after cardiac surgical procedures have been identified, and they include not only intraoperative exposures (eg, cardiopulmonary bypass time or aortic cross-clamp time) [9] but frequently also patient-level factors that are not related to details of the operation itself [9]. Of these patient-level factors, both cardiovascular risk factors (eg, increasing age, sex, hypertension, diabetes, dyslipidemia, and cigarette smoking) and evidence of vascular disease (ascending aorta atherosclerosis, carotid artery stenosis, Accepted for publication Nov 17, 2014. Address correspondence to Dr Arnan, Department of Neurology, Wake Forest Health Sciences, 1 Medical Center Blvd, Winston-Salem, NC 27157; e-mail: [email protected].
Ó 2015 by The Society of Thoracic Surgeons Published by Elsevier
(Ann Thorac Surg 2015;99:1314–20) Ó 2015 by The Society of Thoracic Surgeons
0003-4975/$36.00 http://dx.doi.org/10.1016/j.athoracsur.2014.11.034
Ann Thorac Surg 2015;99:1314–20
ARNAN ET AL ELEVATED BLOOD UREA NITROGEN IS ASSOCIATED WITH STROKE
disease, vascular risk markers, preoperative vital signs, and preoperative and postoperative chemistry laboratory results, were extracted from electronic health records by study team members.
Statistical Analysis Material and Methods Patients Between January 1, 2005 and December 31, 2010, 5498 patients underwent cardiac surgical procedures at the Johns Hopkins Medical Institution in Baltimore, Maryland. The cardiac surgical procedures included the following: coronary artery bypass graft (CABG), including first-time and repeat procedures; valve operations; combined CABG/valve surgical procedures; combined CABG/other surgical procedures; aortic operations; and other surgical procedures (the Appendix at the end of this article contains a complete list of the procedures). Patients who developed strokelike symptoms postoperatively were identified by the primary team and subsequently examined by the neurology consultation team. Stroke was defined as the development of a new neurologic deficit that could not be attributed to other neurologic conditions such as dementia, seizures, or metabolic derangement. All strokes were diagnosed by a vascular neurologist, and in the majority of patients neuroimaging confirmed the diagnosis. Postoperative strokes diagnosed up to 10 days postoperatively were included. Postoperative strokes were classified as “early” if neurologic deficit was present immediately after emergence from anesthesia and “delayed” if the stroke was diagnosed after an initial postoperative period without a neurologic deficit, consistent with earlier definitions of early versus delayed postcardiac surgical stroke [9]. In these analyses, stroke cases were selected from the Johns Hopkins Cardiac Surgery Stroke Database. In total, there were 204 strokes related to cardiac surgical procedures over the study period (2005 to 2010). However, 24 stroke cases were excluded from analyses because these patients had clinically demonstrable neurologic impairment preoperatively, or they developed stroke symptoms beyond postoperative day 10. Controls were selected at random from 5294 cardiac surgical patients who were not diagnosed with a postoperative stroke by using frequency matching based on sex and 20-year age-band to identify appropriate controls, at a 1:1 ratio. The study was approved by the institutional review board of the Johns Hopkins Medical Institutions.
For univariate analyses, c2 comparisons were made for categorical variables and t tests were used for continuous variables. The t tests were determined conservatively by initially testing for equality of variance. Any variable achieving a p value of 0.1 or less in the univariate analysis and factors considered by consensus to be biologically plausible risk factors were included in the multiple logistic regression analysis. We excluded the redundant variable when we identified collinearity in the multivariate logistic regression. The final multivariable model was developed by backward elimination. All statistical analyses were completed using SAS version 9.2 (SAS Institute, Inc, Cary, NC).
