Screening Carotid Ultrasonography and Risk Factors for Stroke in Coronary Artery Surgery Patients1

Screening Carotid Ultrasonography and Risk Factors for Stroke in Coronary Artery Surgery Patients Richard S. D’Agostino, MD, Lars G. Svensson, MD, PhD, Deborah J. Neumann, MD, Husam H. Balkhy, MD, Warren A. Williamson, MD, and David M. Shahian, MD Department of Thoracic and Cardiovascular Surgery, Lahey Hitchcock Medical Center, Burlington, Massachusetts

Background. The role of noninvasive carotid artery screening in relation to other clinical variables in identifying patients at increased risk of stroke after coronary artery bypass grafting was examined. Methods. Preoperative, intraoperative, and postoperative clinical data were prospectively collected for 1,835 consecutive patients undergoing first-time isolated coronary artery bypass grafting between March 1990 and July 1995, 1,279 of whom had screening carotid ultrasonography. All patients with postoperative necrologic events were identified and reviewed in detail. Average patient age was 65.3 years (range, 33 to 92 years), and 9.3?’. (171 patients) had a prior permanent stroke or transient ischemic attack. Hospital and 30-day mortality was 2.2°/0(41 patients). Forty-five patients (2.5°/0) had a transient or permanent postoperative necrologic event. The data were analyzed by stepwise logistic regression to determine the independent predictors of both significant carotid stenosis and stroke. Results. On multivariate analysis, the clinical predictors of significant carotid stenosis were age (p < 0.0001), diabetes (p = 0.0123), female sex (p = 0.0026), left main coronary stenosis greater than 60Y0 (p < 0.0001), prior

c

ontinual improvements in myocardial preservation, cardiopulmonary bypass techniques, and postoperative care have allowed cardiac surgeons to perform increasingly complex procedures on older and sicker patients. Despite these achievements, and ironically in part because of them, postoperative stroke has emerged as a frustrating problem for the surgeon and an often devastating complication for patients. The role of extracranial cerebrovascular disease in the genesis of postoperative stroke and the utility of noninvasive carotid artery screening (NICS) in patients undergoing coronary artery bypass grafting (CABG) remain incompletely defined. This report examines our experience with NICS and stroke in patients undergoing CABG. Presentedat the PosterSessionof the Thirty-secondAnnualMeetingof The Societyof ThoracicSurgeons,Orlando,FL, Jan 29-31, 1996. Addressreprintrequeststo Dr D’Agostino,Departmentof Thoracicand CardiovascularSurge~, Lahey HitchcockMedicalCenter, 41 Mall Rd, Burlington,MA 01805.

stroke or transient ischemic attack (p = 0.0008), peripheral vascular disease (p = 0.0001), prior vascular operation (p = 0.0068), and smoking (p < 0.0001). When all variables were evaluated for those patients who underwent noninvasive carotid artery screenin~ the independent predictors of postoperative necrologic event were prior stroke or transient ischemic attack (p < 0.0001), peripheral vascular disease (p = 0.0037), postinfarction angina pectoris (p = 0.0319), postoperative atrial fibrillation (p = 0.0014), carotid stenosis greater than 50~0 (p = 0.0029), cardiopulmonary bypass time (p = 0.0006), significant aortic atherosclerosis (p = 0.0054), postoperative amrinone or epinephrine use (p = 0.0054), and left ventricular ejection fraction less than 0.30 (p = 0.0744). Conclusions. The etiology of postoperative stroke is multifactorial. Selective use of carotid ultrasonography is of value in identifying patients who are at greater risk of postoperative stroke independent of other variables and should be considered before coronary artery bypass grafting, particularly in patients with a history of necrologic event or peripheral vascular disease. (Ann Thorac Surg 1996;62:1714-23)

Material and Methods Preoperative, intraoperative, and postoperative clinical data in a format similar to that used by The Society of Thoracic Surgeons National Database are prospectively collected and entered into a computerized database for all patients undergoing cardiac operations at the Lahey Hitchcock Medical Center, Burlington, MA. Between March 1990 and August 1995, 1,835 consecutive patients who underwent first-time CABG were the subjects of this study. We included 21 patients who underwent placement of automatic internal defibrillator patches at the time of their CABG, 34 patients who underwent simultaneous carotid endarterectomy (CE) and CABG, and 1 patient each with simultaneous lung biopsy, below-knee amputation, and repair of a femoral artery. Patients who underwent valve, intracardiac, or aortic operations, reoperative CABG, or CABG combined with any intracardiac or ascending aortic procedure were excluded to create a homogeneous group of patients. All patients in whom any necrologic symptoms devel0003-4975/96/$15.00 PII S0003-4975(96)00885-5

01996 by The Society of Thoracic Surgeons Published by Elsevier Science Inc

—.

Ann ThoracSurg

D’AGOSTINOET AL CAROTIDSCREENINGANDSTROKEIN CABG

1996;62:1714-23

oped during the hospitalization or within 30 days of discharge were identified by query of the computer database and by reviewing monthly morbidity and mortality reports. Hospital and outpatient records of all 77 such identified patients were reviewed in detail. Twenty patients who had transient confusion without any focal necrologic finding or new abnormality on computed tomographic or magnetic resonance brain imaging and whose symptoms were clearly attributable to medications, sleep deprivation, alcohol abuse, or metabolic causes were considered not to have sustained a stroke. One patient in whom a global encephalopathy developed without focal findings 1 week postoperatively in the setting of multiorgan system faiIure, hemodynamic instability, and a negative computed tomographic head scan was not regarded as having had a stroke. Also not included in the stroke group were 1 patient with embolic stroke seconda~ to native mitral valve endocarditis 30 days after CABG, 6 patients with ischemic optic neuropathy, 2 patients with visual symptoms secondary to pituitary tumor, and 2 patients with visual symptoms due to complications of diabetic retinopathy. After these exclusions, 45 patients were identified as having had a tempora~ or permanent focal or global necrologic defect. For the purposes of this analysis, patients who had a transient ischemic attack (TIA) or a reversible ischemic necrologic event were considered as having had a stroke (CVA). Screening carotid ultrasonography was performed in 1,279 patients.