Results In this study, 180 postoperative strokes met the case definition. The rate of stroke by type of procedure is presented in Fig 1. Cases and controls are compared in Table 1. Demographics, comorbidities, and preoperative serum tests were not different between cases and controls, except for hematocrit level and frequency of emergency surgical procedures. The mean age for patients with cases and controls was 63 years ( 14). Most surgical candidates were white, and no statistically significant racial/ethnic difference was observed between the groups. History of hypertension, dyslipidemia, and diabetes was similar between the groups. However, preoperative vital signs revealed significantly higher blood pressures among patients with cases. Baseline history of atrial fibrillation or newly detected atrial fibrillation in the postoperative period was similar among cases and controls. Although a history of chronic obstructive pulmonary disease was not different between patients with cases and controls, there were significantly more current smokers among cases (54.4% versus 37.8%).
Data Collection The Cardiac Surgery Database and the Cardiac Surgery Stroke Database have been prospectively managed and include all patients undergoing cardiac surgical procedures at Johns Hopkins and all patients identified with a stroke by the foregoing criteria, respectively. Relevant risk factors, including a history of cardiovascular events, self-reported smoking status, chronic cardiovascular
Fig 1. Stroke rate by procedure type (2005 to 2010). (CABG ¼ coronary artery bypass graft.)
ADULT CARDIAC
have been underused in risk assessment models. In this study, we used an age and sex frequency-matched casecontrol study to determine markers of early and delayed postoperative stroke among immediate postoperative and patient-level risk factors.
1315
1316
ARNAN ET AL ELEVATED BLOOD UREA NITROGEN IS ASSOCIATED WITH STROKE
Ann Thorac Surg 2015;99:1314–20
Table 1. Characteristics of Cases and Frequency-Matched Controls by Age-Band and Sexa
ADULT CARDIAC
Variable Age Race Black White Other Unknown Hypertension Dyslipidemia Diabetes Peripheral vascular disease Chronic atrial fibrillation Incident atrial fibrillation Chronic obstructive pulmonary disease Current smoker Emergency (nonelective) Triple vessel disease Acute myocardial infarction Previous stroke Previous stroke (residual paralysis) Low ejection fraction (35%) Preoperative hematocrit Preoperative systolic blood pressure Preoperative Diastolic Blood Pressure Preoperative mean arterial pressure Log Preoperative BUN Inverse preoperative creatinine Log postoperative BUN Inverse postoperative creatinine
Cases N ¼ 180 Stroke n (%)
Controls N ¼ 180 No Stroke n (%)
63 (13)
Odds Ratio (95% Confidence Interval)
63 (14)
41 127 12 0 130 93 53 32 19 43 26 98 61 37 51 20 12
(22.8) (70.6) (6.7) (0.0) (72.2) (52.0) (29.4) (17.9) (10.6) (23.9) (14.6) (54.4) (33.9) (21.0) (28.3) (11.2) (6.7)
27 135 15 3 125 91 46 13 31 33 21 68 26 31 37 14 2
(15.0) (75.0) (8.3) (1.7) (69.4) (50.6) (25.6) (7.2) (17.2) (18.3) (11.7) (37.8) (14.4) (17.2) (20.6) (7.8) (1.7)
38 35.8 133 73 93 3.0 0.98 3.6 0.76
(22.9) (6.3) (25) (14) (16) (0.5) (0.4) (0.5) (0.6)
35 37.9 124 66 86 3.0 1.0 2.9 1.07
(21.0) (5.4) (26) (14) (16) (0.4) (0.3) (0.4) (0.4)
[20.0] [1.0] [36.6] [1.3]
p Value 0.725 0.091
2.80 (1.41-5.53)
1.97 (1.29-3.00) 3.04 (1.80-5.10)
4.27 (1.18-15.38)
0.562 0.791 0.409 0.002b 0.067 0.197 0.410 0.002b <0.0001b 0.362 0.086 0.272 0.013b 0.670 0.0012b 0.003b <0.0001b <0.0001b 0.364 0.159 <0.0001b <0.0001b
[20.0] [1.0] [18.2] [0.93]
Univariate analysis by c2 test and t test (determined conservatively by first testing for equality of variance). Values are presented as means standard deviation or counts (%). Log and inverse function transformations of BUN and creatinine (respectively) were done because violation of normality assumption was noted on diagnostic testing that included Q-Q plots. Log and inverse transformations of BUN and creatinine normalized the data b appropriately. Values in square brackets are nontransformed BUN and creatinine in mg/dL. Significant at the 95% confidence level.
a
BUN ¼ blood urea nitrogen.