Before

September

1993, patients

were

evaluated with a continuous-wave Doppler scanner. Subsequently, studies were performed with a color duplex scanner (ATL Ultramark 9 HDI; Advanced Technology Laboratories, Bethel, WA). All cardiac operations were performed by five attending surgeons, {hree of whom obtained carotid ultrasonography routinely unless prohibited by the urgency of the patient’s cardiac condition. Two surgeons employed ultrasonography selectively based on a history of prior stroke or TIA, advanced age, carotid bruits, or extensive peripheral vascular disease. The results of noninvasive imaging were graded on the following basis: no significant disease (80Y0 stenosis). The quality of the ascending aorta at operation was recorded by the attending surgeon and was graded as showing no, mild, moderate, or severe disease. In the earlier time period of the study this qualitative assessment was based on the appearance of the aorta, the presence of palpable calcification, or the findings of intramural or intraluminal debris during creation of the proximal anastomoses. During the most recent 2 years, this clinical assessment has been supplemented by intraoperative transesophageal or epiaortic ultrasound in patients with abnormal appearing aortas. All operations were performed using systemic hypothermia and cold blood cardioplegic arrest. Cardiopulmonary bypass flow rates were maintained between 2.0 and 2.4 L “rein-l “m–2 and systemic pressures between 55 and 80 mm Hg. Systemic- perfusion pressures were maintained at a higher range (7o to 80 mm Hg) in patients

1715

with untreated carotid stenoses. All distal coronary anastomoses were created first. Four of five surgeons generally fashioned proximal anastomoses after release of the aortic cross-clamp, and one surgeon uniformly created both proximal and distal anastomoses during a single period of aortic cross-clamping. Left ventricular venting was not routinely employed. Thirty-four patients underwent combined CABG and CE. The endarterectomy was performed by a separate vascular surgical team, before median sternotomy and during harvest of the saphenous vein. Arterial carotid shunts were used in all cases. The following clinical variables were collected for the statistical analyses: preoperative continuous variables were age and left ventricular ejection fraction; preoperative categoric variables were age greater than 70 years, sew hypertension, diabetes, tobacco use, chronic obstructive lung disease, preoperative renal dysfunction, New York Heart Association functional class, urgency of operation, postinfarction angina, peripheral vascular disease, prior vascular operation, history of abdominal aortic aneurysm, prior stroke or transient ischemic attack preoperative NICS, degree of right and left carotid stenoses, presence of unilateral or bilateral NICS stenosis greater than 50%, presence of unilateral or bilateral NICS stenosis greater than 80Y0,aspirin use, intravenous heparin or nitroglycerin use, intraaortic balloon pump, intravenous pressor agents, left main coronary stenosis greater than 60%, and left ventricular ejection fraction less than 0.30. Intraoperative continuous variables were number of distal anastomoses, lowest esophageal temperature during cardiopulmonary bypass, and cardiopulmonary bypass time; intraoperative categoric variables were use of the internal thoracic artery, technique of proximal anastomosis (single period of aortic crossclamping versus partial aortic exclusion with side-biting clamp), ascending aortic quality, and need for femoral arterial cannulation. The only postoperative continuous variable was number of inotropes/pressors used postoperative categoric variables were more than two inotropes/pressers, use of amrinone or epinephrine, perioperative myocardial infarction, renal failure, respiratory failure, need for tracheotomy, atrial fibrillation, stroke, and death. Quantitative data are presented as mean ? standard deviation. Univariate comparisons between groups were calculated by the unpaired t test, Fisher’s exact test, $ test, or the one-way or two-way analysis of variance, as appropriate. All probabilities are two-tailed, with a p value less than O.O5regarded as significant. Independent predictors were determined by stepwise logistic regression analysis. All variables were used for the multivariate analysis, but only variables with a p value less than 0.1 were entered during computational stepping of the model. All computations were performed using the BMDP statistical software package (BMDP Statistical Software, Inc, Los Angeles, CA). Results The preoperative clinical characteristics of all 1,835 patients differentiated by whether NICS was performed are

—

1716

D’AGOSTINOET AL CAROTIDSCREENINGANDSTROKEIN CABG

AnnThoracSurg 1996;62:1714–23

Table 1. PreoperativeClinical Characteristicsof the 1,835 Sfudy PatientsGrouped by NoninvasiveCarotidScreening

NIcs

ClinicalVariable Age (y)b <65 65 to 75 >75 Sex Male Female Hypertension Diabetes Hypercholesterolemia Tobacco use COPD Vascular disease Prior CVA or TIA Postinfarctionangina Renal insficiency NM-IAfunctional class I II III IV LMCA stenosis z 60Y0 Preoperative ASA use Preoperative IV heparin Preoperative IV nitroglycerin Preoperative IABP Preoperative pressers Urgency of operation Elective Urgent Emergent

(n = 1,279)’ 66.3 ? 9.3 484(38) 540 (42) 255 (20)

No NICS

p Value (n= 556)” <0.0001

63.3 * 10.6 276 (50) 195 (35) 85 (15)

NS 918 (72) 361 (28) 770 (60) 390 (31) 711 (56) 772 (61) 143 (11) 270 (21) 155 (12) 297 (23) 45 (3.5)

0.0206 0.0131 NS NS NS <0.0001 <0.0001 NS 0.0254

395 (71) 161 (29) 305 (55) 139 (25) 315 (56) 330 (59) 54 (lo) 72 (13) 16 (3) 137 (2s) 9(1.6)

56 (4) 194(15) 684 (54) 345 (27) 243 (19) 281 (22) 606 (48) 382 (30) 69 (5.4) 14 (1.1)

0.0306 NS 0.0002 <0.001 NS <0.0001 <0.0001 <0.0001 <0.0001 0.0014

13 (2) 70 (13) 245 (44) 228 (41) 100(18) 233 (42) 349 (62) 267(48) 96 (17) 18 (3.2)

537 (42) 697 (54) 45 (4)

<0.0001 0.0282 <0.0001

151 (27) 272 (49) 133 (24)

‘Values in parentheses represent percentage within group. bExpressedas mean I standarddeviation. ASA = aspirin; COPD = chronic obstructivepulmonarydisease; CVA= cerebrovascularaccident; IABP = intraaorticballoonpump; IV = intravenous; LMCA = left main coronaryartery; NICS = NS = not significant(p value > 0.05); noninvasivecarotidscreening NYHA = New York Heart Association; TL4 = transient ischemic attack.

listed in Table 1. The mean age for all patients was 65.4 f 9.8 years. Patients evaluated with NICS were significantly older than those who did not have NICS and were significantly more likely to have diabetes, a history of vascular disease (claudication or peripheral pulse deficits, abdominal aortic aneurysm, or prior vascular operation), or a history of CVA or TIA. Clinically unstable patients were less likely to have had NICS, and this correlated with the more frequent use of aspirin, intravenous nitroglycerin, heparin, pressers, and preoperative intraaortic balloon pumps in these patients. The entire group of 1,835 patients underwent an average of 3.4 ? 0.99 distal anastomoses during a mean cardiopulmonary bypass time of 132 ~ 44 minutes. One or both internal thoracic arteries were used in 88Y0 of cases. There were a total of 41 deaths (2.27.) and 38 (2.170)

perioperative myocardial infarctions. There were no significant differences for death and perioperative myocardial infarction between patient groups based on whether NICS was performed. One thousand two hundred seventy-nine patients underwent NICS. The prevalence of carotid stenoses, prior brain ischemic event, and postoperative stroke is shown in Table 2. The majority of patients (79.5~0, 1,017 of 1,279 patients) had no significant carotid disease. However, 88 of these 1,017 patients (8.77’o)had a history of prior stroke or TIA. Two hundred sixty-two of the 1,279 patients studied by NICS (21).5~0)had a greater than 50’Yostenosis of at least one internal carotid artery, and 23 patients (1.8%) had bilateral stenoses greater than 80%. The prevalence of carotid stenoses in patients undergoing NICS and the occurrence of postoperative stroke by degree of carotid disease is shown in Figure 1. Univariate testing showed the following preoperative clinical characteristics to be predictive of any carotid stenosis greater than 5070: age (p = 0.0001), female sex (P <0.0000, hypertension (P = 0.0019), diabetes (P = 0.0010), smoking (p = 0.0002), renal insufficiency (p = 0.0010), peripheral vascular disease (p < 0.0001), prior CVA or TIA (p < 0.0001), left main coronary stenosis greater than 607. (p < 0.0001), and the need for preoperative pressor agents (p = 0.0006). Multivariate analysis showed age, female sex, diabetes, tobacco use, peripheral vascular disease, prior vascular operation, prior CVA or TIA, left main coronary stenosis greater than 60Y0, and the need for preoperative pressor agents to be significant independent predictors of any internal carotid artery stenosis greater than 50Y0(Table 3). A total of 45 patients (2.5~0) sustained a temporary or permanent necrologic deficit within 30 days of the operation. Their clinical characteristics are contrasted with those of the 1,790 patients who did not have a stroke in Table 4. Patients with postoperative stroke were older and had higher incidence of vascular disease, prior stroke or TIA, and postinfarction angina. Their cardiopulmonary bypass times were longer and they were more prone to show significant atherosclerotic change of the ascending aorta and arch. They were more likely to require postoperative inotropes, to have postoperative atrial fibrillation, and to die. The risk of stroke increased with advancing age (Fig 2). The combination of a previous stroke or TIA and significant carotid stenosis by NICS was indicative of a particularly high risk of postoperative stroke (18.2Y0 versus 4.67.; p < 0.0001) compared with patients with only one or the other of these findings (Fig 3). Specific clinical characteristics of each patient with postoperative stroke are listed in Table 5. We attempted to determine a putative cause of each necrologic event by careful review of all pertinent clinical data. Each patient’s stroke was assigned to one or more of three broad etiologic categories: aortic atheroembolic, cerebrovascular, or cardiac disease. Patients with arrhythmias, significant hypotension, or hemodynamic instability were included within the cardiac category. The resulting distribution of postoperative stroke by probable causative mechanism is shown by Venn diagram in Figure 4. In 8

D’AGOSTINOET AL

Ann ThoracSurg

CAROTID SCREENING ANDSTROKEIN CABG

1996; 62:1714 -23

1717

Table 2. Prevalence of CarotidArtery Disease in the 1,279 PatientsStudied by NoninvasiveCarotidScreening Degree of Carotid Artery Stenosis (ipsilaterallcontralateral) <50%/< 50% <50%/50%-79% <50%/80°/o–100% 50”/o-79yo/50%-79yo 5070-79°/0/80%-10070 80%-100 %/80°/o-100Y0 Total

Event (no. of patients) Number of Patients 1017 120 52 40 27 23 1279

CABG = coronaryarterybypassgraftin~

Percent 79.5 9.5 4.1 3.1 2.1 1.8 100

CE = carotidendarterectomy;

patients (patients 4, 12, 21, 29, 33, 40, 41, and 43) strokes developed due to atheroemboli from the ascending aorta and arch. The aorta was normal appearing at operation in 3 patients (patients 4, 21, and 29) and severe atherosclerosis was discovered after the necrologic event by means of transesophageal echocardiography or autopsy. In 8 patients (patients 8, 11, 16, 17, 20, 32, 34, and 35) stroke developed in temporal association with postoperative dysrhythmias, most notably atrial fibrillation. Patients 11 and 32 had peripheral vascular emboli as well. Patients 34 and 35 also had coexistent carotid stenoses, but in patient 34 an acute necrologic change developed within 15 minutes of the heart’s conversion from atrial fibrillation to sinus rhythm. Patient 42 had preoperative and postoperative cardiogenic shock with significant hypotension; in this patient a stroke developed secondary to cerebral hypoperfusion. Cerebrovascular disease was the dominant finding in 9 patients (patients 6, 13, 18, 23, 27, 28, 30, 36, and 38). Deficits were ipsilateral to known stenoses in all patients. Two patients (patients 23 and 27) had a high-grade stenosis with contralateral occlusion. Four patients had simultaneous CABG and CE (patients

Prior CVA or TIA

Combined CABG and CE

Postoperative Stroke

88 20 12 11 14 10

0 1 6 1 9 16

24 6 3 4 5 3

155

34

45

CVA = cerebrovascularaccident;

TIA = transientischemicattack.

7, 18, 30, and 36). Patient 7 underwent a left CE, and a postoperative computed tomographic brain scan showed diffuse infarcts compatible with multiple emboli, presumably of aortic origin. Patient 18 sustained a stroke contralateral to the side of the endarterectomy and ipsiIateral to a high-grade stenosis. Both patients 30 and 36 sustained strokes ipsilateral to endarterectomy. A total of 262 patients had carotid stenoses greater than 50%, unilateral in 172 (65.7Yo)and bilateral in 90 (34.37.). One hundred forty patients (81Y0)with unilateral stenosis had no prior history of brain ischemic event. Within this subgroup, 3 patients (2.17.) underwent combined CABG and CE, and none had stroke. Of the 137 patients with unilateral stenosis who did not have combined CABG and CE, only 4 (z.g~.) sustained a postoperative stroke. None of these four strokes could be attributed to carotid disease. Two were due to emboli from an atherosclerotic ascending aorta and two were secondary to postoperative atrial fibrillation. Prior stroke or TIA was present in 32 of the 172 patients with unilateral stenosis. Four of these patients (12.5~0) underwent combined CABG and CE, and none had postoperative stroke. Five of 28 patients (17.97.) with unilateral stenosis and prior symptoms who did not undergo combined CABG and CE had a postoperative event. One patient had development of significant post-

Table 3. SignificantIndependent Predictorsof Any Internal CarotidArtew StenosisGreater Than 50!k0 Q/o

c50yo/ “ c50°/o/ “ 50-79%/ 50-790/o 50-79% <50~o

50-79%/ 8O-1OO’YOI 8O-1OOYO80-100’YO

Degree of Carotid Stenoses Fig 1. Prevalenceof carotidstenosesin patientsundergoingnoninvasivecarotidscreeningandoccurrenceof strokebydegreeof carotidstenosis.