Additionally, the frequency of peripheral vascular disease was more than two times higher among cases (17.9 versus 7.2%). A history of cardiovascular disease (defined as myocardial infarction or stroke) was not different between patients with cases and controls. Although a history of stroke was similar between cases and controls, in study participants a history of stroke with residual paralysis was significantly higher among cases (6.7%) than controls (1.7%). Preoperative hematocrit level was significantly lower among patients with cases. Although preoperative blood urea nitrogen (BUN) and creatinine were similar between cases and controls, postoperatively these values were higher in cases than controls. Aortic procedures had the highest postoperative stroke risk of 6.1%, whereas isolated CABG had a stroke rate of approximately 2.5% over the study period. The frequency of aortic procedures was double among patients with cases (Table 2).
In multivariable logistic regression, four factors were found to be strongly associated with increased odds of postoperative stroke (Table 3). Postoperative BUN (odds ratio [OR], 2.37 for every 25% rise in BUN, p < 0.0001), Table 2. Procedure Type and Risk of Stroke Variable Procedure CABG Valve CABG/Valve CABG/Other Aortaa Other a
Cases N ¼ 180 Stroke n (%)
Controls N¼ 180 No Stroke n (%)
p Value 0.010
40 35 20 12 44 29
(22.2) (19.4) (11.1) (6.7) (24.4) (16.1)
59 49 19 7 22 24
(32.8) (27.2) (10.6) (3.9) (12.2) (13.3)
Procedure type associated with highest post-operative stroke risk.
CABG ¼ coronary artery bypass graft.
Ann Thorac Surg 2015;99:1314–20
1317
ARNAN ET AL ELEVATED BLOOD UREA NITROGEN IS ASSOCIATED WITH STROKE
Table 3. Multivariable Logistic Regression on Strokea Model 1
Postoperative BUN Smoker Emergency (nonelective surgery) Diastolic Blood pressure
Odds Ratio
p Value
2.04
<0.0001
Model 3
Model 4
Odds Ratio
p Value
Odds Ratio
p Value
Odds Ratio
p Value
2.18 2.48 2.54 1.76
<0.0001 0.003 0.009 <0.0001
2.29 2.42 2.71 1.76
<0.0001 0.004 0.007 <0.0001
2.37 2.82 2.70 1.77
<0.0001 0.002 0.014 <0.0001
a
Model 1 is an unadjusted univariate logistic regression of post-operative BUN with stroke as outcome. Model 2 was developed through backward elimination. The inclusion criterion into Model 2 was p <0.1 in univariate analysis. Model 3 adjusts for postoperative creatinine level. Model 4 adjusts for postoperative creatinine level, low ejection fraction, acute myocardial infarction as presenting diagnosis, and preoperative hematocrit level. Postoperative blood urea nitrogen (BUN) represents the increased odds of stroke for a 25% rise in BUN. Diastolic blood pressure represents the increased odds of stroke for a 10 mm Hg rise. Model 4 was unchanged even after further adjusting for age, procedure type, and history of stroke (results not shown). BUN ¼ blood urea nitrogen.
emergency surgery (OR, 2.70, p ¼ 0.014), current smoking (OR, 2.82, p ¼ 0.002), and preoperative diastolic blood pressure (DBP) (OR, 1.77 for every 10-point increase in DBP, p < 0.0001) remained significantly associated with heightened stroke risk. The model remained significant even after adjusting for biologically plausible confounders. Receiver operator characteristic (ROC) curves indicated that postoperative BUN (area under the curve [AUC], 0.855, Fig 2) largely explained the postoperative stroke risk. Postoperative BUN values greater than 25 mg/dL predicted stroke with sensitivity of 80% and specificity of 75% (see Fig 2). The ROC curve for DBP, emergency surgical procedures, and current smoking as predictors of postoperative stroke resulted in an AUC of 0.700 (Fig 3). The multiple regression model that included all four factors only marginally improved the model (AUC, 0.891, Fig 4).
Comment In these analyses we identified perioperative risk factors that are associated with increased odds of postoperative stroke: low preoperative hematocrit, history of stroke, peripheral vascular disease, preoperatively elevated blood pressure, smoking status, and elevated postoperative BUN. It has been proposed that the primary mechanisms of perioperative brain injury after cardiac surgical procedures are emboli and hypoperfusion in the setting of ischemia/reperfusion injury [19]. Given the
physical manipulation of the ascending aorta in some cardiac operations, many investigators have argued that perioperative strokes are likely a direct result of embolization of material from large vessels [20]. Although in our study procedures on the aorta had the highest postoperative stroke risk of all surgical types (a finding supporting the hypothesis that direct large vessel atheroembolism may contribute to postoperative stroke), we also observed that postoperative strokes may be delayed up to 10 days after the procedure [9]. Although epiaortic ultrasound examinations were not performed routinely at Johns Hopkins during the period under study, the delay between surgical procedure and postoperative stoke raises the possibility that direct atheroembolism from large vessels is not the only mechanism for perioperative strokes; other mechanisms of injury likely play a significant role as well. Patients with evidence of atherosclerotic disease in one vascular territory are likely to have widespread disease in other territories including the ascending aorta and perhaps the intracranial vessels. Any of these areas of high atherosclerotic burden is vulnerable to embolization or thrombosis. This mechanism of disease may explain why aspirin and statins have been found to be protective in high-risk patients undergoing major vascular operations [21–25]. In our univariate analysis, we found that patients with a diagnosis of peripheral vascular disease were significantly more likely to have a stroke as a Fig 2. Logistic regression of probability of stroke versus log of postoperative blood urea nitrogen (BUN). (A) Predicted probability of new stroke. (B) Receiver operator characteristic curve for model.
ADULT CARDIAC
Variable
Model 2
ADULT CARDIAC
1318
ARNAN ET AL ELEVATED BLOOD UREA NITROGEN IS ASSOCIATED WITH STROKE
Ann Thorac Surg 2015;99:1314–20
Fig 3. Logistic regression of probability of stroke versus preoperative diastolic blood pressure (DBP), current smoker, and emergency surgical status. (A) Predicted probability of new stroke. (B) Receiver operator characteristic curve for model.
surgical complication. These findings suggest that patients who have a diagnosed vascular disease, regardless of the arterial bed involved, are at a high risk of developing other vascular events such as stroke after cardiac surgical procedures. The most interesting finding in our study was that acutely elevated postoperative BUN emerged as by far the most powerful predictor of postcardiac surgical stroke (OR, 2.37 for every 25% rise in BUN, with ROC AUC of 0.855), greatly exceeding the predictive power of the other more commonly recognized risk factors such as smoking, preoperative hypertension, and emergency surgical procedures. Although there was evidence of a higher risk of renal injury in patients with stroke as reflected by significantly higher creatinine levels in cases than in controls, BUN in multivariable regression analyses was more strongly associated with postoperative stroke than was creatinine levels. The impact of BUN persisted even after adjusting for creatinine levels (see Table 3). We do not believe that the relationship between elevated BUN and stroke risk is causal. Still, this novel finding has important implications. Elevated BUN is generally considered to reflect intravascular volume depletion or inadequate renal perfusion (prerenal azotemia). Possibly, patients with elevated postoperative BUN suffered parallel embolic hits to the brain as well as the kidneys. However, the observation that postoperative BUN was elevated among patients with cases but not controls could suggest that the patients identified postoperatively with strokes had suffered from relatively
Fig 4. Logistic regression of probability of stroke versus log of postoperative blood urea nitrogen (BUN), current smoker, emergency surgical status, and preoperative diastolic blood pressure (DBP). (A) Predicted probability of new stroke. (B) Receiver operator characteristic curve for model.