Clinical Variable

DValue

Odds Ratio

Age Female sex Diabetes Tobacco use Peripheral vascular disease Prior vascular operation Prior CVA or TIA LMCA stenosis > 60”/0 Preoperative pressor agents

<0.0001 0.0026 0.0123 <0.0001 0.0001 0.0068 0.0008 <0.0001 0.0541

1.05 1.62 1.47 2.04 2.09 2.72 3.28 4.98 1.84

CVA= cerebrovascular accident; TL4 = transientischemicattack.

LMCA= leftmaincoronaryartery;

1718

D’AGOSTINO ET AL CAROTIDSCREENINGANDSTROKEIN CABG

AnnThoracSurg 1996;62:1714-23

Table 4. Clinical Characteristicsof All PatientsStratifiedby PostoperativeStroke Clinical Variable

No Stroke (n = 1,790)” p Value

Age (y)b

65.3 * 9.9

<65

752 (42)

Stroke (n = 45)”

0.0007 70.4 * 9.0 8 (18)

65 to 75

714 (40)

22 (49)

>75

324 (18)

15 (33) 0.0149

Sex Male Female Hypertension

1288 (72)

25 (56)

502 (28) 1048 (59)

20 (44) NS

27 (60)

515 (29)

NS

14 (31)

Hypercholesterolemia

1003 (56)

NS

23 (51)

Tobacco use

32 (71)

Diabetes

1070 (60)

NS

COPD

189 (11)

NS

8 (18)

Vascular disease

321 (18)

<0,0001

21 (47)

Prior CVA or TIA

154 (8.6)

<0.0001

17 (38)

Postinfarction angina

416 (23)

0.0100

18 (40)

NYHA functional class 67 (4)

NS

II

260 (14)

NS

4 (9)

III

911 (51)

NS

18 (40

IV

21 (47)

1

2 (4)

552 (31)

0.0293

LVEF <0.30

231 (13)

0.0239

11 (24)

LMCA stenosis > 60Y0

332 (19)

NS

11 (24)

1234 (70) 241 (20)

0.0004 <0.0001

21 (50)

78 (4.4)

<0.0001

12 (27)

NICS performed Any carotid stenosis > 50Y0 Bilateral carotid stenoses >5070

45 (100)’

Preoperative ASA

500 (28)

NS

11 (24)

Preoperative IV heparin

926 (52)

NS

29 (64)

Preoperative IV nitroglycerin

628 (35)

NS

21 (47)

Preoperative IABP

161 (9)

NS

4 (9)

Preoperative pressers

31 (1.7)

NS

1 (2.2)

Combined CABG-CE

31 (1.7)

0.0005

4 (8.9)

Table 4. Continued Clinical Variable

No Stroke (n = 1,790)”

p Value

Stroke (n = 45)’

Postoperative atrial fibrillation

536 (30)

<0.0001”

26 (58)

Postoperative amrinone or

316 (18)

<0.0001

20 (44)

35 (2.0)

<0.0001

6 (13)

epinephrine Death

aValues in parentheses represent percentage within group. bExpressedas mean ~ standarddeviation. CIncludes3 patientswhohad NICSpostoperatively afterdetectionof stroke,all ofwhomhadno carotid stenosis. ASA = aspirin; CABG = coronaryarterybypass graftin~ CE = carotidendarterectomy; CPB = cardiopulmonary bypass; COPD = chronicobstructivepulmona~ disease; CVA= cerebrovascularaccident; L4BP = intraaorticballoonpump; ITA = internalthoracic artery; IV = intravenous; LMCA = left main coronarystenosis; LVEF= leftventricularejectionfraction; MI = myocardialinfarction; NICS = noninvasivecarotid screenin~ NS = not significant (p value > 0.05); NYHA = New York Heart Association; TIA = transientischemicattack.

operative hypotension and sustained a stroke contralateral to an occluded carotid artery. The ipsilateral carotid artery showed no significant disease. One patient sustained an occipital stroke in the settkw of atrial fibrillation, aortic at~erosclerosis, and both ~arotid and vertebral artery stenoses. The third patient sustained a stroke contralateral to a moderate carotid stenosis and ipsilateral to a prior stroke. One transient and one permanent event were ipsilateral to a known carotid stenosis. Thus, in only these latter 2 patients would a combined CABG and C-Ehave possibly-prevented a postoperative stroke. Bilateral carotid stenoses greater than .50?’. were present in 90 of the 262 patients. Fifty-five of these patients (61Yo)had no prior symptoms. Of this subgroup, 10 patients underwent combined CABG and CE, and 1 patient (107o) had a postoperative stroke ipsilateral to the CE. Of the 45 patients who did not have combined CABG and CE, 3 (6.770) sustained a postoperative stroke. Two of the strokes (4.4?’o) could be attributed to the carotid disease, and the third patient’s stroke was compatible with an embolic event 4 days postoperatively and tem-

Urgency of operation Elective

679 (38)

0.0289

10 (22)

Urgent

935 (52)

0.0048

33 (73)

Emergent

176 (10) 3.4<.99

0.0486

3.7 t 1.0 36 (80)

Number of distal anastomosesb

NS

2 (5)

Incidence of 6% stroke 4%

ITA use

1569 (88)

NS

Lowest esophageal temperature (“C)’

26,9 t 2.03

0.0241 26.2 t 1.84

CPB time (min)a

132 ~ 43

<0.0001

158 t 44

Severe aortic atherosclerosis

36 (2.0)

<0.0001

6 (13)

2%

Perioperative MI

38 (2.1)

NS

o (o)

0%

Postoperative renal failure

21 (1.2)

NS

2 (4.4)

Postoperative respiratory failure

59 (3.3)

NS

1 (2.2)

AGE (YRS) Fig2. Riskof postoperative stroke in differentagegroups.Error bars indicatestandarderrorof themean.