inadequate renal perfusion during the surgical procedures, even though the same cardiopulmonary bypass perfusion protocol (target mean arterial pressure and cardiac index) was used for both cases and controls. This susceptibility to azotemia among cases may suggest that these patients may have had subclinical renal vascular disease. We suspect that patients with subclinical renal vascular disease may have comorbid overt or subclinical vascular disease in other arterial beds, such as the aorta, the carotid arteries, and possibly intracranial arteries, thus leading to an increased risk of stroke. Elevated BUN in the immediate postoperative period may reflect some underlying vulnerability to vascularly mediated disease. We acknowledge that elevated BUN postoperatively may simply reflect renal injury sustained intraoperatively. However, further analysis of our data (not shown in the Results section) reveals a more complicated picture. We performed a nonparametric correlation procedure (Spearman’s correlation) for BUN and some intraoperative variables. Our analyses revealed that among our patients with stroke, postoperative BUN is only weakly correlated with procedure type (r ¼ 0.11, p ¼ 0.03) and cardiopulmonary bypass time (r ¼ 0.16, p ¼ 0.03) and not correlated with lowest intraoperative mean arterial pressure (r ¼ 0.11, p ¼ 0.13). Additionally, a multiple regression analysis of BUN as a dependent variable, with cardiopulmonary bypass time and lowest intraoperative mean arterial pressure as independent variables, showed that the model was not statistically significant (F ¼ 2.7, p > 0.05). Circulatory arrest time was
Ann Thorac Surg 2015;99:1314–20
ARNAN ET AL ELEVATED BLOOD UREA NITROGEN IS ASSOCIATED WITH STROKE
controls, and this approach precluded our analysis from identifying sex or age as significant risk factors for stroke. Still, we suspect that although female sex and increasing age have been shown to be associated with strokes, the association may be mediated through advanced and widespread vascular disease. In conclusion, we have shown in our analyses that evidence of vascular disease is the likely common mechanism that predisposes cardiac surgical patients to postoperative stroke. We believe that patients with direct evidence of vascular disease such as peripheral arterial disease, with evidence of end-organ damage from advanced vascular disease such as previous stroke, or with subclinical vascular disease (as may be revealed by elevated postoperative BUN) are at highest risk of postoperative stroke regardless of the type of cardiac surgical procedure. Ultimately, learning to identify patients who may be susceptible to adverse outcomes in the setting of routine intraoperative blood pressure goals may help to ameliorate postoperative risk. We wrote and edited this article, and we take full responsibility for its content. Dr Arnan receives research support from a Diversity Supplement to contract N01-HC-95165 from the National Heart, Lung, and Blood Institute. We are grateful to Henry A. Feldman, PhD for his important contributions to this article.
References 1. McKhann GM, Grega MA, Borowicz LM Jr, et al. Stroke and encephalopathy after cardiac surgery: an update. Stroke 2006;37:562–71. 2. Martin WR, Hashimoto SA. Stroke in coronary bypass surgery. Can J Neurol Sci 1982;9:21–6. 3. Coffey CE, Massey EW, Roberts KB, et al. Natural history of cerebral complications of coronary artery bypass graft surgery. Neurology 1983;33:1416–21. 4. Breuer AC, Furlan AJ, Hanson MR, et al. Central nervous system complications of coronary artery bypass graft surgery: prospective analysis of 421 patients. Stroke 1983;14:682–7. 5. Junod FL, Harlan BJ, Payne J, et al. Preoperative risk assessment in cardiac surgery: comparison of predicted and observed results. Ann Thorac Surg 1987;43:59–64. 6. Bucerius J, Gummert JF, Borger MA, et al. Stroke after cardiac surgery: a risk factor analysis of 16,184 consecutive adult patients. Ann Thorac Surg 2003;75:472–8. 7. Dacey LJ, Likosky DS, Leavitt BJ, et al. Perioperative stroke and long-term survival after coronary bypass graft surgery. Ann Thorac Surg 2005;79:532–6; discussion 537. 8. Roach GW, Kanchuger M, Mangano CM, et al. Adverse cerebral outcomes after coronary bypass surgery: Multicenter Study of Perioperative Ischemia Research Group and the Ischemia Research and Education Foundation Investigators. N Engl J Med 1996;335:1857–63. 9. Hogue CW Jr, Murphy SF, Schechtman KB, et al. Risk factors for early or delayed stroke after cardiac surgery. Circulation 1999;100:642–7. 10. McKhann GM, Goldsborough MA, Borowicz LM Jr, et al. Predictors of stroke risk in coronary artery bypass patients. Ann Thorac Surg 1997;63:516–21. 11. Gardner TJ, Horneffer PJ, Manolio TA, et al. Stroke following coronary artery bypass grafting: a ten-year study. Ann Thorac Surg 1985;40:574–81. 12. Tuman KJ, McCarthy RJ, March RJ, et al. Morbidity and duration of ICU stay after cardiac surgery: a model for preoperative risk assessment. Chest 1992;102:36–44.
ADULT CARDIAC
not addressed because only a minority of patients underwent circulatory arrest (n ¼ 26; 14.4%) over the study period. We interpret these analyses to mean that the postoperative BUN level cannot be fully explained by intraoperative factors. This lends some support to our conclusion that the rise in BUN among patients with postoperative stroke could reflect renal vascular vulnerability, which may be associated with stroke susceptibility. Additionally, although our population of patients with stroke includes early and delayed strokes, we detected elevated BUN in both groups. The early postoperative strokes make up only 13.5% (n ¼ 24) of our total population of patients with stroke. Among our patients with stroke, there was no significant difference in immediate postoperative mean BUN levels between early and delayed strokes (29.9 mg/dL versus 36.6 mg/dL, p > 0.05). Thus, we suspect that elevated BUN may be associated with both early and delayed strokes. We also noted that emergency surgical status was associated with stroke. The OR for stroke for emergency surgical procedures is 2.70, consistent with the findings in the Northern New England Cardiovascular Disease Study Group’s prediction model for strokes after CABG [26]. It is not immediately clear why emergency surgical procedures are associated with a higher risk of stroke. The association is unlikely to be caused by a difference in operative technique during emergency surgical procedures. More likely, emergency operations are conducted on patients with acutely active disease who could be in a highly inflammatory state. We propose that in the setting of widespread vascular disease, an acute vascular syndrome or an infectious process requiring emergency surgical procedures can increase the predisposition to a postoperative stroke. Alternatively, those patients requiring emergency operations are more hemodynamically unstable at the time of the procedure, and this instability could also contribute to an increased risk of postoperative stroke. We acknowledge that our study has several limitations. First, it is possible that there are unmeasured confounders. However, given that our findings are consistent with previously published risk stratification schemes we believe our analysis adds value to what is already known in the field of perioperative stroke assessment. Second, neuroimaging was not obtained postoperatively for asymptomatic patients. Therefore, there is a possibility that strokes were underdiagnosed in this study. The institutional postoperative stroke rate at Johns Hopkins is consistent with the rates reported by other institutions. Thus, we are unlikely to be significantly underdiagnosing clinically significant strokes. Third, our findings are specific to one institution and a finite set of surgeons with a certain level of experience. Consequently, the findings from our institution may not be generalizable to other clinical settings. However, the thrust of our findings is that evidence of vascular disease (either overt or subclinical) is associated with postoperative strokes and likely has general applicability. Finally, we used frequency matching by gender and age-band to select
1319
ADULT CARDIAC
1320
ARNAN ET AL ELEVATED BLOOD UREA NITROGEN IS ASSOCIATED WITH STROKE