Ann ThoracSurg 1996;62:1714-23

24%

D’AGOSTINOETAL CAROTIDSCREENINGANDSTROKEIN CABG

11

18.2

P value. 0.0001

20%-

/

Incidence ~670

! / N..%

of Stroke 12%. 8%4%0% J

],8

~

4.6

. N=196

4,6 / 1N=88

N=926 r I I 1 . NICS +Ma - Nrcs +NICS - HISTORY - HISTORY +HISTORY +HISTORY

Fig3. Incidenceof postoperative strokegroupedbythepresenceof anycarotidstenosisgreaterthan50Y0andbya histoyof prior brainischemicevent.Error bars indicatestandarderrorof the mean.(NICS = noninvasivecarotidscreening.) porally associated with conversion from atrial fibrillation to sinus rhythm. Thirty-five of the 90 patients with bilateral significant stenoses had a history of prior stroke or TIA. Sixteen had combined CABG and CE, and postoperative stroke developed in 3 (18.8Y0).One stroke was contralateral to the CE in the setting of an ipsilateral high-grade stenosis, postoperative hypotension, and low output syndrome. One stroke occurred ipsilateral to the CE. The third patient showed diffuse bihemispheric and cerebella infarcts presumably of embolic origin. Nineteen of the 35 patients with bilateral stenoses and prior symptoms did not have CABG and CE, and postoperative stroke devel-

oped in 5 (26.37.). Three of these strokes (15.8’?4.)were attributable to carotid disease. Four separate models were used to evaluate the independent predictors of stroke by multivariate analysis (Table 6). Two models each (all variables versus preoperative variables) were created for the entire group of 1,835 patients and for the group of 1,279 patients who unde~ent NICS. Because of the small number of total strokes, it was not possible to make valid statistical comparisons according to (1) the impact of degree of stenosis and (2) unilateral or bilateral involvement. Hence, during stepwise logistic regression analysis the multivariate models were allowed to select whether either a unilateral stenosis greater than 507. or bilateral carotid stenoses greater than 507. was the best predictor. Postinfarction angina, vascular disease, prior stroke, or TIA were the independent predictors common to all four models. Atrial fibrillation, cardiopulmonary bypass time, degree of aortic atherosclerosis, and the use of amrinone or epinephrine were also significant independent predictors of stroke when all variables were used. Left ventricular ejection fraction less than 0.30 was also a significant independent predictor of stroke in both multivariate models that evaluated preoperative variables only. Comment The central nervous system effects of heart operations can range from the most subtle neuropsychologic

1719

changes to massive brain infarction and irreversible coma. Although careful evaluation can demonstrate some degree of cognitive change in up to 79Y0of patients after a heart operation [1], most studies of postoperative stroke have focused on clinically evident deficits. We did not track symptoms such as forgetfulness, mental slowness, or neurocognitive deficits. Temporary necrologic deficits, however, were purposefully included in our analysis to avoid underestimation of the true event rate because even transient deficits are bothersome to the patient, result in additional testin~ generate consultations, and increase the length of hospital stay. The incidence of a temporary or permanent ischemic necrologic event in our patients was 2.5% and is consistent with previously reported data. With the exception of warm heart surge~ [2], improvements in perfusion techniques have diminished the role of cardiopulmonary bypass in the genesis of clinically evident postoperative necrologic dysfunction. The majority of strokes in the current era are due to atherosclerosis of the ascending aorta, cerebrovascular disease, and macroemboli of cardiac origin. Our multivariate analyses support this multifactorial etiology and have shown the presence of peripheral vascular disease, a prior stroke or TIA, significant carotid stenoses, postinfarction angina, longer cardiopulmonary bypass time, aortic atherosclerosis, postoperative atrial fibrillation, the use of amrinone or epinephrine, and left ventricular ejection fraction less than 0.30 to be significant independent predictors of postoperative stroke in CABG patients. Our patients with a history of CVA or TIA had a 10’XO incidence of postoperative stroke. Rorick and Furlan [3] and Berens and colleagues [4] found similar incidence of postoperative stroke (ls.A~o and .5~0,respectively) in their cardiac surgery patients with a prior necrologic event. Reed and colleagues [5] noted that patients with previous

stroke were at a sixfold increased risk of postoperative stroke. Aortic atherosclerosis is increasingly recognized as a major risk factor for stroke. Although paleness of the ascending aorta or thickening of the adventitial tissue can be an indicator of troublesome atherosclerosis we, as have others [6-8], found aortic palpation to be an unreliable indicator of the degree of disease. We used transesophageal or epiaortic ultrasonography during the last 2 years of our study and then on a selective basis only. Undoubtedly, this has led to an underestimation of the significance of this variable in our patient population. Wareing and associates [9] have demonstrated a lg~o in incidence of moderate to severe aortic atherosclerosis their patients older than 50 years who were evaluated ultrasonographically. Amarenco and colleagues [10] and, most recently, the French Study of Aortic Plaques in Stroke Group [11] have shown aortic arch plaque to be a significant predictor of recurrent brain infarction in patients older than 60 years. Advanced age at operation has been shown to be associated with ascending aortic atherosclerosis and increased incidence of stroke by numerous investigators [8, 12-14]. Although its significance was eclipsed by other factors in our multivariate analyses, we found age to be a

1720

D’AGOSTINO ET AL CAROTIDSCREENINGANDSTROKEIN CABG

Ann ThoracSurg 1996;62:1714-23

Table 5. Selected Clinical Characteristicsof PatientsWith PostoperativeStroke

Patient Prior CVA No. or TIA 1

2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29

No No No No No

No Yes No Yes Yes

No Yes Yes

No No No No Yes Yes

No No No No Yes No No

No Yes

No

30 31 32

Yes Yes No

33 34 35 36 37 38 39 40 41 42 43 44 45

No No No No No Yes Yes Yes

No No No No No

Postoperative Symptoms r

,.