13. Grover FL, Hammermeister KE, Burchfiel C. Initial report of the Veterans Administration Preoperative Risk Assessment Study for Cardiac Surgery. Ann Thorac Surg 1990;50:12–26; discussion 27–8. 14. Djaiani G, Fedorko L, Borger M, et al. Mild to moderate atheromatous disease of the thoracic aorta and new ischemic brain lesions after conventional coronary artery bypass graft surgery. Stroke 2004;35:e356–8. 15. Eagle KA, Guyton RA, Davidoff R, et al. ACC/AHA 2004 guideline update for coronary artery bypass graft surgery: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Committee to Update the 1999 Guidelines for Coronary Artery Bypass Graft Surgery). Circulation 2004;110; p. e340–437. 16. Li Y, Walicki D, Mathiesen C, et al. Strokes after cardiac surgery and relationship to carotid stenosis. Arch Neurol 2009;66:1091–6. 17. Head SJ, Osnabrugge RL, Howell NJ, et al. A systematic review of risk prediction in adult cardiac surgery: considerations for future model development. Eur J Cardiothorac Surg 2013;43:e121–9. 18. Hornero F, Martin E, Paredes F, et al. Stroke after coronary artery bypass grafting: preoperative predictive accuracies of CHADS2 and CHA2DS2VASc stroke risk stratification schemes. J Thorac Cardiovasc Surg 2012;144:1428–35. 19. Shaw PJ, Bates D, Cartlidge NE, et al. An analysis of factors predisposing to neurological injury in patients undergoing coronary bypass operations. Q J Med 1989;72:633–46. 20. Borger MA, Ivanov J, Weisel RD, et al. Stroke during coronary bypass surgery: principal role of cerebral macroemboli. Eur J Cardiothorac Surg 2001;19:627–32. 21. Fleisher LA, Beckman JA, Brown KA, et al. 2009 ACCF/AHA focused update on perioperative beta blockade incorporated into the ACC/AHA 2007 guidelines on perioperative cardiovascular evaluation and care for noncardiac surgery: a Report of the American College of Cardiology Foundation/ American Heart Association Task Force on Practice Guidelines. Circulation 2009;120:e169–276. 22. Kertai MD, Boersma E, Westerhout CM, et al. A combination of statins and beta-blockers is independently associated with a reduction in the incidence of perioperative mortality and nonfatal myocardial infarction in patients undergoing abdominal aortic aneurysm surgery. Eur J Vasc Endovasc Surg 2004;28:343–52. 23. Feringa HH, van Waning VH, Bax JJ, et al. Cardioprotective medication is associated with improved survival in patients with peripheral arterial disease. J Am Coll Cardiol 2006;47: 1182–7. 24. Feringa HH, Bax JJ, Karagiannis SE, et al. Elderly patients undergoing major vascular surgery: risk factors and medication associated with risk reduction. Arch Gerontol Geriatr 2009;48:116–20. 25. Lau WC, Froehlich JB, Jewell ES, et al. Impact of adding aspirin to beta-blocker and statin in high-risk patients
Ann Thorac Surg 2015;99:1314–20
undergoing major vascular surgery. Ann Vasc Surg 2013;27: 537–45. 26. Charlesworth DC, Likosky DS, Marrin CA, et al. Development and validation of a prediction model for strokes after coronary artery bypass grafting. Ann Thorac Surg 2003;76:436–43.
Appendix: List of Other Cardiac Surgical Procedures Pacemaker implantation/removal Pericardial window Sternal wire removal Extracorporeal membrane oxygenator implantation/removal Ventricular septal defect repair Anomalous pulmonary venous return repair Apical conduit Ventricular assist device implantation/removal Intraaortic balloon pump implantation/removal Heart transplantation Evacuation of hematoma Automatic implantable cardioverter-defibrillator implantation/removal Heart/lung transplant Patent foramen ovale closure Pericardiectomy Incision and drainage Debridement and washout Atrial septal defect repair Removal of foreign body Subaortic membrane resection Left ventricular myotomy and myectomy Exploratory median sternotomy Cardiac tumor resection Pulmonary embolectomy Inferior vena cava tumor resection Transmyocardial laser revascularization Ligation of coronary artery fistula Atrial mass removal Radiofrequency ablation Pulmonary artery repair Myxoma resection Atrial septectomy Fontan procedure Unroofing anomalous right coronary artery