Len arm paresis Confusion,homonymous hemianopsia Right hemiplegia Coma Left arm paresis Left retinal artery embolus Left hemiplegia progressing to obtundation Right arm paresis Left hemiparesis Right arm and right facialparesis Homonymous hemianopsia, left foot paresis Temporary obtundation Left visual field deficit Right hemiplegia Right hand weakness, visual field deficit Left arm and right facial paresis Right hemiparesis, aphasia Parapareis Left-sided then right-sided paresis Right hemianopsia Obtundation Left arm paresis, speech ataxia Aphasia, confusion Right arm paresis, confusion Right hemiparesis Left hemiparesis, speech difficulties Right arm paresis, confusion Aphasia Left hemiplegia Acute unresponsivenessand seizures Aphasia, confusion Right facial and body paresis, speech difficulties,visual field deficit Right hemiplegia Right hemiplegia, aphasia Right arm paresis Right hemiparesis Confusion,visual field deficits Speech difficulties Right-sided paresis Paraphasia, confusion Visual deficit Right hemiparesis, speech Coma Right arm paresis, speech difficulties Left hand paresis, diploplia,dysarthria

Left Right Carotid Carotid Stenosis Stenosis Degree of by by Aortic Residual Location of Deficitby NICS NICS Atherosclerotic Necrologic CT or MRI Scan (%) (%) Changea Deficit r..

,

.

.

,-

.

.

. .

Iugnr MCA rermory Right occipital Left frontoparietal Multiple Multiple NA Multiple Left parietal Right parietal Left pontine Left occipital None None Left MCA territoty Left temporal Left MCA territory None Right frontoparietal Left MCA territory None Multiple Right cerebella Left frontal, temporal, parietal None None None None None Right MCA and PCA territory Right MCA territory Left internal capsule Left PCA territory

Left frontoparietal Left MCA territo~ Left frontotemporal NA Left occipitoparietal Left parietal None None NA None Multiple Left frontal None



50-80 <50 <50 <50 80-99 80-99 <50 100 50-80 50-80

<50 <50 <50 <50 <50 50-80 <50 <50 50-80 <50

None Severe None NA NA None None Mild None Moderate None

<50 80-99 <50 <50 <50 <50 80-99 50-80 <50 <50 <50 100

<50 50-80 <50 <50 <50 <50 80-99 50-80 <50 50-80 <50 50-80

Severe None Severeb None None NA None None None NA None None

<50 <50 <50 80-99 80-99 <50

<50 <50 <50 100 <50 <50

None None None None None Mild

Yes

<50 <50 <50

80-99 <50 <50

Mildc None None

Yes Yes Yes

<50 50-80 50-80 80-99 <50 80-99 <50 <50 <50 <50 <50 50-80 <50

<5(I <50 50-80 80-99 <50 <50 50-80 <50 <50 <50 50-80 50-80 <50

Severe None None Mild None None Mild Severe Moderate None Severec None None

Yes Yes Yes Yes Yes Yes

.-A

..O

Yes

No No Yes Yes No Yes No Yes Yes Yes No No Yes Yes Yes No Yes Yes No Yes Yes

Yes No No Yes No

No

No No Yes Yes Yes

No No

“Degree of aortic atheroscleroticchange noted at operationby palpationand visualinspection. bThis patient also had left atrial thrombusby echocardiography. c Echocardiographydemonstratedmobileintraluminalatheroscleroticdebrisin the aorticarch. CT = computedtomographic; CVA = cerebrovascularaccident; MCA = middle cerebral artery; MRI = magneticresonance imagin~ PCA = posteriorcerebral multiple= diffuseor bihemisphericdeficits; NA = test not performed; NICS = noninvasivecarotidscreenin~ artery; TIA = transientischemicattack.

—.

AnnThoracSurg 1996;62:1714-23

ET AL D’AGOSTINO

CAROTID SCREENING AND STROKE IN CABG

therefore, may represent a surrogate for the risk factor of echographic evidence of mural thrombus found by these investigators. Similarly, patients with diminished left ventricular function are at risk for stroke by virtue of mural thrombus or the more frequent requirement for postoperative inotropic support. In our institution, the use of amrirrone and epinephrine is reserved for more pronounced degrees of myocardial dysfunction. The attendant hemodynamic instability and hypoperfusion in such settings as well as the vasoconstriction seen with higher doses of a-adrenergic agents can result in diminished cerebral blood flow. Four of our patients sustained a stroke in such a setting. Postoperative stroke occurred in 4.6% of our patients with atrial fibrillation compared with 1.5Y0 of patients

Cerebrovascular

Aortic ‘=25~~N

1721

= 18

Cardiac N =23

without atrial fibrillation. Taylor and associates [17] reported a 77. incidence of postoperative stroke in patients with atrial fibrillation, and this association has also been noted by others [5, 15]. Whether this risk can be diminished by an aggressive regimen of early anticoagulation for postoperative atrial fibrillation remains to be determined by a prospective study.

Fig4. Putativecausesof postoperative stroke.Thenumbersshown are thenumberof patientsassignedtotheparticularcategoy.Any given patient can be assignedto one or more categoriesbasedon the clinicalcircumstances.

The management of patients with coexisting corona~ and carotid arterial disease remains unsettled. Drawing definitive conclusions from published series is confounded by their varying patient compositions, management strategies, and lengths of postoperative follow-up. In this current era of cost-effective medicine, the practicing cardiac surgeon is faced with two difficult and complementary questions: who should undergo carotid arterial screening, and what should be done with the information obtained? Concerning the first question, our patients had a mean age of 66 years, and a carotid stenosis greater than 50Y0 was found in 207.. Eight percent of patients had a stenosis greater than 80Y0. Berens and associates [4] found similar results in a prospective study of NICS in 1,087 patients older than 65

significant univariate predictor of stroke. Indeed, stroke occurred in 8.67. of our patients older than 80 years. We found postinfarction angina to be a significant independent predictor of stroke. Patients with postinfarction angina can have more perioperative hemodynamic instability, which can potentially influence the stroke rate. But it is also equally plausible that due to their recent infarction, patients with postinfarction angina are at increased risk for occult left ventricular mural thrombus. We were unable to assess this issue because appropriate echographic data were available in a minority of our patients. However, Lynn and associates [15] and Bull and colleagues [16] have shown left ventricular mural thrombus to be an independent risk factor for stroke. Postinfarction angina being a risk factor in our patients,

Table 6. Independent Predictorsof PostoperativeStroke by MultivariateAnalysis Patients With NICS

All Patients All Variables Clinical Variable Female sex Postinfarction angina Vascular disease Any carotid stenosis > 500/o Bilateral carotid stenoses > 50% Prior CVA or TIA Number of distal anastomoses LVEF <0.30 Lowest esophageal temperature Cardiopulmonary bypass time Aortic atherosclerosis Amrinone or epinephrine Postoperative atrial fibrillation CVA= cerebrovascular accident; transient ischemic attack.

Preoperative Variables Only

All Variables

Preoperative Variables Only

p Value

OR

p Value

OR

p Value

OR

p Value

OR

0.0319 0.0037

1.95 2.56

0.0245 0.0736 0.0007

2.13 1.83 3.12

0.0078 0.0090

2.48 2.52

0.0093 0.0081 0.0043

2.39 2.53 2.69

<0.0001

4.49

0.0024 0.0029

3.46 3.01

0.0014

3.13

0.0744

2.12

0.0053

2.97

0.0075 0.0387 0.0157 0.0085

1.01 1.18 2.35 2.44

0.0006 0.0055 0.0054 0.0014

<0.0001 0.0412 0.0360 0.0720

1.01 2.30 2.39 2.65

LVEF= leftventricularejectiontraction;

4.49 1.46 2.29 0.899

NICS = noninvasivecarotidscreenin~

OR = odds ratio;

TLA=

1722

D’AGOSTINO ET AL CAROTID SCREENING AND STROKE IN CABG

years and demonstrated a 17Y0 prevalence of carotid stenosis greater than 50~0 and a 5.9~0 prevalence of stenosis greater than 80V0.Similar incidence of between 17% and 22% have been found by others [18, 19]. Our multivariate analysis demonstrated age, female sex, diabetes, tobacco, peripheral vascular disease or operation, prior CVA or TIA, significant left main coronary artery stenosis, and the need for preoperative inotropes to be the significant independent predictors of a significant carotid stenosis. With the exception of diabetes and inotropes, these risk factors are essentially identical to those reported by Berens and associates [4]. Preoperative noninvasive screening would be expected, therefore, to have the highest yield in patients with one or more of these factors. Until a properly designed, prospective, randomized, multiinstitutional study can provide definitive answers to the second question regarding which patients benefit from combined operations, we must extrapolate from the results of this and other previously reported studies. Akins and associates [20] and Daily and colleagues [21] have recently shown that combined CABG and CE operations can be done safely, although it must be remembered that previously reported stroke rates have ranged from O% to 14% [22, 23]. It is generally, but not universally, accepted that patients with activesymptoms referable to both carotid and coronary disease are best managed by combined CABG and CE. However, the benefits of combined CABG and CE for patients with unilateral or bilateral asymptonzatic carotid stenoses remain unclear as there is no unequivocal evidence to suggest that they are at increased risk of stroke in the perioperative period. Of interest in this regard is our group of 137 patients with an asymptomatic unilateral significant stenosis who did not undergo combined CABG and CE. None of the four strokes that occurred in this group were referable to the carotid stenosis. Two were due to atheroemboli from the ascending aorta and two were due to atrial fibrillation-related cardiac emboli. Also of interest is the fate of the 90 patients who had bilateral significant stenoses. Fifty-five of these 90 had no prior or current symptoms of stroke or TIA. Of the 10 who underwent combined operation, 1 patient (1OYO) had a postoperative stroke ipsilateral to the carotid endarterectomy. A postoperative stroke developed in only 3 of 45 patients (6.77.) who underwent CABG without CE, and we could implicate carotid disease in only 2 (4.Q~o) of these. Hertzer and associates [24] noted a peri-CABG stroke incidence of 6.97. in the subset of patients with unilateral asymptomatic carotid stenosis greater than 707. and unstable coronary disease randomized to CABG followed by delayed CE. Barnes and colleagues [25] and Schultz and associates [26] concluded that peri-CABG stroke risk was not increased in patients with significant asymptomatic carotid stenoses, although Barnes and colleagues found these patients to be at increased risk of necrologic event during late follow-up. Gerraty and associates [27] prospectively evaluated perioperative stroke risk in a

series of 358 patients undergoing a peripheral vascular operation or CABG. A moratorium was placed on CE for

Ann Thorac Surg 1996;62:1714-23

patients with asymptomatic stenoses of 50% or greater. None of the 53 patients (15 of whom had stenosis of 80% or greater) with asymptomatic stenoses of 500/0or greater sustained a stroke. Two of 213 CABG patients sustained small infarcts, and in neither patient was there an associated significant carotid stenosis. Gerraty and associates concluded that the risk of perioperative stroke due to asymptomatic carotid stenosis is low and does not justify prophylactic carotid endarterectomy. Our experience with stroke in patients who have a unilateralasymptomatic carotid stenosis would appear to support this conclusion. The risk of perioperative stroke referable to carotid disease is only 4.4~0 in our patients with asymptomatic bilateral disease managed with CABG alone. However, our data are not sufficient for us to draw any specific conclusions regarding the merits of CABG and CE in this group of patients. In the absence of more definitive information, we generally reserve combined CABG and CE for those asymptomatic patients with high-grade bilateral disease or with a significant stenosis and a contralateral occlusion. In summary, the etiology of postoperative stroke is multifactorial, and any factor or a combination of factors may play a role in any given patient. The majority of strokes are caused by embolic phenomena from an atherosclerotic aorta or from the heart. Cerebrovascular disease is a causative factor of postoperative stroke in approximately soy. of our patients. A history of prior stroke or TIA, peripheral vascular disease, advanced age, female sex, and significant left main corona~ stenosis are associated with a greater likelihood of harboring a significant carotid stenosis. We believe that selective use of carotid ultrasonography can help in identifying patients

who are at greater risk of postoperative stroke and should be considered prior to CABG, particularly in patients with a history of prior necrologic event or peripheral vascular disease. We thank Gerald J. Heatley, MS, for his assistance with the statistical analyses.

References 1. Shaw PJ, Bates D, Cartlidge NEF, et al. Early intellectual dysfunction following coronary bypass surgery. Q J Med 1986;58:59-68. 2. Craver JM, Bufkin BL, Weintraub WS, Guyton RA. Necrologic events after coronary bypass grafting: further observations with warm cardioplegia. Ann Thorac Surg 1995;59: 1429-33. 3. Rorick MB, Furlan AJ. Risk of cardiac surgery in patients with prior stroke. Neurology 1990;40:835–7. 4. Berens ES, Kouchoukos NT, Murphy SF, Wareing TH. Preoperative carotid artery screening in elderly patients undergoing cardiac surgery. J Vase Surg 1992;15:313-23. 5. Reed GL, Singer DE, Picard EH, DeSanctis RW. Stroke following coronary-artery bypass surgery. A case-controlled estimate from the risk of carotid bruits. N Engl J Med 1988; 319:1246-50. 6. Barzilai B, Marshall WG Jr, SaffitzJE, Kouchoukos N. Avoidance of embolic complications by ultrasonographic characterization of the ascending aorta. Circulation 1989;80(Suppl 1):275-9. 7. Culliford AT, Colvin SB, Roher K, Baumann FG, Spencer FC. The atherosclerotic ascending aorta and transverse arch: a

Ann Thorac Surg 1996;62:1714-23

8. 9. 10.

11. 12.

13. 14.

15. 16. 17.

new technique to prevent cerebral injury during bypass: experience with 13 patients. Ann Thorac Surg 1986;41:27-35. Frye RL, Kronmal R, Schaff HV, et al. Stroke in coronary artery bypass graft surgery: an analysis of the CASS experience. Int J Cardiol 1992;36:213-21. Wareing TH, Davila-Roman, Daily BB, et al. Strategy for the reduction of stroke incidence in cardiac surgical patients. Ann Thorac Surg 1993;55:1400-8. Amarenco P, Duyckaerts C, Tzourio C, H6nin D, Bousser MG, Hauw JJ. The prevalence of ulcerated plaques in the aortic arch in patients with stroke. N Engl J Med 1992;326: 221-5. The French Study of Aortic Plaques in Stroke Group. Atherosclerotic disease of the aortic arch as a risk factor for recurrent ischemic stroke. N Engl J Med 1996;334:1216–21. Davila-Roman VG, Barzilai B, Wareing TH, Murphy SF, Kouchoukos NT. Intraoperative ultrasonographic evaluation of the ascending aorta in 100 consecutive patients undergoing cardiac surgery. Circulation 1991;84(Suppl 3):47–53. Gardner TJ, Horneffer PJ, Manolio TA, Hoff SJ, Pearson TA. Major stroke after coronary artery bypass surgery: changing magnitude of the problem. J Vase Surg 1986;3:684-7. Tuman KJ, McCarthy RJ, Najafi H, Ivankovich AD. Differential effectsof advanced age on necrologic and cardiac risks of coronaq+ artery operations. J Thorac Cardiovasc Surg 1992; 104:151G7. Lynn GM, Stefanko K, Reed JF, Gee W, Nicholas G. Risk factors for stroke after coronary bypass. J Thorac Cardiovasc Surg 1992;104:1518-23. Bull DA, Neumayer LA, Hunter GC, et al. Risk factors for stroke in patients undergoing coronary artery bypass grafting. Cardiovasc Surg 1993;1:182–5. Taylor GJ, Malik SA, Colliver JA, et al. Usefulness of atrial fibrillation as a predictor of stroke after isolated coronary artery bypass grafting. Am J Cardiol 1987;60:905–7.

INVITED

D’AGOSTINOETAL

CAROTID SCREENING AND STROKE IN CABG

1723

18, Faggioli GL, Curl GR, Ricicotta JJ. The role of carotid screening before coronary artery bypass. J Vase Surg 1990;12: 724-31. 19. Schwartz LB, Bridgman AH, KiefferRW, et al. Asymptomatic carotid stenosis and stroke in patients undergoing cardiopulmonary bypass. J Vase Surg 1995;21:146-53. 20. Akins CW, Moncure AC, Daggett WM, et al. Safety and efficacy of concomitant carotid and coronary artery operations. Ann Thorac Surg 1995;60:311-8. 21. Daily PO, Freeman RK, Dembitsky WP, et al. Cost reduction by combined carotid endarterectomy and corona~ bypass grafting. J Thorac Cardiovasc Surg 1996;111:1185-93. 22. Bass A, Krupski WC, Dilley RB, Bernstein EF. Combined carotid endarterectomy and coronary artery revascularization: a sobering review. Isr J Med Sci 1992;28:27–32. 23. Gugulakis A, Kalodiki E, Nicolaides AN. Combined carotid endarterectomy and coronary artery bypass grafting: a literature review. Int Angiol 1991;10:167-72. 24. Hertzer N~ Loop FD, Beven EG, O’Hara PJ, Krajewski LP. Surgical staging for simultaneous corona~ and carotid disease: a study including prospective randomization. J Vase Surg 1989;9:455-63. 25. Barnes RW, Nix ML, Sansonetti D, Turley DG, Goldman MR. Late outcome of untreated asymptomatic carotid disease following cardiovascular operations. J Vase Surg 1985;2: 843-9. 26. Schultz RD, Sterpetti AV, Feldhaus RJ. Early and late results in patients with carotid disease undergoing myocardial revascularization. Ann Thorac Surg 1988;45:603–9. 27. Gerraty RP, Gates PC, Doyle JC. Carotid stenosis and perioperative stroke risk in symptomatic and asymptomatic patients undergoing vascular or coronary surgery. Stroke 1993:24:1115-8.

COMMENTARY

From this critical analysis of patients undergoing coronary artery bypass grafting and preoperative noninvasive carotid artery screenin~ D’Agostino and associates have identified risk factors that predict the presence of significant (defined as greater than 507. ) carotid artery stenosis. With few exceptions, they are identical to those reported in other studies of large populations of patients that used preoperative noninvasive carotid artery screening to detect carotid artery disease. The prevalence of carotid artery disease in this population of patients is also similar to that reported from other studies. The study adds strong support to the concept that carotid artery disease is an important predictor of stroke after coronary artery bypass grafting. D’Agostino and associates identified other risk factors predictive of postoperative stroke, the most important being atherosclerosis of the ascending aorta and macroemboli of cardiac origin. Atherosclerosis of the ascending aorta was associated with the largest number of postoperative strokes. Yet, by D’Agostino and associates’ own admission, the incidence of ascending aortic atherosclerosis was underestimated, because only a percentage of the patients underwent epiaortic ultrasonographic scanning, which is currently the most accurate method for identifying ascending aortic atherosclerosis. Thus, the importance of this variable as a cause of stroke was underestimated.

Which patients should be studied preoperatively for carotid artery disease, and how should asymptomatic patients with hemodynamically important carotid artery stenoses be managed during a cardiac operation? These are questions without definitive answers at the present time. What is clear, however, is that any study evaluating the importance of carotid artery disease as a predictor of postoperative stroke must also evaluate the prevalence of atherosclerotic disease in the ascending aorta and arch. Techniques are now available to assess the carotid arteries (preoperative duplex scanning) and the ascending aorta (intraoperative epiaortic ultrasonographic imaging), and to identify intracardiac thrombus (preoperative or intraoperative transesophageal echocardiography). Future studies should focus on the prospective identification of all of these risk factors to more accurately assess their importance as causes of postoperative stroke and, hopefully, to formulate strategies that will reduce the incidence of this important complication.

Nicholas T. Kouchoukos,MD Missouri BaptistMedical Center Suite 266C 3009 North BallasRd St. Louis,MO 63131